@bevel-software/platform-core-backend 0.11.2 → 0.12.0

package/dist/modules/workspace/file-readers/extract-pdf.js

@@ -0,0 +1,176 @@
+import { MAX_DOC_PART_BYTES } from './ooxml-text.js';
+/**
+ * Extract a PDF's TEXT LAYER page by page with pdf.js (`pdfjs-dist`,
+ * Mozilla's maintained renderer — chosen over the unmaintained thin wrappers
+ * around it). The LEGACY build is the one supported under Node; it is loaded
+ * lazily (and once) because it is a heavyweight module most deployments only
+ * need after the first PDF read.
+ *
+ * Layout heuristic: text items on one line are joined with single spaces; a
+ * new line starts when pdf.js flags an EOL or the item's Y position jumps.
+ * A PDF with NO text layer (a scan) extracts to just the `[page N]` markers,
+ * and the summary says "no text layer (scanned document?)" — no OCR in v1.
+ */
+/**
+ * How much DECODED text one PDF may yield before extraction gives up.
+ *
+ * The raw-size cap below bounds what arrives; it does not bound what comes
+ * out. PDF text lives in compressed streams, so a file comfortably under
+ * 50 MB can decode to far more than that, and every character of it is held
+ * in `lines` until the extraction returns. This bound is the decoded
+ * counterpart, checked as the text accumulates rather than after.
+ */
+const MAX_PDF_TEXT_CHARS = 20 * 1024 * 1024; // 20M chars of extracted text
+/**
+ * How many pages one PDF may have before extraction gives up.
+ *
+ * The decoded-text bound does not cover a document whose cost is its PAGE
+ * COUNT rather than its prose: every page costs a `getPage`, a
+ * `getTextContent` and a retained `[page N]` marker even when it holds no
+ * text at all, so a file declaring hundreds of thousands of empty pages spends
+ * minutes and megabytes without ever tripping a character budget. Real
+ * documents do not come close — a 2,000-page manual is an outlier.
+ */
+const MAX_PDF_PAGES = 10_000;
+/**
+ * How many text items one PAGE may hold. Items arrive through
+ * `streamTextContent` in small chunks (~100 items each), so this bound — like
+ * the character budget — fires while the page is still streaming, not after
+ * it has materialized. It exists because item COUNT is its own cost: each
+ * item is a retained heap object, and a page of empty-string items would
+ * never trip the character budget.
+ */
+const MAX_PDF_ITEMS_PER_PAGE = 200_000;
+/** The typed failure both decoded-text bounds return. */
+function overBudget() {
+    return {
+        ok: false,
+        message: `could not be extracted as a PDF (its text decodes to over ${MAX_PDF_TEXT_CHARS} characters — over the extraction limit)`,
+    };
+}
+export async function extractPdf(bytes) {
+    // The same bounded-read guard the zip-based extractors apply per part: a
+    // PDF has no compressed container to pre-scan, so the bound is simply the
+    // file's raw size, checked before pdf.js parses anything.
+    if (bytes.length > MAX_DOC_PART_BYTES) {
+        return {
+            ok: false,
+            message: `could not be extracted as a PDF (the file is ${bytes.length} bytes — over the ${MAX_DOC_PART_BYTES}-byte (50 MB) extraction limit)`,
+        };
+    }
+    let doc;
+    try {
+        doc = await openPdf(bytes);
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a PDF (${err.message})` };
+    }
+    try {
+        if (doc.numPages > MAX_PDF_PAGES) {
+            return {
+                ok: false,
+                message: `could not be extracted as a PDF (it declares ${doc.numPages} pages — over the ${MAX_PDF_PAGES}-page extraction limit)`,
+            };
+        }
+        const lines = [];
+        let textChars = 0;
+        let anyText = false;
+        for (let n = 1; n <= doc.numPages; n++) {
+            lines.push(`[page ${n}]`);
+            // The marker is retained text like any other line: a document whose cost
+            // is its page count must reach the same bound as one whose cost is prose.
+            textChars += n.toString().length + 8;
+            if (textChars > MAX_PDF_TEXT_CHARS)
+                return overBudget();
+            const page = await doc.getPage(n);
+            // `streamTextContent` delivers the page's items in small chunks (~100
+            // items each, `getTextContent` is just this stream materialized), so
+            // both budgets fire WHILE the page streams: a crafted single page can
+            // no longer build its whole item array before a bound trips. Once one
+            // does, the reader is cancelled and pdf.js stops producing.
+            const reader = page.streamTextContent().getReader();
+            let pageItems = 0;
+            let line = '';
+            let lastY;
+            const flush = () => {
+                if (line.trim() !== '') {
+                    lines.push(line);
+                    anyText = true;
+                    // The '\n' the final join emits for this line is retained text too.
+                    textChars += 1;
+                }
+                line = '';
+            };
+            for (;;) {
+                const { done, value: chunk } = await reader.read();
+                if (done)
+                    break;
+                pageItems += chunk.items.length;
+                if (pageItems > MAX_PDF_ITEMS_PER_PAGE) {
+                    await reader.cancel().catch(() => undefined);
+                    page.cleanup();
+                    return {
+                        ok: false,
+                        message: `could not be extracted as a PDF (page ${n} holds more than ${MAX_PDF_ITEMS_PER_PAGE} text items — over the extraction limit)`,
+                    };
+                }
+                for (const item of chunk.items) {
+                    if (!('str' in item))
+                        continue; // marked-content item — no text
+                    const y = item.transform?.[5];
+                    // Y-position jump = new visual line (1pt tolerance for kerning wobble).
+                    if (typeof y === 'number') {
+                        if (lastY !== undefined && Math.abs(y - lastY) > 1)
+                            flush();
+                        lastY = y;
+                    }
+                    if (item.str !== '') {
+                        // COUNTED before it is kept — the join space included: the bound
+                        // exists to stop the decoded text from accumulating, so it must
+                        // fire mid-page and cover every character the result will hold.
+                        textChars += item.str.length + (line === '' ? 0 : 1);
+                        if (textChars > MAX_PDF_TEXT_CHARS) {
+                            await reader.cancel().catch(() => undefined);
+                            return overBudget();
+                        }
+                        line += (line === '' ? '' : ' ') + item.str;
+                    }
+                    if (item.hasEOL)
+                        flush();
+                }
+            }
+            flush();
+            page.cleanup();
+        }
+        const pages = `${doc.numPages} page${doc.numPages === 1 ? '' : 's'}`;
+        return anyText
+            ? { ok: true, summary: `${pages}; layout, images and formatting omitted`, text: lines.join('\n') }
+            : { ok: true, summary: `${pages}; no text layer (scanned document?)`, text: lines.join('\n') };
+    }
+    catch (err) {
+        return { ok: false, message: `could not extract the PDF's text (${err.message})` };
+    }
+    finally {
+        await doc.destroy();
+    }
+}
+let pdfjsPromise;
+async function openPdf(bytes) {
+    pdfjsPromise ??= import('pdfjs-dist/legacy/build/pdf.mjs').catch((err) => {
+        // A FAILED load must not be memoized: left in place, the rejected promise
+        // would answer every later read and disable PDF extraction for the whole
+        // process. Reset so the next read retries the import.
+        pdfjsPromise = undefined;
+        throw err;
+    });
+    const { getDocument } = await pdfjsPromise;
+    return getDocument({
+        // Copy into a fresh Uint8Array: pdf.js TRANSFERS the buffer it is given
+        // (detaching it), and the caller's Buffer must stay usable for hashing.
+        data: new Uint8Array(bytes),
+        // Server side: no font rendering — text content is all we consume.
+        disableFontFace: true,
+        useSystemFonts: true,
+    }).promise;
+}
+//# sourceMappingURL=extract-pdf.js.map

