fix(markdown): one-word bold headings, block ToC entries from headings

Two heading-classification fixes: - Accept single-word headings ("IMPLEMENTATION", "CONTENTS") when the line is all-bold and isolated; the word_count >= 2 gate rejected them unconditionally. - Add is_toc_entry_line: a line ending in a dot-leader group plus page number ("Measurement Lab worksheet ... 3") is a table-of-contents entry, never a heading. has_dot_leaders misses single-group leaders, so entire ToC pages were being promoted to ## headings. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-10 18:14:32 -07:00
27 changed files with 575 additions and 2703 deletions
@@ -1,36 +0,0 @@
-name: Deploy landing page
-
-on:
-  push:
-    branches: [main]
-    paths: ['site/**', '.github/workflows/pages.yml']
-  workflow_dispatch:
-
-permissions:
-  contents: read
-  pages: write
-  id-token: write
-
-# Allow one concurrent deployment; don't cancel an in-progress production deploy.
-concurrency:
-  group: pages
-  cancel-in-progress: false
-
-jobs:
-  deploy:
-    name: Build & deploy to GitHub Pages
-    runs-on: ubuntu-latest
-    environment:
-      name: github-pages
-      url: ${{ steps.deploy.outputs.page_url }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Upload site artifact
-        uses: actions/upload-pages-artifact@v3
-        with:
-          path: site
-
-      - name: Deploy to GitHub Pages
-        id: deploy
-        uses: actions/deploy-pages@v4
@@ -1,166 +0,0 @@
-name: Publish Python package
-
-on:
-  push:
-    branches: [main]
-    paths: ['pyproject.toml']
-  # Manual fallback: re-publish the current version without a version bump
-  # (e.g. first run after PyPI trusted publishing is configured).
-  workflow_dispatch:
-
-permissions:
-  contents: read
-
-jobs:
-  check-version:
-    name: Check version change
-    # Guard manual dispatches too: PyPI trusted publishing matches
-    # repo+workflow+environment but NOT branch, so without this a
-    # workflow_dispatch from any branch could publish unmerged code.
-    if: github.ref == 'refs/heads/main'
-    runs-on: ubuntu-latest
-    outputs:
-      changed: ${{ steps.check.outputs.changed }}
-      published: ${{ steps.check.outputs.published }}
-      version: ${{ steps.check.outputs.version }}
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          fetch-depth: 2
-
-      - name: Check if version changed
-        id: check
-        run: |
-          NEW_VERSION=$(python3 -c 'import pathlib, tomllib; print(tomllib.loads(pathlib.Path("pyproject.toml").read_text())["project"]["version"])')
-          echo "version=$NEW_VERSION" >> "$GITHUB_OUTPUT"
-
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-            # Manual dispatch always rebuilds and publishes. Combined with
-            # skip-existing on the publish step, this repairs partial releases
-            # (PyPI's version endpoint returns 200 even when only some of the
-            # expected wheels were uploaded).
-            echo "manual dispatch: publishing v$NEW_VERSION (skip-existing handles uploaded files)"
-            echo "changed=true" >> "$GITHUB_OUTPUT"
-            echo "published=false" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          # .get(): the parent commit may predate the static version field
-          # (pyproject.toml used dynamic = ["version"]) — treat that as a change
-          # so the very first merge of this workflow publishes.
-          OLD_VERSION=$(git show HEAD~1:pyproject.toml | python3 -c 'import sys, tomllib; print(tomllib.loads(sys.stdin.read())["project"].get("version", ""))')
-          echo "old=$OLD_VERSION new=$NEW_VERSION"
-
-          if [ "$NEW_VERSION" = "$OLD_VERSION" ]; then
-            echo "changed=false" >> "$GITHUB_OUTPUT"
-            echo "published=false" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          echo "changed=true" >> "$GITHUB_OUTPUT"
-
-          HTTP_STATUS=$(curl --silent --show-error --output /tmp/pypi-version.json --write-out "%{http_code}" \
-            "https://pypi.org/pypi/pdf-inspector/$NEW_VERSION/json")
-
-          case "$HTTP_STATUS" in
-            200)
-              echo "published=true" >> "$GITHUB_OUTPUT"
-              echo "pdf-inspector v$NEW_VERSION is already published to PyPI"
-              ;;
-            404)
-              echo "published=false" >> "$GITHUB_OUTPUT"
-              ;;
-            *)
-              cat /tmp/pypi-version.json
-              echo "Unexpected PyPI response: $HTTP_STATUS" >&2
-              exit 1
-              ;;
-          esac
-
-  build:
-    needs: check-version
-    if: needs.check-version.outputs.changed == 'true' && needs.check-version.outputs.published == 'false'
-    name: Build ${{ matrix.target }}
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        include:
-          - os: ubuntu-latest
-            target: x86_64-unknown-linux-gnu
-          - os: ubuntu-latest
-            target: aarch64-unknown-linux-gnu
-          # macos-13 was retired by GitHub; macos-15-intel is the remaining
-          # Intel runner label (available through 2027).
-          - os: macos-15-intel
-            target: x86_64-apple-darwin
-          - os: macos-14
-            target: aarch64-apple-darwin
-          - os: windows-latest
-            target: x86_64-pc-windows-msvc
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-python@v5
-        with:
-          python-version: '3.12'
-
-      - name: Build wheel
-        uses: PyO3/maturin-action@v1
-        with:
-          target: ${{ matrix.target }}
-          args: --release --out dist
-          manylinux: auto
-
-      - name: Upload wheel
-        uses: actions/upload-artifact@v4
-        with:
-          name: wheels-${{ matrix.target }}
-          path: dist/*.whl
-          if-no-files-found: error
-
-  sdist:
-    needs: check-version
-    if: needs.check-version.outputs.changed == 'true' && needs.check-version.outputs.published == 'false'
-    name: Build sdist
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Build sdist
-        uses: PyO3/maturin-action@v1
-        with:
-          command: sdist
-          args: --out dist
-
-      - name: Upload sdist
-        uses: actions/upload-artifact@v4
-        with:
-          name: sdist
-          path: dist/*.tar.gz
-          if-no-files-found: error
-
-  publish:
-    name: Publish to PyPI
-    needs: [check-version, build, sdist]
-    runs-on: ubuntu-latest
-    environment: pypi
-    permissions:
-      contents: read
-      id-token: write
-    steps:
-      - name: Download all artifacts
-        uses: actions/download-artifact@v4
-        with:
-          path: dist
-          merge-multiple: true
-
-      - name: List artifacts
-        run: ls -la dist/
-
-      - name: Publish to PyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: dist
-          # Tolerate already-uploaded files so a manual re-run can complete
-          # a release that previously failed partway through.
-          skip-existing: true
@@ -14,7 +14,7 @@ crate-type = ["lib", "cdylib"]

 [dependencies]
 # Python bindings
-pyo3 = { version = "0.25", features = ["extension-module", "abi3-py38"], optional = true }
+pyo3 = { version = "0.25", features = ["extension-module"], optional = true }

 # PDF parsing
 lopdf = { version = "0.41.0", features = ["rayon"] }
@@ -26,16 +26,16 @@ Evaluated on the [opendataloader-bench](https://github.com/opendataloader-projec

 | Engine | Overall | Reading Order (NID) | Tables (TEDS) | Headings (MHS) | Speed (200 docs) |
 |---|---|---|---|---|---|
-| pdf-inspector | 0.83 | 0.88 | 0.66 | 0.74 | 4s |
+| pdf-inspector | 0.78 | 0.87 | 0.59 | 0.57 | 4s |
 | opendataloader | 0.84 | 0.91 | 0.49 | 0.74 | 11s |
 | pymupdf4llm | 0.73 | 0.89 | 0.40 | 0.41 | 18s |
 | markitdown | 0.58 | 0.88 | 0.00 | 0.00 | 8s |

-For context, engines that use OCR/ML (docling, marker, mineru) score 0.83-0.88 overall but take 2-180 minutes on the same corpus — pdf-inspector reaches the low end of that range without any OCR, in 4 seconds.
+For context, engines that use OCR/ML (docling, marker, mineru) score 0.83-0.88 overall but take 2-180 minutes on the same corpus.

-**Where we do well:** Speed (fastest of all engines), the best table detection of any engine shown, and heading detection now on par with opendataloader. Overall lands within 0.01 of opendataloader at roughly 2.5× the speed.
+**Where we do well:** Speed (fastest of all engines), reading order, table detection vs other direct-text tools.

-**Where we lag:** Reading order still trails opendataloader slightly, and table structure trails OCR-based engines that can see visual layout.
+**Where we lag:** Heading detection trails opendataloader — many PDFs use bold text at body font size for headings, or headings that are only slightly larger than body text. Table detection trails OCR-based engines that can see visual table structure.

 ## Quick start

@@ -1,6 +1,6 @@
 {
  "name": "@firecrawl/pdf-inspector",
-  "version": "1.10.4",
+  "version": "1.10.0",
  "description": "Fast PDF classification and text extraction. Detect text-based vs scanned PDFs, extract text by region with quality checks. Native Rust performance via napi-rs.",
  "main": "index.js",
  "types": "index.d.ts",
@@ -4,9 +4,9 @@ build-backend = "maturin"

