fix(extractor): require a usable ToUnicode mapping to clear the gid flag

A mapping to U+FFFD (or an empty string) is rejected by extraction as
an invalid CMap result, so it must not count as decodable when deciding
whether gid-named Differences codes are resolvable.

.github/workflows/ci.yml

@@ -42,6 +42,9 @@ jobs:
       - name: Check formatting
         run: cargo fmt --all -- --check
 
+      - name: Check WASM formatting
+        run: cargo fmt --manifest-path wasm/Cargo.toml -- --check
+
   clippy:
     name: Clippy
     runs-on: ubuntu-latest
@@ -80,3 +83,37 @@ jobs:
 
       - name: Build
         run: cargo build --release --verbose
+
+  wasm:
+    name: WebAssembly
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install Rust
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: wasm32-unknown-unknown
+          components: clippy
+
+      - name: Cache cargo
+        uses: Swatinem/rust-cache@v2
+        with:
+          workspaces: |
+            wasm -> target
+          key: wasm
+
+      - name: Check WebAssembly bindings
+        run: cargo check --manifest-path wasm/Cargo.toml --target wasm32-unknown-unknown
+
+      - name: Check root package for WebAssembly
+        run: cargo check --target wasm32-unknown-unknown
+
+      - name: Lint WebAssembly bindings
+        run: cargo clippy --manifest-path wasm/Cargo.toml --target wasm32-unknown-unknown -- -D warnings
+
+      - name: Install wasm-pack
+        run: cargo install wasm-pack --version 0.15.0 --locked
+
+      - name: Test WebAssembly package
+        run: wasm-pack test --node --release wasm

.github/workflows/publish-wasm.yml

@@ -0,0 +1,111 @@
+name: Publish WebAssembly package
+
+on:
+  push:
+    branches: [main]
+    paths: ['wasm/Cargo.toml']
+  workflow_dispatch:
+
+permissions:
+  contents: read
+  id-token: write
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  check-version:
+    name: Check version change
+    if: github.ref == 'refs/heads/main'
+    runs-on: ubuntu-latest
+    outputs:
+      changed: ${{ steps.check.outputs.changed }}
+      package_exists: ${{ steps.check.outputs.package_exists }}
+      published: ${{ steps.check.outputs.published }}
+      version: ${{ steps.check.outputs.version }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 2
+
+      - name: Check package version
+        id: check
+        run: |
+          NEW_VERSION=$(python3 -c 'import pathlib, tomllib; print(tomllib.loads(pathlib.Path("wasm/Cargo.toml").read_text())["package"]["version"])')
+          echo "version=$NEW_VERSION" >> "$GITHUB_OUTPUT"
+
+          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+          elif git cat-file -e HEAD~1:wasm/Cargo.toml 2>/dev/null; then
+            OLD_VERSION=$(git show HEAD~1:wasm/Cargo.toml | python3 -c 'import sys, tomllib; print(tomllib.loads(sys.stdin.read())["package"]["version"])')
+            if [ "$NEW_VERSION" = "$OLD_VERSION" ]; then
+              echo "changed=false" >> "$GITHUB_OUTPUT"
+              echo "published=false" >> "$GITHUB_OUTPUT"
+              exit 0
+            fi
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+          fi
+
+          if ! npm view "@firecrawl/pdf-inspector-wasm" name >/dev/null 2>&1; then
+            echo "package_exists=false" >> "$GITHUB_OUTPUT"
+            echo "published=false" >> "$GITHUB_OUTPUT"
+            echo "The initial package must be published once before trusted publishing can be configured."
+            exit 0
+          fi
+
+          echo "package_exists=true" >> "$GITHUB_OUTPUT"
+          if npm view "@firecrawl/pdf-inspector-wasm@$NEW_VERSION" version >/dev/null 2>&1; then
+            echo "published=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "published=false" >> "$GITHUB_OUTPUT"
+          fi
+
+  publish:
+    name: Build and publish
+    needs: check-version
+    if: needs.check-version.outputs.changed == 'true' && needs.check-version.outputs.package_exists == 'true' && needs.check-version.outputs.published == 'false'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      id-token: write
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install Rust
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: wasm32-unknown-unknown
+
+      - uses: actions/setup-node@v6
+        with:
+          node-version: '24'
+          registry-url: 'https://registry.npmjs.org'
+
+      - name: Install wasm-pack
+        run: cargo install wasm-pack --version 0.15.0 --locked
+
+      - name: Build browser package
+        run: wasm-pack build wasm --target web --scope firecrawl --out-dir pkg --release
+
+      - name: Prepare package metadata
+        run: |
+          node -e '
+            const fs = require("fs")
+            const path = "wasm/pkg/package.json"
+            const pkg = JSON.parse(fs.readFileSync(path, "utf8"))
+            pkg.name = "@firecrawl/pdf-inspector-wasm"
+            pkg.description = "Browser WebAssembly bindings for the pdf-inspector Rust PDF parser"
+            pkg.keywords = ["pdf", "pdf-parser", "webassembly", "wasm", "markdown", "rust", "firecrawl"]
+            pkg.repository = { type: "git", url: "https://github.com/firecrawl/pdf-inspector" }
+            pkg.homepage = "https://github.com/firecrawl/pdf-inspector/tree/main/wasm"
+            pkg.publishConfig = { access: "public" }
+            fs.writeFileSync(path, JSON.stringify(pkg, null, 2) + "\n")
+          '
+
+      - name: Inspect package contents
+        run: npm pack --dry-run ./wasm/pkg
+
+      - name: Publish package
+        run: npm publish ./wasm/pkg --provenance --access public

.gitignore

@@ -1,10 +1,13 @@
 # Rust build artifacts
 /target/
+/wasm/target/
+/wasm/pkg/
 debug/
 *.pdb
 
 # Cargo lock (optional for libraries)
 Cargo.lock
+!/wasm/Cargo.lock
 
 # IDE
 .idea/
@@ -39,4 +42,3 @@ test_output/
 __pycache__/
 *.pyc
 .pytest_cache/
-

Cargo.toml

@@ -12,13 +12,13 @@ readme = "docs/rust-api.md"
 # alone exceeds that. external/bcmaps ships in the crate — tounicode.rs
 # loads it at runtime relative to CARGO_MANIFEST_DIR.
 include = [
-    "src/**",
-    "external/bcmaps/**",
-    "docs/rust-api.md",
-    "LICENSE",
+    "/src/**",
+    "/external/bcmaps/**",
+    "/docs/rust-api.md",
+    "/LICENSE",
     # maturin derives the sdist file list from this allowlist; the stub must
     # ship so wheels built from the sdist keep their type hints.
-    "pdf_inspector.pyi",
+    "/pdf_inspector.pyi",
 ]
 
 [lib]
@@ -29,18 +29,11 @@ crate-type = ["lib", "cdylib"]
 # Python bindings
 pyo3 = { version = "0.25", features = ["extension-module", "abi3-py38"], optional = true }
 
-# PDF parsing
-lopdf = { version = "0.41.0", features = ["rayon"] }
-
 # Error handling
 thiserror = "2.0"
 
-# Parallel processing
-rayon = "1.10"
-
 # Logging
 log = "0.4"
-env_logger = "0.11"
 
 # Text processing
 regex = "1.10"
@@ -50,6 +43,19 @@ unicode-normalization = "0.1"
 # TrueType font parsing (for Identity-H CID font cmap extraction)
 ttf-parser = "0.25"
 
+# Native builds keep lopdf's parallel parser and CLI logging. Browser WASM is
+# deliberately single-threaded so it works without cross-origin isolation.
+[target.'cfg(not(target_arch = "wasm32"))'.dependencies]
+lopdf = { version = "0.41.0", features = ["rayon"] }
+rayon = "1.10"
+env_logger = "0.11"
+
+# Browser builds use JavaScript randomness for encrypted PDFs and embed the
+# bundled CMaps because there is no filesystem at runtime.
+[target.'cfg(target_arch = "wasm32")'.dependencies]
+lopdf = { version = "0.41.0", default-features = false, features = ["wasm_js"] }
+include_dir = "0.7"
+
 [dev-dependencies]
 tempfile = "3.3"
 

README.md

@@ -5,7 +5,7 @@
 [![PyPI](https://img.shields.io/pypi/v/pdf-inspector.svg)](https://pypi.org/project/pdf-inspector/)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
 
-Fast Rust library for PDF classification and text extraction. Detects whether a PDF is text-based or scanned, extracts text with position awareness, and converts to clean Markdown — all without OCR. Includes bindings for [Python](docs/python.md) and [Node.js](napi/README.md).
+Fast Rust library for PDF classification and text extraction. Detects whether a PDF is text-based or scanned, extracts text with position awareness, and converts to clean Markdown — all without OCR. Includes bindings for [Python](docs/python.md), [Node.js](napi/README.md), and [browser WebAssembly](wasm/README.md).
 
 Built by [Firecrawl](https://firecrawl.dev) to handle text-based PDFs locally in under 200ms, skipping expensive OCR services for the ~54% of PDFs that don't need them.
 
@@ -19,6 +19,7 @@ Built by [Firecrawl](https://firecrawl.dev) to handle text-based PDFs locally in
 - **Multi-column layout** — Automatic detection of newspaper-style columns, sequential reading order, and RTL text support.
 - **Encoding issue detection** — Automatically flags broken font encodings so callers can fall back to OCR.
 - **Single document load** — The document is parsed once and shared between detection and extraction, avoiding redundant I/O.
+- **Browser WebAssembly** — Run the same Rust parser locally in browsers and Web Workers, with embedded CMaps and no server round trip.
 - **Lightweight** — Pure Rust, no ML models, no external services. Single dependency on `lopdf` for PDF parsing.
 
