@bevel-software/platform-core-backend 0.11.1 → 0.12.0

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

@@ -0,0 +1,67 @@
+import AdmZip from 'adm-zip';
+import { localBlocks, localElementBlocks, localName, paragraphRunText, zipEntryOversize } from './ooxml-text.js';
+/**
+ * Extract the BODY text of a `.docx` (Word) document.
+ *
+ * A docx is a zip whose main part is `word/document.xml`. v1 extracts the body
+ * only — headers/footers are skipped, and the marker summary says so.
+ *
+ *  - Paragraphs become lines. `<w:t>` runs are concatenated with NO separator
+ *    (Word splits runs mid-word on formatting boundaries).
+ *  - Tables become lines with cell text tab-separated, one line per row.
+ */
+export function extractDocx(bytes) {
+    let xml;
+    try {
+        const zip = new AdmZip(bytes);
+        const entry = zip.getEntry('word/document.xml');
+        if (!entry) {
+            return { ok: false, message: 'could not be parsed as a .docx (no word/document.xml inside the archive)' };
+        }
+        // Declared-uncompressed-size bound BEFORE inflation — see zipEntryOversize.
+        const oversize = zipEntryOversize(entry);
+        if (oversize)
+            return { ok: false, message: `could not be extracted as a .docx (${oversize})` };
+        xml = entry.getData().toString('utf8');
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a .docx (${err.message})` };
+    }
+    // The BODY element read by the parser rather than matched by a regex: a
+    // comment or CDATA section mentioning `<w:body>` could answer as the
+    // document body and hand back its text instead of the real one.
+    const body = localBlocks(xml, 'body')[0] ?? xml;
+    const lines = [];
+    let paragraphs = 0;
+    let tables = 0;
+    // Tables and paragraphs in DOCUMENT order, straight from the parser: a match
+    // is never descended into, so a table's own paragraphs stay inside it and are
+    // rendered once, as its rows. The hand-rolled splitter this replaces counted
+    // `<w:tbl>` opens and closes with a regex, which a comment or CDATA section
+    // mentioning either could throw off — splitting a paragraph in half and
+    // dropping its text.
+    for (const block of localElementBlocks(body, ['tbl', 'p'])) {
+        if (localName(block.name) === 'tbl') {
+            tables++;
+            for (const row of localBlocks(block.body ?? '', 'tr')) {
+                const cells = localBlocks(row, 'tc').map((cell) => paragraphRunText(cell, 't'));
+                lines.push(cells.join('	'));
+            }
+        }
+        else {
+            lines.push(paragraphRunText(block.body ?? '', 't'));
+            paragraphs++;
+        }
+    }
+    while (lines.length > 0 && lines[lines.length - 1].trim() === '')
+        lines.pop();
+    const parts = [`${paragraphs} paragraph${paragraphs === 1 ? '' : 's'}`];
+    if (tables > 0)
+        parts.push(`${tables} table${tables === 1 ? '' : 's'}`);
+    return {
+        ok: true,
+        summary: `${parts.join(' + ')}; body only (headers/footers skipped); layout, images and formatting omitted`,
+        text: lines.join('\n'),
+    };
+}
+//# sourceMappingURL=extract-docx.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-docx.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-docx.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,SAAS,CAAC;AAE7B,OAAO,EAAE,WAAW,EAAE,kBAAkB,EAAE,SAAS,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,iBAAiB,CAAC;AAEjH;;;;;;;;;GASG;AACH,MAAM,UAAU,WAAW,CAAC,KAAa;IACvC,IAAI,GAAW,CAAC;IAChB,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC;QAC9B,MAAM,KAAK,GAAG,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC;QAChD,IAAI,CAAC,KAAK,EAAE,CAAC;YACX,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,0EAA0E,EAAE,CAAC;QAC5G,CAAC;QACD,4EAA4E;QAC5E,MAAM,QAAQ,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;QACzC,IAAI,QAAQ;YAAE,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,sCAAsC,QAAQ,GAAG,EAAE,CAAC;QAC/F,GAAG,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;IACzC,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,wEAAwE;IACxE,qEAAqE;IACrE,gEAAgE;IAChE,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC;IAEhD,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,IAAI,MAAM,GAAG,CAAC,CAAC;IACf,6EAA6E;IAC7E,8EAA8E;IAC9E,6EAA6E;IAC7E,4EAA4E;IAC5E,wEAAwE;IACxE,qBAAqB;IACrB,KAAK,MAAM,KAAK,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC;QAC3D,IAAI,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,KAAK,EAAE,CAAC;YACpC,MAAM,EAAE,CAAC;YACT,KAAK,MAAM,GAAG,IAAI,WAAW,CAAC,KAAK,CAAC,IAAI,IAAI,EAAE,EAAE,IAAI,CAAC,EAAE,CAAC;gBACtD,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,gBAAgB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC,CAAC;gBAChF,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;YAC9B,CAAC;QACH,CAAC;aAAM,CAAC;YACN,KAAK,CAAC,IAAI,CAAC,gBAAgB,CAAC,KAAK,CAAC,IAAI,IAAI,EAAE,EAAE,GAAG,CAAC,CAAC,CAAC;YACpD,UAAU,EAAE,CAAC;QACf,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE;QAAE,KAAK,CAAC,GAAG,EAAE,CAAC;IAE9E,MAAM,KAAK,GAAG,CAAC,GAAG,UAAU,aAAa,UAAU,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;IACxE,IAAI,MAAM,GAAG,CAAC;QAAE,KAAK,CAAC,IAAI,CAAC,GAAG,MAAM,SAAS,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;IACxE,OAAO;QACL,EAAE,EAAE,IAAI;QACR,OAAO,EAAE,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,8EAA8E;QAC3G,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;KACvB,CAAC;AACJ,CAAC"}