 [project]
 name = "pdf-inspector"
-# Bump this to publish to PyPI — CI publishes automatically when the version
-# changes on main (same flow as napi/package.json for npm).
-version = "0.2.1"
+# Version is sourced from Cargo.toml [package] version by maturin so the Python
+# artifact always tracks the crate release instead of drifting on its own.
+dynamic = ["version"]
 description = "Fast PDF inspection, classification, and text extraction with smart scanned vs text-based detection"
 license = { text = "MIT" }
 requires-python = ">=3.8"
@@ -1,3 +0,0 @@
-<svg width="200" height="284" viewBox="0 0 200 284" fill="none" xmlns="http://www.w3.org/2000/svg">
-<path d="M166.862 90.7716C155.812 94.0514 147.483 101.471 141.383 109.53C140.073 111.26 137.343 109.96 137.863 107.841C149.543 59.8136 134.113 19.896 86.0157 0.247269C83.5758 -0.752669 81.0359 1.43719 81.6759 3.99704C103.555 91.8416 11.5294 84.432 23.1588 184.016C23.3588 185.726 21.4389 186.896 20.039 185.896C15.6792 182.766 10.8095 176.236 7.46963 171.647C6.48968 170.297 4.36978 170.677 3.9198 172.287C1.25994 181.906 0 190.965 0 199.965C0 234.963 17.9891 265.771 45.2177 283.63C46.7777 284.65 48.7776 283.19 48.2476 281.4C46.8477 276.7 46.0577 271.74 45.9977 266.611C45.9977 263.461 46.1977 260.241 46.6877 257.241C47.8276 249.702 50.4475 242.522 54.8473 235.983C69.9365 213.334 100.185 191.455 95.3552 161.747C95.0453 159.867 97.2651 158.627 98.6651 159.917C119.974 179.386 124.194 205.575 120.694 229.063C120.394 231.103 122.954 232.193 124.244 230.593C127.504 226.513 131.483 222.933 135.813 220.244C136.893 219.574 138.333 220.084 138.743 221.284C141.153 228.293 144.733 234.873 148.113 241.452C152.152 249.362 154.302 258.391 153.962 267.951C153.792 272.6 153.022 277.1 151.732 281.38C151.182 283.19 153.162 284.7 154.752 283.66C182.001 265.801 200 234.993 200 199.975C200 187.806 197.87 175.876 193.84 164.697C185.391 141.248 163.952 123.64 169.372 93.0815C169.632 91.6216 168.282 90.3517 166.862 90.7716Z" fill="#FA5D19" style="fill:#FA5D19;fill:color(display-p3 0.9816 0.3634 0.0984);fill-opacity:1;"/>
-</svg>
@@ -1,12 +0,0 @@
-<svg width="172" height="40" viewBox="0 0 172 40" fill="none" xmlns="http://www.w3.org/2000/svg">
-<path d="M23.3606 12.8281C21.8137 13.2873 20.6476 14.3261 19.7936 15.4544C19.6102 15.6966 19.228 15.5146 19.3008 15.2178C20.936 8.49401 18.7759 2.90556 12.0422 0.154735C11.7006 0.0147436 11.345 0.321324 11.4346 0.679702C14.4977 12.9779 1.61412 11.9406 3.24224 25.8823C3.27024 26.1217 3.00145 26.2855 2.80546 26.1455C2.19509 25.7073 1.51332 24.7932 1.04575 24.1506C0.908555 23.9616 0.611769 24.0148 0.548773 24.2402C0.176391 25.5869 0 26.8553 0 28.1152C0 33.0149 2.51847 37.328 6.33048 39.8283C6.54887 39.9711 6.82886 39.7667 6.75466 39.5161C6.55867 38.8581 6.44808 38.1638 6.43968 37.4456C6.43968 37.0046 6.46768 36.5539 6.53627 36.1339C6.69587 35.0784 7.06265 34.0732 7.67862 33.1577C9.79111 29.9869 14.0259 26.9239 13.3497 22.7647C13.3063 22.5015 13.6171 22.328 13.8131 22.5085C16.7964 25.2342 17.3871 28.9005 16.8972 32.1889C16.8552 32.4745 17.2135 32.6271 17.3941 32.4031C17.8505 31.832 18.4077 31.3308 19.0138 30.9542C19.165 30.8604 19.3666 30.9318 19.424 31.0998C19.7614 32.0811 20.2626 33.0023 20.7358 33.9234C21.3013 35.0308 21.6023 36.2949 21.5547 37.6332C21.5309 38.2842 21.4231 38.9141 21.2425 39.5133C21.1655 39.7667 21.4427 39.9781 21.6653 39.8325C25.4801 37.3322 28 33.0191 28 28.1166C28 26.4129 27.7018 24.7428 27.1376 23.1777C25.9547 19.8949 22.9533 17.4297 23.712 13.1515C23.7484 12.9471 23.5594 12.7693 23.3606 12.8281Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M41 34.0521V10.9618H55.7586V14.3264H44.7969V21.0226H53.8436V24.2882H44.7969V34.0521H41Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M59.9569 14.7882C58.7352 14.7882 57.7777 13.8976 57.7777 12.6441C57.7777 11.3906 58.7352 10.5 59.9569 10.5C61.1785 10.5 62.136 11.3906 62.136 12.6441C62.136 13.8976 61.1785 14.7882 59.9569 14.7882ZM58.1409 34.0521V17.1632H61.7068V34.0521H58.1409Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M73.5885 17.1632H74.3809V20.4948H72.796C69.6264 20.4948 68.6029 22.9687 68.6029 25.5747V34.0521H65.0371V17.1632H68.2067L68.6029 19.7031C69.4613 18.2847 70.815 17.1632 73.5885 17.1632Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M83.632 34.25C78.3163 34.25 74.9816 30.8194 74.9816 25.6406C74.9816 20.4288 78.3163 16.9653 83.3019 16.9653C88.1884 16.9653 91.457 20.066 91.5561 25.0139C91.5561 25.4427 91.5231 25.9045 91.457 26.3663H78.7125V26.5972C78.8116 29.467 80.6275 31.3472 83.4339 31.3472C85.613 31.3472 87.1979 30.2587 87.6931 28.3785H91.2589C90.6646 31.7101 87.8252 34.25 83.632 34.25ZM78.8446 23.7604H87.8582C87.561 21.2535 85.8112 19.8351 83.3349 19.8351C81.0567 19.8351 79.1087 21.3524 78.8446 23.7604Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M102.033 34.25C96.9151 34.25 93.6465 30.9184 93.6465 25.6406C93.6465 20.4288 97.0142 16.9653 102.132 16.9653C106.49 16.9653 109.197 19.3733 109.891 23.1997H106.16C105.698 21.2205 104.278 20 102.066 20C99.1933 20 97.3113 22.309 97.3113 25.6406C97.3113 28.9392 99.1933 31.2153 102.066 31.2153C104.245 31.2153 105.698 29.9618 106.127 28.0156H109.891C109.23 31.842 106.358 34.25 102.033 34.25Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M121.006 17.1632H121.799V20.4948H120.214C117.044 20.4948 116.021 22.9687 116.021 25.5747V34.0521H112.455V17.1632H115.625L116.021 19.7031C116.879 18.2847 118.233 17.1632 121.006 17.1632Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M130.614 16.9653C135.104 16.9653 137.679 19.1094 137.679 23.1007V34.0521H134.576L134.279 31.6441C133.123 33.1615 131.505 34.25 128.831 34.25C125.133 34.25 122.657 32.4358 122.657 29.3021C122.657 25.8385 125.166 23.8924 129.92 23.8924H134.147V22.8698C134.147 20.9896 132.793 19.8351 130.449 19.8351C128.336 19.8351 126.916 20.8247 126.652 22.309H123.152C123.515 19.0104 126.355 16.9653 130.614 16.9653ZM129.425 31.4792C132.397 31.4792 134.114 29.7309 134.147 27.125V26.5312H129.722C127.51 26.5312 126.289 27.3559 126.289 29.0712C126.289 30.4896 127.477 31.4792 129.425 31.4792Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M144.653 34.0521L139.139 17.1632H142.903L146.766 30.0937L150.629 17.1632H153.897L157.595 30.0937L161.59 17.1632H165.222L159.609 34.0521H155.779L152.214 22.5729L148.516 34.0521H144.653Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-<path d="M166.934 34.0521V10.9618H170.5V34.0521H166.934Z" fill="#262626" style="fill:#262626;fill:color(display-p3 0.1500 0.1500 0.1500);fill-opacity:1;"/>
-</svg>
@@ -1,428 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-<head>
-<meta charset="utf-8">
-<meta name="viewport" content="width=device-width, initial-scale=1">
-<title>pdf-inspector — PDF classification &amp; text extraction, no OCR</title>
-<meta name="description" content="Fast Rust library that classifies PDFs (text-based vs scanned) and extracts clean Markdown — no OCR, no ML models. Bindings for Rust, Python, and Node.js.">
-<meta property="og:title" content="pdf-inspector">
-<meta property="og:description" content="Classify PDFs and extract clean Markdown in milliseconds. No OCR. No ML. Pure Rust.">
-<meta property="og:type" content="website">
-<link rel="icon" href="data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 100 100'%3E%3Ctext y='.9em' font-size='90'%3E%F0%9F%93%84%3C/text%3E%3C/svg%3E">
-<link rel="preconnect" href="https://fonts.googleapis.com">
-<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
-<link href="https://fonts.googleapis.com/css2?family=Bricolage+Grotesque:opsz,wght@12..96,400;12..96,600;12..96,800&family=Hanken+Grotesk:wght@400;500;600&family=JetBrains+Mono:wght@400;500;700&display=swap" rel="stylesheet">
-<style>
-  :root {
-    --paper: #f4efe4;
-    --paper-2: #eee7d8;
-    --ink: #1b1712;
-    --ink-soft: #4a433a;
-    --muted: #8b8375;
-    --line: #d9cfbb;
-    --accent: #dd3f22;
-    --accent-deep: #b32d15;
-    --card: #faf6ec;
-    --display: "Bricolage Grotesque", serif;
-    --body: "Hanken Grotesk", sans-serif;
-    --mono: "JetBrains Mono", monospace;
-  }
-  * { box-sizing: border-box; margin: 0; padding: 0; }
-  html { scroll-behavior: smooth; }
-  body {
-    background: var(--paper);
-    color: var(--ink);
-    font-family: var(--body);
-    font-size: 17px;
-    line-height: 1.6;
-    -webkit-font-smoothing: antialiased;
-    overflow-x: hidden;
-    background-image:
-      radial-gradient(circle at 1px 1px, rgba(27,23,18,0.05) 1px, transparent 0);
-    background-size: 22px 22px;
-  }
-  ::selection { background: var(--accent); color: var(--paper); }
-  a { color: inherit; text-decoration: none; }
-
-  .wrap { max-width: 1120px; margin: 0 auto; padding: 0 28px; }
-
-  /* ── nav ── */
-  nav {
-    position: sticky; top: 0; z-index: 50;
-    background: rgba(244,239,228,0.82);
-    backdrop-filter: blur(10px);
-    border-bottom: 1px solid var(--line);
-  }
-  .nav-in { display: flex; align-items: center; gap: 22px; height: 60px; }
-  .brand { font-family: var(--mono); font-weight: 700; font-size: 15px; letter-spacing: -0.02em; display: flex; align-items: center; gap: 9px; }
-  .brand .dot { width: 9px; height: 9px; background: var(--accent); border-radius: 50%; box-shadow: 0 0 0 3px rgba(221,63,34,0.18); }
-  .nav-links { margin-left: auto; display: flex; gap: 24px; align-items: center; font-size: 14.5px; font-weight: 500; }
-  .nav-links a { color: var(--ink-soft); transition: color .15s; }
-  .nav-links a:hover { color: var(--accent); }
-  .nav-gh { border: 1px solid var(--ink); border-radius: 999px; padding: 6px 15px; color: var(--ink) !important; transition: all .15s; }
-  .nav-gh:hover { background: var(--ink); color: var(--paper) !important; }
-  @media (max-width: 680px) { .nav-links .hide-sm { display: none; } }
-
-  /* ── hero ── */
-  header { padding: 74px 0 40px; position: relative; }
-  .eyebrow { font-family: var(--mono); font-size: 12.5px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--accent-deep); margin-bottom: 22px; }
-  h1 {
-    font-family: var(--display);
-    font-weight: 800;
-    font-size: clamp(2.9rem, 8vw, 6.1rem);
-    line-height: 0.96;
-    letter-spacing: -0.035em;
-    max-width: 15ch;
-  }
-  h1 .em { color: var(--accent); font-style: normal; position: relative; }
-  h1 .strike { position: relative; white-space: nowrap; }
-  h1 .strike::after { content: ""; position: absolute; left: -2%; right: -2%; top: 54%; height: 0.09em; background: var(--accent); transform: rotate(-3deg); }
-  .lede { margin-top: 28px; font-size: clamp(1.05rem, 2.2vw, 1.32rem); color: var(--ink-soft); max-width: 46ch; line-height: 1.5; }
-  .lede b { color: var(--ink); font-weight: 600; }
-
-  .hero-grid { display: grid; grid-template-columns: 1.15fr 0.85fr; gap: 48px; align-items: end; }
-  @media (max-width: 880px) { .hero-grid { grid-template-columns: 1fr; gap: 40px; } }
-
-  /* readout card */
-  .readout {
-    background: var(--ink); color: var(--paper);
-    border-radius: 14px; padding: 22px 22px 20px;
-    font-family: var(--mono); font-size: 13px;
-    box-shadow: 14px 14px 0 rgba(27,23,18,0.09);
-    position: relative;
-  }
-  .readout .rlabel { color: #b8ad98; font-size: 11px; letter-spacing: 0.14em; text-transform: uppercase; margin-bottom: 16px; display: flex; justify-content: space-between; }
-  .readout .rrow { display: flex; justify-content: space-between; align-items: center; padding: 9px 0; border-top: 1px solid rgba(255,255,255,0.09); }
-  .readout .rrow:first-of-type { border-top: none; }
-  .readout .k { color: #cfc6b3; }
-  .readout .v { font-weight: 700; }
-  .readout .v.hot { color: var(--accent); }
-  .bar { height: 6px; background: rgba(255,255,255,0.1); border-radius: 3px; overflow: hidden; margin-top: 3px; width: 96px; }
-  .bar > i { display: block; height: 100%; background: var(--accent); border-radius: 3px; }
-
-  /* ── install row ── */
-  .installs { display: grid; grid-template-columns: repeat(3,1fr); gap: 14px; margin-top: 54px; }
-  @media (max-width: 720px) { .installs { grid-template-columns: 1fr; } }
-  .inst {
-    background: var(--card); border: 1px solid var(--line); border-radius: 11px;
-    padding: 15px 17px; transition: transform .16s, border-color .16s, box-shadow .16s;
-    cursor: pointer;
-  }
-  .inst:hover { transform: translateY(-3px); border-color: var(--accent); box-shadow: 0 8px 22px rgba(27,23,18,0.07); }
-  .inst .reg { font-family: var(--mono); font-size: 11px; letter-spacing: 0.12em; text-transform: uppercase; color: var(--muted); margin-bottom: 8px; display: flex; justify-content: space-between; }
-  .inst code { font-family: var(--mono); font-size: 14px; color: var(--ink); font-weight: 500; }
-  .inst .arrow { color: var(--accent); opacity: 0; transition: opacity .16s; }
-  .inst:hover .arrow { opacity: 1; }
-
-  /* ── section scaffold ── */
-  section { padding: 66px 0; border-top: 1px solid var(--line); }
-  .sec-head { display: flex; align-items: baseline; gap: 16px; margin-bottom: 40px; flex-wrap: wrap; }
-  .sec-num { font-family: var(--mono); font-size: 13px; color: var(--accent); font-weight: 700; }
-  .sec-title { font-family: var(--display); font-weight: 600; font-size: clamp(1.7rem, 4vw, 2.6rem); letter-spacing: -0.025em; }
-  .sec-sub { color: var(--ink-soft); max-width: 52ch; font-size: 1.02rem; }
-
-  /* ── features ── */
-  .feat-grid { display: grid; grid-template-columns: repeat(4,1fr); gap: 1px; background: var(--line); border: 1px solid var(--line); border-radius: 14px; overflow: hidden; }
-  @media (max-width: 900px) { .feat-grid { grid-template-columns: repeat(2,1fr); } }
-  @media (max-width: 560px) { .feat-grid { grid-template-columns: 1fr; } }
-  .feat { background: var(--card); padding: 24px 22px; transition: background .16s; }
-  .feat:hover { background: #fff; }
-  .feat .fn { font-family: var(--mono); font-size: 12px; color: var(--accent); font-weight: 700; }
-  .feat h3 { font-family: var(--display); font-weight: 600; font-size: 1.16rem; margin: 12px 0 8px; letter-spacing: -0.01em; }
-  .feat p { font-size: 14.5px; color: var(--ink-soft); line-height: 1.5; }
-
-  /* ── benchmark ── */
-  .bench {
-    border: 1px solid var(--line); border-radius: 14px; overflow: hidden;
-    background: var(--card);
-  }
-  table { width: 100%; border-collapse: collapse; font-size: 15px; }
-  thead th { font-family: var(--mono); font-size: 11px; letter-spacing: 0.08em; text-transform: uppercase; color: var(--muted); text-align: right; padding: 15px 18px; background: var(--paper-2); border-bottom: 1px solid var(--line); font-weight: 500; }
-  thead th:first-child { text-align: left; }
-  tbody td { padding: 14px 18px; text-align: right; font-family: var(--mono); border-bottom: 1px solid var(--line); }
-  tbody td:first-child { text-align: left; font-family: var(--body); font-weight: 500; }
-  tbody tr:last-child td { border-bottom: none; }
-  tbody tr.us { background: rgba(221,63,34,0.06); }
-  tbody tr.us td:first-child { color: var(--accent-deep); font-weight: 700; }
-  tbody tr.us td:first-child::before { content: "▸ "; color: var(--accent); }
-  .bench-foot { padding: 15px 18px; font-size: 13.5px; color: var(--ink-soft); background: var(--paper-2); border-top: 1px solid var(--line); }
-  .bench-wrap { overflow-x: auto; }
-  .callouts { display: grid; grid-template-columns: 1fr 1fr; gap: 16px; margin-top: 22px; }
-  @media (max-width: 640px) { .callouts { grid-template-columns: 1fr; } }
-  .callout { border-left: 3px solid var(--accent); padding: 4px 0 4px 16px; }
-  .callout .ct { font-family: var(--mono); font-size: 11px; letter-spacing: 0.1em; text-transform: uppercase; color: var(--accent-deep); margin-bottom: 5px; }
-  .callout p { font-size: 14.5px; color: var(--ink-soft); }
-
-  /* ── quickstart tabs ── */
-  .tabs input { position: absolute; opacity: 0; pointer-events: none; }
-  .tablist { display: flex; gap: 6px; margin-bottom: 0; }
-  .tablist label {
-    font-family: var(--mono); font-size: 13px; font-weight: 500;
-    padding: 10px 18px; cursor: pointer; color: var(--muted);
-    border: 1px solid var(--line); border-bottom: none;
-    border-radius: 9px 9px 0 0; background: var(--paper-2); transition: all .15s;
-  }
-  .tablist label:hover { color: var(--ink); }
-  .panel { display: none; }
-  .code {
-    background: var(--ink); border-radius: 0 12px 12px 12px;
-    padding: 22px 24px; overflow-x: auto;
-    font-family: var(--mono); font-size: 13.5px; line-height: 1.7;
-    color: #e9e2d3;
-    box-shadow: 12px 12px 0 rgba(27,23,18,0.07);
-  }
-  .code .cm { color: #8a8069; }
-  .code .kw { color: #ff9f7a; }
-  .code .st { color: #cbb78a; }
-  .code .fn { color: #f4efe4; font-weight: 700; }
-  #t-rust:checked ~ .tablist label[for=t-rust],
-  #t-py:checked ~ .tablist label[for=t-py],
-  #t-node:checked ~ .tablist label[for=t-node],
-  #t-cli:checked ~ .tablist label[for=t-cli] {
-    background: var(--ink); color: var(--paper); border-color: var(--ink);
-  }
-  #t-rust:checked ~ .panels #p-rust,
-  #t-py:checked ~ .panels #p-py,
-  #t-node:checked ~ .panels #p-node,
-  #t-cli:checked ~ .panels #p-cli { display: block; }
-  .code a.ref { color: #ff9f7a; border-bottom: 1px dotted #ff9f7a; }
-
-  /* ── closing split (OSS vs hosted) ── */
-  .split { display: grid; grid-template-columns: 1fr 1fr; gap: 18px; }
-  @media (max-width: 780px) { .split { grid-template-columns: 1fr; } }
-  .path { border: 1px solid var(--line); border-radius: 16px; padding: 32px 30px; background: var(--card); display: flex; flex-direction: column; }
-  .path .ptag { font-family: var(--mono); font-size: 11px; letter-spacing: 0.12em; text-transform: uppercase; color: var(--muted); margin-bottom: 15px; }
-  .path h3 { font-family: var(--display); font-weight: 600; font-size: 1.5rem; letter-spacing: -0.02em; line-height: 1.05; margin-bottom: 12px; }
-  .path p { color: var(--ink-soft); font-size: 15px; line-height: 1.5; flex: 1; margin-bottom: 24px; }
-  .pbtns { display: flex; gap: 12px; flex-wrap: wrap; }
-  .path-pro { background: var(--ink); border-color: var(--ink); box-shadow: 14px 14px 0 rgba(27,23,18,0.09); }
-  .path-pro .ptag { color: #b8ad98; }
-  .path-pro .ptag b { color: var(--accent); font-weight: 700; }
-  .path-pro h3 { color: var(--paper); }
-  .path-pro p { color: #cfc6b3; }
-  .btn { font-family: var(--mono); font-size: 14px; font-weight: 500; padding: 13px 24px; border-radius: 999px; transition: all .15s; border: 1px solid var(--ink); }
-  .btn-p { background: var(--accent); border-color: var(--accent); color: var(--paper); }
-  .btn-p:hover { background: var(--accent-deep); border-color: var(--accent-deep); }
-  .btn-s:hover { background: var(--ink); color: var(--paper); }
-  .btn-pro { background: var(--accent); border-color: var(--accent); color: var(--paper); }
-  .btn-pro:hover { background: var(--accent-deep); border-color: var(--accent-deep); }
-  .btn-ghost { color: var(--paper); border-color: rgba(255,255,255,0.3); }
-  .btn-ghost:hover { border-color: var(--paper); background: rgba(255,255,255,0.08); }
-  .fc-mark { height: 34px; width: auto; display: block; margin-bottom: 20px; }
-  .fc-wordmark { height: 15px; width: auto; vertical-align: -2px; transition: opacity .15s; }
-  .fc-wordmark:hover { opacity: 0.65; }
-  footer { border-top: 1px solid var(--line); padding: 34px 0; font-size: 14px; color: var(--muted); }
-  .foot-in { display: flex; justify-content: space-between; align-items: center; gap: 18px; flex-wrap: wrap; }
-  .foot-in a { color: var(--ink-soft); }
-  .foot-in a:hover { color: var(--accent); }
-  .foot-links { display: flex; gap: 20px; font-family: var(--mono); font-size: 13px; }
-
-  /* ── load animation ── */
-  .reveal { opacity: 0; transform: translateY(14px); animation: rise .7s cubic-bezier(.2,.7,.3,1) forwards; }
-  @keyframes rise { to { opacity: 1; transform: none; } }
-  .d1 { animation-delay: .05s; } .d2 { animation-delay: .15s; } .d3 { animation-delay: .25s; }
-  .d4 { animation-delay: .35s; } .d5 { animation-delay: .45s; } .d6 { animation-delay: .55s; }
-  @media (prefers-reduced-motion: reduce) { .reveal { animation: none; opacity: 1; transform: none; } }
-</style>
-</head>
-<body>
-
-<nav>
-  <div class="wrap nav-in">
-    <a href="#top" class="brand"><span class="dot"></span>pdf-inspector</a>
-    <div class="nav-links">
-      <a href="#features" class="hide-sm">Features</a>
-      <a href="#benchmark" class="hide-sm">Benchmark</a>
-      <a href="#start">Quick start</a>
-      <a class="nav-gh" href="https://github.com/firecrawl/pdf-inspector">GitHub ↗</a>
-    </div>
-  </div>
-</nav>
-
-<header id="top">
-  <div class="wrap hero-grid">
-    <div>
-      <div class="eyebrow reveal d1">Rust · Python · Node · CLI</div>
-      <h1 class="reveal d2">Classify PDFs. Extract Markdown. <span class="strike">No OCR.</span></h1>
-      <p class="lede reveal d3">A fast Rust library that tells text-based PDFs from scanned ones, then extracts position-aware text and clean Markdown — <b>locally, in milliseconds</b>. Skip the OCR bill for the ~54% of PDFs that never needed it.</p>
-    </div>
-    <div class="readout reveal d4" aria-hidden="true">
-      <div class="rlabel"><span>classify_pdf()</span><span>~12ms</span></div>
-      <div class="rrow"><span class="k">type</span><span class="v hot">TextBased</span></div>
-      <div class="rrow"><span class="k">confidence</span><span class="v">0.98</span></div>
-      <div class="rrow"><span class="k">needs_ocr</span><span class="v">false</span></div>
-      <div class="rrow"><span class="k">route</span><span class="v">local&nbsp;→&nbsp;md</span></div>
-      <div class="rrow" style="border-top:1px solid rgba(255,255,255,.09);padding-top:13px">
-        <span class="k">signal</span>
-        <span class="bar"><i style="width:98%"></i></span>
-      </div>
-    </div>
-  </div>
-
-  <div class="wrap">
-    <div class="installs">
-      <a class="inst reveal d4" href="https://crates.io/crates/pdf-inspector">
-        <div class="reg"><span>crates.io</span><span class="arrow">↗</span></div>
-        <code>cargo add pdf-inspector</code>
-      </a>
-      <a class="inst reveal d5" href="https://pypi.org/project/pdf-inspector/">
-        <div class="reg"><span>PyPI</span><span class="arrow">↗</span></div>
-        <code>pip install pdf-inspector</code>
-      </a>
-      <a class="inst reveal d6" href="https://www.npmjs.com/package/@firecrawl/pdf-inspector">
-        <div class="reg"><span>npm</span><span class="arrow">↗</span></div>
-        <code>npm i @firecrawl/pdf-inspector</code>
-      </a>
-    </div>
-  </div>
-</header>
-
-<section id="features">
-  <div class="wrap">
-    <div class="sec-head">
-      <span class="sec-num">01</span>
-      <h2 class="sec-title">Built for routing, not just reading</h2>
-    </div>
-    <div class="feat-grid">
-      <div class="feat"><div class="fn">01</div><h3>Smart classification</h3><p>TextBased, Scanned, ImageBased, or Mixed in ~10–50ms by sampling content streams. Returns a confidence score and per-page OCR routing.</p></div>
-      <div class="feat"><div class="fn">02</div><h3>Position-aware text</h3><p>Extraction with font info, X/Y coordinates, and automatic multi-column reading order.</p></div>
-      <div class="feat"><div class="fn">03</div><h3>Markdown conversion</h3><p>Headings, bullet/numbered lists, code blocks, tables, bold/italic, URL linking, and page breaks.</p></div>
-      <div class="feat"><div class="fn">04</div><h3>Table detection</h3><p>Rectangle-based detection from drawing ops plus heuristic alignment detection. Financial tables, footnotes, and cross-page continuations.</p></div>
-      <div class="feat"><div class="fn">05</div><h3>CID font support</h3><p>ToUnicode CMap decoding for Type0/Identity-H fonts, with UTF-16BE, UTF-8, and Latin-1 encodings.</p></div>
-      <div class="feat"><div class="fn">06</div><h3>Multi-column layout</h3><p>Newspaper-style column detection, sequential reading order, and right-to-left text support.</p></div>
-      <div class="feat"><div class="fn">07</div><h3>Encoding checks</h3><p>Flags broken font encodings automatically so callers can fall back to OCR only when it's actually needed.</p></div>
-      <div class="feat"><div class="fn">08</div><h3>Lightweight</h3><p>Pure Rust. No ML models, no external services. A single parse shared between detection and extraction.</p></div>
-    </div>
-  </div>
-</section>
-
-<section id="benchmark">
-  <div class="wrap">
-    <div class="sec-head">
-      <span class="sec-num">02</span>
-      <h2 class="sec-title">Fastest of the direct-text engines</h2>
-      <p class="sec-sub">Evaluated on the <a href="https://github.com/opendataloader-project/opendataloader-bench" style="color:var(--accent-deep);border-bottom:1px solid var(--line)">opendataloader-bench</a> corpus (200 PDFs). Direct text-extraction engines only — no OCR, no ML. Higher is better.</p>
-    </div>
-    <div class="bench">
-      <div class="bench-wrap">
-      <table>
-        <thead>
-          <tr><th>Engine</th><th>Overall</th><th>Reading order</th><th>Tables</th><th>Headings</th><th>200 docs</th></tr>
-        </thead>
-        <tbody>
-          <tr class="us"><td>pdf-inspector</td><td>0.83</td><td>0.88</td><td>0.66</td><td>0.74</td><td>4s</td></tr>
-          <tr><td>opendataloader</td><td>0.84</td><td>0.91</td><td>0.49</td><td>0.74</td><td>11s</td></tr>
-          <tr><td>pymupdf4llm</td><td>0.73</td><td>0.89</td><td>0.40</td><td>0.41</td><td>18s</td></tr>
-          <tr><td>markitdown</td><td>0.58</td><td>0.88</td><td>0.00</td><td>0.00</td><td>8s</td></tr>
-        </tbody>
-      </table>
-      </div>
-      <div class="bench-foot">OCR/ML engines (docling, marker, mineru) score 0.83–0.88 overall — but take 2–180 minutes on the same corpus.</div>
-    </div>
-    <div class="callouts">
-      <div class="callout"><div class="ct">Where we win</div><p>Fastest engine measured, the best table detection of any engine here, and heading quality now on par with opendataloader — at ~2.5× its speed.</p></div>
-      <div class="callout"><div class="ct">Where we're working</div><p>Reading order still trails opendataloader slightly, and tables that need the visual structure only an OCR engine can see.</p></div>
-    </div>
-  </div>
-</section>
-
-<section id="start">
-  <div class="wrap">
-    <div class="sec-head">
-      <span class="sec-num">03</span>
-      <h2 class="sec-title">Three lines to Markdown</h2>
-    </div>
-    <div class="tabs">
-      <input type="radio" name="tab" id="t-rust" checked>
-      <input type="radio" name="tab" id="t-py">
-      <input type="radio" name="tab" id="t-node">
-      <input type="radio" name="tab" id="t-cli">
-      <div class="tablist">
-        <label for="t-rust">Rust</label>
-        <label for="t-py">Python</label>
-        <label for="t-node">Node.js</label>
-        <label for="t-cli">CLI</label>
-      </div>
-      <div class="panels">
-        <div class="panel" id="p-rust"><pre class="code"><span class="kw">use</span> pdf_inspector::process_pdf;
-
-<span class="kw">let</span> result = <span class="fn">process_pdf</span>(<span class="st">"document.pdf"</span>)?;
-<span class="fn">println!</span>(<span class="st">"Type: {:?}"</span>, result.pdf_type);
-<span class="kw">if let</span> <span class="kw">Some</span>(markdown) = &amp;result.markdown {
-    <span class="fn">println!</span>(<span class="st">"{}"</span>, markdown);
-}
-<span class="cm">// full reference → </span><a class="ref" href="https://github.com/firecrawl/pdf-inspector/blob/main/docs/rust-api.md">docs/rust-api.md</a></pre></div>
-        <div class="panel" id="p-py"><pre class="code"><span class="kw">import</span> pdf_inspector
-
-result = pdf_inspector.<span class="fn">process_pdf</span>(<span class="st">"document.pdf"</span>)
-<span class="fn">print</span>(result.pdf_type)   <span class="cm"># "text_based" | "scanned" | "image_based" | "mixed"</span>
-<span class="fn">print</span>(result.markdown)   <span class="cm"># Markdown string or None</span>
-<span class="cm"># full reference → </span><a class="ref" href="https://github.com/firecrawl/pdf-inspector/blob/main/docs/python.md">docs/python.md</a></pre></div>
-        <div class="panel" id="p-node"><pre class="code"><span class="kw">import</span> { readFileSync } <span class="kw">from</span> <span class="st">'fs'</span>;
-<span class="kw">import</span> { processPdf } <span class="kw">from</span> <span class="st">'@firecrawl/pdf-inspector'</span>;
-
-<span class="kw">const</span> result = <span class="fn">processPdf</span>(<span class="fn">readFileSync</span>(<span class="st">'document.pdf'</span>));
-console.<span class="fn">log</span>(result.pdfType);   <span class="cm">// "TextBased" | "Scanned" | ...</span>
-console.<span class="fn">log</span>(result.markdown);  <span class="cm">// Markdown string or null</span>
-<span class="cm">// full reference → </span><a class="ref" href="https://github.com/firecrawl/pdf-inspector/blob/main/napi/README.md">napi/README.md</a></pre></div>
-        <div class="panel" id="p-cli"><pre class="code"><span class="cm"># install the CLI tools</span>
-cargo <span class="fn">install</span> pdf-inspector
-
-<span class="cm"># convert a PDF to Markdown</span>
-<span class="fn">pdf2md</span> document.pdf
-
-<span class="cm"># classify only — is it scanned?</span>
-<span class="fn">detect-pdf</span> document.pdf --analyze --json
-
-<span class="cm"># structured output for pipelines</span>
-<span class="fn">pdf2md</span> document.pdf --json</pre></div>
-      </div>
-    </div>
-  </div>
-</section>
-
-<section class="close">
-  <div class="wrap">
-    <div class="sec-head">
-      <span class="sec-num">04</span>
-      <h2 class="sec-title">Two ways to parse</h2>
-      <p class="sec-sub">Run the classifier locally for text-based PDFs; hand the scanned, OCR, and at-scale work to Firecrawl.</p>
-    </div>
-    <div class="split">
-      <div class="path">
-        <div class="ptag">Open source · runs local</div>
-        <h3>Use pdf-inspector yourself</h3>
-        <p>Pure-Rust library and CLI. Classify and extract text-based PDFs on your own machine in milliseconds — no external calls, no OCR bill, MIT licensed.</p>
-        <div class="pbtns">
-          <a class="btn btn-p" href="https://github.com/firecrawl/pdf-inspector">Get started</a>
-          <a class="btn btn-s" href="https://crates.io/crates/pdf-inspector">crates.io</a>
-        </div>
-      </div>
-      <div class="path path-pro">
-        <img class="fc-mark" src="assets/firecrawl-mark.svg" alt="Firecrawl" width="24" height="34">
-        <div class="ptag"><b>Firecrawl Parse</b> · hosted API</div>
-        <h3>Or let Firecrawl handle the hard ones</h3>
-        <p>Scanned documents, OCR, DOCX / XLSX / HTML, and parsing at scale — clean, LLM-ready Markdown from one API call. The downstream route for everything local parsing can't reach.</p>
-        <div class="pbtns">
-          <a class="btn btn-pro" href="https://docs.firecrawl.dev/api-reference/endpoint/parse">Firecrawl Parse ↗</a>
-          <a class="btn btn-ghost" href="https://firecrawl.dev">firecrawl.dev</a>
-        </div>
-      </div>
-    </div>
-  </div>
-</section>
-
-<footer>
-  <div class="wrap foot-in">
-    <div style="display:flex;align-items:center;gap:7px">Built by <a href="https://firecrawl.dev"><img class="fc-wordmark" src="assets/firecrawl-wordmark.svg" alt="Firecrawl"></a> · MIT licensed</div>
-    <div class="foot-links">
-      <a href="https://github.com/firecrawl/pdf-inspector">GitHub</a>
-      <a href="https://crates.io/crates/pdf-inspector">crates.io</a>
-      <a href="https://pypi.org/project/pdf-inspector/">PyPI</a>
-      <a href="https://www.npmjs.com/package/@firecrawl/pdf-inspector">npm</a>
-    </div>
-  </div>
-</footer>
-
-</body>
-</html>
@@ -11,37 +11,6 @@ use std::process;
 use std::time::Instant;

