@pdf-testkit/core 0.1.0

package/dist/extract/fromFormeLayout.js

@@ -0,0 +1,324 @@
+import { computeOverflow } from '../geometry.js';
+import { buildHeadingLevelMap } from '../text/headingLevels.js';
+import { finalize } from './finalize.js';
+/**
+ * Explicit disposition for every value of `@formepdf/core`'s `ElementNodeType`
+ * union. Every union member has a deliberate role here — none rely on the
+ * fallthrough. Audited against @formepdf/core 0.12.1; re-review whenever that
+ * dependency version bumps (the coverage test in formeNodeTypeCoverage.test.ts
+ * enforces that this table + FORME_INTENTIONALLY_UNMAPPED cover the whole union).
+ *
+ * Judgment calls (confirmed): charts + all graphics → opaque `image` regions;
+ * form controls → structural `container`; `Lbl` (list marker) → `text`.
+ */
+export const FORME_ROLE_BY_NODE_TYPE = {
+    // Text & headings
+    Text: 'text',
+    TextLine: 'text',
+    Lbl: 'text', // list marker (bullet/number) — diffable text
+    H1: 'heading',
+    H2: 'heading',
+    H3: 'heading',
+    H4: 'heading',
+    H5: 'heading',
+    H6: 'heading',
+    // Table primitives (no `Table` wrapper — synthesized from contiguous rows)
+    TableRow: 'row',
+    TableCell: 'cell',
+    // Structural containers / regions
+    View: 'container',
+    List: 'container',
+    ListItem: 'container',
+    FixedHeader: 'container',
+    FixedFooter: 'container',
+    // Interactive form controls — tracked as structural regions (the field's
+    // value is content, not asserted; presence/position is).
+    TextField: 'container',
+    Checkbox: 'container',
+    Dropdown: 'container',
+    RadioButton: 'container',
+    // Opaque graphic regions — presence/position diffed, internals ignored.
+    Image: 'image',
+    Svg: 'image',
+    Canvas: 'image',
+    QrCode: 'image',
+    Barcode: 'image',
+    Watermark: 'image',
+    BarChart: 'image',
+    LineChart: 'image',
+    PieChart: 'image',
+    AreaChart: 'image',
+    DotPlot: 'image',
+    // --- Synthetic aliases NOT in the @formepdf union, kept for consumers that
+    // hand-build a simplified LayoutInfo and for unit-test fixtures.
+    Heading: 'heading',
+    Table: 'table',
+    Row: 'row',
+    Cell: 'cell',
+};
+/**
+ * Union members intentionally given no semantic role (out of scope for v0.1),
+ * with a one-line reason each. Currently empty — every ElementNodeType maps to a
+ * deliberate role above. Kept as the explicit opt-out mechanism and referenced
+ * by the coverage test so a future "don't test this" decision is visible, never
+ * a silent default.
+ */
+export const FORME_INTENTIONALLY_UNMAPPED = new Set([
+// e.g. 'SomeFutureType', // reason it's out of scope
+]);
+function roleFor(nodeType) {
+    // Unknown (future) types fall back to `container`; the coverage test + the
+    // version-pin comment are what keep the known union fully dispositioned.
+    return FORME_ROLE_BY_NODE_TYPE[nodeType] ?? 'container';
+}
+/** Explicit heading level from an `H1`..`H6` tag, else null (fall back to size). */
+function explicitLevel(nodeType) {
+    const m = /^H([1-6])$/.exec(nodeType);
+    return m ? Number(m[1]) : null;
+}
+/** Text-leaf roles fold their subtree into one node (text comes from TextLines). */
+function isTextLeaf(role) {
+    return role === 'heading' || role === 'text' || role === 'cell';
+}
+/**
+ * Leaf roles don't recurse into children. Text leaves fold to collect text;
+ * `image` folds as an OPAQUE graphic (charts, barcodes, etc.) — its internal
+ * labels are deliberately not captured as structural text.
+ */
+function isLeaf(role) {
+    return isTextLeaf(role) || role === 'image';
+}
+/**
+ * Authoritative, heuristic-free extraction from FormePDF layout metadata. Roles,
+ * heading levels, and the content box are explicit, so every node is emitted at
+ * confidence 1.0. This is the reliable path the whole product is validated on
+ * before the fragile pdfjs path is touched.
+ */
+export function fromFormeLayout(layout, opts) {
+    const rawPages = [];
+    const rawNodes = [];
+    let counter = 0;
+    const nextId = () => `f${counter++}`;
+    // Font sizes of headings that lack an explicit H-level, for cluster fallback.
+    const headingSizes = [];
+    for (const page of layout.pages) {
+        walk(page.elements, (el) => {
+            if (roleFor(el.nodeType) === 'heading' && explicitLevel(el.nodeType) == null && el.style?.fontSize) {
+                headingSizes.push(el.style.fontSize);
+            }
+        });
+    }
+    const levelFor = buildHeadingLevelMap(headingSizes);
+    layout.pages.forEach((page, pageIndex) => {
+        const contentBox = {
+            x: page.contentX,
+            y: page.contentY,
+            width: page.contentWidth,
+            height: page.contentHeight,
+        };
+        rawPages.push({ index: pageIndex, width: page.width, height: page.height, contentBox });
+        const emit = (el, parentTempId, parentBBox) => {
+            const role = roleFor(el.nodeType);
+            const tempId = nextId();
+            const bbox = { x: el.x, y: el.y, width: el.width, height: el.height };
+            let overflow = null;
+            if (role === 'text' || role === 'heading') {
+                overflow = computeOverflow(bbox, contentBox, 'page-content');
+            }
+            else if (role === 'cell' && parentBBox) {
+                overflow = computeOverflow(bbox, parentBBox, 'parent');
+            }
+            const node = {
+                tempId,
+                parentTempId,
+                role,
+                pageIndex,
+                bbox,
+                text: isTextLeaf(role) ? collectText(el) : null,
+                font: fontFrom(el.style),
+                headingLevel: role === 'heading' ? explicitLevel(el.nodeType) ?? levelFor(el.style?.fontSize ?? 0) : null,
+                overflow,
+                confidence: 1,
+            };
+            if (role === 'table')
+                node.table = tableShape(el.children ?? []);
+            rawNodes.push(node);
+            // Leaf roles fold their subtree: text leaves already collected their text;
+            // image leaves are opaque graphics. Tables mark their children as
+            // already-in-a-table so rows aren't re-synthesized.
+            if (!isLeaf(role)) {
+                processChildren(el.children ?? [], tempId, bbox, role === 'table');
+            }
+        };
+        // Iterate a sibling list, synthesizing a table node around any run of
+        // contiguous TableRows that aren't already inside a Table.
+        const processChildren = (children, parentTempId, parentBBox, parentIsTable) => {
+            let i = 0;
+            while (i < children.length) {
+                const el = children[i];
+                if (roleFor(el.nodeType) === 'row' && !parentIsTable) {
+                    const run = [];
+                    let j = i;
+                    while (j < children.length && roleFor(children[j].nodeType) === 'row') {