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

@@ -0,0 +1,18 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract a `.eml` (RFC 822 / MIME) email into the shared email text shape
+ * (see `email-text.ts`).
+ *
+ * Parsing is `postal-mime` (postalsys, MIT-0): small, dependency-free, ESM,
+ * and the same code runs in the browser — which keeps the frontend viewer's
+ * story identical to the agent's. It normalizes encoded-word headers and
+ * multipart bodies, and hands the `Date:` header over as ISO when it parses.
+ *
+ * postal-mime is LENIENT — random bytes "parse" to an empty message rather
+ * than throwing — so recognizability is checked here: a file yielding NO
+ * email headers at all (no From/To/Cc/Bcc/Subject/Date/Message-ID) and no
+ * attachments is not an email, and gets the typed could-not-be-parsed
+ * failure instead of an empty extraction.
+ */
+export declare function extractEml(bytes: Buffer): Promise<ExtractResult>;
+//# sourceMappingURL=extract-eml.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-eml.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-eml.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAI5D;;;;;;;;;;;;;;GAcG;AACH,wBAAsB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,CA4BtE"}

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

@@ -0,0 +1,87 @@
+import PostalMime from 'postal-mime';
+import { emailExtraction, htmlToEmailText } from './email-text.js';
+import { MAX_DOC_PART_BYTES } from './ooxml-text.js';
+/**
+ * Extract a `.eml` (RFC 822 / MIME) email into the shared email text shape
+ * (see `email-text.ts`).
+ *
+ * Parsing is `postal-mime` (postalsys, MIT-0): small, dependency-free, ESM,
+ * and the same code runs in the browser — which keeps the frontend viewer's
+ * story identical to the agent's. It normalizes encoded-word headers and
+ * multipart bodies, and hands the `Date:` header over as ISO when it parses.
+ *
+ * postal-mime is LENIENT — random bytes "parse" to an empty message rather
+ * than throwing — so recognizability is checked here: a file yielding NO
+ * email headers at all (no From/To/Cc/Bcc/Subject/Date/Message-ID) and no
+ * attachments is not an email, and gets the typed could-not-be-parsed
+ * failure instead of an empty extraction.
+ */
+export async function extractEml(bytes) {
+    // The same raw-size bound the PDF extractor applies: no container to
+    // pre-scan, so the bound is the file's size, checked before parsing.
+    if (bytes.length > MAX_DOC_PART_BYTES) {
+        return {
+            ok: false,
+            message: `could not be extracted as a .eml (the file is ${bytes.length} bytes — over the ${MAX_DOC_PART_BYTES}-byte (50 MB) extraction limit)`,
+        };
+    }
+    let email;
+    try {
+        email = await PostalMime.parse(bytes);
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a .eml (${err.message})` };
+    }
+    if (email.from === undefined &&
+        email.to === undefined &&
+        email.cc === undefined &&
+        email.bcc === undefined &&
+        email.subject === undefined &&
+        email.date === undefined &&
+        email.messageId === undefined &&
+        email.attachments.length === 0) {
+        return { ok: false, message: 'could not be parsed as a .eml (no email headers found)' };
+    }
+    return { ok: true, ...emailExtraction(emlModel(email)) };
+}
+/** postal-mime's parse, shaped into the format-independent email model. */
+function emlModel(email) {
+    const text = email.text !== undefined && email.text.trim() !== '' ? email.text : undefined;
+    const html = email.html !== undefined && email.html.trim() !== '' ? email.html : undefined;
+    const body = text ?? (html !== undefined ? htmlToEmailText(html) : '');
+    return {
+        from: email.from && addressListText([email.from]),
+        to: email.to && addressListText(email.to),
+        cc: email.cc && addressListText(email.cc),
+        bcc: email.bcc && addressListText(email.bcc),
+        subject: email.subject,
+        date: email.date !== undefined ? isoDate(email.date) : undefined,
+        body: body.replace(/\s+$/, ''),
+        bodySource: text !== undefined ? 'text' : html !== undefined ? 'html' : 'none',