 ## Benchmark
@@ -77,6 +78,26 @@ console.log(result.markdown);  // Markdown string or null
 
 > Full API reference: [napi/README.md](napi/README.md)
 
+### Browser WebAssembly
+
+```bash
+npm install @firecrawl/pdf-inspector-wasm
+```
+
+```javascript
+import init, { processPdf } from '@firecrawl/pdf-inspector-wasm';
+
+await init();
+const response = await fetch('/document.pdf');
+const pdf = new Uint8Array(await response.arrayBuffer());
+const result = processPdf(pdf);
+
+console.log(result.pdfType);
+console.log(result.markdown);
+```
+
+> Full API reference: [wasm/README.md](wasm/README.md)
+
 ### Rust
 
 Install from [crates.io](https://crates.io/crates/pdf-inspector):
@@ -188,6 +209,7 @@ src/
   markdown/             — Markdown conversion and structure detection
   bin/                  — CLI tools (pdf2md, detect_pdf)
 napi/                   — Node.js/Bun bindings (napi-rs)
+wasm/                   — Browser bindings (wasm-bindgen)
 ```
 
 ## How classification works

docs/publishing.md

@@ -19,3 +19,20 @@ The workflow uses `rust-lang/crates-io-auth-action@v1` to exchange GitHub's OIDC
 3. The publish workflow compares the new `Cargo.toml` version with `HEAD~1`, runs `cargo publish --dry-run`, then publishes if that version is not already on crates.io.
 
 If `Cargo.toml` changes without a package version bump, the workflow exits without publishing.
+
+## Browser WebAssembly package
+
+The browser package is published as `@firecrawl/pdf-inspector-wasm`. Its version lives in `wasm/Cargo.toml`, and `.github/workflows/publish-wasm.yml` builds the `web` target with `wasm-pack` before publishing the generated package.
+
+The npm package must exist before a trusted publisher can be configured. For the first release only:
+
+1. Build with `wasm-pack build wasm --target web --scope firecrawl --out-dir pkg --release`.
+2. Inspect with `npm pack --dry-run ./wasm/pkg`.
+3. Publish with `npm publish ./wasm/pkg --access public` from an authorized maintainer session.
+4. In the package settings on npm, configure the GitHub Actions trusted publisher:
+   - Organization: `firecrawl`
+   - Repository: `pdf-inspector`
+   - Workflow: `publish-wasm.yml`
+   - Allowed action: `npm publish`
+
+After that one-time bootstrap, bumping the version in `wasm/Cargo.toml` and merging it to `main` publishes through OIDC. Until the package exists, the workflow exits cleanly without attempting an unauthenticated first publish. See npm's [trusted publishing documentation](https://docs.npmjs.com/trusted-publishers/) for the registry-side setup.

site/index.html

@@ -302,6 +302,171 @@
   .text-link { color: var(--heat-dark); text-decoration: underline; text-decoration-color: var(--heat-16); text-underline-offset: 4px; }
   .text-link:hover, .text-link:focus-visible { text-decoration-color: var(--heat); }
 
+  .demo-shell {
+    border: 1px solid var(--border);
+    border-radius: 12px;
+    background: var(--surface);
+    box-shadow: 0 22px 60px rgba(38,38,38,.07);
+    overflow: hidden;
+  }
+  .demo-toolbar {
+    min-height: 48px;
+    padding: 0 17px;
+    border-bottom: 1px solid var(--border);
+    display: flex;
+    align-items: center;
+    justify-content: space-between;
+    gap: 18px;
+    color: var(--ink-48);
+    font: 11px/1.3 var(--mono);
+  }
+  .demo-engine,
+  .demo-privacy { display: inline-flex; align-items: center; gap: 8px; }
+  .demo-engine strong { color: var(--ink); font-weight: 500; }
+  .demo-status-dot { width: 7px; height: 7px; border-radius: 50%; background: var(--heat); box-shadow: 0 0 0 4px var(--heat-8); }
+  .demo-lock { color: var(--heat); font-size: 12px; }
+  .demo-grid { display: grid; grid-template-columns: minmax(0, .86fr) minmax(0, 1.14fr); min-height: 500px; }
+  .demo-input {
+    padding: 28px;
+    border-right: 1px solid var(--border);
+    display: flex;
+    flex-direction: column;
+    background: var(--lighter);
+  }
+  .drop-zone {
+    min-height: 286px;
+    padding: 30px;
+    border: 1px dashed var(--ink-16);
+    border-radius: 10px;
+    display: flex;
+    flex: 1;
+    flex-direction: column;
+    align-items: center;
+    justify-content: center;
+    text-align: center;
+    background: var(--surface);
+    cursor: pointer;
+    transition: border-color .15s ease, background .15s ease, transform .15s ease;
+  }
+  .drop-zone:hover,
+  .drop-zone:focus-visible,
+  .drop-zone.is-dragging { border-color: var(--heat); background: var(--heat-4); }
+  .drop-zone.is-dragging { transform: scale(.995); }
+  .drop-mark {
+    width: 52px;
+    height: 58px;
+    margin-bottom: 21px;
+    border: 1px solid var(--ink-16);
+    border-radius: 5px 5px 9px 5px;
+    display: grid;
+    place-items: center;
+    position: relative;
+    color: var(--heat-dark);
+    background: var(--heat-4);
+    font: 600 11px/1 var(--mono);
+    letter-spacing: .03em;
+  }
+  .drop-mark::after {
+    content: "";
+    position: absolute;
+    top: -1px;
+    right: -1px;
+    width: 14px;
+    height: 14px;
+    border-left: 1px solid var(--ink-16);
+    border-bottom: 1px solid var(--ink-16);
+    background: var(--surface);
+  }
+  .drop-zone strong { font-size: 18px; font-weight: 500; letter-spacing: -.02em; }
+  .drop-zone > span:not(.drop-mark):not(.demo-choose) { margin-top: 7px; color: var(--ink-48); font-size: 13px; }
+  .demo-choose { margin-top: 20px; color: var(--ink); }
+  .demo-file {
+    margin-top: 14px;
+    padding: 13px 14px;
+    border: 1px solid var(--border);
+    border-radius: 8px;
+    display: grid;
+    grid-template-columns: minmax(0, 1fr) auto;
+    gap: 4px 16px;
+    background: var(--surface);
+  }
+  .demo-file strong { overflow: hidden; text-overflow: ellipsis; font: 500 12px/1.35 var(--mono); white-space: nowrap; }
+  .demo-file span { grid-column: 1; color: var(--ink-32); font: 10px/1.3 var(--mono); }
+  .demo-file button {
+    grid-column: 2;
+    grid-row: 1 / 3;
+    padding: 0;
+    border: 0;
+    color: var(--ink-48);
+    background: transparent;
+    cursor: pointer;
+    font: 11px/1 var(--mono);
+  }
+  .demo-file button:hover,
+  .demo-file button:focus-visible { color: var(--heat-dark); }
+  .demo-message { margin: 13px 2px 0; min-height: 19px; color: var(--ink-48); font: 11px/1.55 var(--mono); }
+  .demo-message[data-tone="error"] { color: #a63520; }
+  .demo-message[data-tone="success"] { color: #34725a; }
+  .demo-output { min-width: 0; display: flex; flex-direction: column; background: var(--code); color: #f5f5f5; }
+  .demo-output-head {
+    min-height: 52px;
+    padding: 0 18px;
+    border-bottom: 1px solid rgba(255,255,255,.08);
+    display: flex;
+    align-items: center;
+    justify-content: space-between;
+    color: rgba(255,255,255,.42);
+    font: 11px/1 var(--mono);
+  }
+  .demo-copy {
+    height: 29px;
+    padding: 0 10px;
+    border: 1px solid rgba(255,255,255,.12);
+    border-radius: 6px;
+    color: rgba(255,255,255,.68);
+    background: rgba(255,255,255,.04);
+    cursor: pointer;
+    font: 10px/1 var(--mono);
+  }
+  .demo-copy:hover,
+  .demo-copy:focus-visible { border-color: var(--heat); color: #fff; }
+  .demo-copy[disabled] { opacity: .35; cursor: not-allowed; }
+  .demo-empty {
+    min-height: 448px;
+    padding: 40px;
+    display: grid;
+    place-items: center;
+    text-align: center;
+    color: rgba(255,255,255,.28);
+    font: 12px/1.7 var(--mono);
+  }
+  .demo-empty span { display: block; color: var(--heat); font-size: 22px; line-height: 1; }
+  .demo-result { min-height: 0; flex: 1; display: flex; flex-direction: column; }
+  .demo-result[hidden],
+  .demo-file[hidden],
+  .demo-empty[hidden] { display: none; }
+  .demo-metrics {
+    display: grid;
+    grid-template-columns: repeat(3, 1fr);
+    border-bottom: 1px solid rgba(255,255,255,.08);
+  }
+  .demo-metric { min-width: 0; padding: 14px 17px; border-right: 1px solid rgba(255,255,255,.08); }
+  .demo-metric:last-child { border-right: 0; }
+  .demo-metric span { display: block; margin-bottom: 6px; color: rgba(255,255,255,.32); font: 9px/1 var(--mono); letter-spacing: .07em; text-transform: uppercase; }
+  .demo-metric strong { display: block; overflow: hidden; text-overflow: ellipsis; color: #fff; font: 500 12px/1.25 var(--mono); white-space: nowrap; }
+  .demo-metric:first-child strong { color: #ff9c70; }
+  .markdown-output {
+    min-height: 0;
+    max-height: 364px;
+    margin: 0;
+    padding: 23px 20px 28px;
+    flex: 1;
+    overflow: auto;
+    color: rgba(255,255,255,.82);
+    font: 12px/1.7 var(--mono);
+    white-space: pre-wrap;
+    word-break: break-word;
+  }
   .capability-grid { display: grid; grid-template-columns: repeat(3, 1fr); border: 1px solid var(--border); }
   .capability {
     min-height: 245px;
@@ -409,6 +574,9 @@
     .hero-main { grid-template-columns: 1fr; gap: 44px; }
     .terminal { max-width: 580px; }
     .section-intro { grid-template-columns: 1fr; gap: 20px; }
