* fix(tables): bound disjoint-rect clustering so overlap tests stay subquadratic
MAX_CLUSTER_RECTS only helped when a component actually merged. Pairwise-disjoint drawing rects never hit that cap, so the all-pairs loop stayed O(n²). Sweep by left edge and cap AABB tests at 1e6.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): cap clustering overlap tests per rect, not globally
A page-wide AABB budget could be spent on a dense stack of disjoint drawings and never reach an independent table at a later X. Limit each rect to 256 later candidates so other X-ranges still cluster.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): cluster overlapping rects with a spatial grid
A per-rect cap in X-sort order could skip a same-X neighbor after 256 junk candidates. Hash rects into 64-pt cells and pair only inside each cell so independent regions still cluster and disjoint drawings stay subquadratic.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): cluster oversized rects via a bounded fallback
A span cap of 64 grid cells could omit the far end of a huge rect. Those rects now compare against every other rect (up to 32 oversized). Grid buckets are visited in sorted key order so union-find is deterministic.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): visit every oversized rect under a per-rect overlap budget
Dropping .take(32) on the oversized-span list so later page-wide rules still
union the cells they overlap. AABB tests stay capped per oversized rect.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): query overlapping grid cells for oversized cluster rects
Index-order scans starved later overlaps once a per-rect check cap filled
with disjoint drawings. Oversized spans now probe the cells they cover,
with Y-banded oversized-to-oversized unions so stacked page-wide rules
stay linear.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): range-query cluster grid cells for oversized rects
Scan only occupied rows in the oversized rect's Y range, then X-partition
those keys, so unrelated drawings are not visited. Band oversized-to-oversized
unions on the short axis instead of a per-rect huge-Y fallback.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): union crossing oversized cluster rects across orientation bands
Wide and tall page-spanning rules are indexed on different axes, so a
crossing pair never shared a bucket. Query the tall X-index from each wide
or dual-oversized rect, and insert dual-oversized spans into every coarse
Y cell they cover.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): skip quadratic wide-by-tall clustering when the product is huge
Cross-orientation union is only needed for a handful of page-spanning rules.
When |wide|×|tall| exceeds the per-cell pair cap, skip that pass so mixed
oversized drawings cannot go quadratic. Pair counts in a range query no
longer reset per band.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): count unique oversized candidates when querying X/Y bands
A tall rule occupying several X cells was charged once per cell against the
pair budget, which could skip a later overlapping partner. Deduplicate `j`
per query so the cap applies to distinct rects.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): bound Tj/TJ operand lookback to the previous operator
A missing `[` before `TJ` walked the entire prefix for every operator, so a compact `] TJ` stream was quadratic. Stop each lookback at the previous text/font operator so total work stays linear.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): skip strings and comments when scanning text operators
A `Tj` token inside a literal string was treated as an operator and pinned the lookback floor, so the real `Tj` could not see its operand. Skip literals, hex strings, and comments before matching operators.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): skip inline image data before string/hex scanning
A `(` or `<` byte in `BI`/`ID` sample data could enter string or hex state and hide every later text operator. Jump from `BI` to `EI` before applying those delimiter states.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): skip inline images by declared size, not the first EI
Sample bytes can contain `EI` followed by a token-like character. When Width/Height are present and the image is uncompressed, jump that many bytes before looking for `EI`; DCT images use JPEG EOI, and the generic scan requires the following bytes to look like PDF content.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): only trust inline-image length when the dict is complete
Require Width, Height, bits-per-component, and a known color space before skipping by size; pad each row to a byte; treat image masks as 1-bit. Drop the post-EI binary heuristic so a following non-ASCII string does not hide later text operators.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): keep a strict EI fallback for filtered inline images
Exact-length skips still accept a following non-ASCII string. Fallback scans require printable PDF after `EI` unless the next token starts a string, name, or array. Boolean image-mask values must end at a token boundary.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): stop the EI printable check at the next string token
A fallback scan of `EI` then `BT (` plus high-byte text was rejected because the 16-byte window included the string payload. Count binary-ness only until `(`, `<`, `[`, or `/`.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(detector): treat Tj/TJ as operators only after a string/array closer
Inline-image EI scanning cannot be made complete in this heuristic, and each attempt produced a new counterexample. Count Tj/TJ only when the previous token is `)`, `>`, or `]`: that keeps `] TJ` lookback linear and ignores `Tj` inside `(Hello Tj World)` without parsing BI/ID/EI.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): bound ToUnicode bfrange expansion during subset remap
Repeated full-width beginbfrange entries were expanded into individual
CID inserts on every copy. Stop after 65,536 assignments, matching the
existing /W and Encoding caps.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): document bfrange remap truncation and assert visit count
The 65,536 cap counts overwrites so repeated ranges cannot keep expanding.
The test now checks the assignment count, not just HashMap size (u16 keys
are always ≤ 65,536).
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): bound Encoding CMap cidrange expansion
Repeating full-width begincidrange declarations re-inserted the entire
16-bit domain on every copy. Stop after 65,536 assignments, matching the
existing /W cap.
Co-authored-by: Cursor <cursoragent@cursor.com>
* docs(extractor): clarify Encoding cidrange cap counts insert operations
The bound includes overwrites so repeated full-width ranges cannot keep
working after the map is full. Unique-key coverage alone would re-open
the CPU blow-up.
Co-authored-by: Cursor <cursoragent@cursor.com>
* docs(extractor): distinguish /W insert vs unique-key CID caps
Encoding cidrange and /W width assignment count every insert; the /W
unicode heuristic caps unique CIDs with the same 65,536 bound.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): cap content-stream decode before allocating operators
The 1M operation limit ran after lopdf materialized the full vector, so a
compact page of q/Q pairs could still abort under memory pressure.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): treat NUL and form-feed as PDF whitespace in op counting
Names must stop on the full PDF whitespace set so a following operator is
not absorbed into /Name, which would undercount and skip the decode cap.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): scan inline-image EI with the full PDF whitespace set
A missed EI terminator used to consume the rest of the stream and drop
later operators from the decode cap. If EI is absent, keep scanning.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(xobjects): track text line matrix and handle T*/TL/'/"/Tc/Tw in Form XObjects
The Form XObject text extractor in xobjects.rs is a separate hand-rolled
implementation of the operator state machine in content_stream.rs, and it had
drifted well out of parity:
- No text line matrix (TLM). `Td`/`TD` were applied to the text matrix already
advanced by `Tj`/`TJ`, so every line began where the previous line *ended*
instead of at the line start. Lines marched off the right edge and were
dropped as off-page.
- `T*` was not handled at all, so it never advanced to the next line.
- `TL`, `'` and `"` were missing, and `TD` never set the leading as a side
effect.
- `Tc`/`Tw` were hardcoded to 0.0 when computing advance widths, drifting
positions and inserting spurious spaces.
- Text state (Tc/Tw/TL/Tf) is part of the graphics state but was not saved or
restored by `q`/`Q`.
This matters well beyond an edge case: producers that emit a page stream of
just `q /X Do Q` and put all content in a Form XObject are common in
print-to-PDF and typesetting workflows, so this parser is on the hot path for
whole classes of real documents.
Measured on nycourts.gov 199AD3d.pdf (1370 pages, PDFlib producer, every page
wrapped in a Form XObject, 1331 pages using T*), word recall against a
pdftotext reference goes from 19.2% to 97.7% — 116k extracted words to 582k
against a 578k-word reference. On a 10-page subset, sequence similarity goes
from 27.3% to 97.7%, against 99.8% for Mistral OCR.
pdf-evals: 195 passed / 7 failed, byte-identical to the origin/main baseline
with the same failure list — no regressions.
Adds 7 unit tests covering the line-matrix-relative `Td`, `T*`, `TD` setting
leading, `'`, `"`, `Tc` advance widths, and `q`/`Q` text-state restore, all
driven through a page whose content is only `q /X1 Do Q`.
* fix(xobjects): restore fill colour across q/Q in Form XObjects
A white fill set inside a q/Q pair leaked past the Q, so all subsequent
text was treated as invisible and dropped. Save and restore fill_is_white
with the rest of the graphics state.
This was the cause of several long-standing extraction failures where
whole passages went missing or degraded into per-character garbage:
cambridge_excerpt (+8.5KB of recovered text), MTUAeroEngines (+7.4KB),
2025_findings-acl_668 (+1.4KB), HTM_02-01_Part_A (+1.3KB), and
HuttoISDWorkPerks / ebgt7isj04ophcq, which both went from exploded
per-character tables to clean prose.
Adds a regression test that fails without the restore (the text after Q
is dropped entirely).
Reported by cubic on #369.
* fix(extractor): merge small-caps runs so they stop reading as table columns
Typesetters render small caps as a full-size capital immediately followed by
shrunken capitals in the same font — `(R) Tj` at 9.98pt, then `(OLANDO) Tj`
at 6.74pt, touching. The 20% font-size band in merge_text_items split those
into separate items, and since table column boundaries cluster on item *start*
positions (find_column_boundaries never consults widths), each fragment
started far enough from the last to become its own column.
The result was garbled pseudo-tables. From 199AD3d.pdf p.5:
|OLANDO|COSTA|NIL|INGH||
|---|---|---|---|---|
|IANNE|ENWICK|ETER|OULTON||
|R|T. A|, P.J.|A|C. S|
|---|---|---|---|---|
now:
|ROLANDO T. ACOSTA, P.J.|ANIL C. SINGH|
|---|---|
|DIANNE T. RENWICK|PETER H. MOULTON|
Adds is_small_caps_continuation, gated tightly enough to exclude the other
reasons a smaller run follows a larger one: superscripts and footnote markers
(requires an uppercase *letter*, so digits never qualify), drop caps (the
following body text is mixed case), and adjacent cells or separate words
(requires the runs to be visually contiguous). A small-caps junction is
mid-word, so it also suppresses the space that would otherwise be inserted
("T. A" + "COSTA" -> "T. ACOSTA", not "T. A COSTA").
On 199AD3d.pdf this drops false-positive tables from 65 to 52 and lifts word
recall against a pdftotext reference from 97.7% to 97.8%.
Verified by running both binaries over all 203 eval PDFs and diffing outputs:
19 differ, 184 byte-identical. Beyond the reporter volume the merge also fixes:
Waters-Edge two more garbled heading-tables — "##### B. C R S" plus
"||OMPLIANCE|EPORTING YSTEM|" became
"##### B. COMPLIANCE REPORTING SYSTEM"
cn-student-handbook six TOC entries had collapsed to their initials;
"U M S H..." is now
"UNIVERSITY MISSION AND STUDENT HANDBOOK..."
546403 "(FAA)" + "ADVISORY CIRCULARS ( )" + a stray "CONT"
became "(FAA) ADVISORY CIRCULARS (CONT)"
ERP-2025 "T ABLE B-1" -> "TABLE B-1"
DMP-Keypad "THINLINE" + stray "TM KEYPADS" -> "THINLINETM KEYPADS"
zhaw / Stijn subscripted math variables: "*T* *G*" -> "*TG*"
One known regression, called out rather than hidden:
PA_PVEM_Sen_Waldo_Fernández (+1689 bytes). Its running footer genuinely is
small caps, so the merge correctly assembles the text (400 items -> 275), but
the now-contiguous footer lines align well enough that the heuristic detector
turns the wrapped document title into a 4-column table where it previously
rendered as bold prose.
Attempting to fix that in the detector was a dead end and is not included
here: gating the `num_cols >= 3` bypass in has_table_like_content on "no cell
has >= 12 words" removed 143 tables across 44 files, including correct ones —
MCF5235RM went 505 -> 483 and its register glossary degraded into a fused
column plus a ~100-word cell, because rejecting a good candidate lets a worse
fallback win. No wordiness threshold separates the two cases: PA_PVEM's
longest cell is 14 words while MCF5235RM's legitimate cells are <= 10. A
positional signal (running-header/footer bands, or cross-page repetition) is
the way in, and belongs in its own change.
Adds 7 unit tests: the two-column small-caps row from the reporter volume, and
rejection of superscript digits, drop caps, separate words, lowercase
continuations, and out-of-band size ratios.
* fix(extractor): tighten small-caps gate to cross-band junctions only
Three findings from cubic on #371, all valid.
1. Space suppression was too broad. `is_small_caps_continuation` accepted size
ratios up to 0.92, which is *inside* the 20% band that merge_text_items
already treats as the same size. Two similarly-sized uppercase words with a
small real word gap therefore had their space suppressed even though the
normal path would have merged them correctly with a space ("SEE" + "ALSO" ->
"SEEALSO"). The helper now requires the junction to *cross* the band, since
rescuing junctions the band would break is its only purpose; within-band
pairs keep the normal word-spacing logic. The band is now a shared
MERGE_FONT_SIZE_BAND constant so the helper and its caller cannot drift.