+                        run.push(children[j]);
+                        j++;
+                    }
+                    const tableTempId = nextId();
+                    const bbox = unionBBox(run);
+                    rawNodes.push({
+                        tempId: tableTempId,
+                        parentTempId,
+                        role: 'table',
+                        pageIndex,
+                        bbox,
+                        text: null,
+                        font: null,
+                        headingLevel: null,
+                        overflow: null,
+                        table: tableShape(run),
+                        confidence: 1,
+                    });
+                    for (const row of run)
+                        emit(row, tableTempId, bbox);
+                    i = j;
+                    continue;
+                }
+                emit(el, parentTempId, parentBBox);
+                i++;
+            }
+        };
+        processChildren(page.elements, null, null, false);
+    });
+    const snapshot = finalize('formepdf', rawPages, rawNodes, opts);
+    if (opts?.assertShape !== false)
+        assertShapeRecognized(layout, snapshot);
+    return snapshot;
+}
+/** Thrown when FormePDF's layout shape no longer matches what this extractor
+ * understands — a broken contract on the authoritative fast path, which must
+ * fail loudly rather than silently produce a wrong diff. */
+export class FormeShapeError extends Error {
+    constructor(message) {
+        super(message);
+        this.name = 'FormeShapeError';
+    }
+}
+/**
+ * Guard against silent misclassification when FormePDF's runtime layout shape
+ * drifts (as it has before: `H1`..`H6`, `TextLine`, and the missing `Table`
+ * wrapper were all undocumented). Rather than allow-listing every nodeType
+ * (FormePDF emits many, and the set isn't fully documented — a strict list would
+ * false-throw on the next benign addition), assert the invariants this extractor
+ * depends on. Verified against @formepdf/core 0.12.0 runtime output (2026-08).
+ */
+function assertShapeRecognized(layout, snapshot) {
+    const fragments = [];
+    const nodeTypesSeen = new Set();
+    let hasRow = false;
+    let hasCell = false;
+    let hasHeading = false;
+    // Custom traversal (not `walk`) so we can STOP at image subtrees: charts and
+    // other graphics are deliberately opaque, so their internal label text is not
+    // expected to be captured and must not count as a dropped fragment.
+    const visit = (el) => {
+        nodeTypesSeen.add(el.nodeType);
+        const role = roleFor(el.nodeType);
+        if (role === 'row')
+            hasRow = true;
+        if (role === 'cell')
+            hasCell = true;
+        if (role === 'heading')
+            hasHeading = true;
+        if (role === 'image')
+            return; // opaque graphic — do not descend or expect its text
+        if (typeof el.textContent === 'string') {
+            const t = el.textContent.replace(/\s+/g, ' ').trim().toLowerCase();
+            if (t)
+                fragments.push(t);
+        }
+        for (const child of el.children ?? [])
+            visit(child);
+    };
+    for (const page of layout.pages)
+        for (const el of page.elements)
+            visit(el);
+    const context = `Seen FormePDF nodeTypes: [${[...nodeTypesSeen].sort().join(', ')}].`;
+    // 1. Every text fragment in the layout must be captured somewhere. Aggregate
+    //    (not per-node) so legitimately empty cells/blocks never false-trip.
+    if (fragments.length > 0) {
+        const haystack = snapshot.nodes.map((n) => n.normText ?? '').join('
');
+        const missing = fragments.filter((f) => !haystack.includes(f));
+        if (missing.length > 0) {
+            throw new FormeShapeError(`fromFormeLayout dropped ${missing.length}/${fragments.length} text fragment(s) from the ` +
+                `FormePDF layout (e.g. ${JSON.stringify(missing.slice(0, 3))}). The layout node shape has ` +
+                `likely changed; the nodeType→role mapping needs updating. ${context}`);
+        }
+    }
+    // 2. Text/table carriers present in the layout must produce their roles.
+    const produced = new Set(snapshot.nodes.map((n) => n.role));
+    if (hasRow && !produced.has('row')) {
+        throw new FormeShapeError(`Layout has TableRow nodes but none were extracted as rows. ${context}`);
+    }
+    if (hasCell && !produced.has('cell')) {
+        throw new FormeShapeError(`Layout has TableCell nodes but none were extracted as cells. ${context}`);
+    }
+    if (hasHeading && !produced.has('heading')) {
+        throw new FormeShapeError(`Layout has H1–H6/Heading nodes but none were extracted as headings. ${context}`);
+    }
+    // 3. Table synthesis sanity: a table with zero rows is malformed.
+    for (const node of snapshot.nodes) {
+        if (node.role === 'table' && node.table && node.table.rows === 0) {
+            throw new FormeShapeError(`Synthesized a table with zero rows — row grouping is broken. ${context}`);
+        }
+    }
+}
+/** Gather text from a node's subtree (TextLine carries it), collapsing space. */
+function collectText(el) {
+    const parts = [];
+    const rec = (n) => {
+        if (typeof n.textContent === 'string' && n.textContent.length)
+            parts.push(n.textContent);
+        for (const c of n.children ?? [])
+            rec(c);
+    };
+    rec(el);
+    const text = parts.join(' ').replace(/\s+/g, ' ').trim();
+    return text.length ? text : null;
+}
+function fontFrom(style) {
+    if (!style)
+        return null;
+    if (style.fontSize == null && style.fontWeight == null && style.fontFamily == null)
+        return null;
+    return {
+        size: style.fontSize ?? null,
+        weight: style.fontWeight ?? null,
+        family: style.fontFamily ?? null,
+        italic: style.fontStyle ? /italic|oblique/i.test(style.fontStyle) : undefined,
+    };
+}
+/** rows = number of row children; cols = max cells across those rows. */
+function tableShape(rows) {
+    const rowEls = rows.filter((c) => roleFor(c.nodeType) === 'row');
+    let cols = 0;
+    for (const row of rowEls) {
+        const cells = (row.children ?? []).filter((c) => roleFor(c.nodeType) === 'cell').length;
+        cols = Math.max(cols, cells);
+    }
+    return { rows: rowEls.length, cols };
+}
+function unionBBox(els) {
+    const minX = Math.min(...els.map((e) => e.x));
+    const maxX = Math.max(...els.map((e) => e.x + e.width));
+    const minY = Math.min(...els.map((e) => e.y));
+    const maxY = Math.max(...els.map((e) => e.y + e.height));
+    return { x: minX, y: minY, width: maxX - minX, height: maxY - minY };
+}
+function walk(elements, fn) {
+    for (const el of elements) {
+        fn(el);
+        if (el.children?.length)
+            walk(el.children, fn);
+    }
+}
+//# sourceMappingURL=fromFormeLayout.js.map