+    .demo-grid { grid-template-columns: 1fr; }
+    .demo-input { border-right: 0; border-bottom: 1px solid var(--border); }
+    .demo-empty { min-height: 360px; }
     .capability-grid { grid-template-columns: repeat(2, 1fr); }
     .capability:nth-child(3n) { border-right: 1px solid var(--border); }
     .capability:nth-child(2n) { border-right: 0; }
@@ -429,6 +597,13 @@
     .package-row { grid-template-columns: 1fr; }
     .package { border-right: 0; border-bottom: 1px solid var(--border); }
     .package:last-child { border-bottom: 0; }
+    .demo-toolbar { padding: 12px 14px; align-items: flex-start; flex-direction: column; gap: 8px; }
+    .demo-input { padding: 18px; }
+    .drop-zone { min-height: 250px; padding: 24px 18px; }
+    .demo-metric { padding: 12px; }
+    .demo-metric strong { font-size: 11px; }
+    .demo-empty { min-height: 320px; padding: 28px; }
+    .markdown-output { max-height: 340px; padding: 20px 16px 24px; font-size: 11px; }
     .capability-grid { grid-template-columns: 1fr; }
     .capability, .capability:nth-child(3n), .capability:nth-child(2n) { min-height: 210px; border-right: 0; border-bottom: 1px solid var(--border); }
     .capability:last-child { border-bottom: 0; }
@@ -463,6 +638,7 @@
     </a>
     <div class="nav-links">
       <a href="#packages">Packages</a>
+      <a href="#demo">Demo</a>
       <a href="#capabilities">Capabilities</a>
       <a href="#architecture">Architecture</a>
       <a href="#benchmark">Benchmark</a>
@@ -484,6 +660,7 @@
         <p class="hero-copy">A Rust-powered, open-source parser that classifies PDFs and turns native text into clean, position-aware Markdown. Use it from Node.js or the bundled CLI, with packages also available from PyPI and crates.io.</p>
         <div class="hero-actions">
           <a class="button button-primary" href="https://github.com/firecrawl/pdf-inspector">View on GitHub <span class="arrow" aria-hidden="true">↗</span></a>
+          <a class="button" href="#demo">Try it locally <span class="arrow" aria-hidden="true">↓</span></a>
           <a class="button" href="https://github.com/firecrawl/pdf-inspector#quick-start">Read the docs <span class="arrow" aria-hidden="true">→</span></a>
         </div>
       </div>
@@ -538,9 +715,61 @@
     </div>
   </div>
 
+  <section id="demo" class="full-rule">
+    <div class="shell">
+      <div class="section-label pad"><span>[ <b>01</b> / 05 ]</span><span>Browser demo</span></div>
+      <div class="section-body pad">
+        <div class="section-intro">
+          <h2>Try it in<br>your browser.</h2>
+          <p>Drop in a native-text PDF to classify it and turn it into Markdown. The Rust core runs locally as WebAssembly in a background worker—your document never leaves this tab.</p>
+        </div>
+        <div class="demo-shell">
+          <div class="demo-toolbar">
+            <span class="demo-engine"><i class="demo-status-dot" aria-hidden="true"></i><strong>Rust core · WebAssembly</strong><span id="demo-engine-status">Loads on first run</span></span>
+            <span class="demo-privacy"><span class="demo-lock" aria-hidden="true">◇</span>PDF bytes stay in your browser</span>
+          </div>
+          <div class="demo-grid">
+            <div class="demo-input">
+              <input id="pdf-input" type="file" accept=".pdf,application/pdf" hidden>
+              <div class="drop-zone" id="drop-zone" role="button" tabindex="0" aria-controls="pdf-input" aria-label="Choose or drop a PDF file">
+                <span class="drop-mark" aria-hidden="true">PDF</span>
+                <strong>Drop a PDF here</strong>
+                <span>Native-text documents · up to 25 MB</span>
+                <span class="button demo-choose" aria-hidden="true">Choose PDF</span>
+              </div>
+              <div class="demo-file" id="demo-file" hidden>
+                <strong id="demo-file-name"></strong>
+                <span id="demo-file-size"></span>
+                <button id="demo-clear" type="button">Remove</button>
+              </div>
+              <p class="demo-message" id="demo-message" role="status" aria-live="polite">Select a PDF to begin.</p>
+            </div>
+            <div class="demo-output" aria-label="PDF parsing result">
+              <div class="demo-output-head">
+                <span>MARKDOWN OUTPUT</span>
+                <button class="demo-copy" id="demo-copy" type="button" disabled>Copy Markdown</button>
+              </div>
+              <div class="demo-empty" id="demo-empty">
+                <p><span aria-hidden="true">_</span><br>Your parsed Markdown will appear here.</p>
+              </div>
+              <div class="demo-result" id="demo-result" hidden>
+                <div class="demo-metrics">
+                  <div class="demo-metric"><span>Document type</span><strong id="demo-type">—</strong></div>
+                  <div class="demo-metric"><span>Pages</span><strong id="demo-pages">—</strong></div>
+                  <div class="demo-metric"><span>Processing</span><strong id="demo-time">—</strong></div>
+                </div>
+                <pre class="markdown-output" id="markdown-output"></pre>
+              </div>
+            </div>
+          </div>
+        </div>
+      </div>
+    </div>
+  </section>
+
   <section id="capabilities" class="full-rule">
     <div class="shell">
-      <div class="section-label pad"><span>[ <b>01</b> / 04 ]</span><span>Core capabilities</span></div>
+      <div class="section-label pad"><span>[ <b>02</b> / 05 ]</span><span>Core capabilities</span></div>
       <div class="section-body pad">
         <div class="section-intro">
           <h2>A focused PDF<br>toolchain.</h2>
@@ -584,7 +813,7 @@
 
   <section id="architecture" class="full-rule">
     <div class="shell">
-      <div class="section-label pad"><span>[ <b>02</b> / 04 ]</span><span>Architecture</span></div>
+      <div class="section-label pad"><span>[ <b>03</b> / 05 ]</span><span>Architecture</span></div>
       <div class="section-body pad">
         <div class="section-intro">
           <h2>One parse.<br>Clear stages.</h2>
@@ -622,7 +851,7 @@
 
   <section id="benchmark" class="full-rule">
     <div class="shell">
-      <div class="section-label pad"><span>[ <b>03</b> / 04 ]</span><span>Benchmark</span></div>
+      <div class="section-label pad"><span>[ <b>04</b> / 05 ]</span><span>Benchmark</span></div>
       <div class="section-body pad">
         <div class="section-intro">
           <h2>Measured on<br>real documents.</h2>
@@ -656,7 +885,7 @@
 
   <section id="usage" class="full-rule">
     <div class="shell">
-      <div class="section-label pad"><span>[ <b>04</b> / 04 ]</span><span>Usage</span></div>
+      <div class="section-label pad"><span>[ <b>05</b> / 05 ]</span><span>Usage</span></div>
       <div class="section-body pad">
         <div class="section-intro">
           <h2>Start with Node.<br>Use the CLI.</h2>
@@ -743,5 +972,201 @@ result = pdf_inspector.<span class="fn">process_pdf</span>(<span class="str">"do
   </div>
 </footer>
 
+<script>
+  (() => {
+    const MAX_FILE_SIZE = 25 * 1024 * 1024;
+    const WASM_MODULE_URL = "https://cdn.jsdelivr.net/npm/@firecrawl/pdf-inspector-wasm@0.1.1/pdf_inspector_wasm.js";
+    const input = document.querySelector("#pdf-input");
+    const dropZone = document.querySelector("#drop-zone");
+    const filePanel = document.querySelector("#demo-file");
+    const fileName = document.querySelector("#demo-file-name");
+    const fileSize = document.querySelector("#demo-file-size");
+    const clearButton = document.querySelector("#demo-clear");
+    const copyButton = document.querySelector("#demo-copy");
+    const message = document.querySelector("#demo-message");
+    const engineStatus = document.querySelector("#demo-engine-status");
+    const emptyOutput = document.querySelector("#demo-empty");
+    const resultPanel = document.querySelector("#demo-result");
+    const markdownOutput = document.querySelector("#markdown-output");
+    const typeOutput = document.querySelector("#demo-type");
+    const pagesOutput = document.querySelector("#demo-pages");
+    const timeOutput = document.querySelector("#demo-time");
+    let selectedFile = null;
+    let currentMarkdown = "";
+    let busy = false;
+
+    const formatBytes = (bytes) => {
+      if (bytes < 1024) return `${bytes} B`;
+      const units = ["KB", "MB", "GB"];
+      let value = bytes / 1024;
+      let unit = units[0];
+      for (let index = 1; value >= 1024 && index < units.length; index += 1) {
+        value /= 1024;
+        unit = units[index];
+      }
+      return `${value >= 10 ? value.toFixed(1) : value.toFixed(2)} ${unit}`;
+    };
+
+    const setMessage = (text, tone = "") => {
+      message.textContent = text;
+      message.dataset.tone = tone;
+    };
+
+    const setBusy = (isBusy) => {
+      busy = isBusy;