2. A trailing digit was skipped. The backward search for "the capital we are
continuing" skipped non-alphabetic characters, so text ending in a footnote
marker ("ANGELA M. MAZZARELLI1") found the earlier uppercase letter and
accepted the join. It now checks the actual trailing character.
Rejecting every trailing digit outright turned out to cost real quality, so
this keeps one narrow exception. In meetings_in_mass_july_2023 the source
reads "TUESDAY, JULY 4TH" with the ordinal suffix set as a smaller run; a
blanket digit rejection reverted that to "TUESDAY, JULY 4" plus a stray "TH"
leaking onto the next line, which is what main produces and what pdftotext
shows is wrong. Only the four English ordinal suffixes (TH/ST/ND/RD) may
follow a digit; anything else after one is treated as a footnote marker and
rejected, so the case cubic raised stays blocked.
3. A test's name did not match its data. `small_caps_merge_does_not_swallow_a_
second_column` claimed to exercise the 72pt column gap but contained only
the second column's items, so it just re-tested the happy-path merge. It now
holds the full nine-item row and asserts exactly two merged results,
"ROLANDO T. ACOSTA, P.J." and "ANIL C. SINGH", which genuinely exercises the
gap.
Both new guards were verified to be load-bearing: removing either one makes its
test fail.
The document that motivated the change is unaffected — small caps there run at
a 0.675 ratio, far outside the band — and 199AD3d.pdf p.5 still produces the
correct two-column justices table.
900 unit tests pass (10 covering small caps), fmt and clippy clean.
* fix(xobjects): track text line matrix and handle T*/TL/'/"/Tc/Tw in Form XObjects
The Form XObject text extractor in xobjects.rs is a separate hand-rolled
implementation of the operator state machine in content_stream.rs, and it had
drifted well out of parity:
- No text line matrix (TLM). `Td`/`TD` were applied to the text matrix already
advanced by `Tj`/`TJ`, so every line began where the previous line *ended*
instead of at the line start. Lines marched off the right edge and were
dropped as off-page.
- `T*` was not handled at all, so it never advanced to the next line.
- `TL`, `'` and `"` were missing, and `TD` never set the leading as a side
effect.
- `Tc`/`Tw` were hardcoded to 0.0 when computing advance widths, drifting
positions and inserting spurious spaces.
- Text state (Tc/Tw/TL/Tf) is part of the graphics state but was not saved or
restored by `q`/`Q`.
This matters well beyond an edge case: producers that emit a page stream of
just `q /X Do Q` and put all content in a Form XObject are common in
print-to-PDF and typesetting workflows, so this parser is on the hot path for
whole classes of real documents.
Measured on nycourts.gov 199AD3d.pdf (1370 pages, PDFlib producer, every page
wrapped in a Form XObject, 1331 pages using T*), word recall against a
pdftotext reference goes from 19.2% to 97.7% — 116k extracted words to 582k
against a 578k-word reference. On a 10-page subset, sequence similarity goes
from 27.3% to 97.7%, against 99.8% for Mistral OCR.
pdf-evals: 195 passed / 7 failed, byte-identical to the origin/main baseline
with the same failure list — no regressions.
Adds 7 unit tests covering the line-matrix-relative `Td`, `T*`, `TD` setting
leading, `'`, `"`, `Tc` advance widths, and `q`/`Q` text-state restore, all
driven through a page whose content is only `q /X1 Do Q`.
* fix(xobjects): restore fill colour across q/Q in Form XObjects
A white fill set inside a q/Q pair leaked past the Q, so all subsequent
text was treated as invisible and dropped. Save and restore fill_is_white
with the rest of the graphics state.
This was the cause of several long-standing extraction failures where
whole passages went missing or degraded into per-character garbage:
cambridge_excerpt (+8.5KB of recovered text), MTUAeroEngines (+7.4KB),
2025_findings-acl_668 (+1.4KB), HTM_02-01_Part_A (+1.3KB), and
HuttoISDWorkPerks / ebgt7isj04ophcq, which both went from exploded
per-character tables to clean prose.
Adds a regression test that fails without the restore (the text after Q
is dropped entirely).
Reported by cubic on #369.
Type0 /W parsing and the Unicode-CID heuristic expanded every CID in every
range. Repeating a full-width [0 65535 w] entry therefore re-materialized
the same 65,536-key domain on every copy, growing a temporary vector and
HashMap work without bound.
Cap expansion at the 16-bit CID domain, collect unique CIDs for the
median heuristic, and stop width assignment once that many entries have
been written. Legitimate compact /W arrays are unchanged.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): bound Form XObject expansion with invocation and operation budgets
Nested Form XObjects were only limited by recursion depth (5). An acyclic
graph where each form invokes the next N times still expands to N^depth
work, so a small PDF can force millions of nested /Do evaluations.
Share a per-page FormWalkBudget that caps 10,000 Form invocations and
1,000,000 operations walked across those expansions. Extraction stops
when either cap is hit. Legitimate shallow nesting is unchanged.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): share Form XObject budget across invisible-layer retry
extract_page_text_items created a fresh FormWalkBudget on every call, so
the invisible-text retry could consume a second full expansion budget for
the same page. Own the budget at the call site and pass it into both
passes.
Also charge form operations independently of the invocation cap so a form
that was already admitted can finish its stream.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Releases fix(regions) #351 — invisible (Tr 3) OCR text layers served from
the region extractor instead of falling back to GPU OCR.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(regions): serve invisible (Tr 3) OCR text layers instead of needs_ocr
Scanned pages (archive.org-style digitizations) carry their text as an
invisible render-mode-3 layer behind the page raster. extract_text_in_regions
extracted only visible items, so such pages yielded nothing but
'[Image: ...]' placeholders and every region fell back to OCR — while the
markdown path already includes the invisible layer for Mixed PDFs. The two
extractors disagreed about the same page.
Mirror the markdown path's gate, page-scoped: when the visible pass is
effectively textless (<40 non-placeholder alphanumerics), retry with
invisible text included and adopt the retry only when it contributes real,
non-garbage text. Pages with real visible text never retry, so double-layer
PDFs (visible text plus an invisible accessibility copy) are unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: strict zero-visible adoption gate, whole-layer garbage check, doc placement, discriminating tests
- Adoption now requires ZERO visible text on the page: the invisible pass
returns visible items too, so adopting alongside any visible text would
duplicate it. Strict gate instead of fuzzy dedupe; the dead
inv_alnum > visible_alnum condition goes with it.
- Garbage check judges the whole recovered layer, not the first 200 items.
- Constant/helper moved above the doc block so rustdoc stays attached to
extract_text_in_regions_mem.
- Tests: any-visible-text-blocks-adoption case (single short visible line)
with exactly-once assertions; mode-0 guard asserts occurrence count.
- Re-verified against real scanned-book pages: all recover (9.2-11.7K chars,
needs_ocr=false) — pure OCR-layer scans carry no visible text.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: gate adoption on visible-item presence, not alphanumeric mass
Punctuation-only visible text (zero alphanumerics) could still adopt the
invisible pass; its OCR twin in the layer would duplicate the glyphs. The
gate is now item-presence: any non-image item with non-whitespace text
blocks adoption (whitespace-only artifacts still tolerated). Pinned by a
punctuation-only fixture test; real scanned-book pages re-verified — all
recover unchanged (they carry no visible items at all).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: punctuation-gate test pins preservation, comment states fixture scope
The fixture's visible punctuation is a separate block, not mirrored in the
invisible layer — so it pins the gate, not the duplication scenario. The
doc comment now says so, and a positive assertion checks the punctuation
survives exactly once.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: retry only when invisible text was actually skipped; negative gate tests
- extract_page_text_items now reports skipped_invisible (4th return): the
Tr-3 suppression sites set it, so the region extractor retries only when
a recoverable layer exists. Blank pages and image-only scans without an
OCR layer — the common scanned case — no longer pay a second
content-stream parse.
- Negative tests: below-floor watermark layer and symbol-garbage layer are
both rejected (region keeps only the raster placeholder). needs_ocr
semantics for placeholder-only regions deliberately unchanged — that
contract predates this PR and downstream pipelines handle it.
- Real scanned-book pages re-verified: recovery unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: flag the ' show-text suppression path; require nonempty strings
- P1: invisible layers shown via the ' operator never set skipped_invisible,
so those pages kept the placeholder and fell back to GPU OCR. The '
suppression path now flags too — pinned by a fixture whose layer is shown
entirely via ' (nothing through Tj/TJ).
- P3: all three flag sites (Tj, TJ, ') require a nonempty string operand —
numeric-only TJ kerning arrays and empty shows no longer trigger the
invisible reparse.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Tagged PDFs carry a structure tree with real heading roles (H1..H6), and
the core already parses it (structure_tree::StructTree) and threads MCIDs
onto TextItem — but neither surfaced through the bindings.
- Expose TextItem.mcid (Option<i64>) through the napi and pyo3 bindings,
matching the core field added with the marked-content extractor.
- Add StructRole::name(), the inverse of from_name, so roles have a
stable string form.
- Add extract_structure_elements / extract_structure_elements_mem to the
core: one (page, mcid, role) entry per marked-content reference, sorted
by (page, mcid), empty for untagged PDFs. Pages are 1-indexed to match
TextItem.page, so results join directly against
extract_text_with_positions output.
- Bind it as extractStructureElements (napi) and
extract_structure_elements / extract_structure_elements_bytes (pyo3),
with type-stub updates in pdf_inspector.pyi.
- Cover the join in Rust integration tests, napi test.mjs, and pytest,
using the existing firecrawl_docs_tagged.pdf fixture (tagged) and
thermo-freon12.pdf (untagged).
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(structure-tree): bound tagged /K parsing against alias/cycle DoS
A struct element that references itself (or an ancestor) through /K — e.g.
/K [n 0 R n 0 R] — made parse_struct_element_dict branch exponentially:
the depth cap (64) alone still permits 2^depth materialized nodes, so a
~830-byte PDF exhausts memory (OOM, exit 134).
Add a StructWalk carrying (1) an active-path set of object IDs so a node
that references itself/an ancestor is not re-expanded (breaks self- and
mutual-reference cycles cheaply), and (2) a global node budget
(MAX_STRUCT_NODES) that caps total materialization for aliased/DAG-shaped
graphs of distinct objects the path guard cannot catch.
Adds regression tests for self-alias, mutual-alias, and the aliased-DAG
budget cap.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): charge /K content refs against the node budget
The per-node budget only covered materialized struct elements and child
recursion; bare MCIDs and MCR dicts in a /K array append to content_refs
without charging it, so one element with a very wide /K array could still
allocate content_refs without bound. Charge every /K array item before
handling it, and stop the top-level /K loop once the budget is spent, so
content refs and loop work are bounded too. Adds a wide-MCID-array test.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): charge /K budget per materialized item, not per array entry
Charging every /K array item double-counted structural children (charged
here and again at their node entry) and charged cycle-skipped references
that materialize nothing, draining the budget up to ~2x faster than the
per-node semantics and risking early truncation of large legitimate trees.
Charge only the unbounded content-ref items (bare MCIDs and MCR dicts);
structural children remain charged once at their node entry.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* refactor(structure-tree): charge every content ref uniformly via helper
Route all budget charges through StructWalk::charge() so every
marked-content reference is charged once, including the single-value /K
branches (bare integer and MCR dict) that previously appended without
charging. charge() also guards against underflow, so charging after the
node-entry charge (which can leave the budget at 0) is safe. Makes the
documented per-item budget contract hold uniformly across all branches.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* feat(structure-tree): log once when the node budget truncates parsing
Add a one-shot truncation flag on StructWalk, set the first time the
budget is exhausted, and emit a single warn! after parsing so an operator
can tell when a (very large or malformed) tagged tree was cut off. Avoids
per-item log spam; negligible overhead on the normal path.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): flag truncation at budget guards, not just in charge
The truncation flag was only set inside charge() on the budget==0 branch,
but the dominant skip paths use budget==0 guards that break/return before
charge() is ever called with an empty budget, so the flag (and the warn!)
almost never fired. Route those guards through a new exhausted() that sets
the flag when it skips remaining work. Adds a parser-level test that would
have caught the missed warning.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): flag cycle/depth skips and charge bare MCIDs fully
Two review follow-ups:
- Cycle-broken and depth-capped /K skips dropped tagged content without
setting the truncation flag, so the one-shot warning never fired for
malformed/over-deep trees. Mark those skips via note_skipped() and
broaden the warning to cover non-budget truncation.
- A bare /K MCID materializes a wrapper node AND a content reference but
charged only one budget unit, allowing ~2x the advertised budget for
such content; charge both.