 /// Escape a string for embedding in a JSON string value.
-fn format_detector_ocr_reasons(reasons: &std::collections::BTreeMap<u32, Vec<String>>) -> String {
-    reasons
-        .iter()
-        .map(|(page, page_reasons)| {
-            let reasons_json = page_reasons
-                .iter()
-                .map(|reason| format!(r#""{}""#, json_escape(reason)))
-                .collect::<Vec<_>>()
-                .join(",");
-            format!(r#"{{"page":{},"reasons":[{}]}}"#, page, reasons_json)
-        })
-        .collect::<Vec<_>>()
-        .join(",")
-}
-
-fn format_ocr_reasons_by_page(reasons: &[pdf_inspector::PageOcrReasons]) -> String {
-    reasons
-        .iter()
-        .map(|entry| {
-            let reasons_json = entry
-                .reasons
-                .iter()
-                .map(|reason| format!(r#""{}""#, json_escape(reason)))
-                .collect::<Vec<_>>()
-                .join(",");
-            format!(r#"{{"page":{},"reasons":[{}]}}"#, entry.page, reasons_json)
-        })
-        .collect::<Vec<_>>()
-        .join(",")
-}
-
 fn json_escape(s: &str) -> String {
    let mut out = String::with_capacity(s.len() + 16);
    for ch in s.chars() {
@@ -148,13 +117,11 @@ fn run_analyze(pdf_path: &str, json_output: bool, start: Instant) {
                    .iter()
                    .map(|p| p.to_string())
                    .collect();
-                let ocr_reasons = format_ocr_reasons_by_page(&result.ocr_reasons_by_page);
                println!(
-                    r#"{{"pdf_type":"{}","page_count":{},"pages_needing_ocr":[{}],"ocr_reasons_by_page":[{}],"is_complex":{},"pages_with_tables":[{}],"pages_with_columns":[{}],"detection_time_ms":{}}}"#,
+                    r#"{{"pdf_type":"{}","page_count":{},"pages_needing_ocr":[{}],"is_complex":{},"pages_with_tables":[{}],"pages_with_columns":[{}],"detection_time_ms":{}}}"#,
                    pdf_type_str(&result.pdf_type),
                    result.page_count,
                    ocr_pages.join(","),
-                    ocr_reasons,
                    result.layout.is_complex,
                    table_pages.join(","),
                    col_pages.join(","),
@@ -177,9 +144,6 @@ fn run_analyze(pdf_path: &str, json_output: bool, start: Instant) {
                println!("Page count: {}", result.page_count);
                if !result.pages_needing_ocr.is_empty() {
                    println!("Pages needing OCR: {:?}", result.pages_needing_ocr);
-                    for entry in &result.ocr_reasons_by_page {
-                        println!("  page {}: {}", entry.page, entry.reasons.join(", "));
-                    }
                }
                println!();
                if result.layout.is_complex {
@@ -220,9 +184,8 @@ fn run_detect_only(pdf_path: &str, json_output: bool, start: Instant) {
                    .iter()
                    .map(|p| p.to_string())
                    .collect();
-                let ocr_reasons = format_detector_ocr_reasons(&result.ocr_reasons_by_page);
                println!(
-                    r#"{{"pdf_type":"{}","page_count":{},"pages_sampled":{},"pages_with_text":{},"confidence":{:.2},"title":{},"ocr_recommended":{},"pages_needing_ocr":[{}],"ocr_reasons_by_page":[{}],"detection_time_ms":{}}}"#,
+                    r#"{{"pdf_type":"{}","page_count":{},"pages_sampled":{},"pages_with_text":{},"confidence":{:.2},"title":{},"ocr_recommended":{},"pages_needing_ocr":[{}],"detection_time_ms":{}}}"#,
                    pdf_type_str(&result.pdf_type),
                    result.page_count,
                    result.pages_sampled,
@@ -235,7 +198,6 @@ fn run_detect_only(pdf_path: &str, json_output: bool, start: Instant) {
                        .unwrap_or_else(|| "null".to_string()),
                    result.ocr_recommended,
                    ocr_pages.join(","),
-                    ocr_reasons,
                    elapsed.as_millis()
                );
            } else {
@@ -270,9 +232,6 @@ fn run_detect_only(pdf_path: &str, json_output: bool, start: Instant) {
                            result.pages_needing_ocr, result.page_count
                        );
                    }
-                    for (page, reasons) in &result.ocr_reasons_by_page {
-                        println!("  page {}: {}", page, reasons.join(", "));
-                    }
                }
                if let Some(title) = &result.title {
                    println!("Title: {}", title);
@@ -208,7 +208,6 @@ fn main() {
        eprintln!("  --raw               Output only markdown (no headers)");
        eprintln!("  --pages             Insert page break markers (<!-- Page N -->)");
        eprintln!("  --select-pages N    Only process specified pages (e.g. 1,3,5-10)");
-        eprintln!("  --password PW       Password for an encrypted PDF");
        eprintln!("  --detect-only       Only detect PDF type (no extraction)");
        eprintln!("  --analyze           Detect + extract + layout analysis (no markdown)");
        process::exit(1);
@@ -222,16 +221,6 @@ fn main() {
    let detect_only = args.iter().any(|a| a == "--detect-only");
    let analyze = args.iter().any(|a| a == "--analyze");

-    // Parse --password value
-    let password = args.iter().position(|a| a == "--password").map(|i| {
-        args.get(i + 1)
-            .unwrap_or_else(|| {
-                eprintln!("Error: --password requires a value");
-                process::exit(1);
-            })
-            .clone()
-    });
-
    // Parse --select-pages value
    let page_filter = args
        .iter()
@@ -280,7 +269,6 @@ fn main() {
    if let Some(pages) = page_filter {
        options.page_filter = Some(pages);
    }
-    options.password = password;

    match process_pdf_with_options(pdf_path, options) {
        Ok(result) => {
@@ -60,10 +60,6 @@ pub struct PdfTypeResult {
    /// 1-indexed page numbers that need OCR (image-only or insufficient text).
    /// Empty for TextBased. All pages for Scanned/ImageBased. Specific pages for Mixed.
    pub pages_needing_ocr: Vec<u32>,
-    /// Per-page explanation for `pages_needing_ocr`: 1-indexed page → reason
-    /// codes (`scanned`, `no_text`, `vector_text`, `suspected_garbled_text`).
-    /// Only contains pages that need OCR.
-    pub ocr_reasons_by_page: std::collections::BTreeMap<u32, Vec<String>>,
 }

 /// Configuration for PDF type detection
@@ -386,12 +382,7 @@ pub(crate) fn detect_from_document(
                let analysis = if let Some(cached) = analysis_cache.get(&page_num) {
                    cached.clone()
                } else if let Some(&page_id) = pages.get(&page_num) {
-                    // Cache the fresh analysis so the reason-classification pass
-                    // below sees the real signals (vector_text, etc.) instead of
-                    // defaulting to "scanned".
-                    let a = analyze_page_content(doc, page_id);
-                    analysis_cache.insert(page_num, a.clone());
-                    a
+                    analyze_page_content(doc, page_id)
                } else {
                    continue;
                };
@@ -438,9 +429,6 @@ pub(crate) fn detect_from_document(
                let analysis = analyze_page_content(doc, page_id);
                if analysis.has_identity_h_no_tounicode || analysis.has_only_type3_fonts {
                    pages_needing_ocr.push(page_num);
-                    // Cache so the reason pass reports suspected_garbled_text
-                    // rather than defaulting to "scanned".
-                    analysis_cache.insert(page_num, analysis);
                }
            }
        }
@@ -448,19 +436,6 @@ pub(crate) fn detect_from_document(
    pages_needing_ocr.sort();
    pages_needing_ocr.dedup();

-    // Explain each OCR-flagged page. Pages we analyzed get a signal-derived
-    // reason; pages flagged only by whole-document classification (unsampled
-    // pages of a Scanned/ImageBased doc) default to `scanned`.
-    let mut ocr_reasons_by_page: std::collections::BTreeMap<u32, Vec<String>> =
-        std::collections::BTreeMap::new();
-    for &page_num in &pages_needing_ocr {
-        let reasons = match analysis_cache.get(&page_num) {
-            Some(analysis) => page_ocr_reasons(analysis),
-            None => vec![crate::OCR_REASON_SCANNED],
-        };
-        ocr_reasons_by_page.insert(page_num, reasons.into_iter().map(String::from).collect());
-    }
-
    // Try to get title from metadata
    let title = get_document_title(doc);

@@ -473,7 +448,6 @@ pub(crate) fn detect_from_document(
        title,
        ocr_recommended,
        pages_needing_ocr,
-        ocr_reasons_by_page,
    })
 }

@@ -513,7 +487,7 @@ fn distribute_pages(n: u32, total: u32) -> Vec<u32> {
 }

 /// Page content analysis result
-#[derive(Clone, Default)]
+#[derive(Clone)]
 struct PageAnalysis {
    text_operator_count: u32,
    has_images: bool,
@@ -549,31 +523,6 @@ struct PageAnalysis {
    has_decodable_text_fonts: bool,
 }

-/// Explain *why* a page needs OCR, from its content analysis. Priority:
-/// undecodable fonts (`suspected_garbled_text`) and vector-outlined text
-/// (`vector_text`) come first because they persist even when a text layer is
-/// present; otherwise a page with no extractable text is `scanned` when an
-/// image backs it or `no_text` when nothing does.
-fn page_ocr_reasons(a: &PageAnalysis) -> Vec<&'static str> {
-    let mut reasons = Vec::new();
-    if a.has_identity_h_no_tounicode || a.has_only_type3_fonts {
-        reasons.push(crate::OCR_REASON_SUSPECTED_GARBLED_TEXT);
-    }
-    if a.has_vector_text {
-        reasons.push(crate::OCR_REASON_VECTOR_TEXT);
-    }
-    if reasons.is_empty() {
-        let has_extractable_text = a.text_operator_count > 0 && a.unique_text_chars > 0;
-        if !has_extractable_text && !a.has_images && !a.has_template_image {
-            reasons.push(crate::OCR_REASON_NO_TEXT);
-        } else {
-            // Image-backed with no usable text, or too little text to trust.
-            reasons.push(crate::OCR_REASON_SCANNED);
-        }
-    }
-    reasons
-}
-
 /// Extracted font information from a Resource dictionary entry.
 /// Stores the properties needed for decodability/identity-h checks
 /// without holding a reference to the document.
@@ -1860,64 +1809,6 @@ fn get_document_title(doc: &Document) -> Option<String> {
 mod tests {
    use super::*;

-    #[test]
-    fn page_ocr_reasons_classify() {
-        // Scanned: no text, full-page image.
-        let scanned = PageAnalysis {
-            has_template_image: true,
-            ..Default::default()
-        };
-        assert_eq!(page_ocr_reasons(&scanned), vec![crate::OCR_REASON_SCANNED]);
-
-        // Image-only page (no template flag, but has an image).
-        let image_only = PageAnalysis {
-            has_images: true,
-            ..Default::default()
-        };
-        assert_eq!(
-            page_ocr_reasons(&image_only),
-            vec![crate::OCR_REASON_SCANNED]
-        );
-
-        // No text, no image → no_text.
-        let blank = PageAnalysis::default();
-        assert_eq!(page_ocr_reasons(&blank), vec![crate::OCR_REASON_NO_TEXT]);
-
-        // Vector-outlined text.
-        let vector = PageAnalysis {
-            has_vector_text: true,
-            ..Default::default()
-        };
-        assert_eq!(
-            page_ocr_reasons(&vector),
-            vec![crate::OCR_REASON_VECTOR_TEXT]
-        );
-
-        // Undecodable fonts → garbled, and it wins over the fall-through.
-        let garbled = PageAnalysis {
-            has_identity_h_no_tounicode: true,
-            has_images: true,
-            ..Default::default()
-        };
-        assert_eq!(
-            page_ocr_reasons(&garbled),
-            vec![crate::OCR_REASON_SUSPECTED_GARBLED_TEXT]
-        );
-
-        // A page with real extractable text and an image is not flagged here
-        // as scanned/no_text (only reached for pages already needing OCR).
-        let text_with_image = PageAnalysis {
-            text_operator_count: 40,
-            unique_text_chars: 120,
-            has_images: true,
-            ..Default::default()
-        };
-        assert_eq!(
-            page_ocr_reasons(&text_with_image),
-            vec![crate::OCR_REASON_SCANNED]
-        );
-    }
-
    #[test]
    fn test_scan_content_operators() {
        let mut uchars = HashSet::new();
@@ -9,7 +9,7 @@ mod links;
 pub(crate) mod underline;
 mod xobjects;

-use crate::text_utils::{is_cjk_char, is_rtl_text};
+use crate::text_utils::is_rtl_text;
 use crate::tounicode::FontCMaps;
 use crate::types::{PageExtraction, PdfLine, PdfRect, TextItem};
 use crate::PdfError;
@@ -252,47 +252,17 @@ fn suppress_table_underlines(
    }

    let mut table_item_indices: HashSet<usize> = HashSet::new();
-    // A detected "table" that swallows nearly every text item on the page
-    // is a detection artifact (prose pages with boxed callouts or stacked
-    // underline rules read as one giant grid), not a real table — letting
-    // it through here erased every legitimate underline on the page
-    // (text_dense__underline: rect detection claimed 52/52 items). Real
-    // ruled tables share the page with headings, captions, and body text.
-    let plausible = |table: &crate::tables::Table| {
-        // Content sanity gate: prose pages with boxed callouts and stacked
-        // underline rules can detect as a structurally rich "table" that
-        // swallows every item on the page (text_dense__underline: a 4x8
-        // grid claiming 52/52 items, one "cell" holding 806 chars of body
-        // text) — suppressing there erased every legitimate underline on
-        // the page. Real data-table cells are short values; a cell with
-        // hundreds of characters means the grid captured flowing prose.
-        let lens: Vec<usize> = table
-            .cells
-            .iter()
-            .flatten()
-            .filter(|cell| !cell.trim().is_empty())
-            .map(|cell| cell.chars().count())
-            .collect();
-        if lens.is_empty() {
-            return false;
-        }
-        let long = lens.iter().filter(|&&n| n > 100).count();
-        (long as f32) < (lens.len() as f32) * 0.3
-    };

    if !rects.is_empty() {
        let (rect_tables, _) = crate::tables::detect_tables_from_rects(items, rects, page);
-        for table in rect_tables.iter().filter(|t| plausible(t)) {
-            table_item_indices.extend(table.item_indices.iter().copied());
+        for table in rect_tables {
+            table_item_indices.extend(table.item_indices);
        }
    }

    if !lines.is_empty() {
-        for table in crate::tables::detect_tables_from_lines(items, lines, page)
-            .iter()
-            .filter(|t| plausible(t))
-        {
-            table_item_indices.extend(table.item_indices.iter().copied());
+        for table in crate::tables::detect_tables_from_lines(items, lines, page) {
+            table_item_indices.extend(table.item_indices);
        }
    }

@@ -557,136 +527,6 @@ fn should_preserve_overlapping_stream_order(group: &[&TextItem]) -> bool {
    saw_backtrack
 }

-/// Detect a tracked (letter-spaced) run of single-glyph items and derive its
-/// run-local space floor.
-///
-/// Display type set with tracking renders one glyph per show op; the merge
-/// loop's fixed thresholds (0.08-0.13 em) then read every letter gap as a
-/// word boundary and emit "H O W" instead of "HOW". Within such a run the
-/// gaps carry the real signal: letter gaps cluster tightly just above the
-/// fixed threshold, word gaps sit clearly higher. Returns (run_end_index,
-/// space_floor) when the run starting at `start` is tracked — spaces are
-/// then inserted only at gaps above the floor (infinity = single word).
-/// Normal text (multi-char items, or single-char runs with sub-threshold
-/// gaps) returns None and keeps the existing behavior.
-/// Han/Kana scripts write without inter-word spaces. Hangul (Korean) DOES
-/// space between words and deliberately stays out of this set — a Korean
-/// tracked run keeps normal word-boundary handling.
-fn is_spaceless_cjk(c: char) -> bool {
-    matches!(c,
-        '\u{3000}'..='\u{303F}'   // CJK Symbols and Punctuation
-        | '\u{3040}'..='\u{309F}' // Hiragana
-        | '\u{30A0}'..='\u{30FF}' // Katakana
-        | '\u{4E00}'..='\u{9FFF}' // CJK Unified Ideographs
-        | '\u{F900}'..='\u{FAFF}' // CJK Compatibility Ideographs
-        | '\u{FF00}'..='\u{FFEF}' // Halfwidth and Fullwidth Forms
-    )
-}
-
-fn tracked_run_space_floor(group: &[&TextItem], start: usize) -> Option<(usize, f32)> {
-    const MIN_GAPS: usize = 4;
-    let first = group[start];
-    if first.text.trim().chars().count() != 1 {
-        return None;
-    }
-    let fs = first.font_size;
-    if fs <= 0.0 {
-        return None;
-    }
-
-    // Walk the run under the SAME break conditions as the merge loop
-    // (size band, style equality, mergeable gap) so indices stay aligned.
-    let mut gaps: Vec<f32> = Vec::new();
-    let mut end_x = first.x + effective_merge_width(first);
-    let mut end = start;
-    for (offset, next) in group[start + 1..].iter().enumerate() {
-        if next.text.trim().chars().count() != 1 {
-            break;
-        }
-        if (next.font_size - fs).abs() > fs * 0.20 {
-            break;
-        }
-        if next.is_bold != first.is_bold
-            || next.is_italic != first.is_italic
-            || next.is_underline != first.is_underline
-            || next.is_strikeout != first.is_strikeout
-        {
-            break;
-        }
-        let gap = next.x - end_x;
-        if gap > fs * 0.5 || gap < -fs * 0.5 {
-            break;
-        }
-        gaps.push(gap / fs);
-        end_x = next.x + effective_merge_width(next);
-        end = start + 1 + offset;
-    }
-    if gaps.len() < 2 {
-        return None;
-    }
-
-    // Tracked signature: the run's TYPICAL gap clears the fixed space
-    // threshold (0.08) — the merge loop would break almost every letter
-    // pair into "words". Short runs (2-3 gaps: "H O W") demand a stricter
-    // shape — clearly wide, uniform, ALL-CAPS — because a genuine spaced
-    // sequence of single letters ("x y z" variables) has the same gap
-    // count; display tracking is a caps convention.
-    let mut sorted = gaps.clone();
-    sorted.sort_by(|a, b| a.total_cmp(b));
-    let median = sorted[sorted.len() / 2];
-    // Typographic convention gate, both tiers: display tracking is an
-    // all-caps convention, and Han/Kana never space between glyphs. Mixed-
-    // or lowercase Latin runs keep their boundaries because geometry alone
-    // cannot distinguish spaced singles ("A b c d e") from a tracked
-    // title-case word ("B u f f a l o").
-    let run_chars = || {
-        group[start..=end]
-            .iter()
-            .flat_map(|it| it.text.trim().chars())
-    };
-    let spaceless_cjk = run_chars().all(|c| is_spaceless_cjk(c) || !c.is_alphanumeric())
-        && run_chars().any(is_spaceless_cjk);
-    let all_caps = run_chars().all(|c| c.is_uppercase() || is_cjk_char(c) || !c.is_alphabetic());
-    if !(spaceless_cjk || all_caps) {
-        return None;
-    }
-
-    if gaps.len() >= MIN_GAPS {
-        if median <= 0.075 {
-            return None;
-        }
-    } else {
-        let uniform = sorted[sorted.len() - 1] <= sorted[0].max(0.01) * 1.4;
-        if median < 0.09 || !uniform {
-            return None;
-        }
-    }
-
-    // Han/Kana: no inter-glyph spaces, period — a nonuniform gap
-    // distribution (punctuation spacing, justification) must not
-    // manufacture word boundaries.
-    if spaceless_cjk {
-        return Some((end, f32::INFINITY));
-    }
-
-    // Word gaps, if present, form a second mode above the letter-gap
-    // cluster: split at the largest relative jump. Unimodal → one word.
-    let mut best_jump = 1.0f32;
-    let mut floor = f32::INFINITY;
-    for pair in sorted.windows(2) {
-        let (lo, hi) = (pair[0].max(0.01), pair[1].max(0.01));
-        let jump = hi / lo;
-        if jump > best_jump {
-            best_jump = jump;
-            floor = (lo + hi) / 2.0;
-        }
-    }
-    if best_jump < 1.4 {
-        floor = f32::INFINITY;
-    }
-    Some((end, floor * fs))
-}
-
 pub(crate) fn merge_text_items(items: Vec<TextItem>) -> Vec<TextItem> {
    if items.is_empty() {
        return items;
@@ -734,14 +574,6 @@ pub(crate) fn merge_text_items(items: Vec<TextItem>) -> Vec<TextItem> {
            let mut text = first.text.clone();
            let mut end_x = first.x + effective_merge_width(first);

-            // Tracked display text: run-local space floor overrides the
-            // fixed thresholds for this run's junctions (see helper).
-            let tracked = if *preserve_stream_order {
-                None
-            } else {
-                tracked_run_space_floor(group, i)
-            };
-
            let mut j = i + 1;
            while j < group.len() {
                let next = group[j];
@@ -796,11 +628,7 @@ pub(crate) fn merge_text_items(items: Vec<TextItem>) -> Vec<TextItem> {
                let needs_bullet_space = *preserve_stream_order
                    && is_standalone_bullet_text(&text)
                    && !next.text.trim().is_empty();
-                let effective_threshold = match tracked {
-                    Some((run_end, floor)) if j <= run_end => floor,
-                    _ => threshold,
-                };
-                if needs_bullet_space || gap > effective_threshold {
+                if needs_bullet_space || gap > threshold {
                    text.push(' ');
                }
                text.push_str(&next.text);
@@ -903,17 +731,9 @@ pub(crate) fn merge_subscript_items(items: Vec<TextItem>) -> Vec<TextItem> {
                        .chars()
                        .last()
                        .is_some_and(|c| c.is_alphabetic());
-                    // Strikeout boundaries block the merge (a struck word
-                    // must not extend its strike over a live footnote digit,
-                    // and a struck digit must not lose its own mark). An
-                    // underlined parent with an unmarked digit DOES merge:
-                    // the drawn rule easily misses the tiny digit's overlap
-                    // window, and refusing costs the whole subscript token
-                    // ("b"+"2" staying split). Visually the rule spans both.
-                    let marks_ok = parent.is_strikeout == item.is_strikeout
-                        && (parent.is_underline == item.is_underline
-                            || (parent.is_underline && !item.is_underline));
-                    if parent.font_size >= sub_threshold && ends_with_letter && marks_ok {
+                    let same_marks = parent.is_underline == item.is_underline
+                        && parent.is_strikeout == item.is_strikeout;
+                    if parent.font_size >= sub_threshold && ends_with_letter && same_marks {
                        let parent_right = parent.x + parent.width;
                        let gap = item.x - parent_right;
                        // Subscripts must be tightly adjacent (within ~1pt)
@@ -974,107 +794,6 @@ mod tests {
    use crate::types::{ItemType, PdfLine, TextLine};
    use layout::{detect_columns, is_newspaper_layout, ColumnRegion};