package/dist/modules/workspace/file-readers/extract-pdf.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-pdf.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pdf.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAErD;;;;;;;;;;;GAWG;AACH;;;;;;;;GAQG;AACH,MAAM,kBAAkB,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,8BAA8B;AAE3E;;;;;;;;;GASG;AACH,MAAM,aAAa,GAAG,MAAM,CAAC;AAE7B;;;;;;;GAOG;AACH,MAAM,sBAAsB,GAAG,OAAO,CAAC;AAEvC,yDAAyD;AACzD,SAAS,UAAU;IACjB,OAAO;QACL,EAAE,EAAE,KAAK;QACT,OAAO,EAAE,6DAA6D,kBAAkB,0CAA0C;KACnI,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,KAAa;IAC5C,yEAAyE;IACzE,0EAA0E;IAC1E,0DAA0D;IAC1D,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACtC,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,gDAAgD,KAAK,CAAC,MAAM,qBAAqB,kBAAkB,iCAAiC;SAC9I,CAAC;IACJ,CAAC;IACD,IAAI,GAAwC,CAAC;IAC7C,IAAI,CAAC;QACH,GAAG,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,CAAC;IAC7B,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,iCAAkC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC5F,CAAC;IACD,IAAI,CAAC;QACH,IAAI,GAAG,CAAC,QAAQ,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,OAAO,EAAE,gDAAgD,GAAG,CAAC,QAAQ,qBAAqB,aAAa,yBAAyB;aACjI,CAAC;QACJ,CAAC;QACD,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YACvC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;YAC1B,yEAAyE;YACzE,0EAA0E;YAC1E,SAAS,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;YACrC,IAAI,SAAS,GAAG,kBAAkB;gBAAE,OAAO,UAAU,EAAE,CAAC;YACxD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YAClC,sEAAsE;YACtE,qEAAqE;YACrE,sEAAsE;YACtE,sEAAsE;YACtE,4DAA4D;YAC5D,MAAM,MAAM,GACV,IAAI,CAAC,iBAAiB,EACvB,CAAC,SAAS,EAAE,CAAC;YACd,IAAI,SAAS,GAAG,CAAC,CAAC;YAClB,IAAI,IAAI,GAAG,EAAE,CAAC;YACd,IAAI,KAAyB,CAAC;YAC9B,MAAM,KAAK,GAAG,GAAS,EAAE;gBACvB,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;oBACvB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;oBACjB,OAAO,GAAG,IAAI,CAAC;oBACf,oEAAoE;oBACpE,SAAS,IAAI,CAAC,CAAC;gBACjB,CAAC;gBACD,IAAI,GAAG,EAAE,CAAC;YACZ,CAAC,CAAC;YACF,SAAS,CAAC;gBACR,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBACnD,IAAI,IAAI;oBAAE,MAAM;gBAChB,SAAS,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC;gBAChC,IAAI,SAAS,GAAG,sBAAsB,EAAE,CAAC;oBACvC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;oBAC7C,IAAI,CAAC,OAAO,EAAE,CAAC;oBACf,OAAO;wBACL,EAAE,EAAE,KAAK;wBACT,OAAO,EAAE,yCAAyC,CAAC,oBAAoB,sBAAsB,0CAA0C;qBACxI,CAAC;gBACJ,CAAC;gBACD,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;oBAC/B,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC;wBAAE,SAAS,CAAC,gCAAgC;oBAChE,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;oBAC9B,wEAAwE;oBACxE,IAAI,OAAO,CAAC,KAAK,QAAQ,EAAE,CAAC;wBAC1B,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC;4BAAE,KAAK,EAAE,CAAC;wBAC5D,KAAK,GAAG,CAAC,CAAC;oBACZ,CAAC;oBACD,IAAI,IAAI,CAAC,GAAG,KAAK,EAAE,EAAE,CAAC;wBACpB,iEAAiE;wBACjE,gEAAgE;wBAChE,gEAAgE;wBAChE,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;wBACrD,IAAI,SAAS,GAAG,kBAAkB,EAAE,CAAC;4BACnC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;4BAC7C,OAAO,UAAU,EAAE,CAAC;wBACtB,CAAC;wBACD,IAAI,IAAI,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC;oBAC9C,CAAC;oBACD,IAAI,IAAI,CAAC,MAAM;wBAAE,KAAK,EAAE,CAAC;gBAC3B,CAAC;YACH,CAAC;YACD,KAAK,EAAE,CAAC;YACR,IAAI,CAAC,OAAO,EAAE,CAAC;QACjB,CAAC;QACD,MAAM,KAAK,GAAG,GAAG,GAAG,CAAC,QAAQ,QAAQ,GAAG,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;QACrE,OAAO,OAAO;YACZ,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,yCAAyC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;YAClG,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,qCAAqC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACnG,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,qCAAsC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAChG,CAAC;YAAS,CAAC;QACT,MAAM,GAAG,CAAC,OAAO,EAAE,CAAC;IACtB,CAAC;AACH,CAAC;AAGD,IAAI,YAAwC,CAAC;AAE7C,KAAK,UAAU,OAAO,CAAC,KAAa;IAClC,YAAY,KAAK,MAAM,CAAC,iCAAiC,CAAC,CAAC,KAAK,CAAC,CAAC,GAAY,EAAE,EAAE;QAChF,0EAA0E;QAC1E,yEAAyE;QACzE,sDAAsD;QACtD,YAAY,GAAG,SAAS,CAAC;QACzB,MAAM,GAAG,CAAC;IACZ,CAAC,CAAC,CAAC;IACH,MAAM,EAAE,WAAW,EAAE,GAAG,MAAM,YAAY,CAAC;IAC3C,OAAO,WAAW,CAAC;QACjB,wEAAwE;QACxE,wEAAwE;QACxE,IAAI,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC;QAC3B,mEAAmE;QACnE,eAAe,EAAE,IAAI;QACrB,cAAc,EAAE,IAAI;KACrB,CAAC,CAAC,OAAO,CAAC;AACb,CAAC"}