+        attachments: email.attachments.map((a) => ({
+            name: a.filename ?? 'unnamed attachment',
+            mimeType: a.mimeType,
+            sizeBytes: typeof a.content === 'string' ? Buffer.byteLength(a.content) : a.content.byteLength,
+        })),
+    };
+}
+/** `Name <addr>, addr2, Group: member, member` — groups flattened inline. */
+function addressListText(list) {
+    const s = list
+        .map(function one(a) {
+        if (a.group !== undefined)
+            return `${a.name}: ${a.group.map(one).join(', ')}`;
+        if (a.address === undefined || a.address === '')
+            return a.name;
+        return a.name !== '' && a.name !== a.address ? `${a.name} <${a.address}>` : a.address;
+    })
+        .filter((t) => t !== '')
+        .join(', ');
+    return s === '' ? undefined : s;
+}
+/** postal-mime already normalizes parseable dates to ISO; keep the raw value when it could not. */
+function isoDate(value) {
+    const d = new Date(value);
+    return Number.isNaN(d.getTime()) ? value : d.toISOString();
+}
+//# sourceMappingURL=extract-eml.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-eml.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-eml.ts"],"names":[],"mappings":"AAAA,OAAO,UAAwC,MAAM,aAAa,CAAC;AAEnE,OAAO,EAAE,eAAe,EAAE,eAAe,EAAyC,MAAM,iBAAiB,CAAC;AAC1G,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAErD;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,KAAa;IAC5C,qEAAqE;IACrE,qEAAqE;IACrE,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACtC,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,iDAAiD,KAAK,CAAC,MAAM,qBAAqB,kBAAkB,iCAAiC;SAC/I,CAAC;IACJ,CAAC;IACD,IAAI,KAAY,CAAC;IACjB,IAAI,CAAC;QACH,KAAK,GAAG,MAAM,UAAU,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IACxC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,kCAAmC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC7F,CAAC;IACD,IACE,KAAK,CAAC,IAAI,KAAK,SAAS;QACxB,KAAK,CAAC,EAAE,KAAK,SAAS;QACtB,KAAK,CAAC,EAAE,KAAK,SAAS;QACtB,KAAK,CAAC,GAAG,KAAK,SAAS;QACvB,KAAK,CAAC,OAAO,KAAK,SAAS;QAC3B,KAAK,CAAC,IAAI,KAAK,SAAS;QACxB,KAAK,CAAC,SAAS,KAAK,SAAS;QAC7B,KAAK,CAAC,WAAW,CAAC,MAAM,KAAK,CAAC,EAC9B,CAAC;QACD,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,wDAAwD,EAAE,CAAC;IAC1F,CAAC;IACD,OAAO,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,eAAe,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;AAC3D,CAAC;AAED,2EAA2E;AAC3E,SAAS,QAAQ,CAAC,KAAY;IAC5B,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC;IAC3F,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC;IAC3F,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IACvE,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,IAAI,IAAI,eAAe,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACjD,EAAE,EAAE,KAAK,CAAC,EAAE,IAAI,eAAe,CAAC,KAAK,CAAC,EAAE,CAAC;QACzC,EAAE,EAAE,KAAK,CAAC,EAAE,IAAI,eAAe,CAAC,KAAK,CAAC,EAAE,CAAC;QACzC,GAAG,EAAE,KAAK,CAAC,GAAG,IAAI,eAAe,CAAC,KAAK,CAAC,GAAG,CAAC;QAC5C,OAAO,EAAE,KAAK,CAAC,OAAO;QACtB,IAAI,EAAE,KAAK,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS;QAChE,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;QAC9B,UAAU,EAAE,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM;QAC9E,WAAW,EAAE,KAAK,CAAC,WAAW,CAAC,GAAG,CAChC,CAAC,CAAC,EAAmB,EAAE,CAAC,CAAC;YACvB,IAAI,EAAE,CAAC,CAAC,QAAQ,IAAI,oBAAoB;YACxC,QAAQ,EAAE,CAAC,CAAC,QAAQ;YACpB,SAAS,EAAE,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU;SAC/F,CAAC,CACH;KACF,CAAC;AACJ,CAAC;AAED,6EAA6E;AAC7E,SAAS,eAAe,CAAC,IAAe;IACtC,MAAM,CAAC,GAAG,IAAI;SACX,GAAG,CAAC,SAAS,GAAG,CAAC,CAAU;QAC1B,IAAI,CAAC,CAAC,KAAK,KAAK,SAAS;YAAE,OAAO,GAAG,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC9E,IAAI,CAAC,CAAC,OAAO,KAAK,SAAS,IAAI,CAAC,CAAC,OAAO,KAAK,EAAE;YAAE,OAAO,CAAC,CAAC,IAAI,CAAC;QAC/D,OAAO,CAAC,CAAC,IAAI,KAAK,EAAE,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC;IACxF,CAAC,CAAC;SACD,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC;SACvB,IAAI,CAAC,IAAI,CAAC,CAAC;IACd,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;AAClC,CAAC;AAED,mGAAmG;AACnG,SAAS,OAAO,CAAC,KAAa;IAC5B,MAAM,CAAC,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC;IAC1B,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;AAC7D,CAAC"}