+      clearButton.disabled = isBusy;
+      input.disabled = isBusy;
+    };
+
+    const clearResult = () => {
+      currentMarkdown = "";
+      copyButton.disabled = true;
+      copyButton.textContent = "Copy Markdown";
+      markdownOutput.textContent = "";
+      resultPanel.hidden = true;
+      emptyOutput.hidden = false;
+    };
+
+    const clearFile = () => {
+      if (busy) return;
+      selectedFile = null;
+      input.value = "";
+      filePanel.hidden = true;
+      clearResult();
+      setMessage("Select a PDF to begin.");
+    };
+
+    const selectFile = (file) => {
+      if (!file) return;
+      const isPdf = file.type === "application/pdf" || file.name.toLowerCase().endsWith(".pdf");
+      if (!isPdf) {
+        clearFile();
+        setMessage("Choose a PDF file to run this demo.", "error");
+        return;
+      }
+      if (file.size > MAX_FILE_SIZE) {
+        clearFile();
+        setMessage("This demo accepts PDFs up to 25 MB.", "error");
+        return;
+      }
+      selectedFile = file;
+      fileName.textContent = file.name;
+      fileSize.textContent = formatBytes(file.size);
+      filePanel.hidden = false;
+      clearResult();
+      parseSelectedFile();
+    };
+
+    const processInWorker = (buffer) => new Promise((resolve, reject) => {
+      const source = `
+        import init, { processPdf, version } from "${WASM_MODULE_URL}";
+        self.onmessage = async ({ data }) => {
+          try {
+            await init();
+            const result = processPdf(new Uint8Array(data.buffer), { profile: "fidelity" });
+            self.postMessage({ ok: true, result, engineVersion: version() });
+          } catch (error) {
+            const detail = error instanceof Error ? error.message : String(error);
+            self.postMessage({ ok: false, error: detail });
+          }
+        };
+      `;
+      const workerUrl = URL.createObjectURL(new Blob([source], { type: "text/javascript" }));
+      let worker;
+
+      try {
+        worker = new Worker(workerUrl, { type: "module" });
+      } catch (error) {
+        URL.revokeObjectURL(workerUrl);
+        reject(error);
+        return;
+      }
+
+      URL.revokeObjectURL(workerUrl);
+      worker.addEventListener("message", ({ data }) => {
+        worker.terminate();
+        if (data.ok) resolve(data);
+        else reject(new Error(data.error));
+      }, { once: true });
+      worker.addEventListener("error", (event) => {
+        worker.terminate();
+        reject(new Error(event.message || "The WebAssembly engine could not be loaded."));
+      }, { once: true });
+      worker.postMessage({ buffer }, [buffer]);
+    });
+
+    const renderResult = ({ result, engineVersion }) => {
+      currentMarkdown = typeof result.markdown === "string" ? result.markdown : "";
+      typeOutput.textContent = result.pdfType || "Unknown";
+      pagesOutput.textContent = String(result.pageCount ?? "—");
+      timeOutput.textContent = Number.isFinite(result.processingTimeMs)
+        ? `${Math.max(1, Math.round(result.processingTimeMs))} ms`
+        : "—";
+      markdownOutput.textContent = currentMarkdown || "No Markdown output was produced for this document.";
+      emptyOutput.hidden = true;
+      resultPanel.hidden = false;
+      copyButton.disabled = !currentMarkdown;
+      engineStatus.textContent = engineVersion ? `Ready · v${engineVersion}` : "Ready";
+    };
+
+    const parseSelectedFile = async () => {
+      if (!selectedFile || busy) return;
+      const file = selectedFile;
+      setBusy(true);
+      setMessage("Loading the Rust engine and parsing in this browser…");
+      engineStatus.textContent = "Loading…";
+
+      try {
+        const buffer = await file.arrayBuffer();
+        const response = await processInWorker(buffer);
+        renderResult(response);
+        setMessage(`Parsed ${file.name} locally.`, "success");
+      } catch (error) {
+        engineStatus.textContent = "Load failed";
+        clearResult();
+        const detail = error instanceof Error ? error.message : String(error);
+        setMessage(detail || "The PDF could not be parsed.", "error");
+      } finally {
+        setBusy(false);
+      }
+    };
+
+    dropZone.addEventListener("click", () => {
+      if (!busy) {
+        input.value = "";
+        input.click();
+      }
+    });
+    dropZone.addEventListener("keydown", (event) => {
+      if ((event.key === "Enter" || event.key === " ") && !busy) {
+        event.preventDefault();
+        input.value = "";
+        input.click();
+      }
+    });
+    dropZone.addEventListener("dragover", (event) => {
+      event.preventDefault();
+      if (!busy) dropZone.classList.add("is-dragging");
+    });
+    dropZone.addEventListener("dragleave", () => dropZone.classList.remove("is-dragging"));
+    dropZone.addEventListener("drop", (event) => {
+      event.preventDefault();
+      dropZone.classList.remove("is-dragging");
+      if (!busy) selectFile(event.dataTransfer.files[0]);
+    });
+    input.addEventListener("change", () => selectFile(input.files[0]));
+    clearButton.addEventListener("click", clearFile);
+    copyButton.addEventListener("click", async () => {
+      if (!currentMarkdown) return;
+      try {
+        await navigator.clipboard.writeText(currentMarkdown);
+        copyButton.textContent = "Copied";
+        window.setTimeout(() => { copyButton.textContent = "Copy Markdown"; }, 1600);
+      } catch {
+        setMessage("Copy is unavailable here. Select the Markdown output instead.", "error");
+      }
+    });
+  })();
+</script>
 </body>
 </html>

src/bin/detect_pdf.rs

@@ -63,6 +63,7 @@ fn json_escape(s: &str) -> String {
 }
 
 fn main() {
+    #[cfg(not(target_arch = "wasm32"))]
     env_logger::init();
     let args: Vec<String> = env::args().collect();
 

src/bin/pdf2md.rs

@@ -190,6 +190,7 @@ fn print_layout_info(layout: &LayoutComplexity) {
 }
 
 fn main() {
+    #[cfg(not(target_arch = "wasm32"))]
     env_logger::init();
     let args: Vec<String> = env::args().collect();
 

src/extractor/content_stream.rs

@@ -162,7 +162,7 @@ pub(crate) fn extract_page_text_items(
     let fonts = doc.get_page_fonts(page_id).unwrap_or_default();
 
     // Build font encoding maps from Differences arrays
-    let (font_encodings, has_gid_fonts) = build_font_encodings(doc, &fonts);
+    let (font_encodings, has_gid_fonts) = build_font_encodings(doc, &fonts, font_cmaps);
 
     // Build font width info for accurate text positioning
     let font_widths = build_font_widths(doc, &fonts);

src/extractor/fonts.rs

@@ -497,9 +497,13 @@ pub(crate) fn get_operand_bytes(obj: &Object) -> Option<&[u8]> {
 /// Build encoding maps for all fonts on a page.
 /// Returns `(encodings, has_gid_fonts)` where `has_gid_fonts` is true when
 /// any font uses raw glyph ID names (gidNNNNN) that can't be decoded.
+/// Gid names whose codes the font's own ToUnicode CMap maps are decodable
+/// and do not set the flag (LibreOffice subsets write /gidNNNN Differences
+/// names alongside a complete ToUnicode CMap).
 pub(crate) fn build_font_encodings(
     doc: &Document,
     fonts: &std::collections::BTreeMap<Vec<u8>, &lopdf::Dictionary>,
+    cmaps: &FontCMaps,
 ) -> (PageFontEncodings, bool) {
     let mut encodings = PageFontEncodings::new();
     let mut has_gid_fonts = false;
@@ -508,7 +512,9 @@ pub(crate) fn build_font_encodings(
         let resource_name = String::from_utf8_lossy(font_name).to_string();
 
         if let Some(result) = parse_font_encoding(doc, font_dict) {
-            if result.gid_glyph_count > 0 {
+            if !result.gid_codes.is_empty()
+                && !tounicode_maps_codes(font_dict, cmaps, &result.gid_codes)
+            {
                 has_gid_fonts = true;
             }
             if !result.map.is_empty() {
@@ -520,6 +526,34 @@ pub(crate) fn build_font_encodings(
     (encodings, has_gid_fonts)
 }
 
+/// True when the font's ToUnicode CMap maps the gid-named character codes,
+/// so the Differences entries still decode through the CMap.
+fn tounicode_maps_codes(font_dict: &lopdf::Dictionary, cmaps: &FontCMaps, codes: &[u8]) -> bool {
+    let Some(obj_ref) = font_dict
+        .get(b"ToUnicode")
+        .ok()
+        .and_then(|o| o.as_reference().ok())
+    else {
+        return false;
+    };
+    let Some(entry) = cmaps.get_by_obj(obj_ref.0) else {
+        return false;
+    };
+    // At least one gid code usably mapped means the CMap addresses these
+    // codes; remaining unmapped codes are subset leftovers (e.g. the
+    // component glyphs of an emoji ZWJ sequence mapped whole on its first
+    // code). A mapping is usable only when extraction would accept it —
+    // empty or U+FFFD results are rejected there as invalid. Fonts whose
+    // CMap ignores the gid codes entirely stay flagged, and the downstream
+    // garbage/encoding checks still catch partial damage.