package/dist/modules/workspace/file-readers/extract-pptx.d.ts

@@ -0,0 +1,37 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract the text of a `.pptx` (PowerPoint) deck.
+ *
+ * Slides live at `ppt/slides/slideN.xml`; each is emitted under a `[slide N]`
+ * marker line, in the PRESENTATION's slide order — `ppt/presentation.xml`'s
+ * `<p:sldIdLst>`, resolved through its rels part (see
+ * `slideOrderFromPresentation`) — with numeric filename order as the fallback
+ * when the package has no readable list. Speaker notes
+ * follow their slide under `[slide N notes]` when non-empty. A slide's notes
+ * part is resolved through the slide's RELATIONSHIPS part (the `_rels` twin of
+ * the slide part's own NAME, relationship type ending `notesSlide`) — the
+ * package is free to number notes parts differently from slides — with the
+ * `notesSlideN.xml` convention as the fallback when the slide has no rels
+ * part at all. Within a slide, each `<a:p>` paragraph is a line;
+ * `<a:t>` runs concatenate with no separator (runs split mid-word).
+ *
+ * Bounded: every entry's DECLARED uncompressed size is checked before
+ * inflation (see `zipEntryOversize`), and the parts read for one deck may not
+ * exceed `MAX_DOC_TOTAL_BYTES` in total — over either bound is a typed
+ * failure, never an allocation.
+ */
+export declare function extractPptx(bytes: Buffer): ExtractResult;
+/**
+ * The Target of the first `notesSlide`-typed Relationship in a rels part, or
+ * undefined.
+ *
+ * Read by the parser: matched on the element's LOCAL name, so a producer that
+ * binds the relationships namespace to a prefix (`<r:Relationship r:Type=…>`)
+ * is read like any other — and a `<Relationship>`-looking fragment written
+ * inside a COMMENT or a CDATA section is text, not live metadata pointing the
+ * notes lookup at a part of its author's choosing.
+ */
+export declare function notesTargetFromRels(relsXml: string): string | undefined;
+/** Resolve an OPC relationship Target against the part's base directory. */
+export declare function resolveRelTarget(baseDir: string, target: string): string;
+//# sourceMappingURL=extract-pptx.d.ts.map

package/dist/modules/workspace/file-readers/extract-pptx.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-pptx.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pptx.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAY5D;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,wBAAgB,WAAW,CAAC,KAAK,EAAE,MAAM,GAAG,aAAa,CA8DxD;AA0KD;;;;;;;;;GASG;AACH,wBAAgB,mBAAmB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAWvE;AAED,4EAA4E;AAC5E,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAWxE"}