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

@@ -0,0 +1,17 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract a `.msg` (Outlook item, CFB container) email into the shared email
+ * text shape (see `email-text.ts`).
+ *
+ * Parsing is `@kenjiuno/msgreader` (HiraokaHyperTools, Apache-2.0) — the
+ * maintained MAPI/CFB reader. It never throws for bad content of its own
+ * accord: unparseable bytes come back as `{ dataType: null, error }`, which
+ * maps onto the typed could-not-be-parsed failure here.
+ *
+ * Body preference mirrors `.eml`: the plain-text `PidTagBody` first, an HTML
+ * body stripped to text second. An Outlook item whose body exists ONLY as
+ * compressed RTF is degraded honestly — the extraction says
+ * "[body is RTF; no plain-text part]" instead of pretending to decode RTF.
+ */
+export declare function extractMsg(bytes: Buffer): ExtractResult;
+//# sourceMappingURL=extract-msg.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-msg.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-msg.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAI5D;;;;;;;;;;;;;GAaG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,aAAa,CAsBvD"}

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

@@ -0,0 +1,121 @@
+import MsgReaderImport from '@kenjiuno/msgreader';
+import { emailExtraction, htmlToEmailText } from './email-text.js';
+import { MAX_DOC_PART_BYTES } from './ooxml-text.js';
+/**
+ * Extract a `.msg` (Outlook item, CFB container) email into the shared email
+ * text shape (see `email-text.ts`).
+ *
+ * Parsing is `@kenjiuno/msgreader` (HiraokaHyperTools, Apache-2.0) — the
+ * maintained MAPI/CFB reader. It never throws for bad content of its own
+ * accord: unparseable bytes come back as `{ dataType: null, error }`, which
+ * maps onto the typed could-not-be-parsed failure here.
+ *
+ * Body preference mirrors `.eml`: the plain-text `PidTagBody` first, an HTML
+ * body stripped to text second. An Outlook item whose body exists ONLY as
+ * compressed RTF is degraded honestly — the extraction says
+ * "[body is RTF; no plain-text part]" instead of pretending to decode RTF.
+ */
+export function extractMsg(bytes) {
+    if (bytes.length > MAX_DOC_PART_BYTES) {
+        return {
+            ok: false,
+            message: `could not be extracted as a .msg (the file is ${bytes.length} bytes — over the ${MAX_DOC_PART_BYTES}-byte (50 MB) extraction limit)`,
+        };
+    }
+    let fields;
+    try {
+        // DataView over the Buffer's exact region — no copy, and msgreader never
+        // sees bytes outside the file.
+        fields = new MsgReader(new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength)).getFileData();
+    }
+    catch (err) {
+        return { ok: false, message: `could not be parsed as a .msg (${err.message})` };
+    }
+    if (fields.dataType !== 'msg') {
+        return {
+            ok: false,
+            message: `could not be parsed as a .msg (${fields.error ?? 'not an Outlook message file'})`,
+        };
+    }
+    return { ok: true, ...emailExtraction(msgModel(fields)) };
+}
+/** msgreader's field data, shaped into the format-independent email model. */
+function msgModel(fields) {
+    const recipients = fields.recipients ?? [];
+    const text = fields.body !== undefined && fields.body.trim() !== '' ? fields.body : undefined;
+    const html = htmlBody(fields);
+    const body = text ?? (html !== undefined ? htmlToEmailText(html) : '');
+    const bodySource = text !== undefined ? 'text' : html !== undefined ? 'html' : fields.compressedRtf !== undefined ? 'rtf-only' : 'none';
+    const date = fields.clientSubmitTime ?? fields.messageDeliveryTime;
+    return {
+        from: mailboxText(fields.senderName, fields.senderSmtpAddress ?? fields.senderEmail),
+        to: recipientList(recipients, 'to'),
+        cc: recipientList(recipients, 'cc'),
+        bcc: recipientList(recipients, 'bcc'),
+        subject: fields.subject,
+        date: date !== undefined ? isoDate(date) : undefined,
+        body: body.replace(/\s+$/, ''),
+        bodySource,
+        attachments: (fields.attachments ?? []).map((a) => ({
+            name: a.fileName ?? a.fileNameShort ?? a.name ?? 'unnamed attachment',
+            mimeType: a.attachMimeTag,
+            sizeBytes: a.contentLength,
+        })),
+    };
+}
+/** The HTML body, whichever MAPI property carries it (string, or utf-8 bytes). */
+function htmlBody(fields) {
+    if (fields.bodyHtml !== undefined && fields.bodyHtml.trim() !== '')
+        return fields.bodyHtml;
+    if (fields.html instanceof Uint8Array && fields.html.length > 0) {
+        return Buffer.from(fields.html).toString('utf8');
+    }
+    return undefined;
+}
+/** `Name <addr>` / `Name` / `addr` — whatever the message carries. */
+function mailboxText(name, address) {
+    const n = name?.trim() ?? '';
+    const a = address?.trim() ?? '';
+    if (n !== '' && a !== '' && n !== a)
+        return `${n} <${a}>`;
+    if (a !== '')
+        return a;
+    return n !== '' ? n : undefined;
+}
+/**
+ * The bucket a recipient belongs to. msgreader maps the MAPI `PidTagRecipientType`
+ * values 1/2/3 to these strings itself (lib/MsgReader.js, the `recipType` case)
+ * — but ONLY those three: any other raw PT_LONG value, e.g. `MAPI_TO | MAPI_P1`
+ * (0x10000001) on a resubmitted message, leaks through as a NUMBER despite the
+ * `'to' | 'cc' | 'bcc'` typing. Mask the resubmit/submitted flag bits and remap