+    codes.iter().any(|&code| {
+        entry
+            .primary
+            .lookup(code as u16)
+            .is_some_and(|s| !s.is_empty() && !s.contains('\u{FFFD}'))
+    })
+}
+
 /// Parse font encoding from a font dictionary
 pub(crate) fn parse_font_encoding(
     doc: &Document,
@@ -558,11 +592,10 @@ pub(crate) fn parse_font_encoding(
 /// Result of parsing an encoding dictionary's Differences array.
 pub(crate) struct EncodingResult {
     pub map: FontEncodingMap,
-    /// Number of glyph names matching the `gidNNNNN` pattern (raw glyph IDs).
-    /// These indicate a font with unresolvable encoding — the glyph IDs
-    /// reference the original font's glyph table, but without the original
-    /// font's cmap there is no way to map them to Unicode.
-    pub gid_glyph_count: u32,
+    /// Character codes whose glyph names match the `gidNNNNN` pattern (raw
+    /// glyph IDs). These reference the original font's glyph table and are
+    /// only decodable when the font's ToUnicode CMap maps the code.
+    pub gid_codes: Vec<u8>,
 }
 
 /// Parse an encoding dictionary with Differences array
@@ -588,7 +621,7 @@ pub(crate) fn parse_encoding_dictionary(
     let mut encoding_map = FontEncodingMap::new();
     let mut current_code: u8 = 0;
     let mut ligature_count = 0u32;
-    let mut gid_glyph_count = 0u32;
+    let mut gid_codes: Vec<u8> = Vec::new();
 
     for item in diff_array {
         match item {
@@ -614,7 +647,7 @@ pub(crate) fn parse_encoding_dictionary(
                     && glyph_name.len() >= 4
                     && glyph_name[3..].chars().all(|c| c.is_ascii_digit())
                 {
-                    gid_glyph_count += 1;
+                    gid_codes.push(current_code);
                 }
                 if let Some(ch) = mapped_char {
                     encoding_map.insert(current_code, ch);
@@ -638,16 +671,16 @@ pub(crate) fn parse_encoding_dictionary(
         );
     }
 
-    if gid_glyph_count > 0 {
+    if !gid_codes.is_empty() {
         debug!(
-            "  Differences: {} gid-encoded glyphs (unresolvable without original font)",
-            gid_glyph_count
+            "  Differences: {} gid-encoded glyphs (decodable only via ToUnicode)",
+            gid_codes.len()
         );
     }
 
     Some(EncodingResult {
         map: encoding_map,
-        gid_glyph_count,
+        gid_codes,
     })
 }
 
@@ -1938,4 +1971,113 @@ mod tests {
             false
         ));
     }
+
+    fn gid_font_doc(bfchar: Option<&str>) -> (Document, lopdf::ObjectId) {
+        use lopdf::Stream;
+        let mut doc = Document::with_version("1.4");
+        let cmap = format!(
+            "/CIDInit /ProcSet findresource begin
+12 dict begin
+begincmap
+1 begincodespacerange
+<00> <FF>
+endcodespacerange
+1 beginbfchar
+{}
+endbfchar
+endcmap
+CMapName currentdict /CMap defineresource pop
+end
+end",
+            bfchar.unwrap_or_default()
+        );
+        let tounicode_id = doc.add_object(Object::Stream(Stream::new(
+            dictionary! {},
+            cmap.into_bytes(),
+        )));
+        let enc_id = doc.add_object(dictionary! {
+            "Type" => "Encoding",
+            "Differences" => vec![
+                1.into(),
+                Object::Name(b"gid1283".to_vec()),
+                Object::Name(b"gid1464".to_vec()),
+            ],
+        });
+        let mut font = dictionary! {
+            "Type" => "Font",
+            "Subtype" => "TrueType",
+            "BaseFont" => "ABCDEF+OpenSymbol",
+            "Encoding" => Object::Reference(enc_id),
+        };
+        if bfchar.is_some() {
+            font.set("ToUnicode", Object::Reference(tounicode_id));
+        }
+        let font_id = doc.add_object(font);
+        let page_id = doc.add_object(dictionary! {
+            "Type" => "Page",
+            "Resources" => dictionary! {
+                "Font" => dictionary! { "F1" => Object::Reference(font_id) },
+            },
+            "MediaBox" => vec![0.into(), 0.into(), 612.into(), 792.into()],
+        });
+        let pages_id = doc.add_object(dictionary! {
+            "Type" => "Pages",
+            "Count" => Object::Integer(1),
+            "Kids" => vec![Object::Reference(page_id)],
+        });
+        let catalog_id = doc.add_object(dictionary! {
+            "Type" => "Catalog",
+            "Pages" => Object::Reference(pages_id),
+        });
+        doc.trailer.set("Root", Object::Reference(catalog_id));
+        (doc, page_id)
+    }
+
+    fn gid_flagged(bfchar: Option<&str>) -> bool {
+        let (doc, page_id) = gid_font_doc(bfchar);
+        let cmaps = FontCMaps::from_doc(&doc);
+        let fonts = doc.get_page_fonts(page_id).unwrap();
+        let (_, has_gid_fonts) = build_font_encodings(&doc, &fonts, &cmaps);
+        has_gid_fonts
+    }
+
+    #[test]
+    fn gid_differences_with_covering_tounicode_are_not_flagged() {
+        // LibreOffice subsets write /gidNNNN Differences names alongside a
+        // ToUnicode CMap that decodes those codes; the page must not be
+        // flagged as unresolvable (which would suppress the whole document's
+        // markdown when every page carries such a font).
+        assert!(!gid_flagged(Some("<01> <2022>\n<02> <25E6>")));
+    }
+
+    #[test]
+    fn gid_differences_with_partial_tounicode_are_not_flagged() {
+        // An emoji ZWJ sequence maps whole on its first code; the remaining
+        // component-glyph codes are subset leftovers, not damage.
+        assert!(!gid_flagged(Some(
+            "<01> <D83DDC68200DD83DDC69200DD83DDC67>"
+        )));
+    }
+
+    #[test]
+    fn gid_differences_without_tounicode_are_flagged() {
+        assert!(
+            gid_flagged(None),
+            "gid glyphs without ToUnicode are unresolvable"
+        );
+    }
+
+    #[test]
+    fn gid_differences_with_disjoint_tounicode_are_flagged() {
+        // A ToUnicode that never addresses the gid codes leaves them
+        // unresolvable.
+        assert!(gid_flagged(Some("<10> <0041>")));
+    }
+
+    #[test]
+    fn gid_differences_with_replacement_char_tounicode_are_flagged() {
+        // A mapping to U+FFFD is not usable — extraction rejects it as an
+        // invalid CMap result — so it must not clear the gid flag.
+        assert!(gid_flagged(Some("<01> <FFFD>\n<02> <FFFD>")));
+    }
 }

src/extractor/layout.rs

@@ -1153,6 +1153,22 @@ pub fn group_into_lines(items: Vec<TextItem>) -> Vec<TextLine> {
     group_into_lines_with_thresholds(items, &HashMap::new(), &HashSet::new())
 }
 
+/// Group text items into lines without removing numeric page headers or footers.
+///
+/// Plain-text extraction uses this path because every extracted item is part of
+/// the API result. Markdown conversion keeps using [`group_into_lines`], where
+/// page-number suppression is an intentional presentation cleanup.
+pub fn group_into_lines_preserving_all_text(items: Vec<TextItem>) -> Vec<TextLine> {
+    group_into_lines_with_thresholds_and_regions_impl(
+        items,
+        &HashMap::new(),
+        &HashSet::new(),
+        &HashMap::new(),
+        &HashMap::new(),
+        false,
+    )
+}
+
 /// Group text items into lines, using pre-computed per-page adaptive thresholds
 /// from Canva-style letter-spacing detection. Falls back to computing the
 /// threshold from item gaps when no pre-computed value is available.
@@ -1195,16 +1211,39 @@ pub(crate) fn group_into_lines_with_thresholds_and_regions(
     table_pages: &HashSet<u32>,
     chart_regions: &HashMap<u32, Vec<(f32, f32, f32, f32)>>,
     image_regions: &HashMap<u32, Vec<super::reading_order::ImageRegion>>,
+) -> Vec<TextLine> {
+    group_into_lines_with_thresholds_and_regions_impl(
+        items,
+        page_thresholds,
+        table_pages,
+        chart_regions,
+        image_regions,
+        true,
+    )
+}
+
+fn group_into_lines_with_thresholds_and_regions_impl(
+    items: Vec<TextItem>,
+    page_thresholds: &HashMap<u32, f32>,
+    table_pages: &HashSet<u32>,
+    chart_regions: &HashMap<u32, Vec<(f32, f32, f32, f32)>>,
+    image_regions: &HashMap<u32, Vec<super::reading_order::ImageRegion>>,
+    filter_page_numbers: bool,
 ) -> Vec<TextLine> {
     if items.is_empty() {
         return Vec::new();
     }
 
-    // Filter out page numbers (standalone numbers at top/bottom of page)
-    let items: Vec<TextItem> = items
-        .into_iter()
-        .filter(|item| !is_page_number(item))
-        .collect();
+    // Markdown output omits standalone numeric headers/footers. Plain-text
+    // callers opt out because dropping extracted text violates that API.
+    let items = if filter_page_numbers {
+        items
+            .into_iter()
+            .filter(|item| !is_page_number(item))
+            .collect()
+    } else {
+        items
+    };
 
     // Get unique pages
     let mut pages: Vec<u32> = items.iter().map(|i| i.page).collect();

