js.documents 14.6.4 → 14.6.6

package/README.md

@@ -859,7 +859,7 @@ The package is layered from generic primitives outward to the two conversion dir
 - **`src/markdown/`** — third adapter family, via `markdown-codec`. `readMarkdownContent` passes `readMarkdownContent`'s (markdown-codec's flat reader, so named since that package's 4.0.0; the bare `readMarkdown` name is now its tree-form `DocumentTree` reader) result through the math-lowering pass (`math.ts` — markdown-codec's preserved `$$` display blocks and `\( \)` inline spans become two-layer formula blocks, with the document's symbol table seeded from its own prose). `buildMarkdownText` wraps `writeMarkdownContent`, reconstructing markdown math syntax from formula blocks carrying a presentation layer. `text.ts` is the byte↔text boundary. `MarkdownEditor` holds a mutable in-memory `ContentDocument`.
 - **`src/csv/`** — fourth adapter family, sharing the spreadsheet variant with xlsx/ods. `records.ts` is the RFC 4180 record parser/writer (one shared `quoteCsvField`, also used by the `.odb` CSV exporter); `text.ts` is the byte↔text boundary, working the character encoding out from the bytes through [byte-codec](../byte-codec/README.md#text-decoding) rather than assuming UTF-8, so Excel's own windows-1252 export reads, and refusing bytes that are not text at all; `read.ts` turns records into a spreadsheet `ContentDocument` (first record as verbatim string header, data cells through the same cell-typing heuristic `pdfToOds` uses); `write.ts` turns one sheet of a spreadsheet `ContentDocument` back into records via each cell's `displayText`. TSV is the same format with `{ delimiter: '\t' }` on either side.
 - **`src/svg/`** — fifth adapter family, sharing the drawing variant with odg. `text.ts` is the byte↔text boundary: unlike csv and markdown, it does not detect the encoding from the bytes, because SVG is XML and XML's own `<?xml ... encoding="..."?>` declaration is authoritative; `xml-encoding.ts` reads that declaration (or a byte order mark) from the leading bytes per XML 1.0's own external-encoding-detection algorithm, ahead of any real decode, with UTF-8 the strict default when neither is present; `read.ts` maps the six SVG shape primitives (rect/circle/ellipse/line/polyline/polygon/path) onto a one-page drawing `ContentDocument`, with transform lists composed as 2×3 affines and CSS lengths and the viewBox map resolved into page points; `write.ts` writes the six primitives back out, one shape element each; `path.ts` is the full SVG path-data grammar (M/L/H/V/C/S/Z plus Q/T/A and the relative forms — S/Q/T convert exactly, A is the one bounded approximation at ≤90° per cubic); `transform.ts` parses and composes the transform attribute and classifies the result by frame representability; `units.ts` resolves CSS length units and the viewBox; `paint.ts` resolves fill/stroke presentation attributes and dash styles; `diagnostics.ts` is the shared scope-limit vocabulary.
-- **`src/layout/`** — the pure conversion algorithms: `engine.ts` (wordprocessing → layout: flow, line-breaking, pagination), `slides.ts` (presentation → layout: direct placement), `sheets.ts` (spreadsheet → layout: grid, print settings, the first algorithm accepting `AbortSignal`), `drawing.ts` (drawing → layout: vector primitives + shape reuse), `reconstruct.ts` (layout → content: baseline clustering for wordprocessing/presentation, near-1:1 mapping for drawing, gridline-lattice-or-text-clustering for spreadsheet; plus the PDF construct surfacing — link reconciliation to `ContentRun.hyperlink` or link constructs, hidden-layer content dropped, anchor/comment and AcroForm contentControl constructs, and tagged-structure semantics where the file states them: heading levels from owning `H1`..`H6` elements over the font-size census, lattice-free table recovery from `/Table`/`/TR`/`/TH`/`/TD` ownership, and `division` constructs around `/Part`/`/Sect`/`/Div` extents).
+- **`src/layout/`** — the pure conversion algorithms: `engine.ts` (wordprocessing → layout: flow, line-breaking, pagination), `slides.ts` (presentation → layout: direct placement), `sheets.ts` (spreadsheet → layout: grid, print settings, the first algorithm accepting `AbortSignal`), `drawing.ts` (drawing → layout: vector primitives + shape reuse), `reconstruct.ts` (layout → content: baseline clustering for wordprocessing/presentation, with a superscript or subscript returned to the line of the run it touches and marked with `ContentRun.verticalAlign` by pdf-codec's `scriptPositionOf`, so `pdfToMarkdown` writes `m<sup>2</sup>` rather than `m` and a stray `2`; near-1:1 mapping for drawing, gridline-lattice-or-text-clustering for spreadsheet; plus the PDF construct surfacing — link reconciliation to `ContentRun.hyperlink` or link constructs, hidden-layer content dropped, anchor/comment and AcroForm contentControl constructs, and tagged-structure semantics where the file states them: heading levels from owning `H1`..`H6` elements over the font-size census, lattice-free table recovery from `/Table`/`/TR`/`/TH`/`/TD` ownership, and `division` constructs around `/Part`/`/Sect`/`/Div` extents).
 - **`src/hsqldb/`** — `.odb` decoders, four tiers: `script.ts` (TEXT-script DDL/DML parser), `rowformat.ts`/`cache.ts` (CACHED binary row-store), `binary-script.ts` (BINARY/COMPRESSED whole-script). All import only `document-schema.js` — no odf.js knowledge.
 - **`src/firebird/`** — Tier 3: gbak logical-backup reader. `reader.ts` (attribute framing + RLE decompression + XDR decoding), `schema.ts`/`data.ts` (table/row walking). No ratified spec — built against Firebird's own engine source.
 - **`src/odb/`** — decoder-selection and pivot-mapping: `read.ts` routes to the right tier, `spreadsheet.ts`/`csv.ts` map to output formats. `odb/sql/` is the bounded SQL engine, `odb/formula/` is the rpt formula engine, `odb/report/` is the renderer, `odb/values.ts` is shared comparison/aggregation semantics.
@@ -934,7 +934,7 @@ To run a single test file: `pnpm vitest run src/path/to/file.test.ts`.
 - **Font resolution uses a real registry, standard 14 as last resort.** A family with no embedded/caller/vendored face (Aptos, third-party typefaces) renders through the nearest standard-14 face with a width-correction factor — expect a visual approximation, not line-identical output. MathML formula rendering is separate: it embeds STIX Two Math, not registry-resolvable.
 - **Justified paragraphs stretch inter-word gaps** in all three layout engines (`engine.ts`, `slides.ts`, `sheets.ts`). `justifyLineGapsPt` divides slack evenly across detected word gaps; final lines stay left-aligned.
 - **Encrypted PDFs and CCITT/JBIG2/JPX images are real capabilities** in pdf-codec — not scope boundaries. The permanent boundary is adversarial/malformed-input robustness.
-- **PDF → docx/pptx/odt/odp table recovery requires a real drawn gridline lattice** — never text alignment (which would invent structure). A lattice with no text inside is rejected.
+- **PDF → docx/pptx/odt/odp table recovery requires drawn rules** — never text alignment alone (which would invent structure). Either a real gridline lattice, or a table ruled between its rows (`src/layout/ruled-rows.ts`): at least three horizontal rules of one span, with the columns taken from the gutters every row holding text in more than one place leaves, a heading row above the first rule read when its lines keep to the columns, a row crossing a gutter read as a spanning cell, and a band whose every line fills its columns split into a row per line. Such a grid is refused when every column reads as running prose or when anything other than horizontal strokes and full-width shading is drawn inside it (a diagram). A lattice with no text inside is rejected.
 - **Merged table cells round-trip as merged.** docx: horizontal merge collapses to one `w:tc` with `w:gridSpan`; vertical merge needs one `w:tc` per covered row with `w:vMerge`. ODF: one entry per grid position, covered cells get `table:covered-table-cell`.
 - **docx headers/footers/comments/footnotes/numbering are readable via `readDocxExtras`** — `readDocxContent` still drops them (`ContentDocument` has nowhere to put them). `PAGE`/`NUMPAGES` field substitution is never read (it's a render-time value).
 - **A docx inline image reads as a real `ContentImageBlock`** — `buildDocxPackage` recognises the flat docx reader's two-block pattern (empty-text paragraph + image) and writes it back as one paragraph, avoiding spurious blank paragraphs on round trip.
@@ -1012,7 +1012,7 @@ Read as **row → column**. `✓` lossless, `~` bounded, `✗` lossy, `✗✗` s
 