+ * so such a recipient keeps its line instead of vanishing; anything else
+ * (including untyped) counts as `to`, matching the frontend's `msgMessage.ts`.
+ */
+function recipientBucket(recipType) {
+    if (recipType === 'to' || recipType === 'cc' || recipType === 'bcc')
+        return recipType;
+    if (typeof recipType === 'number') {
+        const base = recipType & 0x0fffffff; // strip MAPI_SUBMITTED (0x80000000) / MAPI_P1 (0x10000000)
+        if (base === 2)
+            return 'cc';
+        if (base === 3)
+            return 'bcc';
+    }
+    return 'to';
+}
+/** The comma-joined mailboxes of one recipient type. Untyped recipients count as `to`. */
+function recipientList(recipients, type) {
+    const s = recipients
+        .filter((r) => recipientBucket(r.recipType) === type)
+        .map((r) => mailboxText(r.name, r.smtpAddress ?? r.email))
+        .filter((t) => t !== undefined)
+        .join(', ');
+    return s === '' ? undefined : s;
+}
+/** msgreader emits RFC-1123 GMT strings; normalize to ISO, keep raw when unparseable. */
+function isoDate(value) {
+    const d = new Date(value);
+    return Number.isNaN(d.getTime()) ? value : d.toISOString();
+}
+const MsgReader = MsgReaderImport.default ?? MsgReaderImport;
+//# sourceMappingURL=extract-msg.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-msg.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-msg.ts"],"names":[],"mappings":"AAAA,OAAO,eAAe,MAAM,qBAAqB,CAAC;AAElD,OAAO,EAAE,eAAe,EAAE,eAAe,EAAyC,MAAM,iBAAiB,CAAC;AAC1G,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAErD;;;;;;;;;;;;;GAaG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACtC,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,iDAAiD,KAAK,CAAC,MAAM,qBAAqB,kBAAkB,iCAAiC;SAC/I,CAAC;IACJ,CAAC;IACD,IAAI,MAAkB,CAAC;IACvB,IAAI,CAAC;QACH,yEAAyE;QACzE,+BAA+B;QAC/B,MAAM,GAAG,IAAI,SAAS,CAAC,IAAI,QAAQ,CAAC,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;IACvG,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,kCAAmC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC7F,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,KAAK,KAAK,EAAE,CAAC;QAC9B,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,kCAAkC,MAAM,CAAC,KAAK,IAAI,6BAA6B,GAAG;SAC5F,CAAC;IACJ,CAAC;IACD,OAAO,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,eAAe,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;AAC5D,CAAC;AAED,8EAA8E;AAC9E,SAAS,QAAQ,CAAC,MAAkB;IAClC,MAAM,UAAU,GAAG,MAAM,CAAC,UAAU,IAAI,EAAE,CAAC;IAC3C,MAAM,IAAI,GAAG,MAAM,CAAC,IAAI,KAAK,SAAS,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC;IAC9F,MAAM,IAAI,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC;IAC9B,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IACvE,MAAM,UAAU,GACd,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,aAAa,KAAK,SAAS,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,MAAM,CAAC;IACvH,MAAM,IAAI,GAAG,MAAM,CAAC,gBAAgB,IAAI,MAAM,CAAC,mBAAmB,CAAC;IACnE,OAAO;QACL,IAAI,EAAE,WAAW,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,CAAC,iBAAiB,IAAI,MAAM,CAAC,WAAW,CAAC;QACpF,EAAE,EAAE,aAAa,CAAC,UAAU,EAAE,IAAI,CAAC;QACnC,EAAE,EAAE,aAAa,CAAC,UAAU,EAAE,IAAI,CAAC;QACnC,GAAG,EAAE,aAAa,CAAC,UAAU,EAAE,KAAK,CAAC;QACrC,OAAO,EAAE,MAAM,CAAC,OAAO;QACvB,IAAI,EAAE,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS;QACpD,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;QAC9B,UAAU;QACV,WAAW,EAAE,CAAC,MAAM,CAAC,WAAW,IAAI,EAAE,CAAC,CAAC,GAAG,CACzC,CAAC,CAAC,EAAmB,EAAE,CAAC,CAAC;YACvB,IAAI,EAAE,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,aAAa,IAAI,CAAC,CAAC,IAAI,IAAI,oBAAoB;YACrE,QAAQ,EAAE,CAAC,CAAC,aAAa;YACzB,SAAS,EAAE,CAAC,CAAC,aAAa;SAC3B,CAAC,CACH;KACF,CAAC;AACJ,CAAC;AAED,kFAAkF;AAClF,SAAS,QAAQ,CAAC,MAAkB;IAClC,IAAI,MAAM,CAAC,QAAQ,KAAK,SAAS,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,EAAE,KAAK,EAAE;QAAE,OAAO,MAAM,CAAC,QAAQ,CAAC;IAC3F,IAAI,MAAM,CAAC,IAAI,YAAY,UAAU,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAChE,OAAO,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;IACnD,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,sEAAsE;AACtE,SAAS,WAAW,CAAC,IAAwB,EAAE,OAA2B;IACxE,MAAM,CAAC,GAAG,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;IAC7B,MAAM,CAAC,GAAG,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;IAChC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC;IAC1D,IAAI,CAAC,KAAK,EAAE;QAAE,OAAO,CAAC,CAAC;IACvB,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;AAClC,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,eAAe,CAAC,SAAkB;IACzC,IAAI,SAAS,KAAK,IAAI,IAAI,SAAS,KAAK,IAAI,IAAI,SAAS,KAAK,KAAK;QAAE,OAAO,SAAS,CAAC;IACtF,IAAI,OAAO,SAAS,KAAK,QAAQ,EAAE,CAAC;QAClC,MAAM,IAAI,GAAG,SAAS,GAAG,UAAU,CAAC,CAAC,2DAA2D;QAChG,IAAI,IAAI,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QAC5B,IAAI,IAAI,KAAK,CAAC;YAAE,OAAO,KAAK,CAAC;IAC/B,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,0FAA0F;AAC1F,SAAS,aAAa,CAAC,UAAiC,EAAE,IAAyB;IACjF,MAAM,CAAC,GAAG,UAAU;SACjB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,IAAI,CAAC;SACpD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC;SACzD,MAAM,CAAC,CAAC,CAAC,EAAe,EAAE,CAAC,CAAC,KAAK,SAAS,CAAC;SAC3C,IAAI,CAAC,IAAI,CAAC,CAAC;IACd,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;AAClC,CAAC;AAED,yFAAyF;AACzF,SAAS,OAAO,CAAC,KAAa;IAC5B,MAAM,CAAC,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC;IAC1B,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;AAC7D,CAAC;AASD,MAAM,SAAS,GACZ,eAA2D,CAAC,OAAO,IAAI,eAAe,CAAC"}