package/dist/extract/fromFormeLayout.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"fromFormeLayout.js","sourceRoot":"","sources":["../../src/extract/fromFormeLayout.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,eAAe,EAAE,MAAM,gBAAgB,CAAC;AACjD,OAAO,EAAE,oBAAoB,EAAE,MAAM,0BAA0B,CAAC;AAChE,OAAO,EAAE,QAAQ,EAAE,MAAM,eAAe,CAAC;AAgDzC;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAA6B;IAC/D,kBAAkB;IAClB,IAAI,EAAE,MAAM;IACZ,QAAQ,EAAE,MAAM;IAChB,GAAG,EAAE,MAAM,EAAE,8CAA8C;IAC3D,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,2EAA2E;IAC3E,QAAQ,EAAE,KAAK;IACf,SAAS,EAAE,MAAM;IACjB,kCAAkC;IAClC,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,QAAQ,EAAE,WAAW;IACrB,WAAW,EAAE,WAAW;IACxB,WAAW,EAAE,WAAW;IACxB,yEAAyE;IACzE,yDAAyD;IACzD,SAAS,EAAE,WAAW;IACtB,QAAQ,EAAE,WAAW;IACrB,QAAQ,EAAE,WAAW;IACrB,WAAW,EAAE,WAAW;IACxB,wEAAwE;IACxE,KAAK,EAAE,OAAO;IACd,GAAG,EAAE,OAAO;IACZ,MAAM,EAAE,OAAO;IACf,MAAM,EAAE,OAAO;IACf,OAAO,EAAE,OAAO;IAChB,SAAS,EAAE,OAAO;IAClB,QAAQ,EAAE,OAAO;IACjB,SAAS,EAAE,OAAO;IAClB,QAAQ,EAAE,OAAO;IACjB,SAAS,EAAE,OAAO;IAClB,OAAO,EAAE,OAAO;IAChB,4EAA4E;IAC5E,iEAAiE;IACjE,OAAO,EAAE,SAAS;IAClB,KAAK,EAAE,OAAO;IACd,GAAG,EAAE,KAAK;IACV,IAAI,EAAE,MAAM;CACb,CAAC;AAEF;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,4BAA4B,GAAwB,IAAI,GAAG,CAAS;AAC/E,qDAAqD;CACtD,CAAC,CAAC;AAEH,SAAS,OAAO,CAAC,QAAgB;IAC/B,2EAA2E;IAC3E,yEAAyE;IACzE,OAAO,uBAAuB,CAAC,QAAQ,CAAC,IAAI,WAAW,CAAC;AAC1D,CAAC;AAED,oFAAoF;AACpF,SAAS,aAAa,CAAC,QAAgB;IACrC,MAAM,CAAC,GAAG,YAAY,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACtC,OAAO,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AACjC,CAAC;AAED,oFAAoF;AACpF,SAAS,UAAU,CAAC,IAAc;IAChC,OAAO,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,MAAM,IAAI,IAAI,KAAK,MAAM,CAAC;AAClE,CAAC;AAED;;;;GAIG;AACH,SAAS,MAAM,CAAC,IAAc;IAC5B,OAAO,UAAU,CAAC,IAAI,CAAC,IAAI,IAAI,KAAK,OAAO,CAAC;AAC9C,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAC7B,MAAuB,EACvB,IAAgF;IAEhF,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,MAAM,MAAM,GAAG,GAAW,EAAE,CAAC,IAAI,OAAO,EAAE,EAAE,CAAC;IAE7C,8EAA8E;IAC9E,MAAM,YAAY,GAAa,EAAE,CAAC;IAClC,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;QAChC,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,EAAE;YACzB,IAAI,OAAO,CAAC,EAAE,CAAC,QAAQ,CAAC,KAAK,SAAS,IAAI,aAAa,CAAC,EAAE,CAAC,QAAQ,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC;gBACnG,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;YACvC,CAAC;QACH,CAAC,CAAC,CAAC;IACL,CAAC;IACD,MAAM,QAAQ,GAAG,oBAAoB,CAAC,YAAY,CAAC,CAAC;IAEpD,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,EAAE;QACvC,MAAM,UAAU,GAAS;YACvB,CAAC,EAAE,IAAI,CAAC,QAAQ;YAChB,CAAC,EAAE,IAAI,CAAC,QAAQ;YAChB,KAAK,EAAE,IAAI,CAAC,YAAY;YACxB,MAAM,EAAE,IAAI,CAAC,aAAa;SAC3B,CAAC;QACF,QAAQ,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,CAAC,CAAC;QAExF,MAAM,IAAI,GAAG,CAAC,EAAoB,EAAE,YAA2B,EAAE,UAAuB,EAAQ,EAAE;YAChG,MAAM,IAAI,GAAG,OAAO,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAC;YAClC,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC;YACxB,MAAM,IAAI,GAAS,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,EAAE,CAAC;YAE5E,IAAI,QAAQ,GAAG,IAAI,CAAC;YACpB,IAAI,IAAI,KAAK,MAAM,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;gBAC1C,QAAQ,GAAG,eAAe,CAAC,IAAI,EAAE,UAAU,EAAE,cAAc,CAAC,CAAC;YAC/D,CAAC;iBAAM,IAAI,IAAI,KAAK,MAAM,IAAI,UAAU,EAAE,CAAC;gBACzC,QAAQ,GAAG,eAAe,CAAC,IAAI,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;YACzD,CAAC;YAED,MAAM,IAAI,GAAY;gBACpB,MAAM;gBACN,YAAY;gBACZ,IAAI;gBACJ,SAAS;gBACT,IAAI;gBACJ,IAAI,EAAE,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI;gBAC/C,IAAI,EAAE,QAAQ,CAAC,EAAE,CAAC,KAAK,CAAC;gBACxB,YAAY,EACV,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,EAAE,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,EAAE,CAAC,KAAK,EAAE,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI;gBAC7F,QAAQ;gBACR,UAAU,EAAE,CAAC;aACd,CAAC;YACF,IAAI,IAAI,KAAK,OAAO;gBAAE,IAAI,CAAC,KAAK,GAAG,UAAU,CAAC,EAAE,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC;YACjE,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAEpB,2EAA2E;YAC3E,kEAAkE;YAClE,oDAAoD;YACpD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;gBAClB,eAAe,CAAC,EAAE,CAAC,QAAQ,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,KAAK,OAAO,CAAC,CAAC;YACrE,CAAC;QACH,CAAC,CAAC;QAEF,sEAAsE;QACtE,2DAA2D;QAC3D,MAAM,eAAe,GAAG,CACtB,QAA4B,EAC5B,YAA2B,EAC3B,UAAuB,EACvB,aAAsB,EAChB,EAAE;YACR,IAAI,CAAC,GAAG,CAAC,CAAC;YACV,OAAO,CAAC,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC;gBAC3B,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,CAAqB,CAAC;gBAC3C,IAAI,OAAO,CAAC,EAAE,CAAC,QAAQ,CAAC,KAAK,KAAK,IAAI,CAAC,aAAa,EAAE,CAAC;oBACrD,MAAM,GAAG,GAAuB,EAAE,CAAC;oBACnC,IAAI,CAAC,GAAG,CAAC,CAAC;oBACV,OAAO,CAAC,GAAG,QAAQ,CAAC,MAAM,IAAI,OAAO,CAAE,QAAQ,CAAC,CAAC,CAAsB,CAAC,QAAQ,CAAC,KAAK,KAAK,EAAE,CAAC;wBAC5F,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAqB,CAAC,CAAC;wBAC1C,CAAC,EAAE,CAAC;oBACN,CAAC;oBACD,MAAM,WAAW,GAAG,MAAM,EAAE,CAAC;oBAC7B,MAAM,IAAI,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;oBAC5B,QAAQ,CAAC,IAAI,CAAC;wBACZ,MAAM,EAAE,WAAW;wBACnB,YAAY;wBACZ,IAAI,EAAE,OAAO;wBACb,SAAS;wBACT,IAAI;wBACJ,IAAI,EAAE,IAAI;wBACV,IAAI,EAAE,IAAI;wBACV,YAAY,EAAE,IAAI;wBAClB,QAAQ,EAAE,IAAI;wBACd,KAAK,EAAE,UAAU,CAAC,GAAG,CAAC