Adds tests: cycle-skip flags truncation, and bare MCID charges two units.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): charge MCR-dict wrappers the same two units as bare MCIDs
A top-level MCR /K dict flows through parse_kid -> parse_struct_element_dict
and materializes a Span node + one content ref (two items) but was charged
only one unit at node entry, while the bare-MCID path charges two. Charge
the content reference in the MCR branch too so the per-item budget is
uniform across both wrapper paths. Adds a symmetric test.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): reserve leaf-wrapper budget units atomically
A leaf MCID wrapper (bare MCID or MCR dict) materializes a node + one
content ref and charged the two units via separate charge() calls. At the
last unit the first charge succeeded and the second failed, consuming a
unit without emitting the wrapper and denying it to a later element that
would have fit. Add charge_n() to reserve both units atomically (or
neither), and detect MCR before the node charge so it reserves both up
front. Adds a boundary test asserting the leftover unit is preserved.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): stop scanning wide /K once a leaf reservation stalls
The atomic charge_n(2) left budget nonzero (==1) when it failed, so
exhausted() (budget==0) never broke the root /K loop and a crafted wide
array of leaf wrappers was scanned in full after no leaf could fit. Add a
stalled flag set on an insufficient reservation and fold it into
exhausted(); charge()-based (one-unit) loops are unaffected since they
reach budget 0 exactly. Adds a test that a one-unit budget still allows a
one-unit item but a failed two-unit reservation stops the scan.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(structure-tree): add traversal budget and stop charging non-materializing dicts
Two review follow-ups on budget accounting:
- Wide /K arrays of non-materializing items (unsupported value types, OBJR
dicts, cycle back-edges) consumed no node budget, so the loop scanned the
whole array. Add a separate work budget charged per examined /K item and
break the loops when it is spent, bounding traversal even when nothing
materializes.
- OBJR dicts and dicts without a valid /S were charged the node budget before
being recognized and skipped, draining the shared budget and truncating
real content later. Hoist the OBJR check and /S validation above the node
charge so only materializing nodes consume it (matching the MCR hoisting).
Adds tests for the work-budget bound, wide unsupported /K, and non-materializing
dicts not charging the node budget.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* docs(structure-tree): mention traversal budget in truncation warning
The one-shot truncation warning listed the node budget, cycle, and depth
as causes but not the new traversal (work) budget, so a work-budget
truncation printed a misleading message. Include MAX_STRUCT_WORK so
malformed-PDF debugging identifies the actual limit hit.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* feat(napi): add processPdfAsync, classifyPdfAsync, extractPagesMarkdownAsync
The Node bindings are synchronous, so every call parses on the event
loop thread — up to hundreds of milliseconds of dead loop per document
in a server. Add additive AsyncTask-based variants that run the same
shared implementations on the libuv thread pool and return promises.
The existing synchronous exports keep their names, signatures, and
behaviour; each sync/async pair shares one implementation. Panics in
compute() are caught and surfaced as rejections, matching the sync
error contract.
Closes#336
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(napi): read async task buffers in place instead of copying
Review feedback on #337: buffer.to_vec() copied the whole PDF on the
event loop before the task was queued, so large inputs still stalled
the loop and doubled peak memory. The tasks now hold the napi Buffer
itself — its ref pins the JS allocation for the task's lifetime and
the backing store is stable, so compute() reads it directly from the
worker thread. Callers must not mutate the buffer until the promise
settles (same contract as Node's async fs APIs); documented on each
export and in the README.
The suggested removal of ts_return_type was checked and rejected:
without it napi-rs generates Promise<unknown> for AsyncTask returns.
A comment now records that finding.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(napi): copy async task input on the JS thread for soundness
Review feedback on #337: holding the napi Buffer and reading it from
the libuv worker was unsound. Buffer derefs straight to the JS-side
allocation, so a caller mutating it before the promise settled would
race the worker's reads — undefined behavior, not a recoverable error,
and the documented don't-mutate contract was unenforceable. Deferring
the copy to compute() would not help: any off-thread read races the
same way. The JS thread is the only race-free place to take the copy,
because JS is single-threaded and nothing can mutate the buffer during
the synchronous part of the call.
Revert to an owned Vec<u8> copied at call time. The cost is one memcpy,
negligible next to the parse the async variants exist to unblock. Docs
now state the buffer may be reused or mutated immediately, and a test
locks in the copy semantics by mutating the input while a parse is in
flight.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Bound column-detection histogram size
Derive the projection histogram from a clamped bin count and skip
non-finite page widths. Extreme or malformed text-item coordinates
(from the content-stream text matrix) could otherwise drive a very
large allocation. 65,536 bins is ~9x the largest legal page, so real
layouts are unaffected. Adds regression tests.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Exclude non-finite coordinates from page bounds
Items at NaN/inf positions are now skipped when folding the page bounds,
so a malformed coordinate can no longer escape as a ColumnRegion
boundary, and an all-non-finite page returns no columns. Bad items are
dropped individually rather than failing the page, so one stray glyph
does not disable column detection.
Addresses review feedback on the finite-width guard.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Trim far-outlier coordinates from page bounds
Gutter margins, spanning-item width and the XY-cut margin are all
fractions of page_width, so a single far-but-finite item (x=50_000 is
enough) set the scale for the whole page: real gutters fell inside the
rejected margin band and a genuine two-column page collapsed to one
region. When the span exceeds one legal page (14_400 units), re-derive
the bounds from items clustered around the median x. Outliers keep their
text because column assignment buckets by nearest overlap.
The MAX_BINS ceiling stays as an allocation bound that does not depend
on this heuristic.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Harden bounds trimming against widths and wide layouts
Check both item edges when trimming: a malformed width at an ordinary
position poisoned x_max just as a malformed position poisoned x_min, so
a huge width still collapsed a two-column page to one region.
Only trim when the far items are a small minority (<=10%). A genuinely
large-format page has content spread across its full width, so it now
keeps its true bounds instead of being reduced to the median cluster.
Correct the MAX_PAGE_EXTENT comment: 14_400 units is the traditional
Acrobat architectural limit, not a format cap. PDF 2.0 sets no page-size
limit and UserUnit scales physical size, so this is a heuristic.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Scale bin width so the histogram spans the whole page
Clamping the bin count alone left anything past MAX_BINS * BIN_WIDTH
(~131k points) outside the histogram, folded into the final bin. A page
wide enough to hit that lost real gutters: with a visible gutter inside
the covered range the XY-cut fallback never runs, so a three-column
layout silently reported two. Derive bin_width from page_width instead,
keeping the same allocation ceiling and degrading only resolution.
Also anchor the trimming median on the same finite left/right items that
bounds() accepts, so a malformed width cannot shift which items count as
strays.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Require geometric evidence before detaching far content
The median-window trim narrowed any content more than one page from the
centre, so a valid large page with a sparse far sidebar lost the sidebar
from its bounds and its text fell into column 0. An item-count minority
rule cannot tell that layout from malformed coordinates.
Group content into clusters separated by more than a whole page of
continuous emptiness, and only drop a cluster that is both detached by
such a void and a small minority of items. Real content does not leave a
gap that large; a stray coordinate sits alone beyond one.
A single run wider than one page is treated as a malformed width, which
also covers the huge-width case the cluster sweep cannot see (such an
item spans everything and leaves no gap).
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Judge run width against page content, not a fixed extent
Treating any run wider than 14_400 units as malformed penalised valid
large pages: one made entirely of such runs reported no columns at all,
and a mixed page lost the right edge of every long run.
Judge width relative to the page's own content instead. Positions cannot
be inflated by a bogus width, so the spread of the core cluster is a
sound scale: a run wider than that spread plus one page is malformed.
A genuinely large page keeps its genuinely long runs, while a 1e12-wide
run beside ordinary text is still rejected.
Cluster on positions rather than filled intervals, so a bogus width can
no longer merge everything into one cluster, and keep ordinary pages on
an O(n) fast path that skips the sort entirely.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Update SECURITY.md reporting channels
Clarify that email is the only required channel and point the
alternative at Firecrawl's Bugcrowd disclosure engagement instead of
the private-advisory link, which is not enabled on this repo.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* Make Bugcrowd the preferred reporting channel
Bugcrowd's disclosure engagement is the primary channel; email to
help@firecrawl.dev is offered as the alternative.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
Use hex.get(i..i+2) instead of &hex[i..i+2] so a non-hex, non-ASCII
destination in a /ToUnicode CMap can no longer trigger a UTF-8
char-boundary panic. An even byte length does not guarantee the byte
offset falls on a char boundary; get() returns None on a non-boundary
or out-of-range index, folding cleanly into the existing flow.
Add regression tests covering a multi-byte destination char and a
replacement-char byte.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
Use str::get instead of a byte-length check plus slice when parsing the
uniXXXX glyph-name form. The byte-length guard only proved the index was
in bounds, not on a UTF-8 char boundary, so a glyph name containing
non-ASCII bytes could cause a slice on a non-boundary index. Switch to a
checked slice that folds into the existing Option flow, and add tests.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): guard AcroForm /Kids traversal against cycles and huge trees
A crafted PDF whose AcroForm field lists itself (or another ancestor) in
/Kids caused walk_form_fields to recurse indefinitely, overflowing the
stack and aborting pdf2md (exit 134) — an application-level DoS from a
~730-byte input.
Track visited field object IDs to break /Kids cycles, and cap total
field-node traversal at 100k nodes to bound pathologically large trees.
Adds regression tests for self-cycle and mutual-cycle field graphs.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): cap AcroForm /Kids recursion depth to stop deep-chain overflow
The visited-set guard stops cyclic /Kids graphs, but a long *acyclic*
chain of distinct fields still recurses to the chain length and overflows
the stack (a ~1.6MB PDF with 20k linked fields aborts pdf2md, exit 134)
before the 100k node budget is reached.
Add an explicit recursion depth cap (100 levels — far above any legitimate
form hierarchy) so stack usage is bounded independently of node count.
Adds a deep-acyclic-chain regression test.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): enforce form-field node budget before insertion
The node-budget guard inserted each field ID into the visited set before
checking the budget, so the check triggered an early return but never
actually capped the set. A field with a huge /Kids array kept inserting
post-budget IDs, letting visited (memory and work) grow with the crafted
input rather than stopping at MAX_FORM_FIELD_NODES.
Check depth and budget before inserting, so visited can never exceed the
cap. Adds a wide-tree regression test.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): stop /Fields and /Kids iteration once node budget is spent
Checking the budget before insertion capped the visited set, but callers
still iterated every remaining entry of a wide /Fields or /Kids array
after the budget was exhausted — each walk returned immediately, yet the
O(N) sibling iteration let a single multi-million-entry array burn
extraction CPU unbounded. Break out of both the top-level and recursive
loops once visited reaches the cap, making the budget a true
traversal-work cap. Adds a top-level wide-/Fields regression test.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): charge examined entries against the field-node budget
The budget counted only distinct visited nodes, so /Fields or /Kids
arrays full of invalid (non-reference) or duplicate entries never grew
visited and ran to completion regardless of size — the node budget did
not actually cap traversal work.
Introduce FieldWalkBudget tracking both visited nodes and total entries
examined; charge every array entry (valid, invalid, or duplicate) and
stop once either hits MAX_FORM_FIELD_NODES. Adds a regression test with a
huge /Kids array of duplicate + null entries.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(links): iterate /Fields and /Kids arrays by borrow, not clone
Both arrays were cloned in full before the budget check, so a crafted
oversized /Fields or /Kids array forced an O(n) allocation and copy
regardless of the cap. resolve_array already returns a borrow tied to the
document and the walker only needs a shared &Document, so iterate the
borrowed arrays directly — the early break now bounds how many entries
are even touched, before any per-array allocation.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* docs(links): correct wide-array test comments to match range assertions
The two wide-array tests assert item counts within a range near the
budget, not an exact value (charging entries in the entry guard shifts
the boundary by one or two). Fix the stale comments that claimed exact
counts.
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Abimael Martell <abimaelmartell@users.noreply.github.com>
* fix(tables): exclude script attachments and tiny numeric fragments from detection
Split out of #242 (draft) so it can be reviewed on its own evidence.
Display equations with sub/superscripts form phantom small-font table
regions: the subscripts cluster with nearby small text (footnotes, axis
labels) into fake multi-column grids. Two guards:
- Script attachment: a small-font item horizontally adjacent to a
larger-font item at a genuine baseline offset is a sub/superscript,
not a table cell, and is excluded from candidates. A real baseline
offset is required so a small cell beside a larger same-baseline
label is never filtered. Attachment targets are indexed by Y and
scanned through a bounded window rather than a full-page sweep.
The body-font pass applies the same exclusion but only for
heading-sized anchors (>= 1.15x base), so body-size table cells
beside slightly larger labels are untouched.
- Tiny numeric fragments: a <=2-row grid whose every cell is a bare
1-2 digit number carries no tabular information. Restricted to the
small-font pass, where the pattern is overwhelmingly exponent
clusters; body-font numeric grids are unaffected.