 **odf → PDF and embedded formulas** render faithful mathematical typesetting through STIX Two Math: real box-model layout, per-glyph metrics, font-wide constants from the `MATH` table, stretchy fences and braces via `MathVariants`. `pdfToOdf` is not attempted — recovering a semantic operator tree from glyphs is OCR-adjacent.
 
-**PDF → docx/pptx/odt/odp** is best-effort reconstruction from geometry. Reading order, font properties, page count survive; paragraph boundaries are inferred from baseline spacing. Text set in columns is read one column at a time: a vertical strip that no text touches over a run of lines, with running prose on both sides (a measure of at least twenty characters, lines that wrap, word gaps narrower than the strip), splits the page there before lines are clustered, so two columns whose baselines line up no longer interleave line by line. Lines crossing the strip (a full-width heading, a boxed passage) are read in page order between the column bands above and below them, and images and recovered tables are placed by the column they sit in. A table's rows are still read across, since its columns are too narrow or too ragged to pass as prose; a page with no such strip, or with text whose horizontal extent the reader did not report, reads exactly as clustering the whole page always did (`src/layout/reading-order.ts`). Tables recover only from a real gridline lattice. Vector primitives recover into a nested drawing document.
+**PDF → docx/pptx/odt/odp** is best-effort reconstruction from geometry. Reading order, font properties, page count survive; paragraph boundaries are inferred from baseline spacing. Text set in columns is read one column at a time: a vertical strip that no text touches over a run of lines, with running prose on both sides (a measure of at least twenty characters, lines that wrap, word gaps narrower than the strip), splits the page there before lines are clustered, so two columns whose baselines line up no longer interleave line by line. Lines crossing the strip (a full-width heading, a boxed passage) are read in page order between the column bands above and below them, and images and recovered tables are placed by the column they sit in. A table's rows are still read across, since its columns are too narrow or too ragged to pass as prose; a page with no such strip, or with text whose horizontal extent the reader did not report, reads exactly as clustering the whole page always did (`src/layout/reading-order.ts`). Tables recover only from drawn rules: a real gridline lattice, or a table ruled between its rows with its columns set apart by whitespace, which is read cell by cell so a cell's wrapped lines stay together however the producer ordered its drawing (see the table-recovery gotcha below). Vector primitives recover into a nested drawing document.
 
 **PDF → odg** is near-1:1 mapping (no clustering needed). Kind narrows upstream: rotated rects, freeform curves, multi-subpath figures become `path`.
 