;wBACtB,UAAU,EAAE,CAAC;qBACd,CAAC,CAAC;oBACH,KAAK,MAAM,GAAG,IAAI,GAAG;wBAAE,IAAI,CAAC,GAAG,EAAE,WAAW,EAAE,IAAI,CAAC,CAAC;oBACpD,CAAC,GAAG,CAAC,CAAC;oBACN,SAAS;gBACX,CAAC;gBACD,IAAI,CAAC,EAAE,EAAE,YAAY,EAAE,UAAU,CAAC,CAAC;gBACnC,CAAC,EAAE,CAAC;YACN,CAAC;QACH,CAAC,CAAC;QAEF,eAAe,CAAC,IAAI,CAAC,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;IAEH,MAAM,QAAQ,GAAG,QAAQ,CAAC,UAAU,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,CAAC,CAAC;IAChE,IAAI,IAAI,EAAE,WAAW,KAAK,KAAK;QAAE,qBAAqB,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;IACzE,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;4DAE4D;AAC5D,MAAM,OAAO,eAAgB,SAAQ,KAAK;IACxC,YAAY,OAAe;QACzB,KAAK,CAAC,OAAO,CAAC,CAAC;QACf,IAAI,CAAC,IAAI,GAAG,iBAAiB,CAAC;IAChC,CAAC;CACF;AAED;;;;;;;GAOG;AACH,SAAS,qBAAqB,CAAC,MAAuB,EAAE,QAA4B;IAClF,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,MAAM,aAAa,GAAG,IAAI,GAAG,EAAU,CAAC;IACxC,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,IAAI,UAAU,GAAG,KAAK,CAAC;IACvB,6EAA6E;IAC7E,8EAA8E;IAC9E,oEAAoE;IACpE,MAAM,KAAK,GAAG,CAAC,EAAoB,EAAQ,EAAE;QAC3C,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAC;QAC/B,MAAM,IAAI,GAAG,OAAO,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAC;QAClC,IAAI,IAAI,KAAK,KAAK;YAAE,MAAM,GAAG,IAAI,CAAC;QAClC,IAAI,IAAI,KAAK,MAAM;YAAE,OAAO,GAAG,IAAI,CAAC;QACpC,IAAI,IAAI,KAAK,SAAS;YAAE,UAAU,GAAG,IAAI,CAAC;QAC1C,IAAI,IAAI,KAAK,OAAO;YAAE,OAAO,CAAC,qDAAqD;QACnF,IAAI,OAAO,EAAE,CAAC,WAAW,KAAK,QAAQ,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,EAAE,CAAC,WAAW,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;YACnE,IAAI,CAAC;gBAAE,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC3B,CAAC;QACD,KAAK,MAAM,KAAK,IAAI,EAAE,CAAC,QAAQ,IAAI,EAAE;YAAE,KAAK,CAAC,KAAK,CAAC,CAAC;IACtD,CAAC,CAAC;IACF,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK;QAAE,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,QAAQ;YAAE,KAAK,CAAC,EAAE,CAAC,CAAC;IAC3E,MAAM,OAAO,GAAG,6BAA6B,CAAC,GAAG,aAAa,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC;IAEtF,6EAA6E;IAC7E,yEAAyE;IACzE,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,MAAM,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACvE,MAAM,OAAO,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;QAC/D,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,MAAM,IAAI,eAAe,CACvB,2BAA2B,OAAO,CAAC,MAAM,IAAI,SAAS,CAAC,MAAM,6BAA6B;gBACxF,yBAAyB,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,+BAA+B;gBAC3F,6DAA6D,OAAO,EAAE,CACzE,CAAC;QACJ,CAAC;IACH,CAAC;IAED,yEAAyE;IACzE,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IAC5D,IAAI,MAAM,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;QACnC,MAAM,IAAI,eAAe,CAAC,8DAA8D,OAAO,EAAE,CAAC,CAAC;IACrG,CAAC;IACD,IAAI,OAAO,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;QACrC,MAAM,IAAI,eAAe,CAAC,gEAAgE,OAAO,EAAE,CAAC,CAAC;IACvG,CAAC;IACD,IAAI,UAAU,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,eAAe,CAAC,uEAAuE,OAAO,EAAE,CAAC,CAAC;IAC9G,CAAC;IAED,kEAAkE;IAClE,KAAK,MAAM,IAAI,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,IAAI,KAAK,OAAO,IAAI,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YACjE,MAAM,IAAI,eAAe,CAAC,gEAAgE,OAAO,EAAE,CAAC,CAAC;QACvG,CAAC;IACH,CAAC;AACH,CAAC;AAED,iFAAiF;AACjF,SAAS,WAAW,CAAC,EAAoB;IACvC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,GAAG,GAAG,CAAC,CAAmB,EAAQ,EAAE;QACxC,IAAI,OAAO,CAAC,CAAC,WAAW,KAAK,QAAQ,IAAI,CAAC,CAAC,WAAW,CAAC,MAAM;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC;QACzF,KAAK,MAAM,CAAC,IAAI,CAAC,CAAC,QAAQ,IAAI,EAAE;YAAE,GAAG,CAAC,CAAC,CAAC,CAAC;IAC3C,CAAC,CAAC;IACF,GAAG,CAAC,EAAE,CAAC,CAAC;IACR,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;IACzD,OAAO,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;AACnC,CAAC;AAED,SAAS,QAAQ,CAAC,KAAgC;IAChD,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAC;IACxB,IAAI,KAAK,CAAC,QAAQ,IAAI,IAAI,IAAI,KAAK,CAAC,UAAU,IAAI,IAAI,IAAI,KAAK,CAAC,UAAU,IAAI,IAAI;QAAE,OAAO,IAAI,CAAC;IAChG,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,QAAQ,IAAI,IAAI;QAC5B,MAAM,EAAE,KAAK,CAAC,UAAU,IAAI,IAAI;QAChC,MAAM,EAAE,KAAK,CAAC,UAAU,IAAI,IAAI;QAChC,MAAM,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,iBAAiB,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS;KAC9E,CAAC;AACJ,CAAC;AAED,yEAAyE;AACzE,SAAS,UAAU,CAAC,IAAwB;IAC1C,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,KAAK,CAAC,CAAC;IACjE,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QACzB,MAAM,KAAK,GAAG,CAAC,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,MAAM,CAAC,CAAC,MAAM,CAAC;QACxF,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;IAC/B,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC;AACvC,CAAC;AAED,SAAS,SAAS,CAAC,GAAuB;IACxC,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9C,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC;IACxD,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9C,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC;IACzD,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,GAAG,IAAI,EAAE,MAAM,EAAE,IAAI,GAAG,IAAI,EAAE,CAAC;AACvE,CAAC;AAED,SAAS,IAAI,CAAC,QAA4B,EAAE,EAAkC;IAC5E,KAAK,MAAM,EAAE,IAAI,QAAQ,EAAE,CAAC;QAC1B,EAAE,CAAC,EAAE,CAAC,CAAC;QACP,IAAI,EAAE,CAAC,QAAQ,EAAE,MAAM;YAAE,IAAI,CAAC,EAAE,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;IACjD,CAAC;AACH,CAAC"}