Corpus impact: 20 of 186 documents, measured against a control build of
main so main's own drift is excluded. Table rows fall in 17 of 18
inspected documents and no content is lost — 2103_07786 drops all 21
rows, every one a math fragment ('|X 42 43|1|||'); Stijn_SB_doc drops
173 rows of footnote text that had been shredded into cells, with word
count slightly UP and footnote markers intact. M2019_mordeste gains 11
rows from a 3-column grid re-detected as 2-column, neither clearly
better nor worse.
Note the 20 documents is far more than the 3 that per-change ablation
suggested: that figure measured sole-cause attribution inside the
original combined PR, where other heuristics changed the same files and
masked this one. Reach and sole-cause are different measurements.
805 unit + 148 integration tests pass, clippy clean, in-repo snapshots
unchanged.
* fix(tables): suppress script column evidence instead of dropping candidates
Reworked after reviewing the corpus diffs: the first approach removed
sub/superscripts from table candidates entirely, which had two failure
modes beyond the intended fix.
- Legitimate cell content was displaced. citizen-sr-282 is a calculator
manual whose engineering-notation table lists M = 10^6, k = 10^3.
Those exponents are superscripts, so they were dropped from the table
and resurfaced elsewhere in the reading order ('9 mega 6 kilo = 10 3
milli').
- Removing items changed the candidate geometry, so different spurious
structure could form from what remained.
Scripts are now kept as candidates and excluded only from the geometry:
they cannot create a column (find_column_boundaries), cannot qualify a
region on their own (find_table_regions / _strict), but are still
assigned to cells. Column alignment is validated against ALL items
including scripts — validating only the non-script subset would let a
region manufacture alignment by ignoring its awkward items, which is
what block-diagram pages did.
Corpus: 20 of 186 documents, net -359 table rows, no content lost.
Token-level comparison shows the only text changes are merges in the
right direction: 'X' + '10' becomes 'X10', 'L' + 'g' becomes 'Lg' —
subscripts joining their base instead of floating free.
Remaining artifact: MCF5235RM (and its _nxp duplicate) gains a small
spurious table from a block-diagram label line, and M2019_mordeste
gains 11 rows from a 3-column grid re-detected as 2-column. Both are
borderline regions where the previous output was also wrong; documented
rather than tuned away.
805 unit + 148 integration tests pass, clippy clean.
* fix(tables): use a heading-anchored script mask in the body-font pass
Cubic review of #264: a single script mask computed with a 0.0 anchor
was applied to both passes, including body-font region qualification and
geometry. The body pass is supposed to require a heading-sized anchor —
that distinction existed before the geometry rework and was lost in it.
Why it matters: body-pass candidates are themselves body-sized
(0.85..1.05x base). A cell at the low end of that band, say 8.5pt,
sitting beside a 10.5pt label clears the inherent 'anchor >= 1.2x cell'
rule (10.2) and so was flagged as a script attachment. At body sizes a
slightly larger neighbour is a bold label or column header, not the base
of a superscript, so flagging it stripped real cells out of the region
evidence and column geometry and could lose the table entirely.
Two masks now: the small-font pass keeps the 0.0 anchor, the body pass
requires >= 1.15x base. Note the threshold only bites below base size —
for a cell at base, 1.2x-of-cell already exceeds 1.15x-of-base — which
is exactly the 0.85..1.0x band cubic identified.
Corpus: 20 documents, net -367 table rows (was -359 with the single
mask), so the body pass now keeps 8 rows of real table it had been
discarding. 958 tests pass, clippy clean.
* fix(markdown): reject headings that end on a relational verb
A heading candidate ending in 'equals', 'denotes', 'implies' and the
like is the first half of a sentence, not a title. This shows up when a
block dissolves and strands its lead-in ahead of the formula it
introduced — opendataloader 01030000000144 produced
## Note that the exact error equals
M - Q(h) = e - 2.7525... = -0.0342....
Deliberately a very short list. Broader variants were tried and
measured, then rejected:
- Function words (of/and/for/the): a heading that WRAPS across lines
ends on exactly those. Destroyed real IRS Publication 17 headings —
'Casualty and' -> 'Casualty and Theft Losses', 'Rule 10. You Must Be
at' -> '... At Least Age 25'. 52 documents affected, -619 headings.
- Copulas and auxiliaries (is/are/be/have): same failure. 'Rule 15.
Your AGI Must Be', 'What Medical Expenses Are' and 'When Can a Roth
IRA Be' are real wrapped headings, while 'the tax burden should be'
is a genuine fragment. The trailing word cannot separate them; that
needs the next line's context, which this text-only predicate lacks.
The verbs kept never end a heading in any register, so they are safe
without context. Standalone the guard is a no-op on both benchmarks
(0 documents on opendataloader, 4 on pdf-evals with no net heading
change) — its value is as a companion to the table filter in this PR,
which is what strands these lead-ins.
Combined effect on opendataloader (200 docs, vs a control build of
main), where the table filter alone regressed:
table filter + this guard
overall -0.0003 +0.0003
mhs -0.0019 +0.0003
doc ...144 -0.063 +0.053
doc ...144 mhs -0.203 +0.028
* review: gate the dangling-verb veto on sentence case, drop 'yields'
Cubic review of a5a6e8f — both findings valid.
1. 'yields' is also a plural noun. 'Bond Yields', 'Crop Yields' and
'Dividend Yields' are real section titles in financial documents,
which this corpus contains. Removed from the list; my claim that
these verbs 'never end a heading in any register' was wrong for it.
2. A wrapped title-case heading whose first line ends on one of these
verbs would be suppressed if the heading preprocessor failed to
merge it.
Both are fixed by the same gate, which is the discriminator I was
missing: case. A heading is title case ('Bond Yields', 'The Theorem
Implies'); a stranded lead-in is sentence case ('Note that the exact
error equals', 'the method yields'). The veto now applies only when
every content word is NOT capitalized, so titles are spared regardless
of their final word.
This is also why the earlier function-word and copula variants failed:
they had no way to tell 'Rule 15. Your AGI Must Be' from 'the tax
burden should be'. Case separates those two as well.
No measured cost. opendataloader is unchanged from the previous
revision — overall +0.0003, mhs +0.0003, doc 01030000000144 still
0.732 -> 0.785 — and pdf-evals still 20 documents. 963 tests pass,
clippy clean.
* review: exempt section-numbered lines from the dangling-verb veto
Valid ordering bug. heading.rs consults is_heading_fragment at line 282
and only applies its numbered-prefix allowance at line 288, so the veto
pre-empted it: '1. What the model implies' is sentence case and ends on
a listed verb, so it was discarded before numbering could vouch for it.
Numbering is independent evidence of a heading, so the veto now skips
any line opening with a section number.
Acceptance is deliberately a little broader than heading::parse_numbering
(which requires a trailing delimiter) because '2.3 Section Title' is
written without one, and being permissive in a veto exemption can only
avoid suppressing headings. Two guards keep it from swallowing prose:
- a bare single number needs a delimiter ('1.' yes, '3 apples' no)
- roman numerals always need one, since a leading 'I' is the pronoun far
more often than a section number
Not reused from convert::starts_with_section_number, which deliberately
demands two components because it bypasses isolation checks — that would
reject the reviewer's single-'1.' case.
No measured change: opendataloader still overall +0.0003 / mhs +0.0003
with doc 01030000000144 at 0.732 -> 0.785, pdf-evals still 20 documents,
target case still suppressed. 964 tests pass, clippy clean.
* review: share roman_value so the veto exemption matches the parser
Valid. My numbering predicate accepted tokens heading::parse_numbering
rejects — lowercase 'iv)', alphabetical 'd)', over-long 'MMMM.' — because
it case-folded and allowed D and M. Anything the parser rejects is not
numbering, so exempting it let ordinary list items bypass the
dangling-verb veto and reach font-based heading promotion.
Rather than restate the grammar, roman_value is now pub(super) and the
exemption calls it, so the two cannot drift. Its rules apply as written:
uppercase I/V/X/L/C only, at most 8 characters, positive total.
Decimal numbering keeps its slightly broader acceptance (bare '2.3' with
no trailing delimiter), which is deliberate and documented — that form is
common in real headings and being permissive in a veto exemption cannot
manufacture a heading, only decline to suppress one. The roman case is
different because single letters collide with alphabetical list markers.
No measured change: opendataloader overall +0.0003 / mhs +0.0003, doc
01030000000144 still 0.732 -> 0.785. 964 tests pass, clippy clean.
* test: cover the roman length bound with a nine-character token
Valid P3. The 'MMMM.' case fails on the unsupported M, not on length, so
the 8-character bound in roman_value had no coverage and could regress
silently. Added a nine-'I' token, which is rejected only by the bound,
plus an eight-'I' token that must stay exempt to pin the boundary from
both sides.
* fix(tables): stop dropping body-band scripts from the candidate set
Valid: the body-font pass filtered scripts out of body_candidates
itself, so body_script_flags and its two downstream uses were dead. The
mask filters region_evidence and feeds detect_table_in_region's is_script
closure, but neither ever saw a script item because the candidate set no
longer contained any.
Consequences: a body-band sub/superscript attached to a heading-sized
anchor was dropped from the table outright rather than assigned to a
cell, so its text was lost — the opposite of what both the
body_script_flags comment ('they stay candidates') and the
detect_table_in_region docstring ('they remain eligible for cell
assignment') describe, and inconsistent with the small-font pass.
Root cause: the geometry rework removed the candidate-level filter from
the small-font pass, but the body one had been reflowed onto a single
line by rustfmt so the same edit missed it. Adding body_script_flags in
a later review then wired a mask that the surviving filter made
unreachable.
No measured change on either benchmark — pdf-evals still 20 documents
and -367 table rows, opendataloader still overall +0.0003 / mhs +0.0003
with one document changed — because the combination it affects (a
body-sized script attached to a heading-sized anchor) does not occur in
either corpus. The fix is for correctness and consistency between the
two passes, not for a score.
964 tests pass, clippy clean.
* fix(extractor): supply built-in metrics for non-embedded base-14 fonts
PDFs may legally omit /Widths for non-embedded standard fonts (Times,
Helvetica, Courier, Symbol, ZapfDingbats) — the spec requires the reader
to supply the metrics. We returned None, so every glyph advanced 0 and
each text item got width 0, silently breaking every gap-based heuristic
downstream: space synthesis, sub/superscript detection, table column
detection, heading merging.
- src/extractor/base14.rs: Adobe Core-14 AFM width tables keyed by
Unicode char, plus the standard Symbol/ZapfDingbats encoding vectors
(their glyphs sit at byte positions unrelated to Latin text, so widths
must resolve through the built-in encoding, not cp1252)
- Width resolution order: Differences -> built-in encoding -> the same
cp1252-style fallback the text decoder uses, so a code's advance always
matches the character we emit for it
- Type3 visual sizing: PK bitmap fonts (dvips) use FontMatrix
[1 0 0 -1 0 0] with nominal sizes like 0.12pt; scale by FontBBox height
x |matrix_y|. Applied in the page-stream and Form XObject paths.
Indirect numeric array elements are resolved before use.
Effect on Shannon's 'A Mathematical Theory of Communication' (1998
dvips/Distiller, the reported case): glued sentences 95 -> 5. Corpus
impact: 12 of 184 eval documents, e.g. Data-Processing-Agreement
recovers a paragraph that a phantom table had shredded into cells.
Layout heuristics tuned on the same document (indent-based paragraph
breaks, heading reclassification, table script filtering) are held back
for a separate PR — they change ~98 further documents and need to be
justified against the corpus, not against one PDF.
* review: narrow Type3 rescaling to self-inconsistent fonts; dedup + test all width tables
Addresses cubic review on #241, plus a follow-up from a local cubic run.
- Type3 visual scaling was applied to every Type3 font whose FontBBox
height x |matrix_y| deviated >5% from 1.0. FontBBox is the glyph box,
not the em box, so a conventional 1/1000-matrix font with a
descender..ascender bbox (~700 units) computed 0.7 and had every
reported size shrunk by 30% — corrupting the drop-cap, heading-tier,
sub/superscript and table heuristics this is meant to fix.
First attempt gated on the matrix being unit-scale, but a local cubic
run pointed out that wrongly excludes valid non-standard matrices (a
0.005 matrix with a full-em bbox legitimately needs a 5x scale). The
product is the right discriminator, not the matrix: a self-consistent
font lands near 1.0 because the matrix is the reciprocal of the
glyph-space em, so only a wildly inconsistent one (dvips/PK bitmap
fonts sit at ~159) is renormalized. Band widened to [0.25, 4.0].
Corpus effect: 12 -> 7 documents change. The 5 that drop out were
being wrongly rescaled — including Data-Processing-Agreement, whose
phantom-table fix turned out to come from this bug rather than from
the width fallback, so it is correctly given up.
- base14: all 14 width tables now covered by the sort-invariant test via
an ALL_TABLES registry, not a hand-picked subset.