package/dist/modules/workspace/file-readers/extract-pptx.js

@@ -0,0 +1,288 @@
+import AdmZip from 'adm-zip';
+import { MAX_DOC_TOTAL_BYTES, attrByLocalName, decodeXmlEntities, localBlocks, localElementBlocks, localName, paragraphRunText, zipEntryOversize, } from './ooxml-text.js';
+/**
+ * Extract the text of a `.pptx` (PowerPoint) deck.
+ *
+ * Slides live at `ppt/slides/slideN.xml`; each is emitted under a `[slide N]`
+ * marker line, in the PRESENTATION's slide order — `ppt/presentation.xml`'s
+ * `<p:sldIdLst>`, resolved through its rels part (see
+ * `slideOrderFromPresentation`) — with numeric filename order as the fallback
+ * when the package has no readable list. Speaker notes
+ * follow their slide under `[slide N notes]` when non-empty. A slide's notes
+ * part is resolved through the slide's RELATIONSHIPS part (the `_rels` twin of
+ * the slide part's own NAME, relationship type ending `notesSlide`) — the
+ * package is free to number notes parts differently from slides — with the
+ * `notesSlideN.xml` convention as the fallback when the slide has no rels
+ * part at all. Within a slide, each `<a:p>` paragraph is a line;
+ * `<a:t>` runs concatenate with no separator (runs split mid-word).
+ *
+ * Bounded: every entry's DECLARED uncompressed size is checked before
+ * inflation (see `zipEntryOversize`), and the parts read for one deck may not
+ * exceed `MAX_DOC_TOTAL_BYTES` in total — over either bound is a typed
+ * failure, never an allocation.
+ */
+export function extractPptx(bytes) {
+    let slides;
+    let notesBySlide;
+    let presOrder;
+    try {
+        const zip = new AdmZip(bytes);
+        const budget = { remaining: MAX_DOC_TOTAL_BYTES };
+        slides = collectNumbered(zip, /^ppt\/slides\/slide(\d+)\.xml$/, budget);
+        notesBySlide = collectNotes(zip, slides, budget);
+        presOrder = slideOrderFromPresentation(zip, budget);
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a .pptx (${err.message})` };
+    }
+    if (slides.size === 0) {
+        return { ok: false, message: 'could not be parsed as a .pptx (no ppt/slides/slideN.xml inside the archive)' };
+    }
+    // Emission order: the presentation's own slide list when it resolves to
+    // selected parts, numeric filename order otherwise (parts the list does not
+    // name follow it, in filename order). When the LIST orders the deck, the
+    // markers number POSITIONS in it — what a viewer calls slide 1 — because a
+    // reordered deck's part filenames no longer mean anything positional.
+    const byFilename = [...slides.keys()].sort((a, b) => a - b);
+    let order = byFilename;
+    let positional = false;
+    if (presOrder !== undefined) {
+        const numByName = new Map();
+        for (const [n, slide] of slides)
+            numByName.set(slide.name, n);
+        const seen = new Set();
+        const fromList = [];
+        for (const name of presOrder) {
+            const n = numByName.get(name);
+            if (n !== undefined && !seen.has(n)) {
+                seen.add(n);
+                fromList.push(n);
+            }
+        }
+        if (fromList.length > 0) {
+            order = [...fromList, ...byFilename.filter((n) => !seen.has(n))];
+            positional = true;
+        }
+    }
+    const lines = [];
+    let anyNotes = false;
+    order.forEach((n, i) => {
+        const label = positional ? i + 1 : n;
+        lines.push(`[slide ${label}]`);
+        lines.push(...paragraphLines(slides.get(n).xml));
+        const notesXml = notesBySlide.get(n);
+        const noteLines = notesXml !== undefined ? paragraphLines(notesXml) : [];
+        if (noteLines.length > 0) {
+            anyNotes = true;
+            lines.push(`[slide ${label} notes]`);
+            lines.push(...noteLines);
+        }
+    });
+    return {
+        ok: true,
+        summary: `${slides.size} slide${slides.size === 1 ? '' : 's'}${anyNotes ? ' + notes' : ''}; layout, images and formatting omitted`,
+        text: lines.join('\n'),
+    };
+}
+/** Non-empty paragraph texts of one slide/notes part, in document order. */
+function paragraphLines(xml) {
+    const out = [];
+    for (const p of localBlocks(xml, 'p')) {
+        const text = paragraphRunText(p, 't');
+        if (text.trim() !== '')
+            out.push(text);
+    }
+    return out;
+}
+/** `entry`'s bytes as UTF-8, after the per-part and aggregate bounds. Throws over either. */
+function readEntryBounded(entry, budget) {
+    const oversize = zipEntryOversize(entry);
+    if (oversize)
+        throw new Error(oversize);
+    budget.remaining -= entry.header.size;
+    if (budget.remaining < 0) {
+        throw new Error(`the archive's parts exceed the ${MAX_DOC_TOTAL_BYTES}-byte (200 MB) total extraction limit`);
+    }
+    return entry.getData().toString('utf8');
+}
+/**
+ * Entries matching `re` (capture 1 = number), decoded as UTF-8, keyed by
+ * number. Two part names can parse to the SAME number (`slide1.xml` and
+ * `slide01.xml`); the winner is the FIRST in ascending part-name order —
+ * deterministic regardless of zip entry order, and the same policy as the
+ * browser twin (`pptxOutline.ts`), so viewer and `read_file` agree. Losing
+ * duplicates are never inflated (no budget charge).
+ *
+ * The winner's NAME rides along with its bytes because everything else about
+ * a slide hangs off the part name, not the number: see `collectNotes`.
+ */
+function collectNumbered(zip, re, budget) {
+    const matched = [];
+    for (const entry of zip.getEntries()) {
+        const m = re.exec(entry.entryName);
+        if (!m)
+            continue;
+        // A crafted name can spell a number past 2^53 (or Infinity): distinct
+        // parts would collide in the map and one would silently vanish.
+        const n = parseInt(m[1], 10);
+        if (Number.isSafeInteger(n))
+            matched.push([n, entry.entryName, entry]);
+    }
+    // A zip may list the SAME part name twice. Sorting by name leaves those two
+    // in archive order, so which one wins depends on how the file was written —
+    // and two archives with identical parts would extract differently. A part
+    // claimed twice is not a part this reader can resolve, so it is dropped.
+    const claims = new Map();
+    for (const [, name] of matched)
+        claims.set(name, (claims.get(name) ?? 0) + 1);
+    const unique = matched.filter(([, name]) => claims.get(name) === 1);
+    unique.sort((a, b) => (a[1] < b[1] ? -1 : a[1] > b[1] ? 1 : 0));