src/extractor/mod.rs

@@ -27,12 +27,12 @@ pub use crate::text_utils::{is_bold_font, is_italic_font};
 pub use crate::types::{ItemType, TextLine};
 pub(crate) use fonts::FontStyleCache;
 pub(crate) use layout::detect_columns;
-pub use layout::group_into_lines;
 pub(crate) use layout::group_into_lines_with_thresholds;
 pub(crate) use layout::group_into_lines_with_thresholds_and_charts;
 pub(crate) use layout::group_into_lines_with_thresholds_and_regions;
 pub(crate) use layout::is_newspaper_layout;
 pub(crate) use layout::ColumnRegion;
+pub use layout::{group_into_lines, group_into_lines_preserving_all_text};
 
 // ---------------------------------------------------------------------------
 // Public API
@@ -1519,6 +1519,18 @@ mod tests {
         assert_eq!(lines[1].text(), "Next line");
     }
 
+    #[test]
+    fn preserving_all_text_keeps_numeric_page_footer() {
+        let mut page_number = make_merge_item("42", 100.0, 12.0);
+        page_number.y = 50.0;
+
+        assert!(group_into_lines(vec![page_number.clone()]).is_empty());
+
+        let lines = group_into_lines_preserving_all_text(vec![page_number]);
+        assert_eq!(lines.len(), 1);
+        assert_eq!(lines[0].text(), "42");
+    }
+
     #[test]
     fn test_bold_italic_detection() {
         // Test bold detection

src/extractor/xobjects.rs

@@ -162,7 +162,7 @@ fn extract_form_xobject_text_inner(
 
     // Get fonts from the Form's Resources
     let form_fonts = get_form_fonts(doc, &stream.dict);
-    let (font_encodings, _has_gid_fonts) = build_font_encodings(doc, &form_fonts);
+    let (font_encodings, _has_gid_fonts) = build_font_encodings(doc, &form_fonts, font_cmaps);
 
     // Build font width info for the form
     let font_widths = build_font_widths(doc, &form_fonts);

src/lib.rs

@@ -68,6 +68,40 @@ use text_quality::{
 };
 use tounicode::FontCMaps;
 
+#[cfg(not(target_arch = "wasm32"))]
+struct ProcessingTimer(std::time::Instant);
+
+#[cfg(target_arch = "wasm32")]
+struct ProcessingTimer;
+
+impl ProcessingTimer {
+    fn start() -> Self {
+        #[cfg(not(target_arch = "wasm32"))]
+        {
+            Self(std::time::Instant::now())
+        }
+
+        #[cfg(target_arch = "wasm32")]
+        {
+            Self
+        }
+    }
+
+    fn elapsed_ms(&self) -> u64 {
+        #[cfg(not(target_arch = "wasm32"))]
+        {
+            self.0.elapsed().as_millis() as u64
+        }
+
+        #[cfg(target_arch = "wasm32")]
+        {
+            // The wasm32-unknown-unknown standard library has no clock.
+            // Browser bindings measure with JavaScript's host clock.
+            0
+        }
+    }
+}
+
 /// 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";
@@ -250,7 +284,7 @@ pub fn process_pdf_with_options<P: AsRef<Path>>(
     path: P,
     options: PdfOptions,
 ) -> Result<PdfProcessResult, PdfError> {
-    let start = std::time::Instant::now();
+    let start = ProcessingTimer::start();
     validate_pdf_file(&path)?;
 
     // Load the document once — shared by detection AND extraction.
@@ -277,7 +311,7 @@ pub fn process_pdf_mem_with_options(
     buffer: &[u8],
     options: PdfOptions,
 ) -> Result<PdfProcessResult, PdfError> {
-    let start = std::time::Instant::now();
+    let start = ProcessingTimer::start();
     validate_pdf_bytes(buffer)?;
 
     let (doc, page_count) =
@@ -3526,7 +3560,7 @@ fn process_document(
     doc: Document,
     page_count: u32,
     options: PdfOptions,
-    start: std::time::Instant,
+    start: ProcessingTimer,
 ) -> Result<PdfProcessResult, PdfError> {
     // Step 1 — Detection (cheap: scans content streams for text operators)
     let detection = detector::detect_from_document(&doc, page_count, &options.detection)?;
@@ -3542,7 +3576,7 @@ fn process_document(
             pdf_type,
             markdown: None,
             page_count,
-            processing_time_ms: start.elapsed().as_millis() as u64,
+            processing_time_ms: start.elapsed_ms(),
             pages_needing_ocr,
             ocr_reasons_by_page: page_ocr_reasons_vec(detection_ocr_reasons),
             title,
@@ -3558,7 +3592,7 @@ fn process_document(
             pdf_type,
             markdown: None,
             page_count,
-            processing_time_ms: start.elapsed().as_millis() as u64,
+            processing_time_ms: start.elapsed_ms(),
             pages_needing_ocr,
             ocr_reasons_by_page: page_ocr_reasons_vec(detection_ocr_reasons),
             title,
@@ -3824,7 +3858,7 @@ fn process_document(
         pdf_type,
         markdown,
         page_count,
-        processing_time_ms: start.elapsed().as_millis() as u64,
+        processing_time_ms: start.elapsed_ms(),
         pages_needing_ocr,
         ocr_reasons_by_page: {
             // Detector reasons (scanned / no_text / vector_text / garbled) merged

src/tounicode.rs

@@ -4,11 +4,17 @@
 
 use log::{debug, warn};
 use lopdf::{Document, Object, ObjectId};
+use std::borrow::Cow;
 use std::collections::{HashMap, HashSet};
+#[cfg(not(target_arch = "wasm32"))]
 use std::path::{Path, PathBuf};
 
 use crate::glyph_names::glyph_to_char;
 
+#[cfg(target_arch = "wasm32")]
+static BUILTIN_CMAPS: include_dir::Dir<'_> =
+    include_dir::include_dir!("$CARGO_MANIFEST_DIR/external/bcmaps");
+
 /// A parsed ToUnicode CMap mapping CIDs to Unicode strings
 #[derive(Debug, Default, Clone)]
 pub struct ToUnicodeCMap {
@@ -1149,9 +1155,7 @@ fn build_gid_to_unicode(face: &ttf_parser::Face<'_>) -> Option<HashMap<u16, char
 /// Build a ToUnicodeCMap from pdf.js built-in binary CMaps (bcmaps).
 fn build_cmap_from_builtin_cmap(ordering: &str) -> Option<ToUnicodeCMap> {
     let name = format!("Adobe-{}-UCS2.bcmap", ordering);
-    let dir = find_bcmaps_dir()?;
-    let path = dir.join(name);
-    let data = std::fs::read(&path).ok()?;
+    let data = read_builtin_cmap_file(&name)?;
     let mut cmap = parse_binary_cmap(&data).ok()?;
     if cmap.char_map.is_empty() && cmap.ranges.is_empty() {
         return None;
@@ -1159,13 +1163,14 @@ fn build_cmap_from_builtin_cmap(ordering: &str) -> Option<ToUnicodeCMap> {
     cmap.code_byte_length = 2;
     debug!(
         "Built-in CMap {}: char_map={} ranges={}",
-        path.display(),
+        name,
         cmap.char_map.len(),
         cmap.ranges.len()
     );
     Some(cmap)
 }
 
+#[cfg(not(target_arch = "wasm32"))]
 fn find_bcmaps_dir() -> Option<PathBuf> {
     if let Ok(dir) = std::env::var("PDF_INSPECTOR_BCMAPS_DIR") {
         let p = PathBuf::from(dir);
@@ -1182,6 +1187,18 @@ fn find_bcmaps_dir() -> Option<PathBuf> {
     None
 }
 
+#[cfg(not(target_arch = "wasm32"))]
+fn read_builtin_cmap_file(name: &str) -> Option<Cow<'static, [u8]>> {
+    let path = find_bcmaps_dir()?.join(name);
+    std::fs::read(path).ok().map(Cow::Owned)
+}
+
+#[cfg(target_arch = "wasm32")]
+fn read_builtin_cmap_file(name: &str) -> Option<Cow<'static, [u8]>> {
+    let file = BUILTIN_CMAPS.get_file(name)?;
+    Some(Cow::Borrowed(file.contents()))
+}
+
 fn parse_binary_cmap(data: &[u8]) -> Result<ToUnicodeCMap, String> {
     let mut stream = BinaryCMapStream::new(data);
     let _header = stream.read_byte().ok_or("unexpected EOF in bcmap header")?;
@@ -1492,9 +1509,7 @@ fn parse_encoding_cmap_object(obj: &Object, doc: &Document) -> Option<EncodingCM
 }
 
 fn load_builtin_encoding_cmap(name: &str) -> Option<EncodingCMap> {
-    let dir = find_bcmaps_dir()?;
-    let path = dir.join(format!("{}.bcmap", name));
-    let data = std::fs::read(&path).ok()?;
+    let data = read_builtin_cmap_file(&format!("{}.bcmap", name))?;
     parse_binary_cmap_encoding(&data).ok()
 }
 
@@ -1779,9 +1794,7 @@ fn load_builtin_cmap_by_name(name: &str) -> Option<ToUnicodeCMap> {
     if !name.ends_with("UCS2") {
         return None;
     }
-    let dir = find_bcmaps_dir()?;
-    let path = dir.join(format!("{}.bcmap", name));
-    let data = std::fs::read(&path).ok()?;
+    let data = read_builtin_cmap_file(&format!("{}.bcmap", name))?;
     let mut cmap = parse_binary_cmap(&data).ok()?;
     if cmap.char_map.is_empty() && cmap.ranges.is_empty() {
         return None;

wasm/Cargo.toml

@@ -0,0 +1,37 @@
+[package]
+name = "pdf-inspector-wasm"
+version = "0.1.2"
+edition = "2021"
+authors = ["Firecrawl Team"]
+description = "Browser WebAssembly bindings for pdf-inspector"
+license = "MIT"
+repository = "https://github.com/firecrawl/pdf-inspector"
+homepage = "https://github.com/firecrawl/pdf-inspector"
+readme = "README.md"
+publish = false
+
+[lib]
+crate-type = ["cdylib", "rlib"]
+
+[dependencies]
+console_error_panic_hook = "0.1"
+js-sys = "0.3"
+pdf-inspector = { path = ".." }
+serde = { version = "1", features = ["derive"] }
+serde-wasm-bindgen = "0.6"
+wasm-bindgen = "0.2"
+
+[dev-dependencies]
+wasm-bindgen-test = "0.3"
+
+[profile.release]
+codegen-units = 1
+lto = true
+opt-level = "s"
+strip = true
+
+[package.metadata.wasm-pack.profile.release]