-    /// Glyph-per-item run at `fs`=12 with the given inter-glyph gap (pt).
-    fn glyph_run(chars: &str, start_x: f32, glyph_w: f32, gap: f32) -> Vec<TextItem> {
-        let mut x = start_x;
-        let mut out = Vec::new();
-        for c in chars.chars() {
-            out.push(make_merge_item(&c.to_string(), x, glyph_w));
-            x += glyph_w + gap;
-        }
-        out
-    }
-
-    #[test]
-    fn tracked_caps_run_collapses_to_word() {
-        // Display tracking: every letter gap (0.19 em) clears the fixed
-        // space threshold — without the run-local floor this reads "H O W".
-        let items = glyph_run("HOW", 100.0, 10.0, 2.3);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "HOW");
-    }
-
-    #[test]
-    fn tracked_run_keeps_word_gaps_bimodal() {
-        // Letters at 0.19 em, word gaps at 0.42 em (below the 0.5 em item
-        // break): the split must land between the modes. Needs >=4 gaps to
-        // enter the bimodal tier — short runs use the strict uniform gate.
-        let mut items = glyph_run("ITISOK", 100.0, 8.0, 2.3);
-        for i in 2..6 {
-            items[i].x += 2.8; // word gap at T|I
-        }
-        for i in 4..6 {
-            items[i].x += 2.8; // word gap at S|O
-        }
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "IT IS OK");
-    }
-
-    #[test]
-    fn lowercase_spaced_singles_stay_words() {
-        // "x y z" variables: same gap shape but lowercase — the short-run
-        // caps requirement keeps genuine spaced singles apart.
-        let items = glyph_run("xyz", 100.0, 6.0, 2.3);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "x y z");
-    }
-
-    #[test]
-    fn kerned_singles_unaffected() {
-        // Tiny kerning gaps never triggered spaces before and still don't.
-        let items = glyph_run("WORD", 100.0, 8.0, 0.3);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "WORD");
-    }
-
-    #[test]
-    fn long_lowercase_spaced_singles_keep_boundaries() {
-        // Review: a 5+ single-letter lowercase list has the tracked gap
-        // shape at any length — the convention gate must protect it in
-        // the >=4-gap tier too.
-        let items = glyph_run("abcde", 100.0, 6.0, 2.3);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "a b c d e");
-    }
-
-    #[test]
-    fn han_run_with_nonuniform_gaps_never_gains_spaces() {
-        // Review: a bimodal gap distribution (justification, punctuation
-        // spacing) must not manufacture word boundaries in Han text.
-        let mut items = glyph_run("北京时事快报", 100.0, 12.0, 1.4);
-        for item in items.iter_mut().skip(3) {
-            item.x += 3.0; // wide gap after the third glyph
-        }
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "北京时事快报");
-    }
-
-    #[test]
-    fn uppercase_leading_spaced_singles_keep_boundaries() {
-        // "A b c d e" is indistinguishable from a title-case tracked word
-        // without reliable tracking metadata, so preserve its boundaries.
-        let items = glyph_run("Abcde", 100.0, 7.0, 2.3);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "A b c d e");
-    }
-
-    #[test]
-    fn cjk_glyph_run_collapses_without_spaces() {
-        // CJK sets one glyph per item with loose gaps; CJK uses no spaces,
-        // and the non-alphabetic run passes the caps gate.
-        let items = glyph_run("北京时事", 100.0, 12.0, 1.4);
-        let merged = merge_text_items(items);
-        assert_eq!(merged.len(), 1);
-        assert_eq!(merged[0].text, "北京时事");
-    }
-
    fn make_merge_item(text: &str, x: f32, width: f32) -> TextItem {
        TextItem {
            text: text.into(),
@@ -115,13 +115,7 @@ fn rules_from_graphics(rects: &[PdfRect], lines: &[UnderlineLine], page: u32) ->
    rules
 }

-fn discard_repeated_ruling_rules(
-    rules: Vec<Rule>,
-    items: &[TextItem],
-    rects: &[PdfRect],
-    lines: &[UnderlineLine],
-    page: u32,
-) -> Vec<Rule> {
+fn discard_repeated_ruling_rules(rules: Vec<Rule>) -> Vec<Rule> {
    if rules.len() < MIN_REPEATED_RULE_LEVELS {
        return rules;
    }
@@ -129,142 +123,12 @@ fn discard_repeated_ruling_rules(
    rules
        .iter()
        .filter(|rule| {
-            // A rule snugly owned by one text line is an underline even when
-            // span-similar rules repeat down the page — documents that
-            // underline many full-width lines (dense CJK business docs) look
-            // exactly like table rulings to the repetition check, which used
-            // to discard every one of them. Table rulings fail snugness:
-            // row separators extend past their cells' text (or have no text
-            // on the baseline above), and multi-column matches are still
-            // culled by the tabular filter afterwards.
-            // Same-row segmented rules (column-header separators) are
-            // always rulings — each segment snugly owns its column label,
-            // so snugness must not override that check.
-            !is_segmented_row_ruling_rule(rule, &rules)
-                && ((has_snug_text_owner(rule, items)
-                    && !has_flanking_verticals(rule, rects, lines, page))
-                    || !is_repeated_ruling_rule(rule, &rules))
+            !is_repeated_ruling_rule(rule, &rules) && !is_segmented_row_ruling_rule(rule, &rules)
        })
        .cloned()
        .collect()
 }

-/// True when a single text item both matches the rule vertically (baseline
-/// window) and horizontally contains it: the rule may not extend past the
-/// item's span by more than ~0.75em on either side. Underlines are drawn to
-/// the width of the text they decorate; table/form rulings span cells or
-/// full table width and overshoot any single item.
-/// A rule flanked by vertical strokes at its ends is a table/box border
-/// row edge, not an underline — underlined text lines have no vertical
-/// rules rising from their ends. Checked against raw stroked lines: a
-/// near-vertical segment whose x sits at either end of the rule and whose
-/// y-range covers the rule's row.
-fn has_flanking_verticals(
-    rule: &Rule,
-    rects: &[PdfRect],
-    lines: &[UnderlineLine],
-    page: u32,
-) -> bool {
-    // A drawn rect that CONTAINS the rule vetoes rescue only with GRID
-    // EVIDENCE: another drawn rect abutting it vertically (cell rows tile).
-    // Height alone can't separate a table cell from a decorative callout
-    // panel — genuine underlines live inside isolated filled panels, and
-    // multiline table cells can be arbitrarily tall.
-    let norm = |r: &PdfRect| {
-        let (x_lo, x_hi) = if r.width >= 0.0 {
-            (r.x, r.x + r.width)
-        } else {
-            (r.x + r.width, r.x)
-        };
-        let (y_lo, y_hi) = if r.height >= 0.0 {
-            (r.y, r.y + r.height)
-        } else {
-            (r.y + r.height, r.y)
-        };
-        (x_lo, x_hi, y_lo, y_hi)
-    };
-    let page_rects: Vec<(f32, f32, f32, f32)> = rects
-        .iter()
-        .filter(|r| r.page == page && r.height.abs() > 6.0)
-        .map(norm)
-        .collect();
-    let rect_flank = page_rects.iter().any(|&(x_lo, x_hi, y_lo, y_hi)| {
-        let contains = x_lo <= rule.x1 + 2.0
-            && x_hi >= rule.x2 - 2.0
-            && y_lo <= rule.y + 2.0
-            && y_hi >= rule.y - 2.0;
-        if !contains {
-            return false;
-        }
-        // Grid evidence: a vertically abutting neighbor box with x-overlap.
-        page_rects.iter().any(|&(nx_lo, nx_hi, ny_lo, ny_hi)| {
-            let x_overlap = nx_hi.min(x_hi) - nx_lo.max(x_lo);
-            if x_overlap <= 10.0 {
-                return false;
-            }
-            (ny_lo - y_hi).abs() <= 3.0 || (y_lo - ny_hi).abs() <= 3.0
-        })
-    });
-    if rect_flank {
-        return true;
-    }
-    lines.iter().any(|l| {
-        if l.page != page || (l.x1 - l.x2).abs() > 2.0 {
-            return false;
-        }
-        let x = (l.x1 + l.x2) / 2.0;
-        let near_end = (x - rule.x1).abs() <= 6.0 || (x - rule.x2).abs() <= 6.0;
-        if !near_end {
-            return false;
-        }
-        let (y_lo, y_hi) = if l.y1 <= l.y2 {
-            (l.y1, l.y2)
-        } else {
-            (l.y2, l.y1)
-        };
-        y_lo <= rule.y + 2.0 && y_hi >= rule.y - 2.0
-    })
-}
-
-fn has_snug_text_owner(rule: &Rule, items: &[TextItem]) -> bool {
-    // Underlines are drawn to the width of the text they decorate, but the
-    // text may be split into several runs (CJK lines mix scripts and font
-    // switches) — so ownership is judged against the UNION of the runs on
-    // the rule's baseline row. Table/form rulings overshoot their row's
-    // text (row separators span cell padding and empty columns), so they
-    // fail either containment or coverage.
-    let matched: Vec<&TextItem> = items
-        .iter()
-        .filter(|item| is_underline_candidate(item) && rule_matches_item(rule, item))
-        .collect();
-    if matched.is_empty() {
-        return false;
-    }
-    let x1 = matched.iter().map(|i| i.x).fold(f32::INFINITY, f32::min);
-    let x2 = matched
-        .iter()
-        .map(|i| i.x + i.width)
-        .fold(f32::NEG_INFINITY, f32::max);
-    let max_fs = matched.iter().map(|i| i.font_size).fold(0.0, f32::max);
-    let pad = (max_fs * 0.75).max(4.0);
-    if rule.x1 < x1 - pad || rule.x2 > x2 + pad {
-        return false;
-    }
-    let covered: f32 = matched.iter().map(|i| i.width).sum();
-    if covered < rule.width() * 0.6 {
-        return false;
-    }
-    // A table row also unions to the rule's span — but its cells sit apart.
-    // An underlined text line is contiguous runs with word-sized gaps; any
-    // column-sized hole between matched runs means this is a row ruling.
-    let mut sorted = matched;
-    sorted.sort_by(|a, b| a.x.total_cmp(&b.x));
-    sorted.windows(2).all(|pair| {
-        let gap = pair[1].x - (pair[0].x + pair[0].width);
-        gap <= (max_fs * 2.0).max(12.0)
-    })
-}
-
 fn is_repeated_ruling_rule(rule: &Rule, rules: &[Rule]) -> bool {
    let mut y_levels: Vec<f32> = rules
        .iter()
@@ -351,11 +215,9 @@ fn is_underline_candidate(item: &TextItem) -> bool {

 fn rule_matches_item(rule: &Rule, item: &TextItem) -> bool {
    // Vertical window: underlines sit at or slightly below the baseline.
-    // Latin fonts draw them at roughly 5-15% of the em below; CJK layouts
-    // put them under the full em box, measured up to ~0.67em below the
-    // baseline (text_dense__underline). Allow 0.72em (min 3pt) below and
-    // 1pt above for rounding.
-    let below = (item.font_size * 0.72).max(3.0);
+    // Fonts draw them at roughly 5-15% of the em below; allow up to 35%
+    // (min 3pt) below and 1pt above for rounding.
+    let below = (item.font_size * 0.35).max(3.0);
    let y_min = item.y - below;
    let y_max = item.y + 1.0;
    if rule.y < y_min || rule.y > y_max {
@@ -398,48 +260,12 @@ pub(crate) fn mark_underlined_items(
    lines: &[UnderlineLine],
    page: u32,
 ) {
-    let rules = discard_repeated_ruling_rules(
-        rules_from_graphics(rects, lines, page),
-        items,
-        rects,
-        lines,
-        page,
-    );
+    let rules = discard_repeated_ruling_rules(rules_from_graphics(rects, lines, page));
    if rules.is_empty() {
        return;
    }
    let tabular_rules = tabular_row_separator_rule_indices(&rules, items);

-    // Math fraction bars are short horizontal lines with the numerator just
-    // above AND the denominator just below — underline geometry from above,
-    // but no underline has text hanging directly beneath it at fraction
-    // distance. Only narrow rules qualify: real underlines under short
-    // labels have their next text line a full line-pitch away.
-    let fraction_rules: HashSet<usize> = rules
-        .iter()
-        .enumerate()
-        .filter(|(_, rule)| {
-            rule.width() <= 60.0
-                && items.iter().any(|item| {
-                    if !is_underline_candidate(item) {
-                        return false;
-                    }
-                    // A denominator HUGS the bar (fraction typesetting
-                    // leaves ~0.1-0.2em) and is bar-sized. Both bounds
-                    // matter: a short last-line of a paragraph at normal
-                    // leading sits further below, and a full next text
-                    // line is far wider than the rule.
-                    let dy = rule.y - (item.y + item.height);
-                    let overlap = rule.x2.min(item.x + item.width) - rule.x1.max(item.x);
-                    dy > 0.0
-                        && dy <= item.font_size * 0.3
-                        && overlap > rule.width() * 0.5
-                        && item.width <= rule.width() * 1.5
-                })
-        })
-        .map(|(i, _)| i)
-        .collect();
-
    for item in items.iter_mut() {
        if !is_underline_candidate(item) {
            continue;
@@ -449,10 +275,7 @@ pub(crate) fn mark_underlined_items(
            if tabular_rules.contains(&rule_idx) {
                continue;
            }
-            // The fraction guard only gates UNDERLINE marking — a rule that
-            // reads as a fraction bar from below can still legitimately
-            // strike through a line above it.
-            if !fraction_rules.contains(&rule_idx) && rule_matches_item(rule, item) {
+            if rule_matches_item(rule, item) {
                item.is_underline = true;
            }
            if rule_strikes_item(rule, item) {
@@ -500,16 +323,6 @@ mod tests {
        }
    }

-    fn cell_rect(x: f32, y: f32, width: f32, height: f32) -> PdfRect {
-        PdfRect {
-            x,
-            y,
-            width,
-            height,
-            page: 1,
-        }
-    }
-
    fn thin_rect(x: f32, y: f32, width: f32) -> PdfRect {
        PdfRect {
            x,
@@ -730,100 +543,6 @@ mod tests {
        assert!(items.iter().all(|item| !item.is_underline));
    }

-    #[test]
-    fn repeated_snug_underlines_survive_ruling_filter() {
-        // Dense docs underline many full-width lines: span-similar rules at
-        // 3+ y-levels used to be discarded wholesale as table rulings.
-        // Each rule here snugly matches one text line, so all must mark.
-        let mut items = vec![
-            item("first underlined line of text", 50.0, 700.0, 300.0, 11.0),
-            item("second underlined line here", 50.0, 650.0, 300.0, 11.0),
-            item("third underlined line as well", 50.0, 600.0, 300.0, 11.0),
-        ];
-        let lines = vec![
-            hline(50.0, 350.0, 697.0),
-            hline(50.0, 350.0, 647.0),
-            hline(50.0, 350.0, 597.0),
-        ];
-
-        mark_underlined_items(&mut items, &[], &lines, 1);
-
-        assert!(items.iter().all(|item| item.is_underline));
-    }
-
-    #[test]
-    fn snug_rescue_spans_split_runs_on_one_line() {
-        // A single underlined line is often split into several runs (script
-        // or font switches). The union of touching runs owns the rule.
-        let mut items = vec![
-            item("run one", 50.0, 700.0, 100.0, 11.0),
-            item("run two", 150.5, 700.0, 100.0, 11.0),
-            item("run three", 251.0, 700.0, 99.0, 11.0),
-            item("other a", 50.0, 650.0, 300.0, 11.0),
-            item("other b", 50.0, 600.0, 300.0, 11.0),
-        ];
-        let lines = vec![
-            hline(50.0, 350.0, 697.0),
-            hline(50.0, 350.0, 647.0),
-            hline(50.0, 350.0, 597.0),
-        ];
-
-        mark_underlined_items(&mut items, &[], &lines, 1);
-
-        assert!(items[0].is_underline && items[1].is_underline && items[2].is_underline);
-    }
-
-    #[test]
-    fn snug_rescue_denied_for_row_with_cell_gaps() {
-        // A full-width rule whose baseline row is several items separated by
-        // column-sized gaps is a table row separator, not an underline —
-        // even when span-similar rules repeat down the page.
-        let mut items = vec![
-            item("cell a", 50.0, 700.0, 60.0, 11.0),
-            item("cell b", 190.0, 700.0, 60.0, 11.0),
-            item("cell c", 330.0, 700.0, 70.0, 11.0),
-            item("cell d", 50.0, 650.0, 60.0, 11.0),
-            item("cell e", 190.0, 650.0, 60.0, 11.0),
-            item("cell f", 330.0, 650.0, 70.0, 11.0),
-        ];
-        let lines = vec![
-            hline(50.0, 400.0, 697.0),
-            hline(50.0, 400.0, 647.0),
-            hline(50.0, 400.0, 597.0),
-        ];
-
-        mark_underlined_items(&mut items, &[], &lines, 1);
-
-        assert!(items.iter().all(|item| !item.is_underline));
-    }
-
-    #[test]
-    fn snug_rescue_denied_inside_cell_box() {
-        // A rule snugly under one text line but enclosed by a drawn cell
-        // box that TILES with vertical neighbors (grid evidence) is a row
-        // ruling of a rect-grid table. Isolated boxes (callout panels) do
-        // not veto — see repeated_snug_underlines_survive_ruling_filter.
-        let mut items = vec![
-            item("one wide cell row", 50.0, 700.0, 300.0, 11.0),
-            item("second wide cell", 50.0, 650.0, 300.0, 11.0),
-            item("third wide cell", 50.0, 600.0, 300.0, 11.0),
-        ];
-        let lines = vec![
-            hline(50.0, 350.0, 697.0),
-            hline(50.0, 350.0, 647.0),
-            hline(50.0, 350.0, 597.0),
-        ];
-        let boxes = vec![
-            cell_rect(45.0, 690.0, 320.0, 50.0),
-            cell_rect(45.0, 640.0, 320.0, 50.0),
-            cell_rect(45.0, 590.0, 320.0, 50.0),
-        ];
-
-        mark_underlined_items(&mut items, &boxes, &lines, 1);
-
-        assert!(items.iter().all(|item| !item.is_underline));
-    }
-
    #[test]
    fn same_row_spaced_rule_segments_do_not_mark_column_labels() {
        let mut items = vec![
@@ -39,7 +39,6 @@ pub mod markdown;
 pub mod process_mode;
 pub mod structure_tree;
 pub mod tables;
-mod text_quality;
 pub mod text_utils;
 pub mod tounicode;
 pub mod types;
@@ -61,28 +60,12 @@ pub use types::{LayoutComplexity, PdfLine, PdfRect, TextItem};
 use lopdf::Document;
 use std::collections::{BTreeMap, HashMap, HashSet};
 use std::path::Path;
-use text_quality::{
-    analyze_text_quality, detect_encoding_issues, is_cid_garbage, is_garbage_text,
-    region_items_have_decoding_issue,
-};
 use tounicode::FontCMaps;

 /// OCR reason emitted when the extracted text layer appears garbled due to
 /// broken font decoding or mojibake.
 pub const OCR_REASON_SUSPECTED_GARBLED_TEXT: &str = "suspected_garbled_text";

-/// OCR reason: the page is a scanned image (a full-page raster / image-only
-/// page) with no usable text layer.
-pub const OCR_REASON_SCANNED: &str = "scanned";
-
-/// OCR reason: the page has no extractable text and no image to OCR — blank,
-/// or content the parser cannot reach.
-pub const OCR_REASON_NO_TEXT: &str = "no_text";
-
-/// OCR reason: the page's text is drawn as vector outlines (path operators)
-/// rather than real text operators, so it cannot be extracted as characters.
-pub const OCR_REASON_VECTOR_TEXT: &str = "vector_text";
-
 // =========================================================================
 // Result type
 // =========================================================================
@@ -137,7 +120,7 @@ pub struct PdfProcessResult {
 ///     .mode(ProcessMode::Analyze)
 ///     .pages([1, 3, 5]);
 /// ```
-#[derive(Clone)]
+#[derive(Debug, Clone)]
 pub struct PdfOptions {
    /// How far the pipeline should run (default: [`ProcessMode::Full`]).
    pub mode: ProcessMode,
@@ -147,23 +130,6 @@ pub struct PdfOptions {
    pub markdown: MarkdownOptions,
    /// Optional set of 1-indexed pages to process.  `None` = all pages.
    pub page_filter: Option<HashSet<u32>>,
-    /// Password for decrypting an encrypted PDF. `None` falls back to the
-    /// empty password (owner-only encryption).
-    pub password: Option<String>,
-}
-
-// Manual `Debug` so the password is never leaked through debug logging or a
-// panic that formats the options; it renders as `Some("[REDACTED]")`.
-impl std::fmt::Debug for PdfOptions {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.debug_struct("PdfOptions")
-            .field("mode", &self.mode)
-            .field("detection", &self.detection)
-            .field("markdown", &self.markdown)
-            .field("page_filter", &self.page_filter)
-            .field("password", &self.password.as_ref().map(|_| "[REDACTED]"))
-            .finish()
-    }
 }

 impl Default for PdfOptions {
@@ -173,7 +139,6 @@ impl Default for PdfOptions {
            detection: DetectionConfig::default(),
            markdown: MarkdownOptions::default(),
            page_filter: None,
-            password: None,
        }
    }
 }
@@ -215,12 +180,6 @@ impl PdfOptions {
        self.page_filter = Some(pages.into_iter().collect());
        self
    }
-
-    /// Set the password used to decrypt an encrypted PDF.
-    pub fn password(mut self, password: impl Into<String>) -> Self {
-        self.password = Some(password.into());
-        self
-    }
 }

 // =========================================================================
@@ -253,8 +212,7 @@ pub fn process_pdf_with_options<P: AsRef<Path>>(
    validate_pdf_file(&path)?;

    // Load the document once — shared by detection AND extraction.
-    let (doc, page_count) =
-        load_document_from_path_with_password(&path, options.password.as_deref())?;
+    let (doc, page_count) = load_document_from_path(&path)?;

    process_document(doc, page_count, options, start)
 }
@@ -279,8 +237,7 @@ pub fn process_pdf_mem_with_options(
    let start = std::time::Instant::now();
    validate_pdf_bytes(buffer)?;

-    let (doc, page_count) =
-        load_document_from_mem_with_password(buffer, options.password.as_deref())?;
+    let (doc, page_count) = load_document_from_mem(buffer)?;

    process_document(doc, page_count, options, start)
 }
@@ -673,51 +630,18 @@ pub fn extract_text_in_regions_mem(

        let mut page_results = Vec::with_capacity(regions.len());