- base14: identical tables share one static (all four Courier variants
are monospace 600; the oblique Helvetica variants match their upright
forms), removing 5 duplicate copies.
* test: refresh Shannon snapshot after merging main
CI checks out a merge of the PR head with main, and main advanced 8
commits since this branch was cut — including #201 (contextual digit
runs), #240 and #253 (markdown fixes). Those change extraction output,
so a snapshot generated on the unmerged branch could not match; the
Test job failed on the merge commit while passing on the branch itself.
The merged behaviour is better: the footnote marker '2' before
'Hartley, R. V. L.' is now recovered instead of dropped.
950 tests pass on the merged tree, clippy clean.
AGENTS.md was stale (179+ PDFs, missing the semantic-quality bullet).
Both files now match: ~200-PDF corpus, and iteration guidance to prefer
subset runs (bench.py test -q / -s <name>) with the full suite as the
final pre-commit check.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* chore(ci): bump GitHub Actions to current majors
Node 20 action runtimes are deprecated on GitHub runners; bump every
first-party action to its latest major across all workflows:
- actions/checkout v4/v6 -> v7
- actions/cache v4 -> v6
- actions/upload-artifact v4 -> v7, download-artifact v4 -> v8
- actions/setup-node v6 -> v7, setup-python v5 -> v7
- actions/upload-pages-artifact v3 -> v5, deploy-pages v4 -> v5
Third-party pins (dtolnay/rust-toolchain, Swatinem/rust-cache,
setup-zig, setup-bun, taiki-e/install-action, maturin-action) are
already on their latest majors.
* chore(ci): pin all actions to full commit SHAs
Mutable @vN tags can be retagged; in the publish workflows that code
runs with OIDC credentials before npm/PyPI/crates.io publishes. Pin
every action (first- and third-party) to its release commit SHA with
the version in a trailing comment.
dtolnay/rust-toolchain infers the toolchain from its ref name, so the
SHA-pinned invocations pass an explicit toolchain: stable input.
Adds x86_64-unknown-linux-musl, aarch64-unknown-linux-gnu, and
aarch64-unknown-linux-musl to the napi build targets so
@firecrawl/pdf-inspector works on Alpine and ARM64 Linux deployments.
- gnu arm64 cross-compiles with --use-napi-cross (old-glibc sysroot),
musl targets with -x (zig + cargo-zigbuild), per the napi-rs template
- new platform packages carry npm libc metadata (glibc/musl)
- smoke-test job runs napi/test.mjs on all six targets before publish
(Alpine containers for musl, ubuntu-24.04-arm runners for ARM64)
- bump to 1.12.0 to trigger publishing of all platform packages
Closes#216
Single-word bold section headings ('Replace', 'Trash', 'Instructions')
required a paragraph break before AND after, but headings hug their
section's first paragraph — the break-after almost never exists. A
standalone all-bold single word (>=4 chars, paragraph break before or
page top) now classifies; mixed bold lead-ins ('Note: ...') stay
excluded via all_bold.
opendataloader-bench: 0.8567 -> 0.8575, MHS 0.773 -> 0.776; docs 145
+0.118, 069 +0.112 (net of one cover-page layout shuffle at -0.066
where the new output is semantically closer to GT). pdf-evals: 66
snapshots, composite 0.5864 -> 0.5883, sole >0.02 mover positive.
p1244/thermo fixture snapshots regenerated ('Instructions' un-fuses
from its body paragraph — the intended behavior).
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): digit-only lines do not define heading tiers
A large bold page number (14pt folio over 11pt body) claimed tier 0:
every real heading demoted one level document-wide, and the bold-size
fallback (which requires an empty tier list) was blocked for documents
whose headings match body size.
Bench-neutral (MHS scores relative hierarchy); pdf-evals: 18 docs get
their heading levels back (#### -> ###), semantic composite +0.0006,
no percentile down.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): exclude digit-only lines from the bold fallback tier pass too
The exclusion in the main pass wasn't enough: with the page-number
tier gone, the bold fallback re-collected the same bold folio. Also
regenerates the thermo-freon12 snapshot (cosmetic churn on the
scrambled legend fixture).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
IntechOpen-family academic PDFs embed Computer Modern math symbol
subsets whose glyphs are misnamed after Latin lookalikes (equal →
/onequarter, plus → /thorn, parens → /eth //Thorn) — and the generated
ToUnicode faithfully propagates the wrong names, so formulas decode as
'S ¼ kB þ 1' instead of 'S = kB + 1'.
Remap the observed misnames, gated strictly on the TeXCMMathsSymbols
base font (subset prefix stripped) so genuine fractions and thorns in
text fonts are untouched. Known limitation: a sibling subset misnames
the slash as /onequarter too, so an occasional '/' renders as '=' —
still strictly better than the previous mojibake.
Affects 4 bench PDFs (028/031 +0.001-0.004 NID) and zero pdf-evals
snapshots.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): rescue wrapped bold headings on interleaved column pages
Report pages whose columns can't be detected (6pt gutters) interleave
both columns' lines, which breaks every whitespace signal the bold
heading heuristic relies on: para_threshold inflates to ~3x line
height and a wrapped heading's own internal line gap defeats isolation.
'9.5. Adapting to the New Normal: Changing / Business Models' merged
into the following paragraph.
Four changes:
- merge_wrapped_bold_heading_groups: 2-3 consecutive all-bold
body-size lines merge into one line when the group is isolated
(column-locally, judged by x-overlapping lines only) or starts with
a section number.
- Section-numbered all-bold lines ('9.5. ...') classify as headings
without the standalone/isolation score gate.
- Line unfusing extends to uppercase-start continuations, gated on a
bold-style mismatch between the runs (a bold heading beside regular
body text) — same-style label rows stay joined.
- The unfuse line-side wordiness requirement drops to 2 words so a
wrapped heading's short last line ('Business Models') still splits
from the neighboring column.
opendataloader-bench: overall 0.8554 -> 0.8576, MHS 0.769 -> 0.777;
docs 037 +0.161, 111 +0.157, 039 +0.091, 198 +0.028, none down.
pdf-evals: 63 snapshots, composite 0.5952 -> 0.5964, sole >0.02 mover
positive. thermo-freon12 snapshot regenerated (cosmetic churn on an
already-scrambled 3-column legend).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): review follow-ups — multi-component section numbers, wholly-bold line gate
Single '1. ' prefixes are ordered list items and no longer bypass
isolation; the uppercase unfuse requires the whole line bold (a
heading), not merely its last run, so mixed bold-label/value rows
stay joined.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(layout): unfuse independent column runs sharing a baseline
Two-column report pages with charts fused headings into the adjacent
column's body text: the columns' ~6pt gutter is below what histogram
valley detection can safely use, so the page grouped single-column and
same-baseline items from both columns joined into one line ('6.2.
Expectations for Re-Hiring Employees' + mid-sentence text, killing
MHS and NID on the whole survey-report doc family).
Three changes:
- Line grouping splits same-baseline runs separated by a wide void
(>3x font size, >=30pt) when the incoming run starts lowercase
(mid-sentence continuation from another column) and both sides are
multi-word prose. TOC page numbers, dot leaders, and table cells
(numbered/capitalized) stay joined.
- Column detection is blind to chart-region text (tight 2pt bounds —
wider padding ate rows adjacent to charts), via a chart-aware line
grouping variant wired from the markdown pipeline.
- validate_and_build_columns computes its vertical span from
histogram-eligible items only, so full-width captions no longer sink
the overlap ratio for partial-page column regions.
opendataloader-bench: overall 0.8532 -> 0.8554, MHS 0.761 -> 0.769;
doc 038 +0.434, no regressions. pdf-evals: 34 snapshots change,
semantic composite wash (0.5749 -> 0.5748), no per-doc mover >0.015.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(layout): review follow-ups — chart-aware band-split grouping, single chart scan per page
Band-split pages now route through the chart-aware grouping too, and
the band loop reuses the precomputed page_chart_map instead of
re-scanning the rect list per page.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): bold-size fallback tiers when nothing clears the ratio gate
Books often set section headings barely above body size (11pt bold
over 10pt text). Nothing cleared the 1.2x heading-tier ratio gate, so
tiers stayed empty and every bold heading defaulted to H2 — H1 was
unreachable for the whole document.
When no size clears the gate, build tiers from bold line sizes >=1.05x
body, and let tier matches through detect_header_level down to that
ratio. Documents with real (>=1.2x) tiers are untouched.
Bench-neutral by construction (the MHS metric scores relative
hierarchy, not absolute levels); pdf-evals semantic composite +0.004
on the 11 affected docs with all percentiles up.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): require boldness for sub-gate tier matches
Review follow-up: fallback tiers come from bold lines, so honoring
them for non-bold text at the same size would promote captions.
detect_header_level now takes is_bold and only matches tiers below
the 1.2x gate for bold lines; >=1.2x matches stay bold-agnostic.
Also restores the >=1.2x tier-match loop the refactor dropped.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(headings): judge line boldness by character mass
Review follow-up: a heading with an unbold section-number prefix
('4. ' + bold title) failed the first-item boldness test. Judge the
line by bold character mass instead, at all three call sites.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Link items carry an annotation rect, so y is a box edge — unlike text
items, where y is a baseline. Testing rect-bottom dropped partially
visible links whose bottom edge dipped past the tolerance. Follow-up
to a #160 review comment.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(extractor): clip page content to the visible page box
Single-page extracts and imposed spreads keep neighboring pages'
content in the stream, positioned outside the CropBox. Extracting it
appends invisible sections to the page, scrambles NID, and poisons
font statistics (heading tiers built from off-page text).
Clip items (by center), and — only when off-page text was actually
found — rects and lines (by overlap) to CropBox-else-MediaBox, walking
page-tree inheritance. Rotated pages are left unclipped: their item
coordinates are already transformed out of box space. Degenerate boxes
(<1 inch) are ignored.
opendataloader-bench: overall 0.8445 -> 0.8537, NID +0.008,
MHS +0.013; six docs up (best +0.426), none down.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(extractor): guard page-box clipping with coherence and straddle checks
Two real-document counterexamples: curved display text leaves short
glyph fragments with artifact coordinates outside the box (judge by
character mass, not item count), and some PDFs compute inflated
coordinates for visible body text (an off-page item continuing an
on-page baseline means our transform model is wrong there — skip).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(extractor): clip off-box link annotations when page text was clipped
Review follow-up: annotations from the neighboring page bypassed the
filter. Form fields are left as-is — they're document-scoped and rare
on imposed spreads.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Bar charts drawn as filled rects read as cell rects or aligned text:
the cell-rect fallback gridded their axis labels into phantom tables,
and when rect paths rejected them, hint regions and the gap-histogram
heuristic re-gridded the same text. On survey-report pages this
scrambled reading order and swallowed section headings.
Detection (is_chart_bar_cluster): the dominant equal-breadth rect
family arranged in >=2 spaced positions (bars; cell rects touch),
data-driven extent variation (>=1.3x), no same-offset/same-extent
partners across positions (grid rows pair up, chart segments don't),
and only numeric labels inside. Mirrored predicate covers horizontal
bar charts.
Chart clusters are skipped in detect_tables_from_rects (no table, no
hint), and a new detect_chart_regions pass lets the markdown pipeline
pre-claim chart items so heuristic/line/column detection and the
merged-band retry all skip them — the text flows out as plain lines.
opendataloader-bench: overall 0.8389 -> 0.8446, TEDS 0.699 -> 0.708,
MHS 0.742 -> 0.750, NID 0.889 -> 0.894; 7 docs up (best +0.459), none
down. pdf-evals changed-set composite +0.013, TEDS +0.036.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* docs(tables): document the stacked-box 6-rect precision gate
Review follow-up on #157: 3-5 box stacks never reach the stacked-box
fallback (the main loop needs >=6-rect clusters on a >=6-rect page).
Routing smaller clusters through the detector was implemented and
measured: zero opendataloader-bench movement and four pdf-evals
regressions (striped bullet lists, wrapped regulation text, stats-table
columns) across three guard iterations — with 3-5 boxes the anti-prose
guards have too little signal. Keep the gate, document it at the call
site, and pin the behavior with an end-to-end test.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* test: exercise the cluster gate, not just the page gate
Review follow-up: the pinned test's 3-rect page exited at the 6-rect
page gate before reaching the cluster minimum it documents. Scattered
unrelated rects now push the page past the page gate while the 3-box
stack stays below the cluster minimum.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
A framework list drawn as a vertical stack of boxes (one short title
per box) scored TEDS 0 and worse, ran into the surrounding prose as a
single paragraph: the grid path rejects one-column rect structures by
design (needs >=3 x-edges).
Add a stacked-box fallback after grid and row-stripe detection: >=3
x-aligned, same-width, same-height boxes forming a contiguous vertical
stack, each holding one short text run, become a single-column table.