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

@@ -0,0 +1,13 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract the text of a `.odp` (OpenDocument Presentation) deck.
+ *
+ * Slides are the `<draw:page>` elements of `content.xml`, in DOCUMENT order —
+ * ODF orders slides in the file itself, so unlike pptx there is no numeric
+ * filename sort. Each slide is emitted under a `[slide N]` marker (N = 1-based
+ * position); speaker notes (`<presentation:notes>` inside the page) follow
+ * under `[slide N notes]` when non-empty. Within a slide, each `<text:p>` in
+ * its frames is a line; spans concatenate with no separator.
+ */
+export declare function extractOdp(bytes: Buffer): ExtractResult;
+//# sourceMappingURL=extract-odp.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-odp.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-odp.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAI5D;;;;;;;;;GASG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,aAAa,CAmCvD"}

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

@@ -0,0 +1,60 @@
+import { odfParagraphLines, readOdfContentXml } from './odf-text.js';
+import { localBlocks, localElementBlocks, removeLocalElements } from './ooxml-text.js';
+/**
+ * Extract the text of a `.odp` (OpenDocument Presentation) deck.
+ *
+ * Slides are the `<draw:page>` elements of `content.xml`, in DOCUMENT order —
+ * ODF orders slides in the file itself, so unlike pptx there is no numeric
+ * filename sort. Each slide is emitted under a `[slide N]` marker (N = 1-based
+ * position); speaker notes (`<presentation:notes>` inside the page) follow
+ * under `[slide N notes]` when non-empty. Within a slide, each `<text:p>` in
+ * its frames is a line; spans concatenate with no separator.
+ */
+export function extractOdp(bytes) {
+    const content = readOdfContentXml(bytes, '.odp');
+    if (!content.ok)
+        return content;
+    const pages = drawPageBlocks(content.xml);
+    if (pages.length === 0) {
+        return { ok: false, message: 'could not be parsed as a .odp (no draw:page elements in content.xml)' };
+    }
+    const lines = [];
+    let anyNotes = false;
+    pages.forEach((page, i) => {
+        // Split the notes part out FIRST so its paragraphs don't render as slide text.
+        // Notes read by the parser and matched on their LOCAL name: a comment that
+        // resembled `<presentation:notes>` used to be emitted as real speaker notes,
+        // and a deck binding the presentation namespace to another prefix had none.
+        const notesXml = localBlocks(page, 'notes')[0] ?? '';
+        // The slide's own text is the page with the notes ELEMENTS removed by
+        // their parsed boundaries — global string replacement of the notes BODY
+        // also deleted slide text that happened to serialize identically to it.
+        const slideXml = removeLocalElements(page, ['notes']);
+        lines.push(`[slide ${i + 1}]`);
+        lines.push(...odfParagraphLines(slideXml));
+        const noteLines = odfParagraphLines(notesXml);
+        if (noteLines.length > 0) {
+            anyNotes = true;
+            lines.push(`[slide ${i + 1} notes]`);
+            lines.push(...noteLines);
+        }
+    });
+    return {
+        ok: true,
+        summary: `${pages.length} slide${pages.length === 1 ? '' : 's'}${anyNotes ? ' + notes' : ''}; layout, images and formatting omitted`,
+        text: lines.join('\n'),
+    };
+}
+/**
+ * The `<draw:page>…</draw:page>` bodies in document order (pages never nest).
+ * A SELF-CLOSING `<draw:page/>` is a legal, fully blank slide — it yields ''
+ * so the deck's numbering (and a deliberately blank deck) stays correct.
+ */
+function drawPageBlocks(xml) {
+    // The shared quote-aware scanner: a `/>` INSIDE a quoted attribute value (a
+    // page name like `a/>b`) is part of the value, never the self-closing
+    // delimiter, and a page whose close tag is missing costs one scan of the
+    // document rather than one per opener (see `xmlElementBlocks`).
+    return localElementBlocks(xml, ['page']).map((e) => e.body ?? '');
+}
+//# sourceMappingURL=extract-odp.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-odp.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-odp.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,iBAAiB,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AACrE,OAAO,EAAE,WAAW,EAAE,kBAAkB,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AAEvF;;;;;;;;;GASG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,MAAM,OAAO,GAAG,iBAAiB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACjD,IAAI,CAAC,OAAO,CAAC,EAAE;QAAE,OAAO,OAAO,CAAC;IAEhC,MAAM,KAAK,GAAG,cAAc,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAC1C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,sEAAsE,EAAE,CAAC;IACxG,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,QAAQ,GAAG,KAAK,CAAC;IACrB,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE;QACxB,+EAA+E;QAC/E,2EAA2E;QAC3E,6EAA6E;QAC7E,4EAA4E;QAC5E,MAAM,QAAQ,GAAG,WAAW,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACrD,sEAAsE;QACtE,wEAAwE;QACxE,wEAAwE;QACxE,MAAM,QAAQ,GAAG,mBAAmB,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC;QACtD,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAC/B,KAAK,CAAC,IAAI,CAAC,GAAG,iBAAiB,CAAC,QAAQ,CAAC,CAAC,CAAC;QAC3C,MAAM,SAAS,GAAG,iBAAiB,CAAC,QAAQ,CAAC,CAAC;QAC9C,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACzB,QAAQ,GAAG,IAAI,CAAC;YAChB,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,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,KAAK,CAAC,MAAM,SAAS,KAAK,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,GAAG,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,yCAAyC;QACpI,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;KACvB,CAAC;AACJ,CAAC;AAED;;;;GAIG;AACH,SAAS,cAAc,CAAC,GAAW;IACjC,4EAA4E;IAC5E,sEAAsE;IACtE,yEAAyE;IACzE,gEAAgE;IAChE,OAAO,kBAAkB,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC;AACpE,CAAC"}