-        // Exclusive item->region assignment: overlapping layout regions used
-        // to extract shared items into EVERY region they touched (the
-        // 1.5pt inclusion margin makes borders generous), duplicating whole
-        // lines in the final markdown on 21% of bench docs — and downstream
-        // duplicate-handling sometimes dropped the variant holding a
-        // sentence tail, turning duplication into content LOSS. Each item
-        // now belongs to the single region with the largest overlap area;
-        // items are partitioned, never suppressed, so no content can vanish.
-        let all_bounds: Vec<RegionBounds> = regions
-            .iter()
-            .map(|rect| {
-                let [rx1, ry1, rx2, ry2] = *rect;
-                region_bounds(rx1, ry1, rx2, ry2, page_h, coords)
-            })
-            .collect();
-        // Single pass over items: assign each to the best-overlap region and
-        // bucket the clone directly (review: avoid a second O(items x
-        // regions) traversal). `had_candidates` marks regions that touched
-        // at least one item even if every one was assigned elsewhere.
-        let mut region_items: Vec<Vec<TextItem>> = vec![Vec::new(); regions.len()];
-        let mut had_candidates: Vec<bool> = vec![false; regions.len()];
-        if let Some(items) = items {
-            for item in items {
-                let mut best: Option<usize> = None;
-                let mut best_area = 0.0_f32;
-                for (ri, b) in all_bounds.iter().enumerate() {
-                    if !region_overlaps_item(item, *b) {
-                        continue;
-                    }
-                    had_candidates[ri] = true;
-                    let area = region_item_overlap_area(item, *b);
-                    if area > best_area {
-                        best_area = area;
-                        best = Some(ri);
-                    }
-                }
-                if let Some(ri) = best {
-                    region_items[ri].push(item.clone());
-                }
-            }
-        }
+        for rect in regions {
+            let [rx1, ry1, rx2, ry2] = *rect;

-        for (region_idx, _rect) in regions.iter().enumerate() {
-            let matched: Vec<TextItem> = std::mem::take(&mut region_items[region_idx]);
-            let assigned_count = matched.len();
+            let bounds = region_bounds(rx1, ry1, rx2, ry2, page_h, coords);
+            let matched: Vec<TextItem> = match items {
+                Some(items) => items
+                    .iter()
+                    .filter(|item| region_overlaps_item(item, bounds))
+                    .cloned()
+                    .collect(),
+                None => Vec::new(),
+            };
            let has_text_quality_issue = region_items_have_decoding_issue(&matched);
            let text = collect_text_from_matched_items(matched, adaptive_threshold);
            let has_cid_issue = is_cid_garbage(&text);
@@ -731,21 +655,8 @@ pub fn extract_text_in_regions_mem(
            // Check per-region text quality instead of blanket page-level
            // GID rejection. A GID font in a logo elsewhere on the page
            // shouldn't force GPU OCR for clean text regions.
-            // A region whose ONLY overlapping items were assigned to a
-            // better-overlapping neighbor must not fall back to OCR: the
-            // pixels it would re-read belong to that neighbor, and OCR
-            // would reintroduce the duplication exclusivity removed.
-            // Before exclusive assignment these regions were non-empty
-            // native (no OCR), so this preserves the old OCR load too.
-            // Requires ZERO items assigned HERE: a region whose own
-            // assigned items materialize to empty text (whitespace-only,
-            // collector-filtered) keeps its OCR fallback.
-            let lost_to_neighbor = text.trim().is_empty()
-                && ocr_reason.is_none()
-                && assigned_count == 0
-                && had_candidates[region_idx];
-            let needs_ocr = !lost_to_neighbor
-                && (ocr_reason.is_some() || text.trim().is_empty() || is_garbage_text(&text));
+            let needs_ocr =
+                ocr_reason.is_some() || text.trim().is_empty() || is_garbage_text(&text);

            page_results.push(RegionText {
                text,
@@ -3261,26 +3172,8 @@ fn region_bounds(
    }
 }

-/// Inclusion margin shared by the region/item overlap predicates and the
-/// exclusive-assignment area score — these MUST stay in sync: an item that
-/// passes the boolean guard must always have positive overlap area.
-const REGION_MARGIN: f32 = 1.5;
-
-/// Overlap area between an item and region bounds (same margin as the
-/// boolean test) — the exclusive-assignment score.
-fn region_item_overlap_area(item: &TextItem, bounds: RegionBounds) -> f32 {
-    let item_x_max = item.x + text_utils::effective_width(item);
-    let item_y_max = item.y + item.height;
-    let x_overlap = (item_x_max.min(bounds.x_max + REGION_MARGIN)
-        - item.x.max(bounds.x_min - REGION_MARGIN))
-    .max(0.0);
-    let y_overlap = (item_y_max.min(bounds.y_max + REGION_MARGIN)
-        - item.y.max(bounds.y_min - REGION_MARGIN))
-    .max(0.0);
-    x_overlap * y_overlap
-}
-
 fn region_overlaps_item(item: &TextItem, bounds: RegionBounds) -> bool {
+    const REGION_MARGIN: f32 = 1.5;
    let item_x_min = item.x;
    let item_x_max = item.x + text_utils::effective_width(item);
    let item_y_min = item.y;
@@ -3296,6 +3189,7 @@ fn region_overlaps_item(item: &TextItem, bounds: RegionBounds) -> bool {
 }

 fn region_overlaps_rect(rect: &PdfRect, bounds: RegionBounds) -> bool {
+    const REGION_MARGIN: f32 = 1.5;
    let (x_min, y_min, x_max, y_max) = normalized_rect_edges(rect);
    ranges_overlap(
        x_min,
@@ -3311,6 +3205,7 @@ fn region_overlaps_rect(rect: &PdfRect, bounds: RegionBounds) -> bool {
 }

 fn region_overlaps_line(line: &PdfLine, bounds: RegionBounds) -> bool {
+    const REGION_MARGIN: f32 = 1.5;
    let x_min = line.x1.min(line.x2);
    let x_max = line.x1.max(line.x2);
    let y_min = line.y1.min(line.y2);
@@ -3367,40 +3262,24 @@ fn tsr_region_contains_item(item: &TextItem, bounds: RegionBounds) -> bool {
 /// page count from it directly to avoid the metadata-only round-trip.
 pub(crate) fn load_document_from_path<P: AsRef<Path>>(
    path: P,
-) -> Result<(Document, u32), PdfError> {
-    load_document_from_path_with_password(path, None)
-}
-
-/// Load a PDF file, decrypting with `password` if the file is encrypted.
-pub(crate) fn load_document_from_path_with_password<P: AsRef<Path>>(
-    path: P,
-    password: Option<&str>,
 ) -> Result<(Document, u32), PdfError> {
    let buffer = std::fs::read(&path)?;
-    load_document_from_mem_with_password(&buffer, password)
+    load_document_from_mem(&buffer)
 }

 /// Load a PDF from a memory buffer.
 pub(crate) fn load_document_from_mem(buffer: &[u8]) -> Result<(Document, u32), PdfError> {
-    load_document_from_mem_with_password(buffer, None)
-}
-
-/// Load a PDF from a memory buffer, decrypting with `password` if encrypted.
-pub(crate) fn load_document_from_mem_with_password(
-    buffer: &[u8],
-    password: Option<&str>,
-) -> Result<(Document, u32), PdfError> {
    // Fix malformed struct element names before parsing. Some PDF generators
    // write bare names (/S Code) instead of proper PDF names (/S /Code), which
    // causes lopdf to silently drop the entire object.
    let fixed = structure_tree::fix_bare_struct_names(buffer);
    let buf = fixed.as_ref();

-    let doc = match load_document_bytes(buf, password) {
+    let doc = match load_document_bytes(buf) {
        Ok(doc) => doc,
        Err(first_err) => {
            for repaired in repair_pdf_container_candidates(buf) {
-                match load_document_bytes(&repaired, password) {
+                match load_document_bytes(&repaired) {
                    Ok(doc) => {
                        log::debug!("loaded PDF after repairing malformed container bytes");
                        let page_count = doc.get_pages().len() as u32;
@@ -3420,31 +3299,13 @@ pub(crate) fn load_document_from_mem_with_password(
    Ok((doc, page_count))
 }

-fn load_document_bytes(buf: &[u8], password: Option<&str>) -> Result<Document, lopdf::Error> {
+fn load_document_bytes(buf: &[u8]) -> Result<Document, lopdf::Error> {
    match Document::load_mem(buf) {
-        // Some encrypted PDFs load structurally but leave their streams
-        // encrypted (`is_encrypted()` stays true); reading them yields garbage
-        // until we re-load with a password. Others fail load_mem outright with
-        // an encryption error. Handle both by re-loading with the password.
-        Ok(doc) if doc.is_encrypted() => decrypt_document_bytes(buf, password),
        Ok(doc) => Ok(doc),
-        Err(ref e) if is_encrypted_lopdf_error(e) => decrypt_document_bytes(buf, password),
-        Err(e) => Err(e),
-    }
-}
-
-/// Re-load an encrypted PDF, decrypting with `password`. Falls back to the
-/// empty password (owner-only encryption, the common "protected" case) when a
-/// non-empty password was supplied but rejected.
-fn decrypt_document_bytes(buf: &[u8], password: Option<&str>) -> Result<Document, lopdf::Error> {
-    let pw = password.unwrap_or("");
-    match Document::load_mem_with_options(buf, lopdf::LoadOptions::with_password(pw)) {
-        Ok(doc) => Ok(doc),
-        Err(inner) if !pw.is_empty() => {
+        Err(ref e) if is_encrypted_lopdf_error(e) => {
            Document::load_mem_with_options(buf, lopdf::LoadOptions::with_password(""))
-                .map_err(|_| inner)
        }
-        Err(inner) => Err(inner),
+        Err(e) => Err(e),
    }
 }

@@ -3539,7 +3400,6 @@ fn process_document(
    let pages_needing_ocr = detection.pages_needing_ocr;
    let title = detection.title;
    let confidence = detection.confidence;
-    let detection_ocr_reasons = detection.ocr_reasons_by_page;

    // DetectOnly → return immediately
    if options.mode == ProcessMode::DetectOnly {
@@ -3549,7 +3409,7 @@ fn process_document(
            page_count,
            processing_time_ms: start.elapsed().as_millis() as u64,
            pages_needing_ocr,
-            ocr_reasons_by_page: page_ocr_reasons_vec(detection_ocr_reasons),
+            ocr_reasons_by_page: Vec::new(),
            title,
            confidence,
            layout: LayoutComplexity::default(),
@@ -3565,7 +3425,7 @@ fn process_document(
            page_count,
            processing_time_ms: start.elapsed().as_millis() as u64,
            pages_needing_ocr,
-            ocr_reasons_by_page: page_ocr_reasons_vec(detection_ocr_reasons),
+            ocr_reasons_by_page: Vec::new(),
            title,
            confidence,
            layout: LayoutComplexity::default(),
@@ -3831,13 +3691,7 @@ fn process_document(
        page_count,
        processing_time_ms: start.elapsed().as_millis() as u64,
        pages_needing_ocr,
-        ocr_reasons_by_page: {
-            // Detector reasons (scanned / no_text / vector_text / garbled) merged
-            // with the markdown-stage garbled detection, deduped per page.
-            let mut merged = detection_ocr_reasons;
-            merge_ocr_reasons(&mut merged, text_quality_reasons_by_page);
-            page_ocr_reasons_vec(merged)
-        },
+        ocr_reasons_by_page: page_ocr_reasons_vec(text_quality_reasons_by_page),
        title,
        confidence,
        layout,
@@ -3849,15 +3703,283 @@ fn process_document(
 // Internal helpers
 // =========================================================================

+/// Detect broken font encodings in extracted markdown text.
+///
+/// Two heuristics:
+/// 1. **U+FFFD**: Any replacement character indicates decode failures.
+/// 2. **Dollar-as-space**: Pattern like `Word$Word$Word` where `$` is used as a
+///    word separator due to broken ToUnicode CMaps. Triggers when either:
+///    - More than 50% of `$` are between letters (clear substitution pattern), OR
+///    - More than 20 letter-dollar-letter occurrences (even if some `$` are also
+///      used as trailing/leading separators, 20+ is far beyond normal financial text).
+fn detect_encoding_issues(markdown: &str) -> bool {
+    // Heuristic 1: U+FFFD replacement characters
+    if markdown.contains('\u{FFFD}') {
+        return true;
+    }
+
+    // Heuristic 2: dollar-as-space pattern
+    if has_dollar_as_space_pattern(markdown) {
+        return true;
+    }
+
+    // Heuristic 3: substitution-cipher letter statistics (broken ToUnicode)
+    let mut stats = CipherGarbleStats::default();
+    stats.add_text(markdown);
+    stats.looks_garbled()
+}
+
+fn has_dollar_as_space_pattern(markdown: &str) -> bool {
+    let total_dollars = markdown.matches('$').count();
+    if total_dollars > 10 {
+        let bytes = markdown.as_bytes();
+        let mut letter_dollar_letter = 0usize;
+        for i in 1..bytes.len().saturating_sub(1) {
+            if bytes[i] == b'$'
+                && bytes[i - 1].is_ascii_alphabetic()
+                && bytes[i + 1].is_ascii_alphabetic()
+            {
+                letter_dollar_letter += 1;
+            }
+        }
+        if letter_dollar_letter > 20 || letter_dollar_letter * 2 > total_dollars {
+            return true;
+        }
+    }
+
+    false
+}
+
+/// English letter frequencies (percent, a–z). Used as a natural-language
+/// reference: every Latin-script language in the eval corpus (Swedish,
+/// Finnish, Turkish, German, romaji) scores ≥ 0.80 cosine similarity against
+/// it, while substitution-cipher text scores ~0.53.
+const ENGLISH_LETTER_FREQ: [f64; 26] = [
+    8.2, 1.5, 2.8, 4.3, 12.7, 2.2, 2.0, 6.1, 7.0, 0.15, 0.8, 4.0, 2.4, 6.7, 7.5, 1.9, 0.1, 6.0,
+    6.3, 9.1, 2.8, 1.0, 2.4, 0.15, 2.0, 0.07,
+];
+
+/// Letter statistics for detecting substitution-cipher garbling: broken
+/// ToUnicode CMaps that shift every character by a per-range constant (e.g.
+/// `Certificate` extracted as `8VceZWZTReV`). Such text is 100% printable
+/// ASCII with word-like token lengths, so it defeats `is_garbage_text` and
+/// produces no replacement characters — it needs its own discriminator.
+#[derive(Debug, Default)]
+struct CipherGarbleStats {
+    /// Case-folded ASCII letter histogram.
+    letter_counts: [u32; 26],
+    ascii_letters: usize,
+    ascii_vowels: usize,
+    /// Accented Latin letters (Latin-1 Supplement through Latin Extended-B,
+    /// plus Latin Extended Additional). Count toward Latin dominance only.
+    latin_ext_letters: usize,
+    non_latin_letters: usize,
+    /// Adjacent ASCII-letter pairs, and how many of them switch from
+    /// lowercase straight to uppercase mid-word.
+    letter_bigrams: usize,
+    case_shift_bigrams: usize,
+}
+
+impl CipherGarbleStats {
+    fn add_text(&mut self, text: &str) {
+        let mut prev: Option<char> = None;
+        for ch in text.chars() {
+            if ch.is_ascii_alphabetic() {
+                let idx = (ch.to_ascii_lowercase() as u8 - b'a') as usize;
+                self.letter_counts[idx] += 1;
+                self.ascii_letters += 1;
+                if matches!(ch.to_ascii_lowercase(), 'a' | 'e' | 'i' | 'o' | 'u') {
+                    self.ascii_vowels += 1;
+                }
+                if let Some(p) = prev {
+                    self.letter_bigrams += 1;
+                    if p.is_ascii_lowercase() && ch.is_ascii_uppercase() {
+                        self.case_shift_bigrams += 1;
+                    }
+                }
+                prev = Some(ch);
+            } else {
+                if ch.is_alphabetic() {
+                    if matches!(ch as u32, 0xC0..=0x24F | 0x1E00..=0x1EFF) {
+                        self.latin_ext_letters += 1;
+                    } else {
+                        self.non_latin_letters += 1;
+                    }
+                }
+                prev = None;
+            }
+        }
+    }
+
+    /// Cosine similarity between the observed letter histogram and English
+    /// letter frequencies. A shifted alphabet permutes the histogram, which
+    /// destroys the similarity regardless of the shift amount.
+    fn english_cosine(&self) -> f64 {
+        if self.ascii_letters == 0 {
+            return 1.0;
+        }
+        let n = self.ascii_letters as f64;
+        let mut dot = 0.0;
+        let mut norm_obs = 0.0;
+        for (count, freq) in self.letter_counts.iter().zip(ENGLISH_LETTER_FREQ) {
+            let p = *count as f64 / n;
+            dot += p * freq;
+            norm_obs += p * p;
+        }
+        let norm_en = ENGLISH_LETTER_FREQ
+            .iter()
+            .map(|f| f * f)
+            .sum::<f64>()
+            .sqrt();
+        dot / (norm_obs.sqrt() * norm_en)
+    }
+
+    /// Cosine similarity between the observed histogram and English
+    /// frequencies after sorting BOTH descending — i.e. comparing the *shape*
+    /// of the frequency profile, ignoring which letter sits where. A
+    /// substitution cipher is a bijection, so it preserves this shape exactly
+    /// (att10k 0.97, arbitrary shifts 0.99) regardless of case or offset.
+    /// Non-linguistic ASCII has a different profile: a small alphabet is far
+    /// steeper (random DNA 0.74, hex dumps 0.81), so the shape diverges.
+    fn english_shape_cosine(&self) -> f64 {
+        if self.ascii_letters == 0 {
+            return 1.0;
+        }
+        let n = self.ascii_letters as f64;
+        let mut obs: [f64; 26] = std::array::from_fn(|i| self.letter_counts[i] as f64 / n);
+        obs.sort_unstable_by(|a, b| b.total_cmp(a));
+        let mut en = ENGLISH_LETTER_FREQ;
+        en.sort_unstable_by(|a, b| b.total_cmp(a));
+
+        let dot: f64 = obs.iter().zip(en).map(|(o, e)| o * e).sum();
+        let norm_obs = obs.iter().map(|o| o * o).sum::<f64>().sqrt();
+        let norm_en = en.iter().map(|e| e * e).sum::<f64>().sqrt();
+        dot / (norm_obs * norm_en)
+    }
+
+    /// Thresholds validated against the 380-document pdf-evals snapshot
+    /// corpus (0 false positives) and the garbled ParseBench `att10k` page
+    /// (vowel ratio 0.245, case-shift rate 0.225, cosine 0.532). Closest
+    /// legitimate document on each axis: vowel ratio 0.264 (circuit
+    /// schematic), case-shift rate 0.021, cosine 0.801.
+    fn looks_garbled(&self) -> bool {
+        // Need a statistically meaningful, Latin-dominant sample.
+        if self.ascii_letters < 200
+            || self.non_latin_letters > self.ascii_letters + self.latin_ext_letters
+        {
+            return false;
+        }
+
+        // Real Latin-script text keeps vowels above ~30% of letters even in
+        // acronym- and part-number-heavy documents; shifted text starves them.
+        let vowel_ratio = self.ascii_vowels as f64 / self.ascii_letters as f64;
+        if vowel_ratio > 0.30 {
+            return false;
+        }
+
+        // Signal 1: lowercase→uppercase transitions inside words. A shifted
+        // lowercase alphabet straddles the ASCII uppercase block ('i'→'Z',
+        // 't'→'e'), so garbled words flip case constantly. Real documents
+        // stay ≤ 0.02 even with camelCase identifiers.
+        let case_shifts = self.letter_bigrams >= 100
+            && self.case_shift_bigrams as f64 >= self.letter_bigrams as f64 * 0.10;
+
+        // Signal 2: the histogram is a permutation of natural language — an
+        // English-like frequency SHAPE (sorted cosine high) but with letters
+        // in the wrong POSITIONS (unsorted cosine low). This is the signature
+        // of a substitution cipher and is case-independent, so it catches
+        // all-lowercase and all-uppercase shifts as well as case-straddling
+        // ones. Genuinely non-linguistic ASCII that is merely "unlike English"
+        // fails one of the two halves: DNA/hex dumps have too steep a profile
+        // (shape cosine < 0.90), while protein sequences, ticker symbols and
+        // base64 are not sufficiently unlike English in position (unsorted
+        // cosine ≥ 0.60) — so none of them are routed to OCR.
+        let permuted_language = self.english_cosine() < 0.60 && self.english_shape_cosine() >= 0.90;
+
+        case_shifts || permuted_language
+    }
+}
+
+#[derive(Debug, Default)]
+struct TextQualityReport {
+    pages_needing_ocr: Vec<u32>,
+    has_encoding_issues: bool,
+    reasons_by_page: BTreeMap<u32, Vec<String>>,
+}
+
+#[derive(Debug, Default)]
+struct PageTextQualityEvidence {
+    chars: usize,
+    replacement_chars: usize,
+    replacement_spans: usize,
+    longest_replacement_run: usize,
+    cipher_garble: CipherGarbleStats,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+enum TextSpanIssueKind {
+    Replacement,
+    Strong,
+}
+
+fn analyze_text_quality(items: &[TextItem]) -> TextQualityReport {
+    let mut reasons_by_page = BTreeMap::new();
+    let mut evidence_by_page = BTreeMap::<u32, PageTextQualityEvidence>::new();
+
+    for item in items {
+        if !matches!(item.item_type, crate::types::ItemType::Text) {
+            continue;
+        }
+
+        let evidence = evidence_by_page.entry(item.page).or_default();
+        evidence.chars += item.text.chars().filter(|ch| !ch.is_whitespace()).count();
+        evidence.cipher_garble.add_text(&item.text);
+
+        match text_span_decoding_issue_kind(&item.text) {
+            Some(TextSpanIssueKind::Strong) => {
+                add_ocr_reason(
+                    &mut reasons_by_page,
+                    item.page,
+                    OCR_REASON_SUSPECTED_GARBLED_TEXT,
+                );
+            }
+            Some(TextSpanIssueKind::Replacement) => {
+                let stats = replacement_text_stats(&item.text);
+                evidence.replacement_chars += stats.0;
+                evidence.replacement_spans += 1;
+                evidence.longest_replacement_run = evidence.longest_replacement_run.max(stats.1);
+            }
+            None => {}
+        }
+    }
+
+    for (page, evidence) in evidence_by_page {
+        if reasons_by_page.contains_key(&page) {
+            continue;
+        }
+        if page_replacement_evidence_needs_ocr(&evidence) || evidence.cipher_garble.looks_garbled()
+        {
+            add_ocr_reason(
+                &mut reasons_by_page,
+                page,
+                OCR_REASON_SUSPECTED_GARBLED_TEXT,
+            );
+        }
+    }
+
+    let pages_needing_ocr: Vec<u32> = reasons_by_page.keys().copied().collect();
+    TextQualityReport {
+        has_encoding_issues: !pages_needing_ocr.is_empty(),
+        pages_needing_ocr,
+        reasons_by_page,
+    }
+}
+
 fn suspected_garbled_reason() -> String {
    OCR_REASON_SUSPECTED_GARBLED_TEXT.to_string()
 }