Guards against striped prose and grid fragments (all unit-tested):
- boxes flanked by rects or text at their y-level are one column of a
wider structure — bail to the grid/cell-rect paths
- multiple separated text runs per box = striped multi-column content
- prose rows: function-word-dense cells averaging >60 chars
- sentence continuation across rows (trailing comma / open + lowercase)
- numbered/lettered list items stay lists
opendataloader-bench: overall 0.8362 -> 0.8389, TEDS 0.675 -> 0.699,
target doc +0.553, no other doc moved.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
A 4-column compliance table (labels + Small/Medium/Large) was emitted
as two separate tables: split_side_by_side read the text gap between
the ruled and remaining columns as a page-layout gutter, and each band
then detected its own fragment.
Before accepting a side-by-side split, check rect clusters near the
boundary: if a table-shaped cluster spans it (or ends at it with
cell-like text row-aligned beyond), and those table rows account for
the majority of far-side text, the split runs through a table — veto
it. The majority guard keeps legitimate splits on pages where a figure
spans two prose columns.
opendataloader-bench: overall 0.8306 -> 0.8362, TEDS 0.656 -> 0.675,
target doc +0.506, no regressions.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(tables): flatten page-number tables of contents instead of gridding
A title-based contents page ("About the Publisher vii", "Experiment #1
… 3") with no dot leaders and no section numbers was detected as a
2-column data table and rendered as a markdown grid, scrambling the
linear reading order (a top cause of NID loss on affected docs) and
scoring 0 on table structure.
Add is_page_number_toc: a narrow (2-3 col) list whose last column is
mostly page numbers (short integers or roman numerals) that are mostly
non-decreasing, with a text-title first column and NO header row (a
TOC's first row is already an entry). Such tables now route through the
existing flat-list TOC renderer.
The no-header + narrow-width + monotonic guards keep real data tables
intact — e.g. a 4-column regional table, or a 2-column "Mineral | CEC"
table with a header row and ascending values.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* docs: bump reading-order (NID) benchmark to 0.89
Reflects the phantom-TOC fix in this PR: NID 0.88 -> 0.89 on the
200-doc benchmark. Other cells are unchanged at 2-decimal precision.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: canonical roman validation, real-first-row header check, roman page cells
- page_number_value now requires a *canonical* roman numeral (re-encode
and compare), so words like "civil"/"mix"/"ill" are no longer parsed
as page numbers.
- The no-header guard checks the actual first row's last cell instead of
the first non-empty one, so a blank header cell ("Category | ") still
rejects the TOC heuristic.
- format::is_page_number_cell recognizes canonical roman numerals, so
roman front-matter pages (vii, ix) get proper title/page separation in
the flat TOC list.
- Fix the non-monotonic test to use 5 rows so it exercises the
monotonicity guard rather than the row-count early return; add
roman-lookalike and blank-header rejection tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: share roman helper, widen length to 8, require page-span for TOC
- Extract canonical_roman_value + to_roman_lower into tables/mod.rs and
use them from both the TOC detector and the formatter, removing the
duplicated mapping/loop and keeping them in sync. The shared helper
accepts ≤8 chars, so longer front-matter numerals (xxxviii) flatten
consistently on both sides.
- Add a page-span guard to is_page_number_toc: real page numbers skip
through the document (range >> entry count), so a dense consecutive
ordinal/rank/ID column (1,2,3,…) is rejected — monotonicity alone did
not separate those data tables from contents.
Costs ~0.001 aggregate on the benchmark (NID 0.888->0.887) for the added
precision; still a clear win over baseline (NID 0.883, TEDS unchanged).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: recover consecutive-page TOCs via a title signal
The strict page-span rule rejected legitimate one-page-per-entry TOCs
(range ~= entry count). Relax it: accept any page sequence with a gap
(nearly all real contents). Only a *perfectly dense* consecutive run —
which rank/ID/ordinal columns produce, but a chapter-per-page TOC can
too — falls back to a title signal: flatten when the first-column
entries average multi-word headings, keep as a table when they are the
short single-word labels typical of leaderboards/ID lists.
Recovers the ~0.001 the range-only rule cost (NID back to 0.888) while
still rejecting dense ordinal data tables.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Concise Features list and the opendataloader-bench comparison table on
each registry readme, adapted per ecosystem. Bump all three versions
(crate 0.1.6, python 0.2.5, npm 1.11.1) to republish the pages.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(npm): split platform binaries into optionalDependencies (1.11.0)
The single package bundled all three .node binaries (17.6 MB unpacked)
so every install downloaded every platform. Publish one package per
platform (@firecrawl/pdf-inspector-{linux-x64-gnu,darwin-arm64,
win32-x64-msvc}) holding just its binary; the napi-generated loader
already falls back to exactly these names. Main package drops *.node
from files (8.5 kB tarball) and pins the platform packages as
optionalDependencies, re-stamped to the exact version at publish time.
Publish workflow gains a workflow_dispatch fallback and per-package
already-published checks so partial releases can be retried.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* docs(npm): document Windows support and platform packages; drop stale napi.package.name
Review follow-ups: the README claimed only linux-x64 and macOS ARM64
despite the win32-x64-msvc binary shipping, and napi.package.name
(@firecrawl/pdf-inspector-js) contradicts the real platform package
prefix — the loader and workflow derive it from the root package name.
Verified the generated loader is unchanged without the config.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The 0.2.3 sdist was 10.09 MiB (packaged tests/fixtures). maturin derives
the sdist file list from Cargo's include allowlist, so it's now 1.35 MiB
— but the allowlist dropped pdf_inspector.pyi, which would strip type
hints from wheels built from the sdist. Add it back and bump to 0.2.4.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
cargo publish of 0.1.5 failed with 413: the crate packaged everything
(260 files, 10.1MiB compressed) and tests/fixtures alone is 10.2MB.
Add an explicit include list (src, external/bcmaps which tounicode.rs
loads at runtime, readme, license) — 1.3MiB compressed.
Also add a workflow_dispatch fallback to publish-crate.yml so a failed
publish can be retried without a version bump (0.1.5 is already on
main, so a re-push won't register as a version change).
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
crates.io showed the repo README, which leads with Python/Node quick
starts and repo-relative links. Point the crate readme at
docs/rust-api.md, refreshed with an intro, crates.io install, and CLI
install instructions. Bump to 0.1.5 to republish.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The bold-label + comma-list paragraphs render as cramped walls of
inline code on PyPI. A python code block mirroring pdf_inspector.pyi
renders cleanly everywhere and adds field types plus the missing
is_underline/is_strikeout TextItem fields.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
PyPI showed an empty description because pyproject.toml declared no
readme. Point it at docs/python.md (refreshed with pip install now that
wheels exist) and add sidebar URLs. Bump to 0.2.2 to republish.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Ships #145: exclusive item->region assignment in extract_text_in_regions
(overlapping layout regions no longer double-extract shared items —
duplicated lines on 21% of a 2,078-doc bench corpus, with occasional
content loss when downstream dedup kept the wrong variant).
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(regions): exclusive item->region assignment in extract_text_in_regions
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 a 2,078-doc
bench corpus — and downstream duplicate-handling sometimes dropped the
variant holding a sentence tail, turning duplication into content loss.
Each item is now pre-assigned to the single region with the largest
overlap area (same margin as the boolean test); the per-region filter
uses the assignment. Items are partitioned, never suppressed, so no
content can vanish that was previously extracted.
Paired with fire-pdf assembly fixes (neighbor-local sweep dedup +
remainder salvage); verified together on the repro doc: duplicate lines
6 -> 0, the audit's lost sentence recovered (fuzz 78 -> 87.5). Batch
over the worst duplication docs: 185 -> 59 total, 6 of 8 docs to zero.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(regions): review round — single-pass bucketing, no-OCR for lost-to-neighbor empties, shared margin constant
- Assignment and materialization now happen in ONE pass over items
(clone bucketed at argmax time) instead of a second O(items x regions)
traversal.
- A region whose only overlapping items were assigned to a
better-overlapping neighbor no longer flags needs_ocr: the pixels it
would re-read belong to that neighbor, and OCR would reintroduce the
duplication exclusivity removed. Matches the pre-change OCR load
(these regions were non-empty native before).
- REGION_MARGIN hoisted to a module const shared by the boolean
predicates and the area score — they must stay in sync or an item
passing the guard could score zero area.
Repro re-verified after fixes: fuzz 87.5, duplicates 0; 756 tests pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(regions): lost-to-neighbor requires zero items assigned to the region
had_candidates records overlap, not assignment loss: a region whose own
assigned items materialize to empty text (whitespace-only items,
collector filtering) was indistinguishable from one that lost everything
to a neighbor, and wrongly skipped its OCR fallback. The suppression now
also requires assigned_count == 0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
macos-13 runners were retired by GitHub, so the x86_64-apple-darwin
build job queued forever and the publish never ran.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* ci: add PyPI trusted publishing, abi3 wheels, bump to 0.2.1
Adds publish-pypi.yml mirroring the npm/crates.io pattern: triggers on
Cargo.toml version change, builds wheels for 5 platforms via maturin,
publishes with OIDC trusted publishing (no tokens). workflow_dispatch
serves as a manual fallback for the first run after the PyPI project
transfer.
Enables pyo3 abi3-py38 so one wheel per platform covers CPython >=3.8
(previous manual uploads were cp312-only). Bumps version to 0.2.1 since
PyPI already has 0.2.0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* ci(pypi): guard dispatch to main, support partial-release repair
Review feedback: trusted publishing doesn't match on branch, so
workflow_dispatch needed an explicit main-ref guard. Manual dispatch now
always rebuilds and publishes with skip-existing so a release that
failed after uploading only some wheels can be completed by re-running.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* ci(pypi): version PyPI package from pyproject.toml, not Cargo.toml
Decouple the Python package version from the crate version, matching
how npm publishing keys off napi/package.json: bump [project] version
in pyproject.toml manually and CI publishes on merge. Reverts the
Cargo.toml bump so this PR no longer triggers a crates.io release.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: remove accidentally committed uv.lock
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(ci): tolerate missing version key in parent pyproject.toml
The first merge of this workflow has a parent commit where pyproject.toml
still used dynamic = ["version"], so the old-version read would KeyError
and the auto-publish would never fire. Treat a missing key as a change.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(underline): rescue snug-owned underlines from the table-ruling filters
Documents that underline many full-width lines (dense CJK business docs,
legal redlines, 10-K section links) produce span-similar rules at 3+
y-levels — exactly what the repeated-ruling filter treats as table
rulings, so every semantic underline on such pages was discarded. Three
changes fix detection without re-marking real tables:
1. Snug-owner rescue: a rule survives the repeated-ruling filter when
the union of touching text runs on its baseline row owns it (rule
contained within the union's span +0.75em, runs cover >=60% of the
rule, no column-sized gaps between runs). Table row separators fail
ownership: they overshoot their cells' text or match gapped items.
Same-row segmented rules (column-header separators) always stay
discarded, and a rule enclosed by a drawn cell-sized box (rect-grid
tables) is never rescued.
2. Vertical window widened 0.35em -> 0.72em below the baseline: CJK
layouts draw underlines under the full em box, measured at ~0.67em.
3. Prose-table guard in the positions-path suppressor: a detected
'table' whose cells hold flowing prose (>=30% of cells over 100
chars) is a detection artifact of boxed callouts + stacked rules,
not a real table — suppressing there erased every underline on the
page.
Also fixes cluster_x_positions fabricating phantom table columns from
style-split continuation runs (touching items, gap <2pt, now feed one
column start) — the fix that keeps rect-grid table shapes stable while
underlined links inside cells are correctly marked.
Snapshot updates are underline gains on regulation/form fixtures and one
empty spacer-column change in a subscripted header.
Corpus (508-doc public bench sweep): text output byte-identical on all
docs; underlined items +224/-0; strikeout now fires on redline docs.
Item-level GT coverage: is_underline 86->151/405, is_strikeout 0->10/44.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(underline): fraction-bar guard + subscript merge across underline marks
Found by a 202-doc real-world corpus diff (pdf-evals) that exercises the
full markdown pipeline, which the bench-corpus item sweep does not:
1. Math fraction bars and lattice grid lines are underline geometry —
short horizontal rules under digits. Guard: a narrow rule (<=60pt)
with bar-sized text hanging just below it (denominator) never marks.
The below-text width bound matters: tightly-leaded REAL underlines
have a full-width next line below, which must not trip the guard.
2. merge_subscript_items refused to merge when the parent was underlined
but the tiny digit was not (the drawn rule easily misses the digit's
own overlap window) — losing the merge broke subscript tokens inside
table cells (b+2 no longer became b₂). Strikeout boundaries still
block the merge in both directions; only parent-underlined/digit-bare
merges, absorbing with the parent's flags.