package/dist/extract/fromPdf.d.ts

@@ -0,0 +1,15 @@
+import type { StructuralSnapshot } from '../types.js';
+/**
+ * General-producer extraction via pdfjs. Reliable for page-count, added/removed,
+ * moved-to-page, and heading hierarchy; best-effort (lower confidence) for table
+ * regions and overflow, since PDF carries neither a table nor a container concept.
+ * Everything converges through `finalize`, so the diff engine / matcher / CLI are
+ * identical to the FormePDF path.
+ */
+export declare function fromPdf(bytes: Uint8Array, opts?: {
+    source?: {
+        name?: string;
+    };
+    createdAt?: string;
+}): Promise<StructuralSnapshot>;
+//# sourceMappingURL=fromPdf.d.ts.map

package/dist/extract/fromPdf.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"fromPdf.d.ts","sourceRoot":"","sources":["../../src/extract/fromPdf.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAY,kBAAkB,EAAE,MAAM,aAAa,CAAC;AAShE;;;;;;GAMG;AACH,wBAAsB,OAAO,CAC3B,KAAK,EAAE,UAAU,EACjB,IAAI,CAAC,EAAE;IAAE,MAAM,CAAC,EAAE;QAAE,IAAI,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,CAAA;CAAE,GACxD,OAAO,CAAC,kBAAkB,CAAC,CAoG7B"}

package/dist/extract/fromPdf.js

@@ -0,0 +1,120 @@
+import { finalize } from './finalize.js';
+import { loadDocument } from './pdfjs/loadDocument.js';
+import { extractTextRuns } from './pdfjs/textRuns.js';
+import { buildHeadingModel } from './pdfjs/headings.js';
+import { detectTables } from './pdfjs/tables.js';
+import { buildOverflowModel, inferContentBox } from './pdfjs/overflow.js';
+/**
+ * General-producer extraction via pdfjs. Reliable for page-count, added/removed,
+ * moved-to-page, and heading hierarchy; best-effort (lower confidence) for table
+ * regions and overflow, since PDF carries neither a table nor a container concept.
+ * Everything converges through `finalize`, so the diff engine / matcher / CLI are
+ * identical to the FormePDF path.
+ */
+export async function fromPdf(bytes, opts) {
+    const doc = await loadDocument(bytes);
+    const perPage = [];
+    for (let i = 1; i <= doc.numPages; i++) {
+        const page = await doc.getPage(i);
+        perPage.push(await extractTextRuns(page));
+    }
+    // Heading model is doc-wide so levels are consistent across pages.
+    const headingModel = buildHeadingModel(perPage.flatMap((p) => p.runs));
+    const rawPages = [];
+    const rawNodes = [];
+    let counter = 0;
+    const nextId = () => `p${counter++}`;
+    perPage.forEach((pg, pageIndex) => {
+        rawPages.push({
+            index: pageIndex,
+            width: pg.width,
+            height: pg.height,
+            contentBox: inferContentBox(pg.runs, pg.width, pg.height),
+        });
+        const overflowOf = buildOverflowModel(pg.runs, pg.width);
+        const tables = detectTables(pg.runs);
+        const consumed = new Set();
+        for (const table of tables) {
+            const tableTempId = nextId();
+            rawNodes.push({
+                tempId: tableTempId,
+                parentTempId: null,
+                role: 'table',
+                pageIndex,
+                bbox: table.bbox,
+                text: null,
+                font: null,
+                headingLevel: null,
+                overflow: null,
+                table: { rows: table.rows.length, cols: table.cols },
+                confidence: 0.5,
+            });
+            for (const row of table.rows) {
+                const rowTempId = nextId();
+                rawNodes.push({
+                    tempId: rowTempId,
+                    parentTempId: tableTempId,
+                    role: 'row',
+                    pageIndex,
+                    bbox: row.bbox,
+                    text: null,
+                    font: null,
+                    headingLevel: null,
+                    overflow: null,
+                    confidence: 0.5,
+                });
+                for (const cell of row.cells) {
+                    for (const r of cell.runs)
+                        consumed.add(r);
+                    rawNodes.push({
+                        tempId: nextId(),
+                        parentTempId: rowTempId,
+                        role: 'cell',
+                        pageIndex,
+                        bbox: cell.bbox,
+                        text: cell.text,
+                        font: fontOf(cell.runs),
+                        headingLevel: null,
+                        overflow: null,
+                        confidence: 0.5,
+                    });
+                }
+            }
+        }
+        for (const r of pg.runs) {
+            if (consumed.has(r))
+                continue;
+            const heading = headingModel.isHeading(r);
+            rawNodes.push({
+                tempId: nextId(),
+                parentTempId: null,
+                role: heading ? 'heading' : 'text',
+                pageIndex,
+                bbox: { x: r.x, y: r.y, width: r.width, height: r.height },
+                text: r.text,
+                font: {
+                    size: r.fontSize,
+                    weight: r.bold ? 700 : 400,
+                    family: r.fontName || null,
+                    italic: r.italic,
+                },
+                headingLevel: heading ? headingModel.levelOf(r) : null,
+                overflow: overflowOf(r),
+                confidence: heading ? headingModel.confidenceOf(r) : 0.9,
+            });
+        }
+    });
+    return finalize('pdfjs', rawPages, rawNodes, opts);
+}
+function fontOf(runs) {
+    if (runs.length === 0)
+        return null;
+    const dominant = [...runs].sort((a, b) => b.charCount - a.charCount)[0];
+    return {
+        size: dominant.fontSize,
+        weight: dominant.bold ? 700 : 400,
+        family: dominant.fontName || null,
+        italic: dominant.italic,
+    };
+}
+//# sourceMappingURL=fromPdf.js.map