+    const out = new Map();
+    for (const [n, name, entry] of unique) {
+        if (!out.has(n))
+            out.set(n, { name, xml: readEntryBounded(entry, budget) });
+    }
+    return out;
+}
+const NOTES_REL_TYPE_SUFFIX = '/notesSlide';
+const SLIDE_REL_TYPE_SUFFIX = '/slide';
+/**
+ * The value of the attribute whose LOCAL name is `want` AND that carries a
+ * namespace prefix. `<p:sldId>` holds both its own `id` and the relationship
+ * reference `r:id`; plain local-name matching answers with whichever is
+ * written first, so the relationship id must be the PREFIXED one.
+ */
+function prefixedAttrByLocalName(attributes, want) {
+    for (const [key, value] of Object.entries(attributes)) {
+        if (key === 'xmlns' || key.startsWith('xmlns:'))
+            continue;
+        if (key.includes(':') && localName(key) === want)
+            return value;
+    }
+    return undefined;
+}
+/**
+ * Slide part names in PRESENTATION order: `ppt/presentation.xml`'s
+ * `<p:sldIdLst>` entries, each `r:id` resolved through the presentation's own
+ * rels part — or undefined when the package has no readable list. Reordering
+ * slides in PowerPoint rewrites the sldIdLst and leaves the part names alone,
+ * so `slide1.xml` need not be the deck's first slide; the numeric filename
+ * sort is only the fallback for packages without the list.
+ */
+function slideOrderFromPresentation(zip, budget) {
+    const rels = zip.getEntry('ppt/_rels/presentation.xml.rels');
+    const pres = zip.getEntry('ppt/presentation.xml');
+    if (!rels || !pres)
+        return undefined;
+    const targetById = new Map();
+    for (const rel of localElementBlocks(readEntryBounded(rels, budget), ['Relationship'])) {
+        const type = attrByLocalName(rel.attributes, 'Type');
+        if (type === undefined || !type.endsWith(SLIDE_REL_TYPE_SUFFIX))
+            continue;
+        const id = attrByLocalName(rel.attributes, 'Id');
+        // Targets are RAW in the rels (see `notesTargetFromRels`) and relative to
+        // the presentation part's directory.
+        const target = attrByLocalName(rel.attributes, 'Target');
+        if (id !== undefined && target !== undefined) {
+            targetById.set(id, resolveRelTarget('ppt', decodeXmlEntities(target)));
+        }
+    }
+    if (targetById.size === 0)
+        return undefined;
+    const list = localBlocks(readEntryBounded(pres, budget), 'sldIdLst')[0];
+    if (list === undefined)
+        return undefined;
+    const order = [];
+    for (const sld of localElementBlocks(list, ['sldId'])) {
+        const rid = prefixedAttrByLocalName(sld.attributes, 'id');
+        const target = rid !== undefined ? targetById.get(rid) : undefined;
+        if (target !== undefined)
+            order.push(target);
+    }
+    return order.length > 0 ? order : undefined;
+}
+/**
+ * The OPC relationships part of `partName` — `dir/_rels/base.rels`. Derived
+ * from the part NAME, never from the slide number: when a deck ships both
+ * `slide1.xml` and `slide01.xml`, the name-ordering rule picks `slide01.xml`,
+ * whose rels part is `slide01.xml.rels`. Rebuilding the path from the number
+ * asked for `slide1.xml.rels` — a part belonging to the OTHER file — and so
+ * either lost that slide's speaker notes or attached the losing part's.
+ */
+function relsPartName(partName) {
+    const cut = partName.lastIndexOf('/');
+    return `${partName.slice(0, cut)}/_rels/${partName.slice(cut + 1)}.rels`;
+}
+/**
+ * The conventional notes part for a slide part — `slide01.xml` →
+ * `notesSlide01.xml`. Derived from the name for the same reason as the rels
+ * path, so a zero-padded deck's fallback lands on the matching notes part.
+ */
+function conventionalNotesPart(partName) {
+    const base = partName.slice(partName.lastIndexOf('/') + 1);
+    return `ppt/notesSlides/notes${base[0].toUpperCase()}${base.slice(1)}`;
+}
+/**
+ * Slide number → its notes part's XML, resolved through each slide's `.rels`
+ * part; the conventional-name fallback ONLY for a slide without a rels part.
+ * Both paths come from the SELECTED part's name (see `relsPartName`).
+ */
+function collectNotes(zip, slides, budget) {
+    const out = new Map();
+    for (const [n, slide] of slides) {
+        const rels = zip.getEntry(relsPartName(slide.name));
+        let notesPart;
+        if (rels) {
+            const target = notesTargetFromRels(readEntryBounded(rels, budget));
+            notesPart = target !== undefined ? resolveRelTarget('ppt/slides', target) : undefined;
+        }
+        else {
+            notesPart = conventionalNotesPart(slide.name);
+        }
+        if (notesPart === undefined)
+            continue;
+        const entry = zip.getEntry(notesPart);
+        if (entry)
+            out.set(n, readEntryBounded(entry, budget));
+    }
+    return out;
+}
+/**
+ * The Target of the first `notesSlide`-typed Relationship in a rels part, or
+ * undefined.
+ *
+ * Read by the parser: matched on the element's LOCAL name, so a producer that
+ * binds the relationships namespace to a prefix (`<r:Relationship r:Type=…>`)
+ * is read like any other — and a `<Relationship>`-looking fragment written
+ * inside a COMMENT or a CDATA section is text, not live metadata pointing the
+ * notes lookup at a part of its author's choosing.
+ */
+export function notesTargetFromRels(relsXml) {
+    for (const rel of localElementBlocks(relsXml, ['Relationship'])) {
+        const type = attrByLocalName(rel.attributes, 'Type');
+        if (type !== undefined && type.endsWith(NOTES_REL_TYPE_SUFFIX)) {
+            // Attribute values are RAW here (the block reader does not decode), and a
+            // part name may legally contain `&`, written `&amp;` in the rels.
+            const target = attrByLocalName(rel.attributes, 'Target');
+            return target !== undefined ? decodeXmlEntities(target) : undefined;
+        }
+    }
+    return undefined;
+}
+/** Resolve an OPC relationship Target against the part's base directory. */
+export function resolveRelTarget(baseDir, target) {
+    const parts = target.startsWith('/')
+        ? target.slice(1).split('/')
+        : [...baseDir.split('/'), ...target.split('/')];
+    const out = [];
+    for (const p of parts) {
+        if (p === '' || p === '.')
+            continue;
+        if (p === '..')
+            out.pop();
+        else
+            out.push(p);
+    }
+    return out.join('/');
+}
+//# sourceMappingURL=extract-pptx.js.map