Corpus after refinement: underlined items +220/-2 (the 2 are fraction
bars the old code wrongly marked), GT rule-text coverage 149/405
underline + 10/44 strikeout, text output identical on all 508 bench
docs and word-count-identical on the 202 pdf-evals docs.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(underline): address review — grid-evidence veto, strikeout-safe fraction guard, bounded gaps
- Cell-box veto now requires GRID EVIDENCE (a vertically abutting
neighbor rect with x-overlap) instead of a height window: multiline
table cells taller than the old 90pt ceiling veto again, and isolated
filled callout panels (which legitimately contain underlines) no
longer veto at all.
- The fraction guard gates only UNDERLINE marking; rule_strikes_item
still evaluates, so short strikeouts near lower text survive.
- Fraction hug distance tightened to 0.3em so a short last-line at
normal leading is not mistaken for a denominator.
- Continuation-run suppression bounds the negative gap (-4pt): text
overhanging from an adjacent cell keeps its own column start.
Corpus after review fixes: underlined items +222/-2, GT coverage
150/405 underline + 10/44 strikeout, bench text output identical on
all 508 docs, pdf-evals word loss bounded at equation-reflow noise.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* chore: appease clippy (redundant closure)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
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 (opendataloader MHS via fire-pdf's coverage_fallback
path, ENG-5029).
Extends is_heading_fragment, so both bold-heading call sites get the
gate. Corpus sweep (708 opendataloader + ParseBench PDFs): 33 docs
change, all lowercase-fragment demotions from `##`/`#` to plain or
bold ("## caldera" -> "**caldera**", "## of quorum.\"" ->
"**of quorum.\"**") — no real heading is lowercase-initial and that
short in either corpus.
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat: per-page OCR routing reasons (scanned/no_text/vector_text/garbled)
Replaces the single suspected_garbled_text signal with a per-page
explanation for why each OCR-flagged page needs OCR. The detector
classifies each page in pages_needing_ocr from its content analysis:
- scanned — no usable text, image-backed page
- no_text — no text and no image (blank/unreachable)
- vector_text — text drawn as vector outlines, not extractable
- suspected_garbled_text — undecodable Identity-H/Type3 fonts
Exposed on PdfTypeResult.ocr_reasons_by_page and surfaced through
PdfProcessResult and the detect-pdf CLI (JSON + human output). Reasons
only ever explain pages already flagged for OCR — a text page with an
embedded logo stays TextBased, so this doesn't widen the OCR net.
Markdown output is byte-identical across the regression corpus.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: cache freshly-analyzed pages so OCR reasons aren't lost
Under a sampling ScanStrategy, the Mixed per-page loop (Phase 2) and the
garbled-font check (Phase 3) analyze non-sampled pages but dropped the
PageAnalysis after flagging them. The reason-classification pass then
missed the cache and defaulted those pages to "scanned", masking the
real vector_text / suspected_garbled_text cause. Insert the fresh
analyses into analysis_cache so the reason pass classifies them
correctly. No change under the default full-sampling strategy (all
pages are already cached).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(site): add GitHub Pages landing page
Self-contained landing page (single index.html, no build step) plus a
Pages deploy workflow that publishes site/ on push to main. Covers the
pitch, install commands for all three registries, feature grid,
benchmark, and tabbed quick-start for Rust/Python/Node/CLI.
Benchmark numbers mirror the README's current published table; both
should be refreshed together in a follow-up.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(site): add Firecrawl Parse upsell to closing section
Replace the single CTA with a two-path split: run pdf-inspector locally
(OSS) vs. hand scanned/OCR/at-scale documents to Firecrawl Parse
(hosted). Links the OSS library back to the paid product for the cases
local parsing can't cover.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(site): add Firecrawl branding (flame mark + wordmark)
Firecrawl flame mark anchors the hosted-parse card; charcoal wordmark
in the footer credit. Brand SVGs referenced as-is (exact colors).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(site): refresh benchmark with current main numbers
pdf-inspector row updated from a fresh run on latest main: overall
0.78→0.83, tables 0.59→0.66, headings 0.57→0.74. Now within 0.01 of
opendataloader overall, best tables of the group, headings on par.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Re-ran opendataloader-bench on current main. pdf-inspector improved
across the board since the last table: overall 0.78→0.83, tables
0.59→0.66, headings 0.57→0.74 (competitor rows unchanged). Updated the
prose — heading detection no longer lags opendataloader, and overall is
now within 0.01 of it at ~2.5× the speed.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(extractor): run-local space floor for tracked (letter-spaced) glyph runs
Display type set with tracking renders one glyph per show op; the merge
loop's fixed space thresholds (0.08-0.13 em) then read every letter gap
as a word boundary and emit "H O W" / "F U R T H E R" instead of
"HOW" / "FURTHER". The page-level Canva fixer can't help: it requires
>=50% of the page's items to be letter-spaced, and these docs track
only their display headings.
merge_text_items now pre-scans each run of consecutive single-glyph
items (same size band, same style, mergeable gaps — the loop's own
break conditions) and, when the run is tracked, derives the space floor
from the run's own gap distribution:
- runs with >=4 gaps qualify when the median gap clears the fixed
threshold; word gaps, if present, form a second mode — split at the
largest relative jump (>=1.4x), else the run is a single word
("I T I S I M P O R T A N T" -> "IT IS IMPORTANT")
- short runs (2-3 gaps: "H O W") additionally demand uniform gaps and
ALL-CAPS or CJK — a genuine spaced sequence of single letters
("x y z" variables) has the same gap count, and display tracking is
a caps convention; CJK never wants inter-glyph spaces
Corpus sweep (708 opendataloader + ParseBench text PDFs) vs main: 9
docs change — the tracked display titles ("HOW CAN YOU HELP?",
"LUNCHTIME MENU", a tracked email address), and CJK glyph-per-item
docs whose spurious inter-glyph spaces now collapse (GT for those docs
is unspaced CJK; should_join_items already treats no-space CJK as
correct on its path). No other doc moves.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(review): convention gate on both tiers; Han/Kana floor always infinite (PR #133 review)
- Long lowercase spaced-single runs ("a b c d e") had the tracked gap
shape in the >=4-gap tier with no convention guard — word boundaries
lost. The caps/CJK/title-case gate now applies to BOTH tiers; a
title-case single word ("B u f f a l o") also qualifies.
- Han/Kana runs skipped straight to the bimodal split, so a nonuniform
gap distribution (justification, punctuation spacing) could
manufacture a word boundary. Han/Kana now always floors at infinity;
Hangul deliberately keeps word-boundary handling — Korean spaces
between words (is_spaceless_cjk excludes it).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(extractor): preserve mixed-case glyph boundaries
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(markdown): keep isolated headings on sparse pages; gate tagged roles
The isolated-line density guard wiped every isolated line on a page
where they exceeded 25% of lines. On sparse pages (covers, ToC pages
with a lone "CONTENTS" title, section-divider pages) a single heading
is trivially >25%, so the guard erased exactly the line it exists to
find. Require the page to have >=10 lines before the guard runs — the
25% ratio only signals a multi-column misfire on a dense page.
That let more isolated lines through, exposing that the visual heading
heuristic could promote lines already tagged with a non-heading struct
role (list item, blockquote, code, caption, ToC) or set in a monospace
font. Gate the heuristic on those in both converter paths.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: extend non-heading role gate; centralize on StructRole method
Move the non-heading-role check to StructRole::is_non_heading_content
and extend it to the content roles the inline allowlist missed: Quote,
Index, Note, Reference, BibEntry, Formula, Form (in addition to the
existing list/quote/caption/toc/code roles).
Figure is deliberately excluded: cover and banner pages routinely tag
the document title inside a Figure next to a seal/logo, and that title
is a real heading — including Figure demoted the LA County protocol
cover title from headings to bold. Verified against the reference.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: block table roles from heading promotion too
Add Table/TR/TH/TD/THead/TBody/TFoot to is_non_heading_content. When
table reconstruction falls back and cells reach the line loop as plain
text, a short isolated cell (a TH column header in particular) could be
promoted to a heading. Defensive: no change across either regression
corpus, so pure hardening for the fallback path.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The garbage/encoding detectors had accreted as ~490 lines of free
functions scattered through lib.rs (flagged in the #120 review). Move the
whole cluster into src/text_quality.rs behind a module doc that maps the
surface: the two interfaces (markdown-level vs item/span-level) and the
detection classes (replacement runs, private-use/C1 runs, dollar-as-space,
non-alphanumeric dominance, substitution-cipher statistics).
Moved verbatim: detect_encoding_issues, is_garbage_text, is_cid_garbage,
analyze_text_quality, region_items_have_decoding_issue and their helpers,
CipherGarbleStats, and the TextQuality* types. The OCR-reason aggregation
plumbing (add_ocr_reason, merge_ocr_reasons, page_ocr_reason*) stays in
lib.rs since it is shared by the main extraction loops, not detection.
Pure code motion — function bodies are unchanged; only visibility keywords
were added (pub(crate) on the six items lib.rs consumes; add_ocr_reason is
now pub(crate) so the module can call it). Behavior is provably unchanged:
same test counts (565 unit + 139 integration), and release output is
byte-identical to merged main across all 185 eval PDFs. Detector unit tests
stay in lib.rs for now because they share test helpers with the table and
layout tests there.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(markdown): ToC-page suppression, wrapped bold headings, math fragments
Three heading-classification improvements:
- After emitting a "Contents"/"Table of Contents" heading, suppress
heading promotion for the rest of that page: ToC entries are section
titles that look exactly like headings ("1. Overview of OCR Pack")
and whole contents pages came out as stacks of ##.
- merge_heading_lines only merged font-size-tier and struct-tree
headings, so bold-at-body-size headings that wrap emitted two
separate ## lines. Merge a fully-bold line into the previous
fully-bold line when it reads as a wrap continuation (starts
lowercase, tiny Y gap, no terminal punctuation on the previous line).
- Reject display-math fragments from the bold/rarity heading heuristic:
equations ending in an equation number ("S = kB ln W, (2)") and
lead-ins referencing one ("Rearranging Equation (8) gives:"). A bare
trailing colon is deliberately NOT a signal — real headings often end
with colons ("Procedure:").
The p1244 snapshot change is the bold-merge working as intended:
stacked form labels "**Subtotals** **from pages**" now read
"**Subtotals from pages**".
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: guard tier path, protect prev headings in merge, narrow (N) rule
All three review findings applied, calibrated against the corpora:
- is_heading_fragment now gates the font-size-tier path too, not just
the rarity heuristic.
- The bold wrap-merge requires the previous line to be tier-less as
suggested; corpus diff confirmed the old behavior was absorbing a
wrapped list-item fragment into a real heading.
- The bare "(N)" suffix rule suppressed real headings ("Nicaea (325)",
appendix numbering). It now requires math evidence: an operator
(=, <=, <<, ...) in the line or ,/: immediately before the number.
Page-of-total running headers ("PM 2 (10)") get an explicit rule
since the old blanket suffix check had been catching them only by
accident.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(tables): reject row-stripe grids that swallow body text
Charts (bar graphs, axis gridlines) emit fields of drawing rects that
pass the row-stripe shape test; the resulting phantom table then
captures the page's prose in scrambled reading order — losing headings
and paragraph flow with it. The existing max-cell-length gate only
fires for tables with <4 non-empty rows, which these grids exceed.
Add has_dominant_prose_cell: reject when one cell holds >=60 words AND
at least a third of the table's total words. Real tables never
concentrate that much text in a single cell, even with a description
column.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: extend prose-cell guard to merged-cluster path, add boundary test
The merged-cluster fallback had the same unguarded 4+-row gap as
row-stripe; apply has_dominant_prose_cell there too. The cell-rect path
already runs its own function-word prose check and is left unchanged.
Also add the boundary test from review: a 4+-row data table with one
60-word note cell stays accepted because the 1/3-of-total-words
denominator scales with table size.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* review: document intended small-table boundary of prose-cell guard
Investigated the suggested row-count exemption: adding a second-prose-
cell requirement (or a non_empty_rows >= 4 exemption) resurrects
verified phantom grids in 7 corpus documents — scrambled body text and
chart/figure regions, one of which is a 4-row grid. Every observed
single-dominant-cell grid in the corpora is swallowed prose, never a
real note table, and rejection degrades gracefully to prose while
acceptance scrambles reading order. Keep the guard unconditional,
document the rationale, and pin the boundary with a test.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
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>
URW Type 1 fonts abbreviate the Medium weight as "Medi" in the font
name (NimbusRomNo9L-Medi is the Times-Bold substitute embedded by most
LaTeX toolchains; -MediItal is bold italic). is_bold_font only matched
the full word "medium", so bold ran undetected across LaTeX-produced
PDFs — dropping ** emphasis and starving bold-based heading detection
of its signal.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
New in this release (#125): isStrikeout on TextItem (geometric
detection sharing the underline rules pipeline), descriptor/embedded-
font bold+italic recall for subset fonts (FontDescriptor flags,
ttf-parser OS/2+post, bare-CFF Name INDEX), quote-operator advance
width, Ts text-rise handling, ActualText rise/position fixes, and a
document-scoped font style cache.