package/dist/extract/fromPdf.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"fromPdf.js","sourceRoot":"","sources":["../../src/extract/fromPdf.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,QAAQ,EAAE,MAAM,eAAe,CAAC;AAEzC,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AACvD,OAAO,EAAE,eAAe,EAAmB,MAAM,qBAAqB,CAAC;AACvE,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACxD,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACjD,OAAO,EAAE,kBAAkB,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAE1E;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,OAAO,CAC3B,KAAiB,EACjB,IAAyD;IAEzD,MAAM,GAAG,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,CAAC;IAEtC,MAAM,OAAO,GAA4D,EAAE,CAAC;IAC5E,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;QACvC,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QAClC,OAAO,CAAC,IAAI,CAAC,MAAM,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC;IAC5C,CAAC;IAED,mEAAmE;IACnE,MAAM,YAAY,GAAG,iBAAiB,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IAEvE,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,MAAM,MAAM,GAAG,GAAW,EAAE,CAAC,IAAI,OAAO,EAAE,EAAE,CAAC;IAE7C,OAAO,CAAC,OAAO,CAAC,CAAC,EAAE,EAAE,SAAS,EAAE,EAAE;QAChC,QAAQ,CAAC,IAAI,CAAC;YACZ,KAAK,EAAE,SAAS;YAChB,KAAK,EAAE,EAAE,CAAC,KAAK;YACf,MAAM,EAAE,EAAE,CAAC,MAAM;YACjB,UAAU,EAAE,eAAe,CAAC,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC;SAC1D,CAAC,CAAC;QAEH,MAAM,UAAU,GAAG,kBAAkB,CAAC,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC;QACzD,MAAM,MAAM,GAAG,YAAY,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC;QACrC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAc,CAAC;QAEvC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,WAAW,GAAG,MAAM,EAAE,CAAC;YAC7B,QAAQ,CAAC,IAAI,CAAC;gBACZ,MAAM,EAAE,WAAW;gBACnB,YAAY,EAAE,IAAI;gBAClB,IAAI,EAAE,OAAO;gBACb,SAAS;gBACT,IAAI,EAAE,KAAK,CAAC,IAAI;gBAChB,IAAI,EAAE,IAAI;gBACV,IAAI,EAAE,IAAI;gBACV,YAAY,EAAE,IAAI;gBAClB,QAAQ,EAAE,IAAI;gBACd,KAAK,EAAE,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE;gBACpD,UAAU,EAAE,GAAG;aAChB,CAAC,CAAC;YACH,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;gBAC7B,MAAM,SAAS,GAAG,MAAM,EAAE,CAAC;gBAC3B,QAAQ,CAAC,IAAI,CAAC;oBACZ,MAAM,EAAE,SAAS;oBACjB,YAAY,EAAE,WAAW;oBACzB,IAAI,EAAE,KAAK;oBACX,SAAS;oBACT,IAAI,EAAE,GAAG,CAAC,IAAI;oBACd,IAAI,EAAE,IAAI;oBACV,IAAI,EAAE,IAAI;oBACV,YAAY,EAAE,IAAI;oBAClB,QAAQ,EAAE,IAAI;oBACd,UAAU,EAAE,GAAG;iBAChB,CAAC,CAAC;gBACH,KAAK,MAAM,IAAI,IAAI,GAAG,CAAC,KAAK,EAAE,CAAC;oBAC7B,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,IAAI;wBAAE,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;oBAC3C,QAAQ,CAAC,IAAI,CAAC;wBACZ,MAAM,EAAE,MAAM,EAAE;wBAChB,YAAY,EAAE,SAAS;wBACvB,IAAI,EAAE,MAAM;wBACZ,SAAS;wBACT,IAAI,EAAE,IAAI,CAAC,IAAI;wBACf,IAAI,EAAE,IAAI,CAAC,IAAI;wBACf,IAAI,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC;wBACvB,YAAY,EAAE,IAAI;wBAClB,QAAQ,EAAE,IAAI;wBACd,UAAU,EAAE,GAAG;qBAChB,CAAC,CAAC;gBACL,CAAC;YACH,CAAC;QACH,CAAC;QAED,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC;YACxB,IAAI,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC;gBAAE,SAAS;YAC9B,MAAM,OAAO,GAAG,YAAY,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;YAC1C,QAAQ,CAAC,IAAI,CAAC;gBACZ,MAAM,EAAE,MAAM,EAAE;gBAChB,YAAY,EAAE,IAAI;gBAClB,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,MAAM;gBAClC,SAAS;gBACT,IAAI,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;gBAC1D,IAAI,EAAE,CAAC,CAAC,IAAI;gBACZ,IAAI,EAAE;oBACJ,IAAI,EAAE,CAAC,CAAC,QAAQ;oBAChB,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG;oBAC1B,MAAM,EAAE,CAAC,CAAC,QAAQ,IAAI,IAAI;oBAC1B,MAAM,EAAE,CAAC,CAAC,MAAM;iBACjB;gBACD,YAAY,EAAE,OAAO,CAAC,CAAC,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI;gBACtD,QAAQ,EAAE,UAAU,CAAC,CAAC,CAAC;gBACvB,UAAU,EAAE,OAAO,CAAC,CAAC,CAAC,YAAY,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG;aACzD,CAAC,CAAC;QACL,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,QAAQ,CAAC,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,CAAC,CAAC;AACrD,CAAC;AAED,SAAS,MAAM,CAAC,IAAkB;IAChC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IACnC,MAAM,QAAQ,GAAG,CAAC,GAAG,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAe,CAAC;IACtF,OAAO;QACL,IAAI,EAAE,QAAQ,CAAC,QAAQ;QACvB,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG;QACjC,MAAM,EAAE,QAAQ,CAAC,QAAQ,IAAI,IAAI;QACjC,MAAM,EAAE,QAAQ,CAAC,MAAM;KACxB,CAAC;AACJ,CAAC"}