+# Rust 1.95 emits bulk-memory instructions that the binaryen bundled with
+# wasm-pack 0.15.0 does not yet validate. rustc still performs the release,
+# size, and LTO optimizations above.
+wasm-opt = false

wasm/LICENSE

@@ -0,0 +1,58 @@
+MIT License
+
+Copyright (c) 2026 Firecrawl
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+Third-party notices
+===================
+
+Adobe CMaps
+-----------
+
+The WebAssembly binary embeds binary CMaps derived from Adobe CMap resources.
+
+Copyright 1990-2009 Adobe Systems Incorporated.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+Redistributions of source code must retain the above copyright notice, this
+list of conditions and the following disclaimer.
+
+Redistributions in binary form must reproduce the above copyright notice,
+this list of conditions and the following disclaimer in the documentation
+and/or other materials provided with the distribution.
+
+Neither the name of Adobe Systems Incorporated nor the names of its
+contributors may be used to endorse or promote products derived from this
+software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
+LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGE.

wasm/README.md

@@ -0,0 +1,60 @@
+# @firecrawl/pdf-inspector-wasm
+
+Browser WebAssembly bindings for [pdf-inspector](https://github.com/firecrawl/pdf-inspector). Classify PDFs and extract structured Markdown locally from a `Uint8Array`, using the same Rust core as the native Node.js, Python, and Rust packages.
+
+## Install
+
+```bash
+npm install @firecrawl/pdf-inspector-wasm
+```
+
+## Usage
+
+```ts
+import init, { processPdf } from "@firecrawl/pdf-inspector-wasm";
+
+await init();
+
+const response = await fetch("/annual-report.pdf");
+const pdf = new Uint8Array(await response.arrayBuffer());
+const result = processPdf(pdf);
+
+console.log(result.pdfType);
+console.log(result.markdown);
+```
+
+Pass options when you need selected pages or compact Markdown:
+
+```ts
+const result = processPdf(pdf, {
+  pages: [1, 3, 5],
+  profile: "compact",
+  includePageMarkers: true,
+});
+```
+
+The package also exports:
+
+- `detectPdf(pdf, options?)` for detection without extraction.
+- `classifyPdf(pdf)` for the lightweight result shape shared with the native Node.js API.
+- `extractText(pdf)` for plain text.
+- `version()` for the WASM package version.
+
+## Browser behavior
+
+- Parsing runs locally. PDF bytes are not uploaded anywhere.
+- The build is single-threaded and does not require cross-origin isolation.
+- CMaps are embedded so CJK font decoding does not depend on a filesystem.
+- Extraction is synchronous after `init()`. For large documents, call it from a Web Worker to keep the UI responsive.
+- Image-only documents still require a separate OCR step.
+
+## Build from source
+
+```bash
+cargo install wasm-pack --version 0.15.0 --locked
+wasm-pack build wasm --target web --scope firecrawl --release
+```
+
+## License
+
+MIT

wasm/src/lib.rs

@@ -0,0 +1,440 @@
+use pdf_inspector::{
+    LayoutComplexity, MarkdownProfile, PageOcrReasons, PdfOptions, PdfProcessResult, PdfType,
+    ProcessMode,
+};
+use serde::{Deserialize, Serialize};
+use wasm_bindgen::prelude::*;
+
+#[wasm_bindgen(typescript_custom_section)]
+const TYPESCRIPT_TYPES: &str = r#"
+export type PdfType = "TextBased" | "Scanned" | "ImageBased" | "Mixed";
+export type MarkdownProfile = "fidelity" | "compact";
+
+export interface ProcessOptions {
+  /** Restrict extraction to these 1-indexed page numbers. */
+  pages?: number[];
+  /** Password for an encrypted PDF. */
+  password?: string;
+  /** Source-faithful output by default, or compact output for fewer tokens. */
+  profile?: MarkdownProfile;
+  /** Insert `<!-- Page N -->` markers between pages. */
+  includePageMarkers?: boolean;
+  /** Include image placeholders in Markdown output. */
+  includeImages?: boolean;
+}
+
+export interface PageOcrReasons {
+  /** 1-indexed page number. */
+  page: number;
+  reasons: string[];
+}
+
+export interface LayoutComplexity {
+  isComplex: boolean;
+  /** 1-indexed page numbers. */
+  pagesWithTables: number[];
+  /** 1-indexed page numbers. */
+  pagesWithColumns: number[];
+}
+
+export interface PdfProcessResult {
+  pdfType: PdfType;
+  markdown?: string;
+  pageCount: number;
+  processingTimeMs: number;
+  /** 1-indexed page numbers. */
+  pagesNeedingOcr: number[];
+  ocrReasonsByPage: PageOcrReasons[];
+  title?: string;
+  confidence: number;
+  layout: LayoutComplexity;
+  hasEncodingIssues: boolean;
+}
+
+export interface PdfClassification {
+  pdfType: PdfType;
+  pageCount: number;
+  /** 0-indexed page numbers, matching the native Node.js API. */
+  pagesNeedingOcr: number[];
+  confidence: number;
+}
+
+export function processPdf(data: Uint8Array, options?: ProcessOptions): PdfProcessResult;
+export function detectPdf(data: Uint8Array, options?: Pick<ProcessOptions, "password">): PdfProcessResult;
+export function classifyPdf(data: Uint8Array): PdfClassification;
+export function extractText(data: Uint8Array): string;
+export function version(): string;
+"#;
+
+#[derive(Debug, Default, Deserialize)]
+#[serde(default, rename_all = "camelCase", deny_unknown_fields)]
+struct WasmProcessOptions {
+    pages: Option<Vec<u32>>,
+    password: Option<String>,
+    profile: Option<WasmMarkdownProfile>,
+    include_page_markers: Option<bool>,
+    include_images: Option<bool>,
+}
+
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "lowercase")]
+enum WasmMarkdownProfile {
+    Fidelity,
+    Compact,
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+struct WasmPageOcrReasons {
+    page: u32,
+    reasons: Vec<String>,
+}
+
+impl From<PageOcrReasons> for WasmPageOcrReasons {
+    fn from(value: PageOcrReasons) -> Self {
+        Self {
+            page: value.page,
+            reasons: value.reasons,
+        }
+    }
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+struct WasmLayoutComplexity {
+    is_complex: bool,
+    pages_with_tables: Vec<u32>,
+    pages_with_columns: Vec<u32>,
+}
+
+impl From<LayoutComplexity> for WasmLayoutComplexity {
+    fn from(value: LayoutComplexity) -> Self {
+        Self {
+            is_complex: value.is_complex,
+            pages_with_tables: value.pages_with_tables,
+            pages_with_columns: value.pages_with_columns,
+        }
+    }
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+struct WasmPdfProcessResult {
+    pdf_type: &'static str,
+    markdown: Option<String>,
+    page_count: u32,
+    processing_time_ms: f64,
+    pages_needing_ocr: Vec<u32>,
+    ocr_reasons_by_page: Vec<WasmPageOcrReasons>,
+    title: Option<String>,
+    confidence: f64,
+    layout: WasmLayoutComplexity,
+    has_encoding_issues: bool,
+}
+
+impl From<PdfProcessResult> for WasmPdfProcessResult {
+    fn from(value: PdfProcessResult) -> Self {
+        Self {
+            pdf_type: pdf_type_name(value.pdf_type),
+            markdown: value.markdown,
+            page_count: value.page_count,
+            processing_time_ms: value.processing_time_ms as f64,
+            pages_needing_ocr: value.pages_needing_ocr,
+            ocr_reasons_by_page: value
+                .ocr_reasons_by_page
+                .into_iter()
+                .map(Into::into)
+                .collect(),
+            title: value.title,
+            confidence: value.confidence as f64,
+            layout: value.layout.into(),
+            has_encoding_issues: value.has_encoding_issues,
+        }
+    }
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+struct WasmPdfClassification {
+    pdf_type: &'static str,
+    page_count: u32,
+    pages_needing_ocr: Vec<u32>,
+    confidence: f64,
+}
+
+fn pdf_type_name(pdf_type: PdfType) -> &'static str {
+    match pdf_type {
+        PdfType::TextBased => "TextBased",
+        PdfType::Scanned => "Scanned",
+        PdfType::ImageBased => "ImageBased",
+        PdfType::Mixed => "Mixed",
+    }
+}
+
+fn js_error(context: &str, error: impl std::fmt::Display) -> JsValue {
+    js_sys::Error::new(&format!("{context}: {error}")).into()
+}
+
+fn deserialize_options(value: JsValue) -> Result<WasmProcessOptions, JsValue> {
+    if value.is_undefined() || value.is_null() {
+        return Ok(WasmProcessOptions::default());
+    }
+
+    serde_wasm_bindgen::from_value(value).map_err(|error| js_error("invalid options", error))
+}
+
+fn build_options(value: JsValue, mode: ProcessMode) -> Result<PdfOptions, JsValue> {
+    let options = deserialize_options(value)?;
+    if options
+        .pages
+        .as_ref()
+        .is_some_and(|pages| pages.contains(&0))
+    {
+        return Err(js_error(
+            "invalid options",
+            "pages are 1-indexed; page 0 is invalid",
+        ));
+    }
+
+    let mut result = PdfOptions::new().mode(mode);
+    if let Some(pages) = options.pages {
+        result = result.pages(pages);