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

@@ -0,0 +1,10 @@
+import type { ExtractResult } from './doc-extract.types.js';
+/**
+ * Extract a `.ods` (OpenDocument Spreadsheet) workbook: per `<table:table>`
+ * (sheet) a `[sheet: Name]` marker (the `table:name` attribute), then the rows
+ * as tab-separated cell text. A cell's text is its `<text:p>` content
+ * (multiple paragraphs join with a space — a newline would break the row
+ * line); covered cells (under a merge) render empty.
+ */
+export declare function extractOds(bytes: Buffer): ExtractResult;
+//# sourceMappingURL=extract-ods.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-ods.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-ods.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAyB5D;;;;;;GAMG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,aAAa,CAqBvD"}

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

@@ -0,0 +1,173 @@
+import { attrByLocalName, decodeXmlEntities, localElementBlocks, localName, walkLocalElementBlocks, } from './ooxml-text.js';
+import { odfParagraphBlocks, odfParagraphText, readOdfContentXml, } from './odf-text.js';
+/**
+ * Per-sheet extraction caps — the SAME bounds as the xlsx extractor. ODF is
+ * fond of `table:number-columns-repeated="16384"` (or a million empty trailing
+ * rows) to pad a sheet to the grid, so repeats are expanded BOUNDED and the
+ * extraction says when it truncated (a `[sheet truncated …]` line right under
+ * the sheet marker). Trailing EMPTY cells/rows are trimmed before their
+ * repeats are applied at all, so grid padding never counts as truncation.
+ */
+const MAX_ROWS_PER_SHEET = 10_000;
+const MAX_COLS_PER_SHEET = 200;
+/**
+ * Extract a `.ods` (OpenDocument Spreadsheet) workbook: per `<table:table>`
+ * (sheet) a `[sheet: Name]` marker (the `table:name` attribute), then the rows
+ * as tab-separated cell text. A cell's text is its `<text:p>` content
+ * (multiple paragraphs join with a space — a newline would break the row
+ * line); covered cells (under a merge) render empty.
+ */
+export function extractOds(bytes) {
+    const content = readOdfContentXml(bytes, '.ods');
+    if (!content.ok)
+        return content;
+    const tables = tableBlocks(content.xml);
+    if (tables.length === 0) {
+        return { ok: false, message: 'could not be parsed as a .ods (no table:table elements in content.xml)' };
+    }
+    const lines = [];
+    for (const table of tables) {
+        lines.push(`[sheet: ${table.name}]`);
+        const { rows, truncated } = expandRows(table.xml);
+        if (truncated.length > 0)
+            lines.push(`[sheet truncated to the ${truncated.join(' and ')}]`);
+        lines.push(...rows.map((cells) => cells.join('\t')));
+    }
+    return {
+        ok: true,
+        summary: `${tables.length} sheet${tables.length === 1 ? '' : 's'}, rows as tab-separated values; formulas, formatting and charts omitted`,
+        text: lines.join('\n'),
+    };
+}
+/**
+ * The `<table:table>` blocks with their decoded `table:name`, in document
+ * order. Non-greedy close — a nested table (legal in ODF text documents, not
+ * produced by spreadsheets) would end the outer block early, degrading
+ * grouping but never crashing.
+ */
+function tableBlocks(xml) {
+    // Read by the parser and matched on the LOCAL name: a comment or CDATA
+    // section holding a table-looking fragment used to answer as a real sheet,
+    // and a document binding the table namespace to another prefix had none.
+    const out = [];
+    for (const table of localElementBlocks(xml, ['table'])) {
+        const raw = attrByLocalName(table.attributes, 'name');
+        out.push({
+            // Control separators become spaces: a name holding an encoded newline
+            // or tab (`&#10;`) would corrupt the `[sheet: …]` marker's own line and
+            // the TSV structure under it.
+            name: raw ? decodeXmlEntities(raw).replace(/[\t\n\r]+/g, ' ') : `Sheet${out.length + 1}`,
+            xml: table.body ?? '',
+        });
+    }
+    return out;
+}
+/**
+ * Expand a sheet's rows with BOUNDED repeat handling, INCREMENTALLY — a row's
+ * expansion lands in the capped output as it parses, so the caps bound memory
+ * as well as output (materializing every row's cells before consulting the
+ * cap let an accepted ODS allocate its whole expansion first):
+ *
+ *  - all-empty rows are buffered as a COUNT (with their
+ *    `table:number-rows-repeated` applied) and flushed only when a non-empty
+ *    row follows, so a million-row empty tail simply disappears,
+ *  - once the row cap is hit, the remaining rows are never parsed at all.
+ *
+ * `truncated` lists what the caps cut (mirrors the xlsx extractor's note).
+ */
+function expandRows(tableXml) {
+    // The shared quote-aware scanner as a WALK, not an array: materializing
+    // every row block before consulting the cap let a sheet of >10k explicit
+    // rows allocate them all first. Each row lands here as it parses, and the
+    // visitor's `true` stops the scan at the cap — a self-closing row WITH
+    // attributes is still a row, a `/>` inside a quoted attribute value is not
+    // a delimiter, and an UNCLOSED row costs one scan of the sheet rather than
+    // one per opener (see `xmlElementBlocks`).
+    const rows = [];
+    let pendingEmpty = 0;
+    let rowsTruncated = false;
+    let colsTruncated = false;
+    walkLocalElementBlocks(tableXml, ['table-row'], (row) => {
+        const repeat = repeatCount(attrByLocalName(row.attributes, 'number-rows-repeated'));