package/dist/extract/pdfjs/headings.d.ts

@@ -0,0 +1,16 @@
+import type { PdfTextRun } from './textRuns.js';
+export interface HeadingModel {
+    bodySize: number;
+    isHeading(run: PdfTextRun): boolean;
+    levelOf(run: PdfTextRun): number;
+    confidenceOf(run: PdfTextRun): number;
+}
+/**
+ * Infer heading levels for raw pdfjs runs by font-size clustering. The
+ * char-count-weighted modal size is the body baseline; runs meaningfully larger
+ * (or bold and slightly larger) are headings, ranked into H1..H6. Confidence is
+ * 0.8 when a run is size-separated, 0.5 when only weight-distinguished — lower
+ * than the FormePDF path, which knows heading roles outright.
+ */
+export declare function buildHeadingModel(runs: PdfTextRun[]): HeadingModel;
+//# sourceMappingURL=headings.d.ts.map

package/dist/extract/pdfjs/headings.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"headings.d.ts","sourceRoot":"","sources":["../../../src/extract/pdfjs/headings.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,eAAe,CAAC;AAEhD,MAAM,WAAW,YAAY;IAC3B,QAAQ,EAAE,MAAM,CAAC;IACjB,SAAS,CAAC,GAAG,EAAE,UAAU,GAAG,OAAO,CAAC;IACpC,OAAO,CAAC,GAAG,EAAE,UAAU,GAAG,MAAM,CAAC;IACjC,YAAY,CAAC,GAAG,EAAE,UAAU,GAAG,MAAM,CAAC;CACvC;AAKD;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,UAAU,EAAE,GAAG,YAAY,CA6BlE"}

package/dist/extract/pdfjs/headings.js

@@ -0,0 +1,39 @@
+import { round1 } from '../../geometry.js';
+import { buildHeadingLevelMap } from '../../text/headingLevels.js';
+const HEADING_RATIO = 1.15;
+const BOLD_RATIO = 1.05;
+/**
+ * Infer heading levels for raw pdfjs runs by font-size clustering. The
+ * char-count-weighted modal size is the body baseline; runs meaningfully larger
+ * (or bold and slightly larger) are headings, ranked into H1..H6. Confidence is
+ * 0.8 when a run is size-separated, 0.5 when only weight-distinguished — lower
+ * than the FormePDF path, which knows heading roles outright.
+ */
+export function buildHeadingModel(runs) {
+    const weightBySize = new Map();
+    for (const r of runs) {
+        const s = round1(r.fontSize);
+        weightBySize.set(s, (weightBySize.get(s) ?? 0) + r.charCount);
+    }
+    let bodySize = 0;
+    let bestWeight = -1;
+    for (const [size, w] of weightBySize) {
+        if (w > bestWeight) {
+            bestWeight = w;
+            bodySize = size;
+        }
+    }
+    const candidateSizes = [
+        ...new Set(runs.filter((r) => r.fontSize > bodySize * HEADING_RATIO).map((r) => round1(r.fontSize))),
+    ];
+    const levelFor = buildHeadingLevelMap(candidateSizes);
+    const sizeSeparated = (r) => r.fontSize > bodySize * HEADING_RATIO;
+    const boldSeparated = (r) => r.bold && r.fontSize >= bodySize * BOLD_RATIO;
+    return {
+        bodySize,
+        isHeading: (r) => sizeSeparated(r) || boldSeparated(r),
+        levelOf: (r) => (sizeSeparated(r) ? levelFor(round1(r.fontSize)) : 6),
+        confidenceOf: (r) => (sizeSeparated(r) ? 0.8 : 0.5),
+    };
+}
+//# sourceMappingURL=headings.js.map

package/dist/extract/pdfjs/headings.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"headings.js","sourceRoot":"","sources":["../../../src/extract/pdfjs/headings.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;AAC3C,OAAO,EAAE,oBAAoB,EAAE,MAAM,6BAA6B,CAAC;AAUnE,MAAM,aAAa,GAAG,IAAI,CAAC;AAC3B,MAAM,UAAU,GAAG,IAAI,CAAC;AAExB;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,IAAkB;IAClD,MAAM,YAAY,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC/C,KAAK,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;QACrB,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;QAC7B,YAAY,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,SAAS,CAAC,CAAC;IAChE,CAAC;IACD,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,UAAU,GAAG,CAAC,CAAC,CAAC;IACpB,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,YAAY,EAAE,CAAC;QACrC,IAAI,CAAC,GAAG,UAAU,EAAE,CAAC;YACnB,UAAU,GAAG,CAAC,CAAC;YACf,QAAQ,GAAG,IAAI,CAAC;QAClB,CAAC;IACH,CAAC;IAED,MAAM,cAAc,GAAG;QACrB,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,GAAG,QAAQ,GAAG,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC;KACrG,CAAC;IACF,MAAM,QAAQ,GAAG,oBAAoB,CAAC,cAAc,CAAC,CAAC;IAEtD,MAAM,aAAa,GAAG,CAAC,CAAa,EAAW,EAAE,CAAC,CAAC,CAAC,QAAQ,GAAG,QAAQ,GAAG,aAAa,CAAC;IACxF,MAAM,aAAa,GAAG,CAAC,CAAa,EAAW,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,QAAQ,IAAI,QAAQ,GAAG,UAAU,CAAC;IAEhG,OAAO;QACL,QAAQ;QACR,SAAS,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,aAAa,CAAC,CAAC,CAAC;QACtD,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACrE,YAAY,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC;KACpD,CAAC;AACJ,CAAC"}