package/dist/{composition-plan-DMTkmk8n.js → composition-plan-BLhc55_x.js}

@@ -23,7 +23,7 @@ import { buildCsvText } from "./csv/write.js";
 import { i as readSvgContent } from "./read-BSH--XB4.js";
 import { buildSvgText } from "./svg/write.js";
 import { throwIfAborted } from "./ports/abort.js";
-import { _ as reconstructWordprocessing, g as reconstructSpreadsheet, i as reconstructPresentation, r as reconstructDrawing } from "./reconstruct-presentations-prod-D1KbGPxa.js";
+import { _ as reconstructWordprocessing, g as reconstructSpreadsheet, i as reconstructPresentation, r as reconstructDrawing } from "./reconstruct-presentations-prod-DV7UQqMT.js";
 import { stampPdfPackageTables } from "./convert/pdf-package-tables.js";
 import { drawingToPresentation, presentationToDrawing, presentationToWordprocessing, spreadsheetToWordprocessing, wordprocessingToPresentation } from "./convert/variant-bridges.js";
 import { UnsupportedConversionError } from "./convert/capability.js";

package/dist/{composition-plan-DJQgPvBj.cjs → composition-plan-CgjS-XZU.cjs}

@@ -23,7 +23,7 @@ const require_csv_write = require("./csv/write.cjs");
 const require_read$1 = require("./read-CahCZMwG.cjs");
 const require_svg_write = require("./svg/write.cjs");
 const require_ports_abort = require("./ports/abort.cjs");
-const require_reconstruct_presentations_prod = require("./reconstruct-presentations-prod-CsvGlvP2.cjs");
+const require_reconstruct_presentations_prod = require("./reconstruct-presentations-prod-DtpKKgRY.cjs");
 const require_convert_pdf_package_tables = require("./convert/pdf-package-tables.cjs");
 const require_convert_variant_bridges = require("./convert/variant-bridges.cjs");
 const require_convert_capability = require("./convert/capability.cjs");

package/dist/convert/composition-plan.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_composition_plan = require("../composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("../composition-plan-CgjS-XZU.cjs");
 exports.buildCompositionGraph = require_composition_plan.buildCompositionGraph;
 exports.isContentFormat = require_composition_plan.isContentFormat;
 exports.isSourceContentFormat = require_composition_plan.isSourceContentFormat;

package/dist/convert/composition-plan.js

@@ -1,2 +1,2 @@
-import { a as shortestPath, i as resolveCompositionPlan, n as isContentFormat, r as isSourceContentFormat, t as buildCompositionGraph } from "../composition-plan-DMTkmk8n.js";
+import { a as shortestPath, i as resolveCompositionPlan, n as isContentFormat, r as isSourceContentFormat, t as buildCompositionGraph } from "../composition-plan-BLhc55_x.js";
 export { buildCompositionGraph, isContentFormat, isSourceContentFormat, resolveCompositionPlan, shortestPath };

package/dist/convert/composition-to-pdf.cjs

@@ -7,7 +7,7 @@ const require_ports_abort = require("../ports/abort.cjs");
 const require_sheets = require("../sheets-BQCpNXsE.cjs");
 const require_layout_drawing = require("../layout/drawing.cjs");
 const require_convert_capability = require("./capability.cjs");
-const require_composition_plan = require("../composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("../composition-plan-CgjS-XZU.cjs");
 let document_schema_js = require("document-schema.js");
 let pdf_codec = require("pdf-codec");
 //#region src/convert/composition-to-pdf.ts

package/dist/convert/composition-to-pdf.js

@@ -6,7 +6,7 @@ import { throwIfAborted } from "../ports/abort.js";
 import { l as convertSpreadsheetToLayout } from "../sheets-DWDvGN-g.js";
 import { convertDrawingToLayout } from "../layout/drawing.js";
 import { UnsupportedConversionError } from "./capability.js";
-import { _ as executeFromPdf, b as runCompositionPlan, c as LAYOUT_CAPABLE, d as READ_ONLY_FORMAT_NODES, g as executeBridge, i as resolveCompositionPlan, s as FORMAT_NODES, v as isReadOnlyContentFormat, y as isTextFormatNode } from "../composition-plan-DMTkmk8n.js";
+import { _ as executeFromPdf, b as runCompositionPlan, c as LAYOUT_CAPABLE, d as READ_ONLY_FORMAT_NODES, g as executeBridge, i as resolveCompositionPlan, s as FORMAT_NODES, v as isReadOnlyContentFormat, y as isTextFormatNode } from "../composition-plan-BLhc55_x.js";
 import { assembleTree } from "document-schema.js";
 import { createFontMeasurer, createFontRegistry, loadMathFont, writePdf } from "pdf-codec";
 //#region src/convert/composition-to-pdf.ts

package/dist/convert/composition.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_composition_plan = require("../composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("../composition-plan-CgjS-XZU.cjs");
 exports.CONTENT_FORMATS = require_composition_plan.CONTENT_FORMATS;
 exports.FORMAT_NODES = require_composition_plan.FORMAT_NODES;
 exports.LAYOUT_CAPABLE = require_composition_plan.LAYOUT_CAPABLE;

package/dist/convert/composition.js

@@ -1,2 +1,2 @@
-import { _ as executeFromPdf, b as runCompositionPlan, c as LAYOUT_CAPABLE, d as READ_ONLY_FORMAT_NODES, f as RECONSTRUCTORS, g as executeBridge, h as convertDocumentFromPdf, l as MAX_CONVERSION_PATH_NODES, m as TRANSFORMS, o as CONTENT_FORMATS, p as THROUGH_PDF_EDGE_COST, s as FORMAT_NODES, u as READ_ONLY_CONTENT_FORMATS, v as isReadOnlyContentFormat, y as isTextFormatNode } from "../composition-plan-DMTkmk8n.js";
+import { _ as executeFromPdf, b as runCompositionPlan, c as LAYOUT_CAPABLE, d as READ_ONLY_FORMAT_NODES, f as RECONSTRUCTORS, g as executeBridge, h as convertDocumentFromPdf, l as MAX_CONVERSION_PATH_NODES, m as TRANSFORMS, o as CONTENT_FORMATS, p as THROUGH_PDF_EDGE_COST, s as FORMAT_NODES, u as READ_ONLY_CONTENT_FORMATS, v as isReadOnlyContentFormat, y as isTextFormatNode } from "../composition-plan-BLhc55_x.js";
 export { CONTENT_FORMATS, FORMAT_NODES, LAYOUT_CAPABLE, MAX_CONVERSION_PATH_NODES, READ_ONLY_CONTENT_FORMATS, READ_ONLY_FORMAT_NODES, RECONSTRUCTORS, THROUGH_PDF_EDGE_COST, TRANSFORMS, convertDocumentFromPdf, executeBridge, executeFromPdf, isReadOnlyContentFormat, isTextFormatNode, runCompositionPlan };