-pub(crate) fn add_ocr_reason(
-    reasons_by_page: &mut BTreeMap<u32, Vec<String>>,
-    page: u32,
-    reason: &str,
-) {
+fn add_ocr_reason(reasons_by_page: &mut BTreeMap<u32, Vec<String>>, page: u32, reason: &str) {
    let reasons = reasons_by_page.entry(page).or_default();
    if !reasons.iter().any(|existing| existing == reason) {
        reasons.push(reason.to_string());
@@ -3889,6 +4011,225 @@ fn page_ocr_reasons_vec(reasons_by_page: BTreeMap<u32, Vec<String>>) -> Vec<Page
        .collect()
 }

+fn region_items_have_decoding_issue(items: &[TextItem]) -> bool {
+    items.iter().any(|item| {
+        matches!(item.item_type, crate::types::ItemType::Text)
+            && text_span_has_decoding_issue(&item.text)
+    })
+}
+
+fn text_span_has_decoding_issue(text: &str) -> bool {
+    text_span_decoding_issue_kind(text).is_some()
+}
+
+fn text_span_decoding_issue_kind(text: &str) -> Option<TextSpanIssueKind> {
+    let text = text.trim();
+    if text.is_empty() {
+        return None;
+    }
+
+    if has_dollar_as_space_pattern(text)
+        || has_private_use_text_run(text)
+        || is_cid_garbage(text)
+        || has_cid_control_token(text)
+    {
+        return Some(TextSpanIssueKind::Strong);
+    }
+
+    if has_replacement_text_run(text) {
+        return Some(TextSpanIssueKind::Replacement);
+    }
+
+    None
+}
+
+fn replacement_text_stats(text: &str) -> (usize, usize) {
+    let mut replacement = 0usize;
+    let mut current_run = 0usize;
+    let mut longest_run = 0usize;
+
+    for ch in text.chars() {
+        if ch == '\u{FFFD}' {
+            replacement += 1;
+            current_run += 1;
+            longest_run = longest_run.max(current_run);
+        } else {
+            current_run = 0;
+        }
+    }
+
+    (replacement, longest_run)
+}
+
+fn page_replacement_evidence_needs_ocr(evidence: &PageTextQualityEvidence) -> bool {
+    if evidence.replacement_chars == 0 || evidence.chars == 0 {
+        return false;
+    }
+
+    // If the entire page is only a short broken text layer, even a short
+    // replacement run is enough evidence. On otherwise text-heavy pages,
+    // require density so math formulas do not force full-page OCR.
+    if evidence.chars <= 80 && evidence.longest_replacement_run >= 2 {
+        return true;
+    }
+
+    let replacement_density_bps = evidence.replacement_chars * 10_000 / evidence.chars;
+    let enough_bad_text = evidence.replacement_chars >= 12 && replacement_density_bps >= 500;
+    let repeated_bad_spans = evidence.replacement_spans >= 3 && replacement_density_bps >= 250;
+    let long_bad_run = evidence.longest_replacement_run >= 8 && replacement_density_bps >= 250;
+
+    enough_bad_text || repeated_bad_spans || long_bad_run
+}
+
+fn has_replacement_text_run(text: &str) -> bool {
+    let (replacement, longest_run) = replacement_text_stats(text);
+    longest_run >= 2 || replacement >= 3
+}
+
+fn has_private_use_text_run(text: &str) -> bool {
+    let mut total = 0usize;
+    let mut private_use = 0usize;
+    let mut current_run = 0usize;
+    let mut longest_run = 0usize;
+
+    for ch in text.chars() {
+        if ch.is_whitespace() {
+            current_run = 0;
+            continue;
+        }
+        total += 1;
+        if is_private_use_char(ch) {
+            private_use += 1;
+            current_run += 1;
+            longest_run = longest_run.max(current_run);
+        } else {
+            current_run = 0;
+        }
+    }
+
+    if private_use == 0 {
+        return false;
+    }
+
+    longest_run >= 3 || (total >= 5 && private_use >= 2 && private_use * 2 >= total)
+}
+
+fn has_cid_control_token(text: &str) -> bool {
+    text.split_whitespace().any(token_has_cid_control)
+}
+
+fn token_has_cid_control(token: &str) -> bool {
+    let mut total = 0usize;
+    let mut c1_control = 0usize;
+
+    for ch in token.chars() {
+        total += 1;
+        if ('\u{0080}'..='\u{009F}').contains(&ch) {
+            c1_control += 1;
+        }
+    }
+
+    total >= 5 && c1_control >= 2 && c1_control * 20 >= total
+}
+
+fn is_private_use_char(ch: char) -> bool {
+    matches!(
+        ch as u32,
+        0xE000..=0xF8FF | 0xF0000..=0xFFFFD | 0x100000..=0x10FFFD
+    )
+}
+
+/// Check if extracted text is predominantly garbage (non-alphanumeric).
+///
+/// Broken font encodings produce text like "----1-.-.-.___  --.-. .._ I_---."
+/// where most characters are punctuation/symbols. Real text in any language
+/// has >50% alphanumeric characters.
+fn is_garbage_text(markdown: &str) -> bool {
+    let mut alphanum = 0usize;
+    let mut non_alphanum = 0usize;
+
+    let chars: Vec<char> = markdown.chars().collect();
+    let mut i = 0usize;
+    while i < chars.len() {
+        let ch = chars[i];
+        let mut run_end = i + 1;
+        while run_end < chars.len() && chars[run_end] == ch {
+            run_end += 1;
+        }
+
+        let is_decorative_leader = matches!(ch, '.' | '_' | '·') && run_end - i >= 3;
+        if !is_decorative_leader {
+            for &run_ch in &chars[i..run_end] {
+                if run_ch.is_whitespace() {
+                    continue;
+                }
+                // Skip markdown syntax chars that we add (not from the PDF)
+                if matches!(run_ch, '#' | '*' | '|' | '-' | '\n') {
+                    continue;
+                }
+                if run_ch.is_alphanumeric() {
+                    alphanum += 1;
+                } else {
+                    non_alphanum += 1;
+                }
+            }
+        }
+        i = run_end;
+    }
+
+    let total = alphanum + non_alphanum;
+    total >= 50 && alphanum * 2 < total
+}
+
+/// Detect garbage from failed CID-to-Unicode mapping on Identity-H fonts.
+///
+/// When CID values don't correspond to Unicode codepoints, the raw bytes often
+/// produce characters in the C1 control range (U+0080–U+009F) or Private Use
+/// Area, mixed with random Latin Extended characters.  Valid text in any
+/// language almost never contains C1 controls.  We also fall back to the
+/// general `is_garbage_text` check for non-alphanumeric-heavy patterns.
+fn is_cid_garbage(text: &str) -> bool {
+    if is_garbage_text(text) {
+        return true;
+    }
+    let mut total = 0usize;
+    let mut c1_control = 0usize;
+    let mut high_latin = 0usize;
+    for ch in text.chars() {
+        if ch.is_whitespace() {
+            continue;
+        }
+        total += 1;
+        // C1 control characters (U+0080–U+009F) — almost never in real text
+        if ch == '·' {
+            continue;
+        }
+        if ('\u{0080}'..='\u{009F}').contains(&ch) {
+            c1_control += 1;
+        }
+        // High Latin-1 (U+00A0–U+00FF) — legitimate in Western European text
+        // but when combined with ASCII in CID passthrough, indicates mojibake
+        // from CID values being misinterpreted as Latin-1 characters.
+        if ('\u{00A0}'..='\u{00FF}').contains(&ch) {
+            high_latin += 1;
+        }
+    }
+    if total < 5 {
+        return false;
+    }
+    // If ≥5% of non-whitespace chars are C1 controls, it's garbage
+    if c1_control >= 2 && c1_control * 20 >= total {
+        return true;
+    }
+    // If ≥40% of non-whitespace chars are high Latin-1 AND the text has few
+    // ASCII letters, it's likely CID-as-Latin-1 mojibake (Japanese/CJK PDFs
+    // where CID values 0x80-0xFF become accented Latin characters).  Keep a
+    // minimum length so short math tokens like "2×()×" do not route a clean
+    // page to OCR.
+    let ascii_letters = text.chars().filter(|c| c.is_ascii_alphabetic()).count();
+    total >= 20 && high_latin * 5 >= total * 2 && ascii_letters * 3 < total
+}
+
 /// Detect markdown tables with suspicious structure that suggest the heuristic
 /// missed/mangled rows or columns. Returns true when the caller should treat
 /// the result as `needs_ocr` and fall back to GPU OCR.
@@ -153,100 +153,6 @@ pub(crate) fn is_toc_entry_line(text: &str) -> bool {
    dots >= 3
 }

-/// A heading that announces a table of contents ("Contents", "Table of
-/// Contents"). Lines after it on the same page are ToC entries — section
-/// titles that look exactly like headings but must not be promoted.
-pub(crate) fn is_toc_marker_heading(text: &str) -> bool {
-    let t = text.trim().trim_end_matches(':').trim().to_lowercase();
-    matches!(t.as_str(), "contents" | "table of contents")
-}
-
-/// Lines that resemble headings structurally but are display-math fragments:
-/// equations ending in an equation number ("S = kB ln W, (2)") or equation
-/// lead-ins ("Rearranging Equation (8) gives:"). Both carry an "(N)" equation
-/// reference — but a trailing "(N)" alone is not enough: real headings end
-/// with parenthesized numbers too ("Nicaea (325)", appendix numbering), so
-/// the suffix form additionally requires math evidence — an "=" in the line
-/// or a comma immediately before the number, both present in every display
-/// equation and absent from name-plus-number headings. A bare trailing colon
-/// is NOT a fragment signal either: real headings frequently end with colons
-/// ("Procedure:", "Steps for Using the Microscope:").
-pub(crate) fn is_heading_fragment(text: &str) -> bool {
-    let t = text.trim_end();
-
-    // A lowercase-initial one-or-two-word "heading" is a mid-sentence
-    // fragment beside display math ("or inversely", "and therefore") —
-    // real headings that short start uppercase. Measured as spurious
-    // headings on academic docs (fire-pdf ENG-5029 / opendataloader MHS).
-    {
-        let words: Vec<&str> = t.split_whitespace().collect();
-        if words.len() <= 2 {
-            if let Some(first_alpha) = t.chars().find(|c| c.is_alphabetic()) {
-                if first_alpha.is_lowercase() {
-                    return true;
-                }
-            }
-        }
-    }
-
-    fn is_equation_number(s: &str) -> bool {
-        s.strip_prefix('(')
-            .and_then(|r| r.strip_suffix(')'))
-            .is_some_and(|inner| {
-                !inner.is_empty() && inner.len() <= 3 && inner.chars().all(|c| c.is_ascii_digit())
-            })
-    }
-
-    // Equation-number suffix with math evidence: "S = kB ln W, (2)"
-    let mut rev = t.rsplit(' ');
-    let last = rev.next().unwrap_or("");
-    if is_equation_number(last) {
-        // Page-of-total running headers: "LIVSMEDELSVERKET PM 2 (10)"
-        if let Some(prev_word) = t.rsplit(' ').nth(1) {
-            if let (Ok(page), Some(total)) = (
-                prev_word.parse::<u32>(),
-                last.trim_start_matches('(')
-                    .trim_end_matches(')')
-                    .parse::<u32>()
-                    .ok(),
-            ) {
-                if page <= total {
-                    return true;
-                }
-            }
-        }
-        let punct_before = rev
-            .next()
-            .is_some_and(|w| w.ends_with(',') || w.ends_with(':'));
-        let has_math_op = t.chars().any(|c| {
-            matches!(
-                c,
-                '=' | '<'
-                    | '>'
-                    | '≤'
-                    | '≥'
-                    | '≪'
-                    | '≫'
-                    | '≈'
-                    | '≠'
-                    | '±'
-                    | '∑'
-                    | '∫'
-                    | '√'
-                    | '∝'
-            )
-        });
-        if punct_before || has_math_op {
-            return true;
-        }
-    }
-    // Lead-in: ends with a colon AND references an equation number inline
-    if t.ends_with(':') && t.split_whitespace().any(is_equation_number) {
-        return true;
-    }
-    false
-}
-
 /// Compute the Y-gap threshold for paragraph break detection.
 ///
 /// Instead of using a fixed multiple of base_size (which fails for double-spaced
@@ -461,45 +367,4 @@ mod tests {
        // Long numbers are data, not page refs
        assert!(!is_toc_entry_line("ISBN ... 97814"));
    }
-
-    #[test]
-    fn toc_marker_headings() {
-        assert!(is_toc_marker_heading("Contents"));
-        assert!(is_toc_marker_heading("CONTENTS"));
-        assert!(is_toc_marker_heading("Table of Contents"));
-        assert!(is_toc_marker_heading("Table of contents:"));
-        assert!(!is_toc_marker_heading("Contents of the Shipment"));
-        assert!(!is_toc_marker_heading("Introduction"));
-    }
-
-    #[test]
-    fn heading_fragments() {
-        // Equation lead-ins: colon ending + inline equation reference
-        assert!(is_heading_fragment("or inversely"));
-        assert!(is_heading_fragment("and therefore"));
-        assert!(!is_heading_fragment("Introduction"));
-        assert!(!is_heading_fragment("iPhone Sales Strategy Overview")); // 4 words, exempt
-        assert!(is_heading_fragment("Rearranging Equation (8) gives:"));
-        // Display-equation neighbours ending in an equation number
-        assert!(is_heading_fragment("S = kB ln W, (2)"));
-        assert!(is_heading_fragment("E = mc2 (12)"));
-        assert!(is_heading_fragment("x + y = z, (3)"));
-        // Page-of-total running headers
-        assert!(is_heading_fragment("LIVSMEDELSVERKET PM 2 (10)"));
-        // Comparison-operator evidence and colon-before-number
-        assert!(is_heading_fragment(
-            "PLL\u{fe} PHH\u{226a} PLH\u{fe} PHL: (12)"
-        ));
-        // Real headings pass — including name-plus-number and colon-ended ones
-        assert!(!is_heading_fragment("Nicaea (325)"));
-        assert!(!is_heading_fragment(
-            "\u{627}\u{644}\u{645}\u{644}\u{62d}\u{642} \u{631}\u{642}\u{645} (1)"
-        ));
-        assert!(!is_heading_fragment("4. Entropy"));
-        assert!(!is_heading_fragment("Procedure:"));
-        assert!(!is_heading_fragment("Steps for Using the Microscope:"));
-        assert!(!is_heading_fragment("Changing objectives:"));
-        assert!(!is_heading_fragment("Sales by Region (2024)"));
-        assert!(!is_heading_fragment("Results (preliminary)"));
-    }
 }
@@ -7,8 +7,7 @@ use crate::types::TextLine;

 use super::analysis::{
    bold_heading_level, calculate_font_stats, compute_heading_tiers, compute_paragraph_threshold,
-    detect_header_level, font_size_rarity, has_dot_leaders, is_heading_fragment, is_toc_entry_line,
-    is_toc_marker_heading,
+    detect_header_level, font_size_rarity, has_dot_leaders, is_toc_entry_line,
 };
 use super::classify::{
    format_list_item, is_caption_line, is_list_item, is_monospace_font, starts_with_bullet_marker,
@@ -141,11 +140,8 @@ fn find_isolated_lines(lines: &[TextLine], base_size: f32, para_threshold: f32)
        }
    }
    for (&page, &(total, isolated)) in &page_line_counts {
-        // The ratio only means something on pages dense enough for a
-        // multi-column misfire; on sparse pages (covers, ToC pages with a
-        // lone title) one isolated line is 25%+ of the page and exactly the
-        // line isolation exists to find.
-        if total >= 10 && isolated as f32 / total as f32 > 0.25 {
+        if total > 0 && isolated as f32 / total as f32 > 0.25 {
+            // Too many isolated lines on this page — remove them all
            set.retain(|&i| lines[i].page != page);
        }
    }
@@ -490,7 +486,6 @@ pub(super) fn to_markdown_from_lines_with_tables_and_images(
    let mut in_code_block = false;
    let mut prev_had_dot_leaders = false;
    let mut paragraph_in_wrapped_bold_run = false;
-    let mut toc_suppress_page: Option<u32> = None;
    let mut inserted_tables: HashSet<(u32, usize)> = HashSet::new();
    let mut inserted_images: HashSet<(u32, usize)> = HashSet::new();

@@ -703,22 +698,12 @@ pub(super) fn to_markdown_from_lines_with_tables_and_images(
                _ => false,
            };

-        // Lines explicitly tagged with a non-heading content role must never
-        // be promoted by the visual heuristic — a tagged list item, quote, or
-        // code line can look exactly like a heading (short, isolated).
-        let non_heading_role = struct_role
-            .as_ref()
-            .is_some_and(StructRole::is_non_heading_content);
        let heuristic_heading = if options.detect_headers
-            && !non_heading_role
-            && !is_code_line
            && !looks_like_list_continuation
            && plain_trimmed.len() > 3
            && plain_trimmed.split_whitespace().count() <= 15
            && !starts_with_bullet_marker(plain_trimmed)
            && !is_toc_entry_line(plain_trimmed)
-            && !is_heading_fragment(plain_trimmed)
-            && toc_suppress_page != Some(line.page)
        {
            let line_font_size = line.items.first().map(|i| i.font_size).unwrap_or(base_size);
            detect_header_level(line_font_size, base_size, &heading_tiers).or_else(|| {
@@ -783,9 +768,6 @@ pub(super) fn to_markdown_from_lines_with_tables_and_images(
                plain_text.clone()
            };
            output.push_str(&format!("{} {}\n\n", prefix, heading_text.trim()));
-            if is_toc_marker_heading(plain_trimmed) {
-                toc_suppress_page = Some(line.page);
-            }
            in_list = false;
            continue;
        }
@@ -982,7 +964,6 @@ pub fn to_markdown_from_lines(lines: Vec<TextLine>, options: MarkdownOptions) ->
    let mut last_list_x: Option<f32> = None;
    let mut prev_had_dot_leaders = false;
    let mut paragraph_in_wrapped_bold_run = false;
-    let mut toc_suppress_page: Option<u32> = None;

    for (line_idx, line) in lines.iter().enumerate() {
        // Page break
@@ -1062,9 +1043,6 @@ pub fn to_markdown_from_lines(lines: Vec<TextLine>, options: MarkdownOptions) ->
            && plain_trimmed.len() > 3
            && plain_trimmed.split_whitespace().count() <= 15
            && !is_toc_entry_line(plain_trimmed)
-            && !is_heading_fragment(plain_trimmed)
-            && toc_suppress_page != Some(line.page)
-            && !(options.detect_code && line.items.iter().any(|i| is_monospace_font(&i.font)))
        {
            let line_font_size = line.items.first().map(|i| i.font_size).unwrap_or(base_size);
            if let Some(header_level) =
@@ -1108,9 +1086,6 @@ pub fn to_markdown_from_lines(lines: Vec<TextLine>, options: MarkdownOptions) ->
                    plain_text.clone()
                };
                output.push_str(&format!("{} {}\n\n", prefix, heading_text.trim()));
-                if is_toc_marker_heading(plain_trimmed) {
-                    toc_suppress_page = Some(line.page);
-                }
                in_list = false;
                continue;
            }
@@ -1239,42 +1214,6 @@ mod tests {
        }
    }

-    fn line_at(text: &str, page: u32, y: f32) -> TextLine {
-        let mut item = make_item(text, page, None);
-        item.y = y;
-        make_line(vec![item])
-    }
-
-    #[test]
-    fn isolated_lines_kept_on_sparse_pages() {
-        // A ToC page with a lone title and one entry far below: the density
-        // ratio is 50% but the page is too sparse for the multi-column
-        // misfire the guard targets — the title must stay isolated.
-        let lines = vec![
-            line_at("CONTENTS", 1, 700.0),
-            line_at("Chapter One 5", 1, 500.0),
-        ];
-        let isolated = find_isolated_lines(&lines, 12.0, 20.0);
-        assert!(
-            isolated.contains(&0),
-            "sparse-page title must stay isolated"
-        );
-    }
-
-    #[test]
-    fn isolated_lines_wiped_on_dense_pages() {
-        // 12 short lines all with paragraph gaps — the multi-column misfire
-        // shape. The guard must clear them all.
-        let lines: Vec<TextLine> = (0..12)
-            .map(|i| line_at("Short column line", 1, 700.0 - i as f32 * 50.0))
-            .collect();
-        let isolated = find_isolated_lines(&lines, 12.0, 20.0);
-        assert!(
-            isolated.is_empty(),
-            "dense page of isolated lines must be wiped"
-        );
-    }
-
    #[test]
    fn test_struct_role_heading() {
        let lines = vec![
@@ -3,7 +3,7 @@
 use std::collections::{HashMap, HashSet};

 use crate::structure_tree::StructRole;
-use crate::types::{TextItem, TextLine};
+use crate::types::TextLine;

 use super::analysis::detect_header_level;

@@ -87,41 +87,6 @@ pub(crate) fn merge_heading_lines(
            false
        };