package/dist/modules/workspace/file-readers/extract-pptx.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-pptx.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pptx.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,SAAS,CAAC;AAE7B,OAAO,EACL,mBAAmB,EACnB,eAAe,EACf,iBAAiB,EACjB,WAAW,EACX,kBAAkB,EAClB,SAAS,EACT,gBAAgB,EAChB,gBAAgB,GACjB,MAAM,iBAAiB,CAAC;AAEzB;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,UAAU,WAAW,CAAC,KAAa;IACvC,IAAI,MAA8B,CAAC;IACnC,IAAI,YAAiC,CAAC;IACtC,IAAI,SAA+B,CAAC;IACpC,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC;QAC9B,MAAM,MAAM,GAAG,EAAE,SAAS,EAAE,mBAAmB,EAAE,CAAC;QAClD,MAAM,GAAG,eAAe,CAAC,GAAG,EAAE,gCAAgC,EAAE,MAAM,CAAC,CAAC;QACxE,YAAY,GAAG,YAAY,CAAC,GAAG,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;QACjD,SAAS,GAAG,0BAA0B,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;IACtD,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,mCAAoC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC9F,CAAC;IACD,IAAI,MAAM,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;QACtB,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,8EAA8E,EAAE,CAAC;IAChH,CAAC;IAED,wEAAwE;IACxE,4EAA4E;IAC5E,yEAAyE;IACzE,2EAA2E;IAC3E,sEAAsE;IACtE,MAAM,UAAU,GAAG,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IAC5D,IAAI,KAAK,GAAG,UAAU,CAAC;IACvB,IAAI,UAAU,GAAG,KAAK,CAAC;IACvB,IAAI,SAAS,KAAK,SAAS,EAAE,CAAC;QAC5B,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAC;QAC5C,KAAK,MAAM,CAAC,CAAC,EAAE,KAAK,CAAC,IAAI,MAAM;YAAE,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAC/B,MAAM,QAAQ,GAAa,EAAE,CAAC;QAC9B,KAAK,MAAM,IAAI,IAAI,SAAS,EAAE,CAAC;YAC7B,MAAM,CAAC,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;YAC9B,IAAI,CAAC,KAAK,SAAS,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;gBACpC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;gBACZ,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YACnB,CAAC;QACH,CAAC;QACD,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxB,KAAK,GAAG,CAAC,GAAG,QAAQ,EAAE,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACjE,UAAU,GAAG,IAAI,CAAC;QACpB,CAAC;IACH,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,QAAQ,GAAG,KAAK,CAAC;IACrB,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;QACrB,MAAM,KAAK,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACrC,KAAK,CAAC,IAAI,CAAC,UAAU,KAAK,GAAG,CAAC,CAAC;QAC/B,KAAK,CAAC,IAAI,CAAC,GAAG,cAAc,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAE,CAAC,GAAG,CAAC,CAAC,CAAC;QAClD,MAAM,QAAQ,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QACrC,MAAM,SAAS,GAAG,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QACzE,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACzB,QAAQ,GAAG,IAAI,CAAC;YAChB,KAAK,CAAC,IAAI,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC;YACrC,KAAK,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC,CAAC,CAAC;IACH,OAAO;QACL,EAAE,EAAE,IAAI;QACR,OAAO,EAAE,GAAG,MAAM,CAAC,IAAI,SAAS,MAAM,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,GAAG,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,yCAAyC;QAClI,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;KACvB,CAAC;AACJ,CAAC;AAED,4EAA4E;AAC5E,SAAS,cAAc,CAAC,GAAW;IACjC,MAAM,GAAG,GAAa,EAAE,CAAC;IACzB,KAAK,MAAM,CAAC,IAAI,WAAW,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,CAAC;QACtC,MAAM,IAAI,GAAG,gBAAgB,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;QACtC,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE;YAAE,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACzC,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAOD,6FAA6F;AAC7F,SAAS,gBAAgB,CAAC,KAAuB,EAAE,MAAkB;IACnE,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;IACzC,IAAI,QAAQ;QAAE,MAAM,IAAI,KAAK,CAAC,QAAQ,CAAC,CAAC;IACxC,MAAM,CAAC,SAAS,IAAI,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC;IACtC,IAAI,MAAM,CAAC,SAAS,GAAG,CAAC,EAAE,CAAC;QACzB,MAAM,IAAI,KAAK,CAAC,kCAAkC,mBAAmB,uCAAuC,CAAC,CAAC;IAChH,CAAC;IACD,OAAO,KAAK,CAAC,OAAO,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;AAC1C,CAAC;AASD;;;;;;;;;;GAUG;AACH,SAAS,eAAe,CAAC,GAAW,EAAE,EAAU,EAAE,MAAkB;IAClE,MAAM,OAAO,GAA8C,EAAE,CAAC;IAC9D,KAAK,MAAM,KAAK,IAAI,GAAG,CAAC,UAAU,EAAE,EAAE,CAAC;QACrC,MAAM,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;QACnC,IAAI,CAAC,CAAC;YAAE,SAAS;QACjB,sEAAsE;QACtE,gEAAgE;QAChE,MAAM,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QAC7B,IAAI,MAAM,CAAC,aAAa,CAAC,CAAC,CAAC;YAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC;IACzE,CAAC;IACD,4EAA4E;IAC5E,4EAA4E;IAC5E,0EAA0E;IAC1E,yEAAyE;IACzE,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAC;IACzC,KAAK,MAAM,CAAC,EAAE,IAAI,CAAC,IAAI,OAAO;QAAE,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;IACpE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAChE,MAAM,GAAG,GAAG,IAAI,GAAG,EAAqB,CAAC;IACzC,KAAK,MAAM,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,EAAE,CAAC;QACtC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;YAAE,GAAG,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE,gBAAgB,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;IAC9E,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,MAAM,qBAAqB,GAAG,aAAa,CAAC;AAC5C,MAAM,qBAAqB,GAAG,QAAQ,CAAC;AAEvC;;;;;GAKG;AACH,SAAS,uBAAuB,CAAC,UAAkC,EAAE,IAAY;IAC/E,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,CAAC;QACtD,IAAI,GAAG,KAAK,OAAO,IAAI,GAAG,CAAC,UAAU,CAAC,QAAQ,CAAC;YAAE,SAAS;QAC1D,IAAI,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,SAAS,CAAC,GAAG,CAAC,KAAK,IAAI;YAAE,OAAO,KAAK,CAAC;IACjE,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,0BAA0B,CAAC,GAAW,EAAE,MAAkB;IACjE,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,iCAAiC,CAAC,CAAC;IAC7D,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,sBAAsB,CAAC,CAAC;IAClD,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAC;IACrC,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC7C,KAAK,MAAM,GAAG,IAAI,kBAAkB,CAAC,gBAAgB,CAAC,IAAI,EAAE,MAAM,CAAC,EAAE,CAAC,cAAc,CAAC,CAAC,EAAE,CAAC;QACvF,MAAM,IAAI,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;QACrD,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,qBAAqB,CAAC;YAAE,SAAS;QAC1E,MAAM,EAAE,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;QACjD,0EAA0E;QAC1E,qCAAqC;QACrC,MAAM,MAAM,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;QACzD,IAAI,EAAE,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;YAC7C,UAAU,CAAC,GAAG,CAAC,EAAE,EAAE,gBAAgB,CAAC,KAAK,EAAE,iBAAiB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;QACzE,CAAC;IACH,CAAC;IACD,IAAI,UAAU,CAAC,IAAI,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC;IAC5C,MAAM,IAAI,GAAG,WAAW,CAAC,gBAAgB,CAAC,IAAI,EAAE,MAAM,CAAC,EAAE,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC;IACxE,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,SAAS,CAAC;IACzC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,GAAG,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;QACtD,MAAM,GAAG,GAAG,uBAAuB,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;QAC1D,MAAM,MAAM,GAAG,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACnE,IAAI,MAAM,KAAK,SAAS;YAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC/C,CAAC;IACD,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC;AAC9C,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,YAAY,CAAC,QAAgB;IACpC,MAAM,GAAG,GAAG,QAAQ,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;IACtC,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,UAAU,QAAQ,CAAC,KAAK,CAAC,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC;AAC3E,CAAC;AAED;;;;GAIG;AACH,SAAS,qBAAqB,CAAC,QAAgB;IAC7C,MAAM,IAAI,GAAG,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC;IAC3D,OAAO,wBAAwB,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;AACzE,CAAC;AAED;;;;GAIG;AACH,SAAS,YAAY,CAAC,GAAW,EAAE,MAA8B,EAAE,MAAkB;IACnF,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAC;IACtC,KAAK,MAAM,CAAC,CAAC,EAAE,KAAK,CAAC,IAAI,MAAM,EAAE,CAAC;QAChC,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;QACpD,IAAI,SAA6B,CAAC;QAClC,IAAI,IAAI,EAAE,CAAC;YACT,MAAM,MAAM,GAAG,mBAAmB,CAAC,gBAAgB,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC;YACnE,SAAS,GAAG,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,gBAAgB,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACxF,CAAC;aAAM,CAAC;YACN,SAAS,GAAG,qBAAqB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAChD,CAAC;QACD,IAAI,SAAS,KAAK,SAAS;YAAE,SAAS;QACtC,MAAM,KAAK,GAAG,GAAG,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;QACtC,IAAI,KAAK;YAAE,GAAG,CAAC,GAAG,CAAC,CAAC,EAAE,gBAAgB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,mBAAmB,CAAC,OAAe;IACjD,KAAK,MAAM,GAAG,IAAI,kBAAkB,CAAC,OAAO,EAAE,CAAC,cAAc,CAAC,CAAC,EAAE,CAAC;QAChE,MAAM,IAAI,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;QACrD,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,CAAC,QAAQ,CAAC,qBAAqB,CAAC,EAAE,CAAC;YAC/D,0EAA0E;YAC1E,kEAAkE;YAClE,MAAM,MAAM,GAAG,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;YACzD,OAAO,MAAM,KAAK,SAAS,CAAC,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACtE,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,4EAA4E;AAC5E,MAAM,UAAU,gBAAgB,CAAC,OAAe,EAAE,MAAc;IAC9D,MAAM,KAAK,GAAG,MAAM,CAAC,UAAU,CAAC,GAAG,CAAC;QAClC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC;QAC5B,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;IAClD,MAAM,GAAG,GAAa,EAAE,CAAC;IACzB,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACtB,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,GAAG;YAAE,SAAS;QACpC,IAAI,CAAC,KAAK,IAAI;YAAE,GAAG,CAAC,GAAG,EAAE,CAAC;;YACrB,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IACnB,CAAC;IACD,OAAO,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACvB,CAAC"}