package/dist/convert/from-pdf.cjs

@@ -1,7 +1,7 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-CsvGlvP2.cjs");
+const require_reconstruct_presentations_prod = require("../reconstruct-presentations-prod-DtpKKgRY.cjs");
 const require_convert_pdf_package_tables = require("./pdf-package-tables.cjs");
-const require_composition_plan = require("../composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("../composition-plan-CgjS-XZU.cjs");
 const require_codecs_read = require("../codecs/read.cjs");
 let document_schema_js = require("document-schema.js");
 let pdf_codec_read = require("pdf-codec/read");

package/dist/convert/from-pdf.js

@@ -1,6 +1,6 @@
-import { _ as reconstructWordprocessing } from "../reconstruct-presentations-prod-D1KbGPxa.js";
+import { _ as reconstructWordprocessing } from "../reconstruct-presentations-prod-DV7UQqMT.js";
 import { stampPdfPackageTables } from "./pdf-package-tables.js";
-import { h as convertDocumentFromPdf } from "../composition-plan-DMTkmk8n.js";
+import { h as convertDocumentFromPdf } from "../composition-plan-BLhc55_x.js";
 import { CONTENT_READERS } from "../codecs/read.js";
 import { assembleTree } from "document-schema.js";
 import { readPdf } from "pdf-codec/read";

package/dist/convert/local.cjs

@@ -1,7 +1,7 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
 const require_convert_capability = require("./capability.cjs");
 const require_convert_port = require("./port.cjs");
-const require_composition_plan = require("../composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("../composition-plan-CgjS-XZU.cjs");
 const require_convert_composition_to_pdf = require("./composition-to-pdf.cjs");
 const require_convert_convert = require("./convert.cjs");
 //#region src/convert/local.ts

package/dist/convert/local.js

@@ -1,6 +1,6 @@
 import { UnsupportedConversionError } from "./capability.js";
 import { DOCUMENT_FORMATS } from "./port.js";
-import { i as resolveCompositionPlan } from "../composition-plan-DMTkmk8n.js";
+import { i as resolveCompositionPlan } from "../composition-plan-BLhc55_x.js";
 import { convertDocument } from "./composition-to-pdf.js";
 import { odfToPdf } from "./convert.js";
 //#region src/convert/local.ts

package/dist/index.cjs

@@ -93,7 +93,7 @@ const require_ports_abort = require("./ports/abort.cjs");
 const require_sheets = require("./sheets-BQCpNXsE.cjs");
 const require_layout_drawing = require("./layout/drawing.cjs");
 const require_layout_lattice = require("./layout/lattice.cjs");
-const require_reconstruct_presentations_prod = require("./reconstruct-presentations-prod-CsvGlvP2.cjs");
+const require_reconstruct_presentations_prod = require("./reconstruct-presentations-prod-DtpKKgRY.cjs");
 const require_hsqldb_script = require("./hsqldb/script.cjs");
 const require_odb_csv = require("./odb/csv.cjs");
 const require_hsqldb_rowformat = require("./hsqldb/rowformat.cjs");
@@ -107,7 +107,7 @@ const require_firebird_backup = require("./firebird/backup.cjs");
 const require_odb_read = require("./odb/read.cjs");
 const require_odb_spreadsheet = require("./odb/spreadsheet.cjs");
 const require_convert_port = require("./convert/port.cjs");
-const require_composition_plan = require("./composition-plan-DJQgPvBj.cjs");
+const require_composition_plan = require("./composition-plan-CgjS-XZU.cjs");
 const require_convert_composition_to_pdf = require("./convert/composition-to-pdf.cjs");
 const require_convert_convert = require("./convert/convert.cjs");
 const require_package_codec = require("./package-codec.cjs");

package/dist/index.js

@@ -92,7 +92,7 @@ import { throwIfAborted } from "./ports/abort.js";
 import { l as convertSpreadsheetToLayout } from "./sheets-DWDvGN-g.js";
 import { convertDrawingToLayout } from "./layout/drawing.js";
 import { detectGridLattice } from "./layout/lattice.js";
-import { _ as reconstructWordprocessing, g as reconstructSpreadsheet, i as reconstructPresentation, r as reconstructDrawing } from "./reconstruct-presentations-prod-D1KbGPxa.js";
+import { _ as reconstructWordprocessing, g as reconstructSpreadsheet, i as reconstructPresentation, r as reconstructDrawing } from "./reconstruct-presentations-prod-DV7UQqMT.js";
 import { HsqldbScriptParseError, displayTextFor, parseHsqldbScript } from "./hsqldb/script.js";
 import { OdbTableNotFoundError, OdbTableNotSpecifiedError } from "./odb/csv.js";
 import { HsqldbRowFormatError } from "./hsqldb/rowformat.js";
@@ -106,7 +106,7 @@ import { FirebirdBackupFormatError, SUPPORTED_BACKUP_FORMAT_VERSION, readFirebir
 import { OdbNoEmbeddedDataSourceError, OdbUnsupportedFormatError, readOdbTables } from "./odb/read.js";
 import { odbTablesToSpreadsheetDocument } from "./odb/spreadsheet.js";
 import { DOCUMENT_FORMATS, DocumentFormatSchema } from "./convert/port.js";
-import { i as resolveCompositionPlan } from "./composition-plan-DMTkmk8n.js";
+import { i as resolveCompositionPlan } from "./composition-plan-BLhc55_x.js";
 import { convertDocument } from "./convert/composition-to-pdf.js";
 import { OdmUnresolvedSectionError, csvToMarkdown, csvToOds, csvToPdf, csvToXlsx, docToMarkdown, docToPdf, docxToMarkdown, docxToOdt, docxToPdf, docxToPptx, epubToPdf, markdownToCsv, markdownToDocx, markdownToOdt, markdownToPdf, markdownToXlsx, odbReportToDocx, odbReportToOdt, odbReportToPdf, odbToCsv, odbToXlsx, odfToPdf, odgToPdf, odgToSvg, odmToPdf, odpToOdt, odpToPdf, odpToPptx, odsToCsv, odsToPdf, odsToXlsx, odtToDocx, odtToMarkdown, odtToOdp, odtToPdf, pptToMarkdown, pptToPdf, pptxToDocx, pptxToOdp, pptxToPdf, rtfToPdf, svgToOdg, svgToPdf, xlsToMarkdown, xlsToPdf, xlsxToCsv, xlsxToMarkdown, xlsxToOds, xlsxToPdf } from "./convert/convert.js";
 import { UnsupportedPackageFormatError, decodeDocumentPackage, decodeOdbPackage, encodeDocumentPackage } from "./package-codec.js";