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(extractor): descriptor/embedded-font style flags + geometric strikeout detection
Two style-recall gaps, both invisible to the existing name-based
heuristics:
1. Subset fonts with opaque BaseFont names ("Tc1", "AAAAAB+Amplitude")
defeat is_italic_font/is_bold_font. New descriptor_style_flags reads
the FontDescriptor (ItalicAngle beyond 4 degrees, Flags bit 7 Italic,
bit 19 ForceBold) and, when the descriptor claims upright, falls back
to the embedded font file: ttf-parser's OS/2 fsSelection + post
italicAngle for sfnt fonts, and the CFF Name INDEX PostScript name
for bare-CFF FontFile3 (descriptor rewritten to ItalicAngle 0 while
embedding "Amplitude-LightItalic" was observed in the wild).
ORed into is_bold/is_italic at item creation (content streams and
form XObjects).
2. No strikeout signal existed. New is_strikeout on TextItem, detected
in the same pass as underline: same rules pipeline (stroked lines /
thin filled rects, table-ruling suppression), different vertical
window — a rule crossing the glyphs at 12-55% of the em above the
baseline instead of sitting at it. Exposed through napi and python
bindings and pdf2md --items-json.
Verified on public ParseBench corpus docs: previously-missed italic
council titles and bold CJK itinerary headings now flagged (render-
checked); 24/508 docs gain flags, none lose any; 35 strikeout items
detected corpus-wide, disjoint from underline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(review): quote-op advance width, Ts text rise, doc-level font style cache (PR #125 review)
Address three valid findings from review:
- The ' (move-to-next-line-and-show-text) operator emitted zero-width
items and never advanced the text matrix, so geometric underline/
strikeout detection (which requires width > 0) could never mark its
text, and following show ops overlapped it. Reuse Tj's advance-width
computation and matrix advance.
- Ts (text rise) was dropped entirely: raised/lowered runs kept the
unshifted baseline, so rules drawn at the risen glyph position missed
the strike/underline windows. Track rise in the text state (saved and
restored with q/Q) and shift the rendering position through the text
matrix's y column; advances stay on the unshifted matrix per spec.
- descriptor_style_flags re-decompressed and re-parsed the same embedded
font program on every page whenever the descriptor left a style flag
unset (the common case). Add a document-scoped FontStyleCache keyed by
the FontFile2/FontFile3 object id, threaded through page and form
extraction alongside the existing CMapDecisionCache.
The fourth finding (Form XObject rules never reach geometric detection)
is real but pre-existing for underline and needs the form walker to grow
path/paint tracking plus a new return type; deferred as a follow-up.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(review): ActualText items render at their glyphs' text rise (PR #125 review)
The EMC-built ActualText item used the captured text matrix without the
rise adjustment the ordinary Tj/TJ/' emission sites apply, so a tagged
run shown with Ts landed on the unshifted baseline — off the strikeout/
underline windows and inconsistent with untagged runs. The rise is
captured together with the first-glyph matrix (and at BDC for the
entry-position fallback): the item must render at the rise of its
GLYPHS, not whatever rise is set by EMC time.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(review): capture ActualText glyph position after the quote op's line move (PR #125 review)
The `'` handler skipped the entire suppressed-extraction block, so a
tagged span whose show op is `'` never captured its glyph matrix/rise —
the EMC item fell back to the BDC-entry matrix, which sits on the
PREVIOUS line (the `'` line move happens after BDC) with no rise. The
capture now happens right after the line move, matching the Tj/TJ
paths.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
* fix(review): style-boundary gate on subscript merge + strikeout suppression coverage (PR #125 review)
merge_subscript_items absorbed a script digit into its parent
regardless of underline/strikeout flags — dropping the digit's own mark
or widening the parent's over it. The merged item carries one flag, so
differing marks now break the merge, mirroring merge_text_items'
style-boundary rule (pre-existing for underline as well).
Also extends the table-suppression test to assert is_strikeout is
cleared alongside is_underline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L3U6BKYCS73DVA83odAfYB
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Cargo.toml and pyproject.toml already declare MIT but the repo had no
LICENSE file, so GitHub and package registries couldn't display it.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* fix(lib): detect substitution-cipher garbled text from broken ToUnicode CMaps
ParseBench text_simple__att10k.pdf (issue #118) ships Type0/Identity-H
fonts whose ToUnicode CMaps are authored garbled: every bfrange maps with
a wrong constant delta, so text extracts as pure-ASCII ciphertext
("Certificate" -> "8VceZWZTReV"). The embedded subset font has no cmap
table and no glyph names, so no decode source can recover the real text
(poppler and mupdf emit the same ciphertext). The only correct behavior
is to flag the page for OCR instead of serving the garbage silently --
but the text is 100% printable ASCII with word-like tokens, so it slipped
past is_garbage_text and detect_encoding_issues.
Add CipherGarbleStats, a letter-statistics discriminator that flags a
Latin-dominant sample (>=200 ASCII letters) when vowels are starved
(<=30% of letters) AND either:
- lowercase->uppercase transitions inside words exceed 10% of letter
bigrams (a shifted lowercase alphabet straddles the ASCII uppercase
block), or
- the letter histogram's cosine similarity against English letter
frequencies drops below 0.60 (catches shifts that stay within case
blocks).
Wired into analyze_text_quality (per-page, item-level) and
detect_encoding_issues (markdown-level), so extract_pages_markdown
reports needs_ocr + suspected_garbled_text and suppresses the garbage.
Thresholds validated against the 380-document pdf-evals snapshot corpus
(Swedish, Finnish, Turkish, German, romaji, schematics, all-caps and
camelCase-heavy docs): zero false positives, and byte-identical eval
output vs main. Garbled page measures vowel ratio 0.245 / case-shift
rate 0.225 / cosine 0.532; closest legitimate document on each axis is
0.264 / 0.021 / 0.801.
Fixes#118
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: bump pdf-inspector to 0.1.4, npm package to 1.9.11
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(lib): exempt uniform-case structured content from cipher detection
Address PR review (cubic P2): the frequency branch (english_cosine < 0.60)
fired on any Latin-dominant, low-vowel letter distribution unlike English,
so non-linguistic ASCII — DNA/protein sequences, ticker symbols, hex dumps —
could be suppressed and routed to OCR despite not being garbled. Measured:
DNA cosine 0.428 / vowel ratio 0.260, protein 0.738, tickers 0.747, hex
0.549 — all would have flagged.
Add a mixed-case guard to looks_garbled: garbled English is a permutation of
natural language and carries sentence capitalization (block-straddling shifts
invert the ratio — att10k is 60% uppercase; in-case Caesar shifts preserve it
at ~3%), so both keep some of each case. The exempted structured content is
uniform case (all upper or all lower). Requiring the minority case to be >=1%
of ASCII letters exempts single-case sequences while preserving both garble
signals, including the in-case-shift scenario the frequency branch exists for.
Strictly tightens the detector: it can only remove flags, so the eval corpus
stays at zero false positives (verified byte-identical to a baseline main
binary across all 185 PDFs) and att10k remains flagged. Adds regression tests
for DNA, protein, tickers, and an in-case Caesar shift.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(lib): make cipher detection case-agnostic via sorted-histogram shape
Address PR review follow-up: the mixed-case guard from the previous commit
returned before the vowel/frequency checks, creating a blind spot — a
uniform-case (all-lower or all-upper) substitution cipher is a plausible
broken-CMap output and would bypass OCR entirely.
Replace the case proxy with the actual invariant. A substitution cipher is
a bijection over a real language's alphabet, so it preserves the frequency
SHAPE (the sorted histogram) while scrambling letter POSITIONS (the unsorted
histogram). Signal 2 now flags when english_cosine < 0.60 (positions unlike
English) AND english_shape_cosine >= 0.90 (profile is still English-shaped).
This is independent of case, so it catches all-lower, all-upper, and
case-straddling shifts alike.
The exempted structured content fails one half: DNA/hex dumps have too steep
a profile (shape cosine 0.74 / 0.81 < 0.90), while protein sequences, ticker
symbols and base64 are not sufficiently unlike English in position (unsorted
cosine 0.74 / 0.75 / 0.77 >= 0.60). All stay out of OCR.
Still strictly corpus-safe: every real Latin document scores unsorted cosine
>= 0.70 (min 0.80), far above the 0.60 gate, so none can reach Signal 2.
Re-verified byte-identical to a baseline main binary across all 185 eval
PDFs; att10k remains flagged. Drops the now-unused case counters and adds
all-lowercase / all-uppercase shifted-prose regression tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: source Python package version from Cargo.toml via maturin
Address PR review (cubic P2): pyproject.toml pinned version = "0.1.0",
which overrides Cargo.toml, so a maturin build produced a 0.1.0 Python
artifact regardless of the crate version (it had drifted since the PyO3
bindings were added). Switch to dynamic = ["version"] so maturin sources
the version from Cargo.toml [package] version and the two can no longer
diverge. No workflow auto-publishes the Python package, so this is metadata
hygiene rather than a release-path fix.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(markdown): underline emission, Unicode scripts, style-preserving merges (ENG-5015 2b)
Three formatting losses in the direct-extraction markdown path:
1. text_with_formatting gains <u> run emission (detect_underline option,
default on) using the geometric is_underline flag from 1.9.9.
Underline runs stay free of nested bold/italic markers — consumers
match tag content literally. Heading lines keep plain text for
bold/italic but preserve <u>: the tag carries meaning `#` doesn't.
2. merge_subscript_items now maps absorbed digit scripts to Unicode
sub/superscript forms with direction from the baseline offset
("H"+"2" -> "H₂", "word"+raised "2" -> "word²", "m"+"3" -> "m³").
NFKC/NFKD folds these back to plain digits so text matching
downstream is unaffected; renderers keep the script semantics.
3. merge_text_items no longer merges across bold/italic boundaries —
absorbing a styled run into a plain neighbor erased the styling
before markdown emission ever saw it. On eval docs this recovers
20-82 italic runs per document that previously emitted as plain.
Snapshots regenerated (diffs are the features: CCl₂F₂, m³, underlined
legal section headings, finer bold runs). pdf-evals regression suite:
202/202 real PDFs pass. napi 1.9.9 -> 1.9.10.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): break merges at underline boundaries too (review)
OR-merging underline stretched the eventual <u> span over neighboring
plain fragments. Merge runs now break on any style-flag change, the
redundant accumulator is gone, and format_list_item learned to move
bullet markers outside <u> wrappers so fully-underlined bullet lines
still render as markdown lists. td9264 snapshot regenerated — spans are
tighter (trailing periods correctly outside the tag).
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(markdown): strip stray spaces before sentence punctuation (review)
Style-boundary item splits can strand a trailing period in its own
fragment, and multiple assembly paths join fragments with spaces,
yielding "word ." artifacts. Rather than chasing every join site, a
postprocess pass removes a space before `.`/`,`/`;` when the mark ends
its token (whitespace, cell boundary `|`, or end of text follows).
Dot leaders/ellipses and mid-token periods are untouched.
Fixes the td9264 "companies ." artifacts and two pre-existing
"armoring ," artifacts in the 2013-app2 snapshot. pdf-evals: zero
markdown diffs across all 203 corpus PDFs vs committed baselines.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(tables): trim spaces inside parenthetical cell fragments
* fix(tables): reject sparse prose row-stripe tables
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* feat(extractor): geometric underline detection on TextItem (ENG-5015)
PDFs carry no underline font flag — underlines are stroked horizontal
lines or thin filled rects drawn under the baseline. Correlate those
graphics (already parsed from the content stream) with text items in a
post-pass: a rule within ~0.35em below the baseline covering >=60% of
an item's width marks is_underline.
Exposed through the napi and python bindings. Verified on real docs:
4/4 underlined sentences flagged on a Japanese report, links/headings
flagged on 8 of 10 underline-bearing eval docs, zero flags on docs
without underlines. Known FP source (table cell borders) documented —
downstream applies inline styling only to plain-text regions.
napi 1.9.8 -> 1.9.9.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): underline rules only from painted rects, normalized extents (review)
Two review fixes: (1) normalize rect extents before the thickness/width
checks — `re` operands pass through the CTM so width/height can be
negative, which missed negative-width rules and let negative-height
bands pass as thin; (2) only feed painted rects to underline detection —
`re` rects now wait in a pending list until a paint operator (S/s, f/F/
f*, B/B*/b/b*) confirms them, and `re W n` clip-only paths are discarded
at `n`, so invisible clip boundaries no longer underline nearby text.
Marking moved into content_stream where paint state lives (pre-rotation,
consistent device space).
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(extractor): harden underline detection
* feat(cli): export positioned text item json
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