package/dist/modules/workspace/file-readers/extract-xlsx.d.ts

@@ -0,0 +1,10 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract a `.xlsx` (Excel) workbook via SheetJS: per sheet a `[sheet: Name]`
+ * marker, then the rows as tab-separated values (each cell's FORMATTED value,
+ * e.g. dates as dates; tabs/newlines INSIDE a cell become single spaces so a
+ * cell's line break never reads as a row boundary), with trailing empty rows
+ * trimmed.
+ */
+export declare function extractXlsx(bytes: Buffer): ExtractResult;
+//# sourceMappingURL=extract-xlsx.d.ts.map

package/dist/modules/workspace/file-readers/extract-xlsx.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-xlsx.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-xlsx.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAY5D;;;;;;GAMG;AACH,wBAAgB,WAAW,CAAC,KAAK,EAAE,MAAM,GAAG,aAAa,CA0ExD"}

package/dist/modules/workspace/file-readers/extract-xlsx.js

@@ -0,0 +1,98 @@
+import AdmZip from 'adm-zip';
+import * as XLSX from 'xlsx';
+import { MAX_DOC_TOTAL_BYTES, zipEntryOversize } from './ooxml-text.js';
+/**
+ * Per-sheet extraction caps. A worksheet's declared range can be enormous
+ * (a stray cell at XFD1048576 makes the range 16k x 1M); the caps bound the
+ * text an agent gets to something readable, and the extraction SAYS when it
+ * truncated (a `[sheet truncated …]` line right under the sheet marker).
+ */
+const MAX_ROWS_PER_SHEET = 10_000;
+const MAX_COLS_PER_SHEET = 200;
+/**
+ * Extract a `.xlsx` (Excel) workbook via SheetJS: per sheet a `[sheet: Name]`
+ * marker, then the rows as tab-separated values (each cell's FORMATTED value,
+ * e.g. dates as dates; tabs/newlines INSIDE a cell become single spaces so a
+ * cell's line break never reads as a row boundary), with trailing empty rows
+ * trimmed.
+ */
+export function extractXlsx(bytes) {
+    // A real .xlsx is a zip. Check the signature OURSELVES because SheetJS
+    // helpfully falls back to parsing arbitrary bytes as CSV/HTML — which would
+    // turn a corrupt upload into confident nonsense instead of an honest error.
+    if (bytes.length < 4 || bytes[0] !== 0x50 || bytes[1] !== 0x4b) {
+        return { ok: false, message: 'could not be parsed as a .xlsx (not a zip archive)' };
+    }
+    // Zip-bomb bound BEFORE SheetJS inflates anything: the central directory
+    // declares every entry's uncompressed size, so per-part (50 MB) and
+    // aggregate (200 MB) limits cost one directory scan. A zip AdmZip cannot
+    // read falls through — SheetJS then reports its own parse failure.
+    try {
+        const zip = new AdmZip(bytes);
+        let total = 0;
+        for (const entry of zip.getEntries()) {
+            const oversize = zipEntryOversize(entry);
+            if (oversize)
+                return { ok: false, message: `could not be extracted as a .xlsx (${oversize})` };
+            total += entry.header.size;
+            if (total > MAX_DOC_TOTAL_BYTES) {
+                return {
+                    ok: false,
+                    message: `could not be extracted as a .xlsx (the archive's parts exceed the ${MAX_DOC_TOTAL_BYTES}-byte (200 MB) total extraction limit)`,
+                };
+            }
+        }
+    }
+    catch {
+        // not AdmZip-readable — let SheetJS produce the typed parse failure below
+    }
+    let wb;
+    try {
+        wb = XLSX.read(bytes, { type: 'buffer' });
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a .xlsx (${err.message})` };
+    }
+    const lines = [];
+    for (const name of wb.SheetNames) {
+        // Control separators become spaces — the same rule as the ods extractor:
+        // a crafted workbook's sheet name holding a tab or newline would split
+        // the `[sheet: …]` marker's own line and break grep line numbers.
+        lines.push(`[sheet: ${name.replace(/[\t\n\r]+/g, ' ')}]`);
+        const ws = wb.Sheets[name];
+        const ref = ws?.['!ref'];
+        if (!ws || !ref)
+            continue; // empty sheet — marker only
+        const range = XLSX.utils.decode_range(ref);
+        const truncated = [];
+        if (range.e.r - range.s.r + 1 > MAX_ROWS_PER_SHEET) {
+            range.e.r = range.s.r + MAX_ROWS_PER_SHEET - 1;
+            truncated.push(`first ${MAX_ROWS_PER_SHEET} rows`);
+        }
+        if (range.e.c - range.s.c + 1 > MAX_COLS_PER_SHEET) {
+            range.e.c = range.s.c + MAX_COLS_PER_SHEET - 1;
+            truncated.push(`first ${MAX_COLS_PER_SHEET} columns`);
+        }
+        if (truncated.length > 0) {
+            lines.push(`[sheet truncated to the ${truncated.join(' and ')}]`);
+            ws['!ref'] = XLSX.utils.encode_range(range);
+        }
+        // Rows via sheet_to_json (formatted cell text, one array per worksheet
+        // row): sheet_to_csv CSV-quotes a cell containing a line break, so
+        // splitting its output on newlines turned ONE worksheet row into several
+        // extracted rows. Cell-internal tabs/newlines become single spaces
+        // instead — the same one-line-per-row contract as the ods extractor.
+        const rows = XLSX.utils
+            .sheet_to_json(ws, { header: 1, raw: false, defval: '', blankrows: true })
+            .map((cells) => cells.map((c) => String(c).replace(/[\t\n\r]+/g, ' ')).join('\t'));
+        while (rows.length > 0 && rows[rows.length - 1].replace(/\t/g, '') === '')
+            rows.pop();
+        lines.push(...rows);
+    }
+    return {
+        ok: true,
+        summary: `${wb.SheetNames.length} sheet${wb.SheetNames.length === 1 ? '' : 's'}, rows as tab-separated values; formulas, formatting and charts omitted`,
+        text: lines.join('\n'),
+    };
+}
+//# sourceMappingURL=extract-xlsx.js.map