package/dist/layout/lattice.cjs

@@ -29,7 +29,7 @@ function extractLineCandidates(items) {
 			continue;
 		}
 		if (item.kind === "rect") {
-			const isThinEnoughToBeALine = Math.min(item.widthPt, item.heightPt) <= RECT_LINE_THICKNESS_TOLERANCE_PT;
+			const isThinEnoughToBeALine = Math.min(item.widthPt, item.heightPt) <= 3;
 			if (isThinEnoughToBeALine && item.widthPt > item.heightPt) {
 				const midYPt = item.yPt + item.heightPt / 2;
 				segments.push({
@@ -123,7 +123,7 @@ function hasClosedOuterRectangle(rowLines, columnLines) {
 	const rightXPt = columnLines[columnLines.length - 1].position;
 	const topYPt = rowLines[0].position;
 	const bottomYPt = rowLines[rowLines.length - 1].position;
-	return unionCoverageRatio(rowLines[0].ranges, leftXPt, rightXPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(rowLines[rowLines.length - 1].ranges, leftXPt, rightXPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(columnLines[0].ranges, bottomYPt, topYPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(columnLines[columnLines.length - 1].ranges, bottomYPt, topYPt) >= EDGE_COVERAGE_RATIO;
+	return unionCoverageRatio(rowLines[0].ranges, leftXPt, rightXPt) >= .9 && unionCoverageRatio(rowLines[rowLines.length - 1].ranges, leftXPt, rightXPt) >= .9 && unionCoverageRatio(columnLines[0].ranges, bottomYPt, topYPt) >= .9 && unionCoverageRatio(columnLines[columnLines.length - 1].ranges, bottomYPt, topYPt) >= .9;
 }
 function unionFind(size) {
 	return Array.from({ length: size }, (_, i) => i);
@@ -154,7 +154,7 @@ function findCellRegions(rowLines, columnLines) {
 		const rowBottomPt = rowLines[i + 1].position;
 		for (let j = 0; j < colCount - 1; j++) {
 			const divider = columnLines[j + 1];
-			if (unionCoverageRatio(divider.ranges, rowBottomPt, rowTopPt) < EDGE_COVERAGE_RATIO) union(parents, indexOf(i, j), indexOf(i, j + 1));
+			if (unionCoverageRatio(divider.ranges, rowBottomPt, rowTopPt) < .9) union(parents, indexOf(i, j), indexOf(i, j + 1));
 		}
 	}
 	for (let j = 0; j < colCount; j++) {
@@ -162,7 +162,7 @@ function findCellRegions(rowLines, columnLines) {
 		const colRightPt = columnLines[j + 1].position;
 		for (let i = 0; i < rowCount - 1; i++) {
 			const divider = rowLines[i + 1];
-			if (unionCoverageRatio(divider.ranges, colLeftPt, colRightPt) < EDGE_COVERAGE_RATIO) union(parents, indexOf(i, j), indexOf(i + 1, j));
+			if (unionCoverageRatio(divider.ranges, colLeftPt, colRightPt) < .9) union(parents, indexOf(i, j), indexOf(i + 1, j));
 		}
 	}
 	const cellsByRoot = /* @__PURE__ */ new Map();
@@ -220,6 +220,25 @@ function clusterSegments(horizontal, vertical) {
 	}
 	return [...clusters.values()];
 }
+/**
+* Every horizontal rule on the page that its drawn segments cover across its own span, top to bottom.
+* @param items - the page's items; lines, paths and thin rectangles are read as segments, as for a lattice
+* @returns one rule per boundary position, merging collinear segments the way a lattice boundary does, and only those whose segments cover at least EDGE_COVERAGE_RATIO of the span from their leftmost to their rightmost point, so a rule broken at a table's column gutters counts while a row of separate short rules (underlines under a few words) does not
+*/
+function horizontalRules(items) {
+	const horizontal = extractLineCandidates(items).map(classifyAxisLine).filter((segment) => segment?.axis === "horizontal");
+	if (horizontal.length === 0) return [];
+	return dedupeAxisLines(horizontal, true).flatMap((line) => {
+		const leftPt = Math.min(...line.ranges.map((range) => range.startPt));
+		const rightPt = Math.max(...line.ranges.map((range) => range.endPt));
+		return unionCoverageRatio(line.ranges, leftPt, rightPt) >= .9 ? [{
+			position: line.position,
+			leftPt,
+			rightPt,
+			items: line.items
+		}] : [];
+	});
+}
 function detectGridLattice(items) {
 	const horizontal = [];
 	const vertical = [];
@@ -235,7 +254,7 @@ function detectGridLattice(items) {
 		const clusterVertical = cluster.filter((s) => s.axis === "vertical");
 		const rowLines = dedupeAxisLines(clusterHorizontal, true);
 		const columnLines = dedupeAxisLines(clusterVertical, false);
-		if (rowLines.length < MIN_GRIDLINE_COUNT_PER_AXIS || columnLines.length < MIN_GRIDLINE_COUNT_PER_AXIS) continue;
+		if (rowLines.length < 3 || columnLines.length < 3) continue;
 		if (!hasClosedOuterRectangle(rowLines, columnLines)) continue;
 		const regions = findCellRegions(rowLines, columnLines);
 		if (regions === void 0) continue;
@@ -267,8 +286,12 @@ function findColumnIndex(columnBoundariesAscPt, xPt) {
 	return lastIndex - 1;
 }
 //#endregion
+exports.EDGE_COVERAGE_RATIO = EDGE_COVERAGE_RATIO;
+exports.MIN_GRIDLINE_COUNT_PER_AXIS = MIN_GRIDLINE_COUNT_PER_AXIS;
+exports.RECT_LINE_THICKNESS_TOLERANCE_PT = RECT_LINE_THICKNESS_TOLERANCE_PT;
 exports.detectGridLattice = detectGridLattice;
 exports.extractLineCandidates = extractLineCandidates;
 exports.findCellRegions = findCellRegions;
 exports.findColumnIndex = findColumnIndex;
 exports.findRowIndex = findRowIndex;
+exports.horizontalRules = horizontalRules;