+    }
+    if let Some(password) = options.password {
+        result = result.password(password);
+    }
+    if let Some(profile) = options.profile {
+        result.markdown.profile = match profile {
+            WasmMarkdownProfile::Fidelity => MarkdownProfile::Fidelity,
+            WasmMarkdownProfile::Compact => MarkdownProfile::Compact,
+        };
+    }
+    if let Some(include_page_markers) = options.include_page_markers {
+        result.markdown.include_page_numbers = include_page_markers;
+    }
+    if let Some(include_images) = options.include_images {
+        result.markdown.include_images = include_images;
+    }
+    Ok(result)
+}
+
+fn serialize<T: Serialize>(value: &T) -> Result<JsValue, JsValue> {
+    serde_wasm_bindgen::to_value(value).map_err(|error| js_error("serialize result", error))
+}
+
+fn initialize() {
+    console_error_panic_hook::set_once();
+}
+
+/// Process PDF bytes entirely inside WebAssembly.
+#[wasm_bindgen(js_name = processPdf, skip_typescript)]
+pub fn process_pdf(data: &[u8], options: JsValue) -> Result<JsValue, JsValue> {
+    initialize();
+    let options = build_options(options, ProcessMode::Full)?;
+    let started = js_sys::Date::now();
+    let mut result = pdf_inspector::process_pdf_mem_with_options(data, options)
+        .map_err(|error| js_error("process PDF", error))?;
+    result.processing_time_ms = (js_sys::Date::now() - started).max(0.0) as u64;
+    serialize(&WasmPdfProcessResult::from(result))
+}
+
+/// Classify PDF bytes without extracting text or producing Markdown.
+#[wasm_bindgen(js_name = detectPdf, skip_typescript)]
+pub fn detect_pdf(data: &[u8], options: JsValue) -> Result<JsValue, JsValue> {
+    initialize();
+    let options = build_options(options, ProcessMode::DetectOnly)?;
+    let started = js_sys::Date::now();
+    let mut result = pdf_inspector::process_pdf_mem_with_options(data, options)
+        .map_err(|error| js_error("detect PDF", error))?;
+    result.processing_time_ms = (js_sys::Date::now() - started).max(0.0) as u64;
+    serialize(&WasmPdfProcessResult::from(result))
+}
+
+/// Return the lightweight classification shape used by the native Node API.
+#[wasm_bindgen(js_name = classifyPdf, skip_typescript)]
+pub fn classify_pdf(data: &[u8]) -> Result<JsValue, JsValue> {
+    initialize();
+    let result =
+        pdf_inspector::classify_pdf_mem(data).map_err(|error| js_error("classify PDF", error))?;
+    serialize(&WasmPdfClassification {
+        pdf_type: pdf_type_name(result.pdf_type),
+        page_count: result.page_count,
+        pages_needing_ocr: result.pages_needing_ocr,
+        confidence: result.confidence as f64,
+    })
+}
+
+/// Extract plain text from PDF bytes without Markdown conversion.
+#[wasm_bindgen(js_name = extractText, skip_typescript)]
+pub fn extract_text(data: &[u8]) -> Result<String, JsValue> {
+    initialize();
+    let items = pdf_inspector::extractor::extract_text_with_positions_mem(data)
+        .map_err(|error| js_error("extract text", error))?;
+    Ok(
+        pdf_inspector::extractor::group_into_lines_preserving_all_text(items)
+            .into_iter()
+            .map(|line| line.text())
+            .filter(|line| !line.trim().is_empty())
+            .collect::<Vec<_>>()
+            .join("\n"),
+    )
+}
+
+/// Return the WebAssembly package version.
+#[wasm_bindgen(skip_typescript)]
+pub fn version() -> String {
+    env!("CARGO_PKG_VERSION").to_string()
+}
+
+#[cfg(all(test, target_arch = "wasm32"))]
+mod tests {
+    use super::*;
+    use js_sys::Reflect;
+    use wasm_bindgen_test::*;
+
+    const TEXT_PDF: &[u8] = include_bytes!("../../tests/fixtures/thermo-freon12.pdf");
+    const ENCRYPTED_PDF: &[u8] = include_bytes!("../../tests/fixtures/encrypted-secret123.pdf");
+
+    fn synthetic_korea1_pdf() -> Vec<u8> {
+        let mut pdf = b"%PDF-1.4\n".to_vec();
+        let mut offsets = vec![0usize];
+
+        fn add_object(pdf: &mut Vec<u8>, offsets: &mut Vec<usize>, id: usize, body: &str) {
+            offsets.push(pdf.len());
+            pdf.extend_from_slice(format!("{id} 0 obj\n").as_bytes());
+            pdf.extend_from_slice(body.as_bytes());
+            pdf.extend_from_slice(b"\nendobj\n");
+        }
+
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            1,
+            "<< /Type /Catalog /Pages 2 0 R >>",
+        );
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            2,
+            "<< /Type /Pages /Kids [3 0 R] /Count 1 >>",
+        );
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            3,
+            "<< /Type /Page /Parent 2 0 R /MediaBox [0 0 200 200] /Resources << /Font << /F0 5 0 R >> >> /Contents 4 0 R >>",
+        );
+
+        // Adobe-Korea1 CID 1086 (0x043E) maps to U+AC00 (Korean syllable GA).
+        // There is deliberately no ToUnicode stream: decoding must use the
+        // embedded predefined CMap rather than lopdf's plain-text fallback.
+        // Korea1 CIDs 21 and 19 map to ASCII "4" and "2". Place them near
+        // the bottom edge so they look exactly like a numeric page footer.
+        let content = "BT /F0 12 Tf 50 100 Td <043E> Tj 0 -60 Td <00150013> Tj ET";
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            4,
+            &format!(
+                "<< /Length {} >>\nstream\n{}\nendstream",
+                content.len(),
+                content
+            ),
+        );
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            5,
+            "<< /Type /Font /Subtype /Type0 /BaseFont /SyntheticKorea1 /Encoding /Identity-H /DescendantFonts [6 0 R] >>",
+        );
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            6,
+            "<< /Type /Font /Subtype /CIDFontType2 /BaseFont /SyntheticKorea1 /CIDSystemInfo << /Registry (Adobe) /Ordering (Korea1) /Supplement 2 >> /FontDescriptor 7 0 R /DW 1000 >>",
+        );
+        add_object(
+            &mut pdf,
+            &mut offsets,
+            7,
+            "<< /Type /FontDescriptor /FontName /SyntheticKorea1 /Flags 4 /FontBBox [-100 -200 1000 900] /ItalicAngle 0 /Ascent 800 /Descent -200 /CapHeight 700 /StemV 80 >>",
+        );
+
+        let xref_start = pdf.len();
+        pdf.extend_from_slice(format!("xref\n0 {}\n", offsets.len()).as_bytes());
+        pdf.extend_from_slice(b"0000000000 65535 f \n");
+        for offset in offsets.iter().skip(1) {
+            pdf.extend_from_slice(format!("{offset:010} 00000 n \n").as_bytes());
+        }
+        pdf.extend_from_slice(
+            format!(
+                "trailer\n<< /Size {} /Root 1 0 R >>\nstartxref\n{}\n%%EOF",
+                offsets.len(),
+                xref_start
+            )
+            .as_bytes(),
+        );
+        pdf
+    }
+
+    #[wasm_bindgen_test]
+    fn processes_pdf_to_markdown() {
+        let result = process_pdf(TEXT_PDF, JsValue::UNDEFINED).expect("process PDF");
+        let pdf_type = Reflect::get(&result, &JsValue::from_str("pdfType"))
+            .expect("pdfType")
+            .as_string()
+            .expect("pdfType string");
+        let markdown = Reflect::get(&result, &JsValue::from_str("markdown"))
+            .expect("markdown")
+            .as_string()
+            .expect("markdown string");
+
+        assert_eq!(pdf_type, "TextBased");
+        assert!(!markdown.is_empty());
+    }
+
+    #[wasm_bindgen_test]
+    fn rejects_non_pdf_bytes() {
+        assert!(process_pdf(b"not a PDF", JsValue::UNDEFINED).is_err());
+    }
+
+    #[wasm_bindgen_test]
+    fn classifies_and_extracts_plain_text() {
+        let classification = classify_pdf(TEXT_PDF).expect("classify PDF");
+        let pdf_type = Reflect::get(&classification, &JsValue::from_str("pdfType"))
+            .expect("pdfType")
+            .as_string()
+            .expect("pdfType string");
+        let text = extract_text(TEXT_PDF).expect("extract text");
+
+        assert_eq!(pdf_type, "TextBased");
+        assert!(!text.is_empty());
+    }
+
+    #[wasm_bindgen_test]
+    fn extracts_cjk_and_preserves_numeric_page_footer() {
+        let text = extract_text(&synthetic_korea1_pdf()).expect("extract predefined CMap text");
+
+        assert_eq!(text, "가\n42");
+    }
+
+    #[wasm_bindgen_test]
+    fn opens_encrypted_pdf_with_password() {
+        assert!(process_pdf(ENCRYPTED_PDF, JsValue::UNDEFINED).is_err());
+
+        let options = js_sys::Object::new();
+        Reflect::set(
+            &options,
+            &JsValue::from_str("password"),
+            &JsValue::from_str("secret123"),
+        )
+        .expect("set password");
+        let result = process_pdf(ENCRYPTED_PDF, options.into()).expect("process encrypted PDF");
+        let markdown = Reflect::get(&result, &JsValue::from_str("markdown"))
+            .expect("markdown")
+            .as_string()
+            .expect("markdown string");
+
+        assert!(!markdown.is_empty());
+    }
+}