package/dist/extract/pdfjs/loadDocument.d.ts

@@ -0,0 +1,32 @@
+/**
+ * The single most fragile integration point in the codebase: loading pdfjs in a
+ * Node context without a worker or DOM. Kept deliberately small and isolated so
+ * the rest of the extractor never touches pdfjs internals. `pdfjs-dist` is an
+ * optional peer dep, loaded lazily via a non-literal specifier so consumers that
+ * only diff FormePDF `LayoutInfo` never need it installed.
+ */
+export interface LoadedPdfPage {
+    getViewport(opts: {
+        scale: number;
+    }): {
+        width: number;
+        height: number;
+    };
+    getTextContent(): Promise<{
+        items: PdfTextItem[];
+    }>;
+}
+export interface PdfTextItem {
+    str: string;
+    transform: number[];
+    width: number;
+    height: number;
+    fontName?: string;
+    hasEOL?: boolean;
+}
+export interface LoadedPdf {
+    numPages: number;
+    getPage(pageNumber: number): Promise<LoadedPdfPage>;
+}
+export declare function loadDocument(data: Uint8Array): Promise<LoadedPdf>;
+//# sourceMappingURL=loadDocument.d.ts.map

package/dist/extract/pdfjs/loadDocument.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"loadDocument.d.ts","sourceRoot":"","sources":["../../../src/extract/pdfjs/loadDocument.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AA2BH,MAAM,WAAW,aAAa;IAC5B,WAAW,CAAC,IAAI,EAAE;QAAE,KAAK,EAAE,MAAM,CAAA;KAAE,GAAG;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;IACxE,cAAc,IAAI,OAAO,CAAC;QAAE,KAAK,EAAE,WAAW,EAAE,CAAA;KAAE,CAAC,CAAC;CACrD;AAED,MAAM,WAAW,WAAW;IAC1B,GAAG,EAAE,MAAM,CAAC;IACZ,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,KAAK,EAAE,MAAM,CAAC;IACd,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,MAAM,WAAW,SAAS;IACxB,QAAQ,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,CAAC;CACrD;AAED,wBAAsB,YAAY,CAAC,IAAI,EAAE,UAAU,GAAG,OAAO,CAAC,SAAS,CAAC,CAevE"}

package/dist/extract/pdfjs/loadDocument.js

@@ -0,0 +1,45 @@
+/**
+ * The single most fragile integration point in the codebase: loading pdfjs in a
+ * Node context without a worker or DOM. Kept deliberately small and isolated so
+ * the rest of the extractor never touches pdfjs internals. `pdfjs-dist` is an
+ * optional peer dep, loaded lazily via a non-literal specifier so consumers that
+ * only diff FormePDF `LayoutInfo` never need it installed.
+ */
+let cached;
+async function getPdfjs() {
+    if (cached)
+        return cached;
+    const candidates = ['pdfjs-dist/legacy/build/pdf.mjs', 'pdfjs-dist'];
+    let lastErr;
+    for (const spec of candidates) {
+        try {
+            // Non-literal specifier: keeps TS from statically resolving an optional dep.
+            const mod = (await import(/* @vite-ignore */ spec));
+            cached = mod.getDocument ? mod : mod.default;
+            if (cached?.getDocument)
+                return cached;
+        }
+        catch (err) {
+            lastErr = err;
+        }
+    }
+    throw new Error('pdf-testkit: could not load "pdfjs-dist". Install it to diff non-FormePDF PDFs ' +
+        `(npm i pdfjs-dist). Original error: ${String(lastErr)}`);
+}
+export async function loadDocument(data) {
+    const pdfjs = await getPdfjs();
+    // In Node, the legacy build runs on the main thread via a built-in fake
+    // worker when workerSrc is left unset — so we deliberately don't touch
+    // GlobalWorkerOptions here (setting it to undefined fails pdfjs's validation).
+    const task = pdfjs.getDocument({
+        // pdfjs transfers (detaches) the `data` buffer to its worker; copy so the
+        // caller's Uint8Array stays usable and re-extraction never crashes.
+        data: data.slice(),
+        isEvalSupported: false,
+        disableFontFace: true,
+        useSystemFonts: false,
+        verbosity: 0, // errors only
+    });
+    return (await task.promise);
+}
+//# sourceMappingURL=loadDocument.js.map

package/dist/extract/pdfjs/loadDocument.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"loadDocument.js","sourceRoot":"","sources":["../../../src/extract/pdfjs/loadDocument.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAKH,IAAI,MAA4B,CAAC;AAEjC,KAAK,UAAU,QAAQ;IACrB,IAAI,MAAM;QAAE,OAAO,MAAM,CAAC;IAC1B,MAAM,UAAU,GAAG,CAAC,iCAAiC,EAAE,YAAY,CAAC,CAAC;IACrE,IAAI,OAAgB,CAAC;IACrB,KAAK,MAAM,IAAI,IAAI,UAAU,EAAE,CAAC;QAC9B,IAAI,CAAC;YACH,6EAA6E;YAC7E,MAAM,GAAG,GAAG,CAAC,MAAM,MAAM,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAa,CAAC;YAChE,MAAM,GAAG,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC;YAC7C,IAAI,MAAM,EAAE,WAAW;gBAAE,OAAO,MAAM,CAAC;QACzC,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,OAAO,GAAG,GAAG,CAAC;QAChB,CAAC;IACH,CAAC;IACD,MAAM,IAAI,KAAK,CACb,iFAAiF;QAC/E,uCAAuC,MAAM,CAAC,OAAO,CAAC,EAAE,CAC3D,CAAC;AACJ,CAAC;AAqBD,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAgB;IACjD,MAAM,KAAK,GAAG,MAAM,QAAQ,EAAE,CAAC;IAC/B,wEAAwE;IACxE,uEAAuE;IACvE,+EAA+E;IAC/E,MAAM,IAAI,GAAG,KAAK,CAAC,WAAW,CAAC;QAC7B,0EAA0E;QAC1E,oEAAoE;QACpE,IAAI,EAAE,IAAI,CAAC,KAAK,EAAE;QAClB,eAAe,EAAE,KAAK;QACtB,eAAe,EAAE,IAAI;QACrB,cAAc,EAAE,KAAK;QACrB,SAAS,EAAE,CAAC,EAAE,cAAc;KAC7B,CAAC,CAAC;IACH,OAAO,CAAC,MAAM,IAAI,CAAC,OAAO,CAAc,CAAC;AAC3C,CAAC"}