package/dist/layout/lattice.d.cts

@@ -1,5 +1,6 @@
 import { LayoutItem } from "pdf-codec";
 //#region src/layout/lattice.d.ts
+declare const RECT_LINE_THICKNESS_TOLERANCE_PT = 3;
 interface LineSegment {
   readonly item: LayoutItem;
   readonly x1Pt: number;
@@ -17,6 +18,8 @@ interface AxisLine {
   readonly ranges: readonly Range[];
   readonly items: readonly LayoutItem[];
 }
+declare const MIN_GRIDLINE_COUNT_PER_AXIS = 3;
+declare const EDGE_COVERAGE_RATIO = 0.9;
 interface TableRegion {
   readonly rowStart: number;
   readonly rowEnd: number;
@@ -30,8 +33,20 @@ interface GridLattice {
   readonly regions: readonly TableRegion[];
   readonly sourceItems: ReadonlySet<LayoutItem>;
 }
+interface HorizontalRule {
+  readonly position: number;
+  readonly leftPt: number;
+  readonly rightPt: number;
+  readonly items: readonly LayoutItem[];
+}
+/**
+ * Every horizontal rule on the page that its drawn segments cover across its own span, top to bottom.
+ * @param items - the page's items; lines, paths and thin rectangles are read as segments, as for a lattice
+ * @returns one rule per boundary position, merging collinear segments the way a lattice boundary does, and only those whose segments cover at least EDGE_COVERAGE_RATIO of the span from their leftmost to their rightmost point, so a rule broken at a table's column gutters counts while a row of separate short rules (underlines under a few words) does not
+ */
+declare function horizontalRules(items: readonly LayoutItem[]): HorizontalRule[];
 declare function detectGridLattice(items: readonly LayoutItem[]): GridLattice | undefined;
 declare function findRowIndex(rowBoundariesDescPt: readonly number[], yPt: number): number | undefined;
 declare function findColumnIndex(columnBoundariesAscPt: readonly number[], xPt: number): number | undefined;
 //#endregion
-export { GridLattice, LineSegment, TableRegion, detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex };
+export { EDGE_COVERAGE_RATIO, GridLattice, HorizontalRule, LineSegment, MIN_GRIDLINE_COUNT_PER_AXIS, RECT_LINE_THICKNESS_TOLERANCE_PT, TableRegion, detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex, horizontalRules };

package/dist/layout/lattice.d.ts

@@ -1,5 +1,6 @@
 import { LayoutItem } from "pdf-codec";
 //#region src/layout/lattice.d.ts
+declare const RECT_LINE_THICKNESS_TOLERANCE_PT = 3;
 interface LineSegment {
   readonly item: LayoutItem;
   readonly x1Pt: number;
@@ -17,6 +18,8 @@ interface AxisLine {
   readonly ranges: readonly Range[];
   readonly items: readonly LayoutItem[];
 }
+declare const MIN_GRIDLINE_COUNT_PER_AXIS = 3;
+declare const EDGE_COVERAGE_RATIO = 0.9;
 interface TableRegion {
   readonly rowStart: number;
   readonly rowEnd: number;
@@ -30,8 +33,20 @@ interface GridLattice {
   readonly regions: readonly TableRegion[];
   readonly sourceItems: ReadonlySet<LayoutItem>;
 }
+interface HorizontalRule {
+  readonly position: number;
+  readonly leftPt: number;
+  readonly rightPt: number;
+  readonly items: readonly LayoutItem[];
+}
+/**
+ * Every horizontal rule on the page that its drawn segments cover across its own span, top to bottom.
+ * @param items - the page's items; lines, paths and thin rectangles are read as segments, as for a lattice
+ * @returns one rule per boundary position, merging collinear segments the way a lattice boundary does, and only those whose segments cover at least EDGE_COVERAGE_RATIO of the span from their leftmost to their rightmost point, so a rule broken at a table's column gutters counts while a row of separate short rules (underlines under a few words) does not
+ */
+declare function horizontalRules(items: readonly LayoutItem[]): HorizontalRule[];
 declare function detectGridLattice(items: readonly LayoutItem[]): GridLattice | undefined;
 declare function findRowIndex(rowBoundariesDescPt: readonly number[], yPt: number): number | undefined;
 declare function findColumnIndex(columnBoundariesAscPt: readonly number[], xPt: number): number | undefined;
 //#endregion
-export { GridLattice, LineSegment, TableRegion, detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex };
+export { EDGE_COVERAGE_RATIO, GridLattice, HorizontalRule, LineSegment, MIN_GRIDLINE_COUNT_PER_AXIS, RECT_LINE_THICKNESS_TOLERANCE_PT, TableRegion, detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex, horizontalRules };