-        // Bold headings at body font size never reach a tier, so wrapped ones
-        // split into two output headings ("…of wood pellets and cost" /
-        // "structure in Japan"). Merge a fully-bold line into the previous
-        // fully-bold line when it reads as a wrap continuation: starts
-        // lowercase, tiny Y gap, and the previous line has no terminal
-        // punctuation. Kept deliberately narrow — bold list labels and bold
-        // sentences start with markers or capitals and are unaffected.
-        let should_merge = should_merge
-            || if let Some(prev) = result.last() {
-                let all_bold = |l: &TextLine| {
-                    !l.items.is_empty() && l.items.iter().all(|i: &TextItem| i.is_bold)
-                };
-                let prev_text = prev.text();
-                let prev_trim = prev_text.trim_end();
-                let curr_text = line.text();
-                let curr_trim = curr_text.trim();
-                let y_gap = prev.y - line.y;
-                // Both lines must be tier-less: a tiered/tagged bold heading
-                // followed by bold body text must not absorb it.
-                line_level.is_none()
-                    && effective_heading_level(prev, base_size, heading_tiers, struct_roles)
-                        .is_none()
-                    && prev.page == line.page
-                    && all_bold(prev)
-                    && all_bold(&line)
-                    && y_gap > 0.0
-                    && y_gap < line_font * 1.6
-                    && curr_trim.chars().next().is_some_and(|c| c.is_lowercase())
-                    && !prev_trim.ends_with(['.', ':', ';', '!', '?'])
-                    && prev_trim.split_whitespace().count() + curr_trim.split_whitespace().count()
-                        <= 20
-            } else {
-                false
-            };
-
        if should_merge {
            // Append this line's items to the previous line
            let prev = result.last_mut().unwrap();
@@ -720,68 +685,4 @@ mod tests {
            .unwrap();
        assert_eq!(first_header.page, 1, "first occurrence should be on page 1");
    }
-
-    fn make_bold_line(text: &str, page: u32, y: f32) -> TextLine {
-        let mut item = make_item(text, 12.0, None);
-        item.is_bold = true;
-        TextLine {
-            items: vec![item],
-            y,
-            page,
-            adaptive_threshold: 0.10,
-        }
-    }
-
-    #[test]
-    fn merge_wrapped_bold_heading_lowercase_continuation() {
-        // Bold-at-body-size heading wrapped across two lines: the second line
-        // starts lowercase and must merge into the first.
-        let lines = vec![
-            make_bold_line(
-                "3. Perspective of supply and demand balance and cost",
-                1,
-                700.0,
-            ),
-            make_bold_line("structure in Japan", 1, 686.0),
-            make_line("Body text paragraph follows here.", 12.0, 1, 660.0, None),
-        ];
-        let result = merge_heading_lines(lines, 12.0, &[], None);
-        assert_eq!(result.len(), 2, "wrapped bold heading should merge");
-        assert!(result[0].text().contains("cost structure in Japan"));
-    }
-
-    #[test]
-    fn no_merge_for_bold_sentences_or_new_headings() {
-        // Second bold line starts with a capital — a new heading or label,
-        // not a wrap continuation.
-        let lines = vec![
-            make_bold_line("Replace", 1, 700.0),
-            make_bold_line("Trash", 1, 686.0),
-        ];
-        let result = merge_heading_lines(lines, 12.0, &[], None);
-        assert_eq!(result.len(), 2, "distinct bold lines must not merge");
-
-        // Previous line ends a sentence — continuation must not merge.
-        let lines = vec![
-            make_bold_line("This is a bold sentence.", 1, 700.0),
-            make_bold_line("another bold line", 1, 686.0),
-        ];
-        let result = merge_heading_lines(lines, 12.0, &[], None);
-        assert_eq!(result.len(), 2, "sentence-final bold line must not merge");
-    }
-
-    #[test]
-    fn tiered_bold_heading_does_not_absorb_bold_body() {
-        // Previous line is a tier-level bold heading (16pt vs 12pt body);
-        // a following lowercase bold body line must NOT merge into it.
-        let mut heading = make_bold_line("Section Title", 1, 700.0);
-        heading.items[0].font_size = 16.0;
-        heading.items[0].height = 16.0;
-        let lines = vec![
-            heading,
-            make_bold_line("emphasized body text continues here", 1, 686.0),
-        ];
-        let result = merge_heading_lines(lines, 12.0, &[16.0], None);
-        assert_eq!(result.len(), 2, "tiered heading must not absorb bold body");
-    }
 }
@@ -76,52 +76,6 @@ pub enum StructRole {
 }

 impl StructRole {
-    /// Content roles whose text must never be promoted to a heading by the
-    /// visual heuristic. These carry an explicit non-heading meaning in the
-    /// struct tree (lists, quotes, notes, references, captions, formulas,
-    /// forms, ToC entries), yet their text is often short and visually
-    /// isolated — exactly what the heuristic keys on. Heading roles (H, H1–H6)
-    /// and generic container/flow roles (P, Div, Sect, Span, …) are excluded
-    /// so the heuristic can still fire there.
-    ///
-    /// `Figure` is deliberately NOT in this set: cover/banner pages routinely
-    /// tag the document title inside a Figure (alongside a seal or logo), and
-    /// that title is a real heading. `Formula` and `Form` stay — a line
-    /// explicitly tagged as an equation or form field is never a heading.
-    ///
-    /// Table roles (Table/TR/TH/TD/THead/TBody/TFoot) are included so that
-    /// when table reconstruction falls back and cells reach the line loop as
-    /// plain text, a short isolated cell — a `TH` column header especially —
-    /// is not promoted to a heading.
-    pub(crate) fn is_non_heading_content(&self) -> bool {
-        matches!(
-            self,
-            Self::L
-                | Self::LI
-                | Self::Lbl
-                | Self::LBody
-                | Self::BlockQuote
-                | Self::Quote
-                | Self::Caption
-                | Self::TOC
-                | Self::TOCI
-                | Self::Index
-                | Self::Note
-                | Self::Reference
-                | Self::BibEntry
-                | Self::Code
-                | Self::Formula
-                | Self::Form
-                | Self::Table
-                | Self::TR
-                | Self::TH
-                | Self::TD
-                | Self::THead
-                | Self::TBody
-                | Self::TFoot
-        )
-    }
-
    fn from_name(name: &str) -> Self {
        match name {
            "Document" => Self::Document,
@@ -902,54 +856,6 @@ fn contains_bytes(haystack: &[u8], needle: &[u8]) -> bool {
 mod tests {
    use super::*;

-    #[test]
-    fn non_heading_content_roles() {
-        for r in [
-            StructRole::L,
-            StructRole::LI,
-            StructRole::BlockQuote,
-            StructRole::Quote,
-            StructRole::Caption,
-            StructRole::TOC,
-            StructRole::TOCI,
-            StructRole::Index,
-            StructRole::Note,
-            StructRole::Reference,
-            StructRole::BibEntry,
-            StructRole::Code,
-            StructRole::Formula,
-            StructRole::Form,
-            StructRole::Table,
-            StructRole::TR,
-            StructRole::TH,
-            StructRole::TD,
-            StructRole::THead,
-            StructRole::TBody,
-            StructRole::TFoot,
-        ] {
-            assert!(
-                r.is_non_heading_content(),
-                "{r:?} should block heading promotion"
-            );
-        }
-        // Heading and generic container/flow roles must NOT block promotion
-        for r in [
-            StructRole::H,
-            StructRole::H1,
-            StructRole::H3,
-            StructRole::P,
-            StructRole::Div,
-            StructRole::Sect,
-            StructRole::Span,
-            StructRole::Figure,
-        ] {
-            assert!(
-                !r.is_non_heading_content(),
-                "{r:?} should allow heading promotion"
-            );
-        }
-    }
-
    #[test]
    fn test_struct_role_from_name() {
        assert_eq!(StructRole::from_name("H1"), StructRole::H1);
@@ -1477,10 +1477,6 @@ fn detect_row_stripe_table(
        debug!("  row-stripe rejected: sparse outline/prose continuation shape");
        return None;
    }
-    if has_dominant_prose_cell(&cells) {
-        debug!("  row-stripe rejected: dominant prose cell (chart/figure region over body text)");
-        return None;
-    }

    let column_centers: Vec<f32> = (0..num_cols)
        .map(|c| (col_edges[c] + col_edges[c + 1]) / 2.0)
@@ -1499,34 +1495,6 @@ fn detect_row_stripe_table(
    Some(Table::new(column_centers, row_centers, cells, item_indices))
 }

-/// Detect a grid that swallowed body text instead of tabular data.
-///
-/// Charts (bar graphs, axis gridlines) emit fields of drawing rects that can
-/// pass the row-stripe shape test; the resulting "table" then captures the
-/// page's prose. The signature: one cell holds an entire paragraph — ≥60 words
-/// AND at least a third of all words in the table.
-///
-/// There is deliberately no row-count exemption. A small table whose single
-/// long cell dominates its word count is indistinguishable by content from a
-/// phantom grid over body text, and across the regression corpora every such
-/// grid observed has been swallowed prose, never a real note table. The costs
-/// are also asymmetric: rejecting a real table degrades it to readable prose,
-/// while accepting a phantom scrambles the page into Y-interleaved cells.
-/// Larger legitimate tables are safe because the one-third-of-total threshold
-/// scales with table size.
-fn has_dominant_prose_cell(cells: &[Vec<String>]) -> bool {
-    let mut total_words = 0usize;
-    let mut max_cell_words = 0usize;
-    for row in cells {
-        for cell in row {
-            let words = cell.split_whitespace().count();
-            total_words += words;
-            max_cell_words = max_cell_words.max(words);
-        }
-    }
-    max_cell_words >= 60 && max_cell_words * 3 >= total_words
-}
-
 fn row_stripe_is_sparse_prose_outline(cells: &[Vec<String>]) -> bool {
    let Some(num_cols) = cells.first().map(|row| row.len()) else {
        return false;
@@ -2326,12 +2294,6 @@ fn detect_merged_cluster_table(
        );
        return None;
    }
-    if has_dominant_prose_cell(&cells) {
-        debug!(
-            "  merged-cluster rejected: dominant prose cell (chart/figure region over body text)"
-        );
-        return None;
-    }

    // No empty columns
    for col in 0..num_cols {
@@ -2367,29 +2329,7 @@ fn detect_merged_cluster_table(
 /// suitable for rect-backed tables where we already know tabular structure exists
 /// (no need for anti-paragraph safeguards).
 fn cluster_x_positions(items: &[(usize, &TextItem)], min_threshold: f32) -> Vec<f32> {
-    // Column edges come from where text STARTS. An item whose left edge hugs
-    // the previous item's right edge on the same line is a continuation run
-    // (style boundary, script change, underline split) — feeding its x-start
-    // in here fabricates a phantom column mid-cell.
-    let mut sorted: Vec<&TextItem> = items.iter().map(|&(_, i)| i).collect();
-    sorted.sort_by(|a, b| a.y.total_cmp(&b.y).then(a.x.total_cmp(&b.x)));
-    let mut x_positions: Vec<f32> = Vec::with_capacity(sorted.len());
-    for (idx, item) in sorted.iter().enumerate() {
-        let is_continuation = idx > 0 && {
-            let prev = sorted[idx - 1];
-            // Style/underline splits leave runs that TOUCH (gap ~0); real
-            // cell boundaries in even the tightest tables keep a visible
-            // gap. 2pt separates the two without eating dense-table columns.
-            // The negative side is bounded too: text overhanging from an
-            // adjacent cell overlaps by far more than italic kerning ever
-            // does, and must still start its own column.
-            let gap = item.x - (prev.x + prev.width);
-            (prev.y - item.y).abs() <= 2.0 && gap < 2.0 && gap > -4.0 && item.x >= prev.x
-        };
-        if !is_continuation {
-            x_positions.push(item.x);
-        }
-    }
+    let mut x_positions: Vec<f32> = items.iter().map(|(_, i)| i.x).collect();
    x_positions.sort_by(|a, b| a.total_cmp(b));

    if x_positions.is_empty() {
@@ -2458,89 +2398,6 @@ mod tests {
        }
    }

-    // --- has_dominant_prose_cell ---
-
-    fn cells_of(rows: &[&[&str]]) -> Vec<Vec<String>> {
-        rows.iter()
-            .map(|r| r.iter().map(|c| c.to_string()).collect())
-            .collect()
-    }
-
-    #[test]
-    fn dominant_prose_cell_rejects_swallowed_paragraph() {
-        // Two cells hold paragraphs (the shape every observed phantom grid
-        // has: swallowed body text spans multiple cells), rest are chart labels
-        let para = ["word"; 70].join(" ");
-        let para2 = ["word"; 35].join(" ");
-        let cells = cells_of(&[
-            &[para.as_str(), "81", "76"],
-            &[para2.as_str(), "56", "9"],
-            &["2019", "2020", ""],
-        ]);
-        assert!(has_dominant_prose_cell(&cells));
-    }
-
-    #[test]
-    fn dominant_prose_cell_rejects_small_table_dominated_by_one_cell() {
-        // Boundary case, documented as INTENDED: a small grid whose single
-        // long cell dominates the word count is rejected even at 4+ rows.
-        // By content alone this shape is indistinguishable from a phantom
-        // grid over body text, and every observed instance in the regression
-        // corpora was swallowed prose (chart/figure regions), not a real
-        // note table. Rejection degrades gracefully — the text is still
-        // extracted as prose — while accepting a phantom scrambles reading
-        // order.
-        let note = ["word"; 70].join(" ");
-        let cells = cells_of(&[
-            &["Purpose", note.as_str()],
-            &["Owner", "Facilities team"],
-            &["Date", "2024-06-01"],
-            &["Status", "Active"],
-        ]);
-        assert!(has_dominant_prose_cell(&cells));
-    }
-
-    #[test]
-    fn dominant_prose_cell_allows_description_column() {
-        // Long-ish description cells, but text is spread across the table
-        let desc = ["word"; 25].join(" ");
-        let cells = cells_of(&[
-            &["Item A", desc.as_str(), "100"],
-            &["Item B", desc.as_str(), "200"],
-            &["Item C", desc.as_str(), "300"],
-            &["Item D", desc.as_str(), "400"],
-        ]);
-        assert!(!has_dominant_prose_cell(&cells));
-    }
-
-    #[test]
-    fn dominant_prose_cell_allows_short_tables() {
-        let cells = cells_of(&[&["Name", "Value"], &["Total", "42"]]);
-        assert!(!has_dominant_prose_cell(&cells));
-    }
-
-    #[test]
-    fn dominant_prose_cell_allows_data_table_with_long_note() {
-        // A real 4+ row table with one verbose remark cell: the note is ≥60
-        // words but the table's other content carries more than 2× its word
-        // count, so concentration stays below the 1/3 threshold. The
-        // denominator scales with table size — this is what keeps large
-        // legitimate tables safe where a bare length cap would not.
-        let note = ["word"; 60].join(" ");
-        let row_text = ["data"; 12].join(" ");
-        let mut rows: Vec<Vec<String>> = (0..11)
-            .map(|i| {
-                vec![
-                    format!("Item {i}"),
-                    row_text.clone(),
-                    format!("{}", i * 100),
-                ]
-            })
-            .collect();
-        rows.push(vec!["Note".into(), note, String::new()]);
-        assert!(!has_dominant_prose_cell(&rows));
-    }
-
    // --- rects_overlap ---