package/dist/modules/workspace/file-readers/extract-xlsx.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-xlsx.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-xlsx.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,SAAS,CAAC;AAC7B,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAE7B,OAAO,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,MAAM,iBAAiB,CAAC;AAExE;;;;;GAKG;AACH,MAAM,kBAAkB,GAAG,MAAM,CAAC;AAClC,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAE/B;;;;;;GAMG;AACH,MAAM,UAAU,WAAW,CAAC,KAAa;IACvC,uEAAuE;IACvE,4EAA4E;IAC5E,4EAA4E;IAC5E,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QAC/D,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,oDAAoD,EAAE,CAAC;IACtF,CAAC;IACD,yEAAyE;IACzE,oEAAoE;IACpE,yEAAyE;IACzE,mEAAmE;IACnE,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC;QAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,MAAM,KAAK,IAAI,GAAG,CAAC,UAAU,EAAE,EAAE,CAAC;YACrC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;YACzC,IAAI,QAAQ;gBAAE,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,sCAAsC,QAAQ,GAAG,EAAE,CAAC;YAC/F,KAAK,IAAI,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC;YAC3B,IAAI,KAAK,GAAG,mBAAmB,EAAE,CAAC;gBAChC,OAAO;oBACL,EAAE,EAAE,KAAK;oBACT,OAAO,EAAE,qEAAqE,mBAAmB,wCAAwC;iBAC1I,CAAC;YACJ,CAAC;QACH,CAAC;IACH,CAAC;IAAC,MAAM,CAAC;QACP,0EAA0E;IAC5E,CAAC;IACD,IAAI,EAAiB,CAAC;IACtB,IAAI,CAAC;QACH,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;IAC5C,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,mCAAoC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC9F,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,IAAI,IAAI,EAAE,CAAC,UAAU,EAAE,CAAC;QACjC,yEAAyE;QACzE,uEAAuE;QACvE,kEAAkE;QAClE,KAAK,CAAC,IAAI,CAAC,WAAW,IAAI,CAAC,OAAO,CAAC,YAAY,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC;QAC1D,MAAM,EAAE,GAAG,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAC3B,MAAM,GAAG,GAAG,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC;QACzB,IAAI,CAAC,EAAE,IAAI,CAAC,GAAG;YAAE,SAAS,CAAC,4BAA4B;QACvD,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC;QAC3C,MAAM,SAAS,GAAa,EAAE,CAAC;QAC/B,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,kBAAkB,EAAE,CAAC;YACnD,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,kBAAkB,GAAG,CAAC,CAAC;YAC/C,SAAS,CAAC,IAAI,CAAC,SAAS,kBAAkB,OAAO,CAAC,CAAC;QACrD,CAAC;QACD,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,kBAAkB,EAAE,CAAC;YACnD,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,kBAAkB,GAAG,CAAC,CAAC;YAC/C,SAAS,CAAC,IAAI,CAAC,SAAS,kBAAkB,UAAU,CAAC,CAAC;QACxD,CAAC;QACD,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACzB,KAAK,CAAC,IAAI,CAAC,2BAA2B,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;YAClE,EAAE,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,KAAK,CAAC,CAAC;QAC9C,CAAC;QACD,uEAAuE;QACvE,mEAAmE;QACnE,yEAAyE;QACzE,mEAAmE;QACnE,qEAAqE;QACrE,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK;aACpB,aAAa,CAAY,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC;aACpF,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,YAAY,EAAE,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;QACrF,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,EAAE;YAAE,IAAI,CAAC,GAAG,EAAE,CAAC;QACtF,KAAK,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC;IACtB,CAAC;IACD,OAAO;QACL,EAAE,EAAE,IAAI;QACR,OAAO,EAAE,GAAG,EAAE,CAAC,UAAU,CAAC,MAAM,SAAS,EAAE,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,yEAAyE;QACvJ,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;KACvB,CAAC;AACJ,CAAC"}