package/dist/layout/lattice.js

@@ -28,7 +28,7 @@ function extractLineCandidates(items) {
 			continue;
 		}
 		if (item.kind === "rect") {
-			const isThinEnoughToBeALine = Math.min(item.widthPt, item.heightPt) <= RECT_LINE_THICKNESS_TOLERANCE_PT;
+			const isThinEnoughToBeALine = Math.min(item.widthPt, item.heightPt) <= 3;
 			if (isThinEnoughToBeALine && item.widthPt > item.heightPt) {
 				const midYPt = item.yPt + item.heightPt / 2;
 				segments.push({
@@ -122,7 +122,7 @@ function hasClosedOuterRectangle(rowLines, columnLines) {
 	const rightXPt = columnLines[columnLines.length - 1].position;
 	const topYPt = rowLines[0].position;
 	const bottomYPt = rowLines[rowLines.length - 1].position;
-	return unionCoverageRatio(rowLines[0].ranges, leftXPt, rightXPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(rowLines[rowLines.length - 1].ranges, leftXPt, rightXPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(columnLines[0].ranges, bottomYPt, topYPt) >= EDGE_COVERAGE_RATIO && unionCoverageRatio(columnLines[columnLines.length - 1].ranges, bottomYPt, topYPt) >= EDGE_COVERAGE_RATIO;
+	return unionCoverageRatio(rowLines[0].ranges, leftXPt, rightXPt) >= .9 && unionCoverageRatio(rowLines[rowLines.length - 1].ranges, leftXPt, rightXPt) >= .9 && unionCoverageRatio(columnLines[0].ranges, bottomYPt, topYPt) >= .9 && unionCoverageRatio(columnLines[columnLines.length - 1].ranges, bottomYPt, topYPt) >= .9;
 }
 function unionFind(size) {
 	return Array.from({ length: size }, (_, i) => i);
@@ -153,7 +153,7 @@ function findCellRegions(rowLines, columnLines) {
 		const rowBottomPt = rowLines[i + 1].position;
 		for (let j = 0; j < colCount - 1; j++) {
 			const divider = columnLines[j + 1];
-			if (unionCoverageRatio(divider.ranges, rowBottomPt, rowTopPt) < EDGE_COVERAGE_RATIO) union(parents, indexOf(i, j), indexOf(i, j + 1));
+			if (unionCoverageRatio(divider.ranges, rowBottomPt, rowTopPt) < .9) union(parents, indexOf(i, j), indexOf(i, j + 1));
 		}
 	}
 	for (let j = 0; j < colCount; j++) {
@@ -161,7 +161,7 @@ function findCellRegions(rowLines, columnLines) {
 		const colRightPt = columnLines[j + 1].position;
 		for (let i = 0; i < rowCount - 1; i++) {
 			const divider = rowLines[i + 1];
-			if (unionCoverageRatio(divider.ranges, colLeftPt, colRightPt) < EDGE_COVERAGE_RATIO) union(parents, indexOf(i, j), indexOf(i + 1, j));
+			if (unionCoverageRatio(divider.ranges, colLeftPt, colRightPt) < .9) union(parents, indexOf(i, j), indexOf(i + 1, j));
 		}
 	}
 	const cellsByRoot = /* @__PURE__ */ new Map();
@@ -219,6 +219,25 @@ function clusterSegments(horizontal, vertical) {
 	}
 	return [...clusters.values()];
 }
+/**
+* Every horizontal rule on the page that its drawn segments cover across its own span, top to bottom.
+* @param items - the page's items; lines, paths and thin rectangles are read as segments, as for a lattice
+* @returns one rule per boundary position, merging collinear segments the way a lattice boundary does, and only those whose segments cover at least EDGE_COVERAGE_RATIO of the span from their leftmost to their rightmost point, so a rule broken at a table's column gutters counts while a row of separate short rules (underlines under a few words) does not
+*/
+function horizontalRules(items) {
+	const horizontal = extractLineCandidates(items).map(classifyAxisLine).filter((segment) => segment?.axis === "horizontal");
+	if (horizontal.length === 0) return [];
+	return dedupeAxisLines(horizontal, true).flatMap((line) => {
+		const leftPt = Math.min(...line.ranges.map((range) => range.startPt));
+		const rightPt = Math.max(...line.ranges.map((range) => range.endPt));
+		return unionCoverageRatio(line.ranges, leftPt, rightPt) >= .9 ? [{
+			position: line.position,
+			leftPt,
+			rightPt,
+			items: line.items
+		}] : [];
+	});
+}
 function detectGridLattice(items) {
 	const horizontal = [];
 	const vertical = [];
@@ -234,7 +253,7 @@ function detectGridLattice(items) {
 		const clusterVertical = cluster.filter((s) => s.axis === "vertical");
 		const rowLines = dedupeAxisLines(clusterHorizontal, true);
 		const columnLines = dedupeAxisLines(clusterVertical, false);
-		if (rowLines.length < MIN_GRIDLINE_COUNT_PER_AXIS || columnLines.length < MIN_GRIDLINE_COUNT_PER_AXIS) continue;
+		if (rowLines.length < 3 || columnLines.length < 3) continue;
 		if (!hasClosedOuterRectangle(rowLines, columnLines)) continue;
 		const regions = findCellRegions(rowLines, columnLines);
 		if (regions === void 0) continue;
@@ -266,4 +285,4 @@ function findColumnIndex(columnBoundariesAscPt, xPt) {
 	return lastIndex - 1;
 }
 //#endregion
-export { detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex };
+export { EDGE_COVERAGE_RATIO, MIN_GRIDLINE_COUNT_PER_AXIS, RECT_LINE_THICKNESS_TOLERANCE_PT, detectGridLattice, extractLineCandidates, findCellRegions, findColumnIndex, findRowIndex, horizontalRules };

package/dist/layout/reading-order.cjs

@@ -199,4 +199,6 @@ function firstWordWidthPt(item) {
 }
 //#endregion
 exports.compareReadingPosition = compareReadingPosition;
+exports.isRunningProse = isRunningProse;
 exports.readingOrder = readingOrder;
+exports.spaceAdvancePt = spaceAdvancePt;