    #[test]
@@ -1,520 +0,0 @@
-//! Text-quality detection: deciding when an extracted text layer is too broken
-//! to serve and a page should fall back to OCR.
-//!
-//! Extraction can produce plausible-looking bytes that are actually garbage —
-//! failed CID→Unicode mappings, broken ToUnicode CMaps, mojibake. These
-//! detectors catch that and let callers set `needs_ocr`. They come in two
-//! layers, sharing the same primitives:
-//!
-//! - **Markdown-level** ([`detect_encoding_issues`], [`is_garbage_text`],
-//!   [`is_cid_garbage`]) run on a page's final markdown string. Used as a
-//!   backstop on the region-extraction and whole-document paths.
-//! - **Item/span-level** ([`analyze_text_quality`],
-//!   [`region_items_have_decoding_issue`]) run on individual `TextItem`s and
-//!   accumulate per-page evidence, so localized garbled spans on an otherwise
-//!   clean page are caught without a single span having to condemn the page.
-//!
-//! Detection classes, roughly by signal:
-//! - **Replacement runs**: U+FFFD clusters ([`has_replacement_text_run`]).
-//! - **Private-use / C1-control runs**: CID passthrough landing in PUA or the
-//!   C1 block ([`has_private_use_text_run`], [`has_cid_control_token`]).
-//! - **Dollar-as-space**: `Word$Word$Word` from broken CMaps
-//!   ([`has_dollar_as_space_pattern`]).
-//! - **Non-alphanumeric dominance**: symbol soup ([`is_garbage_text`]).
-//! - **Substitution-cipher letter statistics**: pure-ASCII output whose letter
-//!   distribution is a permutation of natural language ([`CipherGarbleStats`]).
-
-use crate::types::TextItem;
-use crate::{add_ocr_reason, OCR_REASON_SUSPECTED_GARBLED_TEXT};
-use std::collections::BTreeMap;
-
-/// Detect broken font encodings in extracted markdown text.
-///
-/// Two heuristics:
-/// 1. **U+FFFD**: Any replacement character indicates decode failures.
-/// 2. **Dollar-as-space**: Pattern like `Word$Word$Word` where `$` is used as a
-///    word separator due to broken ToUnicode CMaps. Triggers when either:
-///    - More than 50% of `$` are between letters (clear substitution pattern), OR
-///    - More than 20 letter-dollar-letter occurrences (even if some `$` are also
-///      used as trailing/leading separators, 20+ is far beyond normal financial text).
-pub(crate) fn detect_encoding_issues(markdown: &str) -> bool {
-    // Heuristic 1: U+FFFD replacement characters
-    if markdown.contains('\u{FFFD}') {
-        return true;
-    }
-
-    // Heuristic 2: dollar-as-space pattern
-    if has_dollar_as_space_pattern(markdown) {
-        return true;
-    }
-
-    // Heuristic 3: substitution-cipher letter statistics (broken ToUnicode)
-    let mut stats = CipherGarbleStats::default();
-    stats.add_text(markdown);
-    stats.looks_garbled()
-}
-
-fn has_dollar_as_space_pattern(markdown: &str) -> bool {
-    let total_dollars = markdown.matches('$').count();
-    if total_dollars > 10 {
-        let bytes = markdown.as_bytes();
-        let mut letter_dollar_letter = 0usize;
-        for i in 1..bytes.len().saturating_sub(1) {
-            if bytes[i] == b'$'
-                && bytes[i - 1].is_ascii_alphabetic()
-                && bytes[i + 1].is_ascii_alphabetic()
-            {
-                letter_dollar_letter += 1;
-            }
-        }
-        if letter_dollar_letter > 20 || letter_dollar_letter * 2 > total_dollars {
-            return true;
-        }
-    }
-
-    false
-}
-
-/// English letter frequencies (percent, a–z). Used as a natural-language
-/// reference: every Latin-script language in the eval corpus (Swedish,
-/// Finnish, Turkish, German, romaji) scores ≥ 0.80 cosine similarity against
-/// it, while substitution-cipher text scores ~0.53.
-const ENGLISH_LETTER_FREQ: [f64; 26] = [
-    8.2, 1.5, 2.8, 4.3, 12.7, 2.2, 2.0, 6.1, 7.0, 0.15, 0.8, 4.0, 2.4, 6.7, 7.5, 1.9, 0.1, 6.0,
-    6.3, 9.1, 2.8, 1.0, 2.4, 0.15, 2.0, 0.07,
-];
-
-/// Letter statistics for detecting substitution-cipher garbling: broken
-/// ToUnicode CMaps that shift every character by a per-range constant (e.g.
-/// `Certificate` extracted as `8VceZWZTReV`). Such text is 100% printable
-/// ASCII with word-like token lengths, so it defeats `is_garbage_text` and
-/// produces no replacement characters — it needs its own discriminator.
-#[derive(Debug, Default)]
-struct CipherGarbleStats {
-    /// Case-folded ASCII letter histogram.
-    letter_counts: [u32; 26],
-    ascii_letters: usize,
-    ascii_vowels: usize,
-    /// Accented Latin letters (Latin-1 Supplement through Latin Extended-B,
-    /// plus Latin Extended Additional). Count toward Latin dominance only.
-    latin_ext_letters: usize,
-    non_latin_letters: usize,
-    /// Adjacent ASCII-letter pairs, and how many of them switch from
-    /// lowercase straight to uppercase mid-word.
-    letter_bigrams: usize,
-    case_shift_bigrams: usize,
-}
-
-impl CipherGarbleStats {
-    fn add_text(&mut self, text: &str) {
-        let mut prev: Option<char> = None;
-        for ch in text.chars() {
-            if ch.is_ascii_alphabetic() {
-                let idx = (ch.to_ascii_lowercase() as u8 - b'a') as usize;
-                self.letter_counts[idx] += 1;
-                self.ascii_letters += 1;
-                if matches!(ch.to_ascii_lowercase(), 'a' | 'e' | 'i' | 'o' | 'u') {
-                    self.ascii_vowels += 1;
-                }
-                if let Some(p) = prev {
-                    self.letter_bigrams += 1;
-                    if p.is_ascii_lowercase() && ch.is_ascii_uppercase() {
-                        self.case_shift_bigrams += 1;
-                    }
-                }
-                prev = Some(ch);
-            } else {
-                if ch.is_alphabetic() {
-                    if matches!(ch as u32, 0xC0..=0x24F | 0x1E00..=0x1EFF) {
-                        self.latin_ext_letters += 1;
-                    } else {
-                        self.non_latin_letters += 1;
-                    }
-                }
-                prev = None;
-            }
-        }
-    }
-
-    /// Cosine similarity between the observed letter histogram and English
-    /// letter frequencies. A shifted alphabet permutes the histogram, which
-    /// destroys the similarity regardless of the shift amount.
-    fn english_cosine(&self) -> f64 {
-        if self.ascii_letters == 0 {
-            return 1.0;
-        }
-        let n = self.ascii_letters as f64;
-        let mut dot = 0.0;
-        let mut norm_obs = 0.0;
-        for (count, freq) in self.letter_counts.iter().zip(ENGLISH_LETTER_FREQ) {
-            let p = *count as f64 / n;
-            dot += p * freq;
-            norm_obs += p * p;
-        }
-        let norm_en = ENGLISH_LETTER_FREQ
-            .iter()
-            .map(|f| f * f)
-            .sum::<f64>()
-            .sqrt();
-        dot / (norm_obs.sqrt() * norm_en)
-    }
-
-    /// Cosine similarity between the observed histogram and English
-    /// frequencies after sorting BOTH descending — i.e. comparing the *shape*
-    /// of the frequency profile, ignoring which letter sits where. A
-    /// substitution cipher is a bijection, so it preserves this shape exactly
-    /// (att10k 0.97, arbitrary shifts 0.99) regardless of case or offset.
-    /// Non-linguistic ASCII has a different profile: a small alphabet is far
-    /// steeper (random DNA 0.74, hex dumps 0.81), so the shape diverges.
-    fn english_shape_cosine(&self) -> f64 {
-        if self.ascii_letters == 0 {
-            return 1.0;
-        }
-        let n = self.ascii_letters as f64;
-        let mut obs: [f64; 26] = std::array::from_fn(|i| self.letter_counts[i] as f64 / n);
-        obs.sort_unstable_by(|a, b| b.total_cmp(a));
-        let mut en = ENGLISH_LETTER_FREQ;
-        en.sort_unstable_by(|a, b| b.total_cmp(a));
-
-        let dot: f64 = obs.iter().zip(en).map(|(o, e)| o * e).sum();
-        let norm_obs = obs.iter().map(|o| o * o).sum::<f64>().sqrt();
-        let norm_en = en.iter().map(|e| e * e).sum::<f64>().sqrt();
-        dot / (norm_obs * norm_en)
-    }
-
-    /// Thresholds validated against the 380-document pdf-evals snapshot
-    /// corpus (0 false positives) and the garbled ParseBench `att10k` page
-    /// (vowel ratio 0.245, case-shift rate 0.225, cosine 0.532). Closest
-    /// legitimate document on each axis: vowel ratio 0.264 (circuit
-    /// schematic), case-shift rate 0.021, cosine 0.801.
-    fn looks_garbled(&self) -> bool {
-        // Need a statistically meaningful, Latin-dominant sample.
-        if self.ascii_letters < 200
-            || self.non_latin_letters > self.ascii_letters + self.latin_ext_letters
-        {
-            return false;
-        }
-
-        // Real Latin-script text keeps vowels above ~30% of letters even in
-        // acronym- and part-number-heavy documents; shifted text starves them.
-        let vowel_ratio = self.ascii_vowels as f64 / self.ascii_letters as f64;
-        if vowel_ratio > 0.30 {
-            return false;
-        }
-
-        // Signal 1: lowercase→uppercase transitions inside words. A shifted
-        // lowercase alphabet straddles the ASCII uppercase block ('i'→'Z',
-        // 't'→'e'), so garbled words flip case constantly. Real documents
-        // stay ≤ 0.02 even with camelCase identifiers.
-        let case_shifts = self.letter_bigrams >= 100
-            && self.case_shift_bigrams as f64 >= self.letter_bigrams as f64 * 0.10;
-
-        // Signal 2: the histogram is a permutation of natural language — an
-        // English-like frequency SHAPE (sorted cosine high) but with letters
-        // in the wrong POSITIONS (unsorted cosine low). This is the signature
-        // of a substitution cipher and is case-independent, so it catches
-        // all-lowercase and all-uppercase shifts as well as case-straddling
-        // ones. Genuinely non-linguistic ASCII that is merely "unlike English"
-        // fails one of the two halves: DNA/hex dumps have too steep a profile
-        // (shape cosine < 0.90), while protein sequences, ticker symbols and
-        // base64 are not sufficiently unlike English in position (unsorted
-        // cosine ≥ 0.60) — so none of them are routed to OCR.
-        let permuted_language = self.english_cosine() < 0.60 && self.english_shape_cosine() >= 0.90;
-
-        case_shifts || permuted_language
-    }
-}
-
-#[derive(Debug, Default)]
-pub(crate) struct TextQualityReport {
-    pub(crate) pages_needing_ocr: Vec<u32>,
-    pub(crate) has_encoding_issues: bool,
-    pub(crate) reasons_by_page: BTreeMap<u32, Vec<String>>,
-}
-
-#[derive(Debug, Default)]
-struct PageTextQualityEvidence {
-    chars: usize,
-    replacement_chars: usize,
-    replacement_spans: usize,
-    longest_replacement_run: usize,
-    cipher_garble: CipherGarbleStats,
-}
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
-enum TextSpanIssueKind {
-    Replacement,
-    Strong,
-}
-
-pub(crate) fn analyze_text_quality(items: &[TextItem]) -> TextQualityReport {
-    let mut reasons_by_page = BTreeMap::new();
-    let mut evidence_by_page = BTreeMap::<u32, PageTextQualityEvidence>::new();
-
-    for item in items {
-        if !matches!(item.item_type, crate::types::ItemType::Text) {
-            continue;
-        }
-
-        let evidence = evidence_by_page.entry(item.page).or_default();
-        evidence.chars += item.text.chars().filter(|ch| !ch.is_whitespace()).count();
-        evidence.cipher_garble.add_text(&item.text);
-
-        match text_span_decoding_issue_kind(&item.text) {
-            Some(TextSpanIssueKind::Strong) => {
-                add_ocr_reason(
-                    &mut reasons_by_page,
-                    item.page,
-                    OCR_REASON_SUSPECTED_GARBLED_TEXT,
-                );
-            }
-            Some(TextSpanIssueKind::Replacement) => {
-                let stats = replacement_text_stats(&item.text);
-                evidence.replacement_chars += stats.0;
-                evidence.replacement_spans += 1;
-                evidence.longest_replacement_run = evidence.longest_replacement_run.max(stats.1);
-            }
-            None => {}
-        }
-    }
-
-    for (page, evidence) in evidence_by_page {
-        if reasons_by_page.contains_key(&page) {
-            continue;
-        }
-        if page_replacement_evidence_needs_ocr(&evidence) || evidence.cipher_garble.looks_garbled()
-        {
-            add_ocr_reason(
-                &mut reasons_by_page,
-                page,
-                OCR_REASON_SUSPECTED_GARBLED_TEXT,
-            );
-        }
-    }
-
-    let pages_needing_ocr: Vec<u32> = reasons_by_page.keys().copied().collect();
-    TextQualityReport {
-        has_encoding_issues: !pages_needing_ocr.is_empty(),
-        pages_needing_ocr,
-        reasons_by_page,
-    }
-}
-
-pub(crate) fn region_items_have_decoding_issue(items: &[TextItem]) -> bool {
-    items.iter().any(|item| {
-        matches!(item.item_type, crate::types::ItemType::Text)
-            && text_span_has_decoding_issue(&item.text)
-    })
-}
-
-fn text_span_has_decoding_issue(text: &str) -> bool {
-    text_span_decoding_issue_kind(text).is_some()
-}
-
-fn text_span_decoding_issue_kind(text: &str) -> Option<TextSpanIssueKind> {
-    let text = text.trim();
-    if text.is_empty() {
-        return None;
-    }
-
-    if has_dollar_as_space_pattern(text)
-        || has_private_use_text_run(text)
-        || is_cid_garbage(text)
-        || has_cid_control_token(text)
-    {
-        return Some(TextSpanIssueKind::Strong);
-    }
-
-    if has_replacement_text_run(text) {
-        return Some(TextSpanIssueKind::Replacement);
-    }
-
-    None
-}
-
-fn replacement_text_stats(text: &str) -> (usize, usize) {
-    let mut replacement = 0usize;
-    let mut current_run = 0usize;
-    let mut longest_run = 0usize;
-
-    for ch in text.chars() {
-        if ch == '\u{FFFD}' {
-            replacement += 1;
-            current_run += 1;
-            longest_run = longest_run.max(current_run);
-        } else {
-            current_run = 0;
-        }
-    }
-
-    (replacement, longest_run)
-}
-
-fn page_replacement_evidence_needs_ocr(evidence: &PageTextQualityEvidence) -> bool {
-    if evidence.replacement_chars == 0 || evidence.chars == 0 {
-        return false;
-    }
-
-    // If the entire page is only a short broken text layer, even a short
-    // replacement run is enough evidence. On otherwise text-heavy pages,
-    // require density so math formulas do not force full-page OCR.
-    if evidence.chars <= 80 && evidence.longest_replacement_run >= 2 {
-        return true;
-    }
-
-    let replacement_density_bps = evidence.replacement_chars * 10_000 / evidence.chars;
-    let enough_bad_text = evidence.replacement_chars >= 12 && replacement_density_bps >= 500;
-    let repeated_bad_spans = evidence.replacement_spans >= 3 && replacement_density_bps >= 250;
-    let long_bad_run = evidence.longest_replacement_run >= 8 && replacement_density_bps >= 250;
-
-    enough_bad_text || repeated_bad_spans || long_bad_run
-}
-
-fn has_replacement_text_run(text: &str) -> bool {
-    let (replacement, longest_run) = replacement_text_stats(text);
-    longest_run >= 2 || replacement >= 3
-}
-
-fn has_private_use_text_run(text: &str) -> bool {
-    let mut total = 0usize;
-    let mut private_use = 0usize;
-    let mut current_run = 0usize;
-    let mut longest_run = 0usize;
-
-    for ch in text.chars() {
-        if ch.is_whitespace() {
-            current_run = 0;
-            continue;
-        }
-        total += 1;
-        if is_private_use_char(ch) {
-            private_use += 1;
-            current_run += 1;
-            longest_run = longest_run.max(current_run);
-        } else {
-            current_run = 0;
-        }
-    }
-
-    if private_use == 0 {
-        return false;
-    }
-
-    longest_run >= 3 || (total >= 5 && private_use >= 2 && private_use * 2 >= total)
-}
-
-fn has_cid_control_token(text: &str) -> bool {
-    text.split_whitespace().any(token_has_cid_control)
-}
-
-fn token_has_cid_control(token: &str) -> bool {
-    let mut total = 0usize;
-    let mut c1_control = 0usize;
-
-    for ch in token.chars() {
-        total += 1;
-        if ('\u{0080}'..='\u{009F}').contains(&ch) {
-            c1_control += 1;
-        }
-    }
-
-    total >= 5 && c1_control >= 2 && c1_control * 20 >= total
-}
-
-fn is_private_use_char(ch: char) -> bool {
-    matches!(
-        ch as u32,
-        0xE000..=0xF8FF | 0xF0000..=0xFFFFD | 0x100000..=0x10FFFD
-    )
-}
-
-/// Check if extracted text is predominantly garbage (non-alphanumeric).
-///
-/// Broken font encodings produce text like "----1-.-.-.___  --.-. .._ I_---."
-/// where most characters are punctuation/symbols. Real text in any language
-/// has >50% alphanumeric characters.
-pub(crate) fn is_garbage_text(markdown: &str) -> bool {
-    let mut alphanum = 0usize;
-    let mut non_alphanum = 0usize;
-
-    let chars: Vec<char> = markdown.chars().collect();
-    let mut i = 0usize;
-    while i < chars.len() {
-        let ch = chars[i];
-        let mut run_end = i + 1;
-        while run_end < chars.len() && chars[run_end] == ch {
-            run_end += 1;
-        }
-
-        let is_decorative_leader = matches!(ch, '.' | '_' | '·') && run_end - i >= 3;
-        if !is_decorative_leader {
-            for &run_ch in &chars[i..run_end] {
-                if run_ch.is_whitespace() {
-                    continue;
-                }
-                // Skip markdown syntax chars that we add (not from the PDF)
-                if matches!(run_ch, '#' | '*' | '|' | '-' | '\n') {
-                    continue;
-                }
-                if run_ch.is_alphanumeric() {
-                    alphanum += 1;
-                } else {
-                    non_alphanum += 1;
-                }
-            }
-        }
-        i = run_end;
-    }
-
-    let total = alphanum + non_alphanum;
-    total >= 50 && alphanum * 2 < total
-}
-
-/// Detect garbage from failed CID-to-Unicode mapping on Identity-H fonts.
-///
-/// When CID values don't correspond to Unicode codepoints, the raw bytes often
-/// produce characters in the C1 control range (U+0080–U+009F) or Private Use
-/// Area, mixed with random Latin Extended characters.  Valid text in any
-/// language almost never contains C1 controls.  We also fall back to the
-/// general `is_garbage_text` check for non-alphanumeric-heavy patterns.
-pub(crate) fn is_cid_garbage(text: &str) -> bool {
-    if is_garbage_text(text) {
-        return true;
-    }
-    let mut total = 0usize;
-    let mut c1_control = 0usize;
-    let mut high_latin = 0usize;
-    for ch in text.chars() {
-        if ch.is_whitespace() {
-            continue;
-        }
-        total += 1;
-        // C1 control characters (U+0080–U+009F) — almost never in real text
-        if ch == '·' {
-            continue;
-        }
-        if ('\u{0080}'..='\u{009F}').contains(&ch) {
-            c1_control += 1;
-        }
-        // High Latin-1 (U+00A0–U+00FF) — legitimate in Western European text
-        // but when combined with ASCII in CID passthrough, indicates mojibake
-        // from CID values being misinterpreted as Latin-1 characters.
-        if ('\u{00A0}'..='\u{00FF}').contains(&ch) {
-            high_latin += 1;
-        }
-    }
-    if total < 5 {
-        return false;
-    }
-    // If ≥5% of non-whitespace chars are C1 controls, it's garbage
-    if c1_control >= 2 && c1_control * 20 >= total {
-        return true;
-    }
-    // If ≥40% of non-whitespace chars are high Latin-1 AND the text has few
-    // ASCII letters, it's likely CID-as-Latin-1 mojibake (Japanese/CJK PDFs
-    // where CID values 0x80-0xFF become accented Latin characters).  Keep a
-    // minimum length so short math tokens like "2×()×" do not route a clean
-    // page to OCR.
-    let ascii_letters = text.chars().filter(|c| c.is_ascii_alphabetic()).count();
-    total >= 20 && high_latin * 5 >= total * 2 && ascii_letters * 3 < total
-}
@@ -93,9 +93,6 @@ pub fn is_bold_font(font_name: &str) -> bool {
        || lower.contains("extrabold")
        || lower.contains("ultrabold")
        || lower.contains("medium") && !lower.contains("mediumitalic") // Some fonts use Medium for semi-bold
-        // URW Type 1 fonts abbreviate Medium as "Medi" (e.g. NimbusRomNo9L-Medi,
-        // the Times-Bold substitute in LaTeX documents; -MediItal is bold italic).
-        || lower.contains("-medi") && !lower.contains("mediumital")
 }

 /// Detect if a font name indicates italic/oblique style
@@ -765,17 +762,6 @@ mod tests {
    use super::*;
    use crate::types::ItemType;

-    #[test]
-    fn bold_font_urw_medi_abbreviation() {
-        // URW Type 1 fonts (LaTeX default Times) abbreviate Medium as "Medi"
-        assert!(is_bold_font("NROFIU+NimbusRomNo9L-Medi"));
-        assert!(is_bold_font("NimbusRomNo9L-MediItal"));
-        assert!(!is_bold_font("DSSZWN+NimbusRomNo9L-Regu"));
-        assert!(!is_bold_font("NimbusRomNo9L-ReguItal"));
-        // Medium-Italic exclusion still holds
-        assert!(!is_bold_font("Foo-MediumItalic"));
-    }
-
    #[test]
    fn strip_soft_hyphen() {
        assert_eq!(expand_ligatures("con\u{00AD}tent"), "content");
@@ -1102,7 +1102,6 @@ fn test_pages_needing_ocr_field_accessible() {
        title: None,
        ocr_recommended: false,
        pages_needing_ocr: Vec::new(),
-        ocr_reasons_by_page: std::collections::BTreeMap::new(),
    };
    assert!(detection_result.pages_needing_ocr.is_empty());

@@ -3607,45 +3606,3 @@ fn test_markdown_options_default_has_include_images_false() {
    let opts = MarkdownOptions::default();
    assert!(!opts.include_images);
 }
-
-#[test]
-fn encrypted_pdf_decrypts_with_correct_password() {
-    let path = "tests/fixtures/encrypted-secret123.pdf";
-
-    // No password: the file is encrypted and can't be read.
-    let no_pw = process_pdf_with_options(path, PdfOptions::new());
-    assert!(
-        matches!(no_pw, Err(PdfError::Encrypted)),
-        "expected Encrypted without a password, got {no_pw:?}"
-    );
-
-    // Wrong password: still rejected.
-    let wrong = process_pdf_with_options(path, PdfOptions::new().password("wrong"));
-    assert!(
-        matches!(wrong, Err(PdfError::Encrypted)),
-        "expected Encrypted with a wrong password, got {wrong:?}"
-    );
-
-    // Correct password: decrypts and extracts real content.
-    let ok = process_pdf_with_options(path, PdfOptions::new().password("secret123"))
-        .expect("correct password should decrypt");
-    let md = ok.markdown.unwrap_or_default();
-    // Assert a stable fixture token so a garbled-but-long extraction (the
-    // encrypted-stream regression this guards) still fails the test.
-    assert!(
-        md.contains("Procurement"),
-        "decrypted markdown should contain the fixture's real text, got {} chars",
-        md.len()
-    );
-}
-
-#[test]
-fn pdf_options_debug_redacts_password() {
-    let opts = PdfOptions::new().password("secret123");
-    let dbg = format!("{opts:?}");
-    assert!(
-        !dbg.contains("secret123"),
-        "password leaked in Debug: {dbg}"
-    );
-    assert!(dbg.contains("REDACTED"), "expected redaction marker: {dbg}");
-}
@@ -48,7 +48,7 @@ tips of directly from customers received other employees paid tips rec’d. entr

 **Page 3**

-27 28 29 30 31 **Subtotals from pages** **1, 2, and 3** **Totals**
+27 28 29 30 31 **Subtotals** **from pages** **1, 2, and 3** **Totals**

 **1.** Report total cash tips (col. **a**) on Form 4070, line **1.**
 **2.** Report total credit card tips (col. **b**) on Form 4070, line **2.**
@@ -72,11 +72,11 @@ Month or shorter period in which tips were received **4** Net tips (lines **1 +

 forms simpler, we would be happy to hear from you. You can write to the Tax Forms Committee, Western Area Distribution Center, Rancho Cordova, CA 95743-0001. **Purpose.—**Use this form to report tips you receive to your employer. This includes cash tips, tips you receive from other employees, and credit card tips. You must report tips every month regardless of your total wages and tips for the year. However, you do not have to report tips to your employer for any month you received less than $20 in tips while working for that employer. Report tips by the 10th day of the month following the month that you receive them. If the 10th day is a Saturday, Sunday, or legal holiday, report tips by the next day that is not a Saturday, Sunday, or legal holiday. See **Pub. 531**, Reporting Tip Income, for more information. You can get additional copies of **Pub. 1244**, Employee’s Daily Record of Tips and Report to Employer, which contains both Forms 4070A and 4070, by calling 1-800-TAX-FORM (1-800-829-3676).

-<u>Instructions (continued)</u>
+**Instructions** *(continued)*

 **Unreported Tips.—**If you received tips of $20 or more for any month while working for one employer but did not report them to your employer, you must figure and pay social security and Medicare taxes on the unreported tips when you file your tax return. If you have unreported tips, you **must** use Form 1040 and **Form 4137,** Social Security and Medicare Tax on Unreported Tip Income, to report them. You may **not** use Form 1040A or 1040EZ. Employees subject to the Railroad Retirement Tax Act **cannot** use Form 4137 to pay railroad retirement tax on unreported tips. To get railroad retirement credit, you must report tips to your employer. If you do not report tips to your employer as required, you may be charged a penalty of 50% of the social security and Medicare taxes (or railroad retirement tax) due on the unreported tips unless there was reasonable cause for not reporting them. **Additional Information.—**Get **Pub. 531,** Reporting Tip Income, and Form 4137 for more information on tips. If you are an employee of certain large food or beverage establishments, see Pub. 531 for tip allocation rules. **Recordkeeping.—**If you do not keep a daily record of tips, you must keep other reliable proof of the tip income you received. This proof includes copies of restaurant bills and credit card charges that show amounts customers added as tips. Keep your tip income records for as long as the information on them may be needed in the administration of any Internal Revenue law.

-### Instructions (continued)
+**Instructions** *(continued)*

 Use this space to total your tips for the year

@@ -1,8 +1,17 @@
-||||(e) [Reserved]. For further guidance, see §1.1563-3T(e)(1). Par. 50. Section 1.1563-3T is added to read as follows: §1.1563-3T Rules for determining stock ownership (temporary). (a) through (d)(2)(iii) [Reserved]. For further guidance, see §1.1563-3(a)|
-|---|---|---|---|
-||through (d)(2)(iii).|||
-||(iv)|Statement|. If the application of paragraph (d)(2)(ii) or (iii) of §1.1563-3 does not result in a corporation being treated as a component member of only one controlled group of corporations on a December 31, then such corporation will be treated as a component member of only one such group on such date. Such corporation may elect the group in which it is to be included by including on or with its income tax return a statement entitled, “STATEMENT TO ELECT CONTROLLED GROUP PURSUANT TO §1.1563-3T(d)(2)(iv).” The statement must include-- (A) A description of each of the controlled groups in which the corporation could be included. The description must include the name and employer identification number of each component member of each such group and the stock ownership of the component members of each such group; and (B) The following representation: [INSERT NAME AND EMPLOYER IDENTIFICATION NUMBER OF CORPORATION] ELECTS TO BE TREATED AS A COMPONENT MEMBER OF THE [INSERT DESIGNATION OF GROUP].|
-||(v)|Election|-- (A) Election filed. An election filed under paragraph (d)(2)(iv) of this section is irrevocable and effective until paragraph (d)(2)(ii) or (iii) of §1.1563-3 applies or until a change in the stock ownership of the corporation results in|
+(e) [Reserved]. For further guidance, see §1.1563-3T(e)(1). Par. 50. Section 1.1563-3T is added to read as follows:
+<u>§1.1563-3T Rules for determining stock ownership (temporary)</u>.
+
+(a) through (d)(2)(iii) [Reserved]. For further guidance, see §1.1563-3(a)
+through (d)(2)(iii). (iv) <u>Statement</u>. If the application of paragraph (d)(2)(ii) or (iii) of §1.1563-3 does not result in a corporation being treated as a component member of only one controlled group of corporations on a December 31, then such corporation will be treated as a component member of only one such group on such date. Such corporation may elect the group in which it is to be included by including on or with its income tax return a statement entitled, “STATEMENT TO ELECT CONTROLLED GROUP PURSUANT TO §1.1563-3T(d)(2)(iv).” The statement must include--
+
+(A) A description of each of the controlled groups in which the corporation
+could be included. The description must include the name and employer identification number of each component member of each such group and the stock ownership of the component members of each such group; and
+
+(B) The following representation: [INSERT NAME AND EMPLOYER
+IDENTIFICATION NUMBER OF CORPORATION] ELECTS TO BE TREATED AS A COMPONENT MEMBER OF THE [INSERT DESIGNATION OF GROUP].
+
+(v) <u>Election</u>-- (A) <u>Election filed</u>. An election filed under paragraph (d)(2)(iv) of
+this section is irrevocable and effective until paragraph (d)(2)(ii) or (iii) of §1.1563-3 applies or until a change in the stock ownership of the corporation results in

 |termination of membership in the controlled group in which such corporation has||
 |---|---|
@@ -20,7 +29,7 @@
 Federal income tax return (including any amended return filed on or before the due date (including extensions) of such original return) timely filed on or after May 30,

 2006.
-(2) <u>Expiration date</u>. The applicability of this section will expire on May 26,
+(2) Expiration date. The applicability of this section will expire on May 26,
 2009. Par. 51. Section 1.6012-2 is amended by revising paragraph (c) and adding paragraph (k) to read as follows: <u>§1.6012-2 Corporations required to make returns of income</u>.
 * * * * *
 (c) [Reserved]. For further guidance, see §1.6012-2T(c).
@@ -43,9 +43,9 @@ l

 **Freon** **®** **12 Saturation Properties-Temperature Table**

-|Temp|Pressure||Volume||Density||Enthalpy|||Entropy|Temp|
-|---|---|---|---|---|---|---|---|---|---|---|---|
-|°C|[kPa]|[m³ Liquid v f|/kg] Vapour v g|[kg/m³ Liquid d f|] Vapour d g|Liquid H f|[kJ/kg] Latent H fg|Vapour H g|Liquid S f|[kJ/K-kg] Vapour S g|°C|
+|Temp|Pressure||Volume|||Density||Enthalpy|||Entropy|Temp|
+|---|---|---|---|---|---|---|---|---|---|---|---|---|
+|°C|[kPa]|[m³ Liquid v f|/kg]|Vapour v g|Liquid d f|[kg/m³] Vapour d g|Liquid H f|[kJ/kg] Latent H fg|Vapour H g|Liquid S f|[kJ/K-kg] Vapour S g|°C|

 |-100|1.2|0.0006|10.0000|1679.0|0.100|113.3|192.8|306.1|0.6077|1.7210|-100|
 |---|---|---|---|---|---|---|---|---|---|---|---|