package/dist/layout/reading-order.d.cts

@@ -36,5 +36,7 @@ declare function compareReadingPosition(a: {
   readonly path: readonly number[];
   readonly yPt: number;
 }): number;
+declare function spaceAdvancePt(item: LayoutText): number;
+declare function isRunningProse(items: readonly LayoutText[], gutterWidthPt: number): boolean;
 //#endregion
-export { ReadingOrder, ReadingRegion, compareReadingPosition, readingOrder };
+export { ReadingOrder, ReadingRegion, compareReadingPosition, isRunningProse, readingOrder, spaceAdvancePt };

package/dist/layout/reading-order.d.ts

@@ -36,5 +36,7 @@ declare function compareReadingPosition(a: {
   readonly path: readonly number[];
   readonly yPt: number;
 }): number;
+declare function spaceAdvancePt(item: LayoutText): number;
+declare function isRunningProse(items: readonly LayoutText[], gutterWidthPt: number): boolean;
 //#endregion
-export { ReadingOrder, ReadingRegion, compareReadingPosition, readingOrder };
+export { ReadingOrder, ReadingRegion, compareReadingPosition, isRunningProse, readingOrder, spaceAdvancePt };

package/dist/layout/reading-order.js

@@ -197,4 +197,4 @@ function firstWordWidthPt(item) {
 	return text.length === 0 ? 0 : (item.widthPt ?? 0) * firstWord.length / item.text.length;
 }
 //#endregion
-export { compareReadingPosition, readingOrder };
+export { compareReadingPosition, isRunningProse, readingOrder, spaceAdvancePt };

package/dist/layout/reconstruct-lines.cjs

@@ -28,13 +28,14 @@ function dropDuplicatePaints(items) {
 	return kept;
 }
 function dropOverlappingRepeatsWithinLine(items) {
-	const rightEdgeByText = /* @__PURE__ */ new Map();
+	const keptByText = /* @__PURE__ */ new Map();
 	const kept = [];
 	for (const item of items) {
 		const textKey = item.text.trim();
-		const priorRightEdge = rightEdgeByText.get(textKey);
-		if (priorRightEdge !== void 0 && item.xPt < priorRightEdge) continue;
-		rightEdgeByText.set(textKey, item.xPt + (item.widthPt ?? 0));
+		const earlier = keptByText.get(textKey);
+		if (earlier?.some((prior) => item.xPt < prior.xPt + (prior.widthPt ?? 0) && Math.abs(item.yPt - prior.yPt) <= DUPLICATE_PAINT_BUCKET_PT) === true) continue;
+		if (earlier === void 0) keptByText.set(textKey, [item]);
+		else earlier.push(item);
 		kept.push(item);
 	}
 	return kept;
@@ -127,12 +128,23 @@ function clusterIntoLines(items) {
 		});
 		else line.items.push(item);
 	}
-	for (const line of working) {
+	const lines = (0, pdf_codec_text_group.attachScriptRuns)(working, (line) => line.items).filter(({ runs }) => runs.length > 0).map(({ line, runs }) => ({
+		...line,
+		items: runs
+	}));
+	for (const line of lines) {
 		line.items.sort((a, b) => a.xPt - b.xPt);
 		line.items = dropOverlappingRepeatsWithinLine(line.items);
 	}
-	working.sort((a, b) => b.baselineY - a.baselineY);
-	return working;
+	lines.sort((a, b) => b.baselineY - a.baselineY);
+	return lines;
+}
+function scriptPositionOnLine(item, line) {
+	for (const base of line.items) {
+		if (base === item) continue;
+		const position = (0, pdf_codec_text_group.scriptPositionOf)(base, item);
+		if (position !== void 0) return position;
+	}
 }
 function textItemVerticalExtent(item) {
 	const { standardName } = (0, pdf_codec_fonts.resolveStandardFont)(item.font.family, item.font.weight === "bold", item.font.style === "italic");
@@ -192,13 +204,32 @@ function pushRunsForLine(runs, line, pageIndex) {
 		if (itemIndex > 0) {
 			const prevItem = line.items[itemIndex - 1];
 			if (gapExceeds(prevItem, item, 2 * item.sizePt)) runs.push({ text: "	" });
-			else if (gapExceeds(prevItem, item, .5)) runs[runs.length - 1].text += " ";
+			else if (gapExceeds(prevItem, item, .5)) appendWordSpace(runs);
 		}
 		const run = textItemToContentRun(item);
 		require_layout_shared.stampFrame(run, pageIndex, textBoxOfItem(item, pageIndex));
-		runs.push(run);
+		const verticalAlign = scriptPositionOnLine(item, line);
+		const core = item.text.trim();
+		if (verticalAlign === void 0 || core.length === 0) {
+			runs.push(run);
+			return;
+		}
+		const leading = item.text.slice(0, item.text.length - item.text.trimStart().length);
+		const trailing = item.text.slice(item.text.trimEnd().length);
+		if (leading.length > 0) runs.push({ text: leading });
+		runs.push({
+			...run,
+			text: core,
+			verticalAlign
+		});
+		if (trailing.length > 0) runs.push({ text: trailing });
 	});
 }
+function appendWordSpace(runs) {
+	const last = runs[runs.length - 1];
+	if (last.verticalAlign === void 0) last.text += " ";
+	else runs.push({ text: " " });
+}
 function bucketCounts(values, bucketSize) {
 	const counts = /* @__PURE__ */ new Map();
 	for (const v of values) {
@@ -281,8 +312,7 @@ function paragraphToContentParagraph(paragraph, pageIndex, headingLevel) {
 	paragraph.lines.forEach((line, lineIndex) => {
 		require_layout_shared.stampFrame(result, pageIndex, lineBox(line, pageIndex));
 		if (lineIndex > 0) {
-			const lastRun = result.runs[result.runs.length - 1];
-			if (lastRun !== void 0) lastRun.text += " ";
+			if (result.runs.length > 0) appendWordSpace(result.runs);
 		}
 		pushRunsForLine(result.runs, line, pageIndex);
 	});