+        const cells = expandCells(row.body ?? '');
+        if (cells.cells.length === 0) {
+            // Empty rows are interior padding until a non-empty row proves it —
+            // trailing ones are dropped with their repeats (grid padding, not data).
+            pendingEmpty += repeat;
+            return false;
+        }
+        if (cells.truncated)
+            colsTruncated = true;
+        for (; pendingEmpty > 0 && rows.length < MAX_ROWS_PER_SHEET; pendingEmpty--)
+            rows.push([]);
+        let i = 0;
+        for (; i < repeat && rows.length < MAX_ROWS_PER_SHEET; i++)
+            rows.push(cells.cells);
+        if (pendingEmpty > 0 || i < repeat) {
+            // The cap cut real content (a sheet that merely FILLS it is not truncated).
+            rowsTruncated = true;
+            return true;
+        }
+        return false;
+    });
+    const truncated = [];
+    if (rowsTruncated)
+        truncated.push(`first ${MAX_ROWS_PER_SHEET} rows`);
+    if (colsTruncated)
+        truncated.push(`first ${MAX_COLS_PER_SHEET} columns`);
+    return { rows, truncated };
+}
+/**
+ * One row's cell texts: `<table:table-cell>` / `<table:covered-table-cell>`
+ * in order, expanded INCREMENTALLY like the rows above — trailing EMPTY cells
+ * are buffered as a count (their `table:number-columns-repeated` never
+ * expands) and the parse stops at the column cap.
+ */
+function expandCells(rowXml) {
+    // Same walking scanner as the rows above — a row spelling a million
+    // explicit cells stops parsing at the column cap too.
+    const cells = [];
+    let pendingEmpty = 0;
+    let truncated = false;
+    walkLocalElementBlocks(rowXml, ['table-cell', 'covered-table-cell'], (cell) => {
+        const repeat = repeatCount(attrByLocalName(cell.attributes, 'number-columns-repeated'));
+        // Covered cells carry no own text anyway. Element-produced newlines/tabs
+        // INSIDE a cell (<text:line-break/>, <text:tab/>) become single spaces:
+        // the extraction's contract is one row per line with tab-separated cells,
+        // and a literal \n or \t inside a cell's text would silently break both.
+        // A COVERED cell is the hidden half of a merge: the visible cell carries
+        // the text. Such a cell may still hold stale content, and emitting it put
+        // a value in the grid where the sheet shows none.
+        const text = localName(cell.name) === 'covered-table-cell'
+            ? ''
+            : odfParagraphBlocks(cell.body ?? '')
+                .map(odfParagraphText)
+                .join(' ')
+                .replace(/[\t\n\r]+/g, ' ');
+        if (text === '') {
+            pendingEmpty += repeat;
+            return false;
+        }
+        for (; pendingEmpty > 0 && cells.length < MAX_COLS_PER_SHEET; pendingEmpty--)
+            cells.push('');
+        let i = 0;
+        for (; i < repeat && cells.length < MAX_COLS_PER_SHEET; i++)
+            cells.push(text);
+        if (pendingEmpty > 0 || i < repeat) {
+            // The cap cut real content (a row that merely FILLS it is not truncated).
+            truncated = true;
+            return true;
+        }
+        return false;
+    });
+    return { cells, truncated };
+}
+/** A `…-repeated="N"` attribute value, clamped to a sane positive integer. */
+function repeatCount(raw) {
+    // Decoded first: the block scanner hands attribute values RAW, and a repeat
+    // legally written with character references (`&#49;&#48;`) must count as
+    // 10, not silently fall back to 1.
+    const n = raw !== undefined ? parseInt(decodeXmlEntities(raw), 10) : 1;
+    return Number.isFinite(n) && n >= 1 ? n : 1;
+}
+//# sourceMappingURL=extract-ods.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"extract-ods.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-ods.ts"],"names":[],"mappings":"AACA,OAAO,EACL,eAAe,EACf,iBAAiB,EACjB,kBAAkB,EAClB,SAAS,EACT,sBAAsB,GACvB,MAAM,iBAAiB,CAAC;AACzB,OAAO,EACL,kBAAkB,EAClB,gBAAgB,EAChB,iBAAiB,GAClB,MAAM,eAAe,CAAC;AAEvB;;;;;;;GAOG;AACH,MAAM,kBAAkB,GAAG,MAAM,CAAC;AAClC,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAE/B;;;;;;GAMG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,MAAM,OAAO,GAAG,iBAAiB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACjD,IAAI,CAAC,OAAO,CAAC,EAAE;QAAE,OAAO,OAAO,CAAC;IAEhC,MAAM,MAAM,GAAG,WAAW,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IACxC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACxB,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,wEAAwE,EAAE,CAAC;IAC1G,CAAC;IAED,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,KAAK,CAAC,IAAI,CAAC,WAAW,KAAK,CAAC,IAAI,GAAG,CAAC,CAAC;QACrC,MAAM,EAAE,IAAI,EAAE,SAAS,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAClD,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,2BAA2B,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;QAC5F,KAAK,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IACvD,CAAC;IACD,OAAO;QACL,EAAE,EAAE,IAAI;QACR,OAAO,EAAE,GAAG,MAAM,CAAC,MAAM,SAAS,MAAM,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,yEAAyE;QACzI,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;KACvB,CAAC;AACJ,CAAC;AAED;;;;;GAKG;AACH,SAAS,WAAW,CAAC,GAAW;IAC9B,uEAAuE;IACvE,2EAA2E;IAC3E,yEAAyE;IACzE,MAAM,GAAG,GAAyC,EAAE,CAAC;IACrD,KAAK,MAAM,KAAK,IAAI,kBAAkB,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;QACvD,MAAM,GAAG,GAAG,eAAe,CAAC,KAAK,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;QACtD,GAAG,CAAC,IAAI,CAAC;YACP,sEAAsE;YACtE,wEAAwE;YACxE,8BAA8B;YAC9B,IAAI,EAAE,GAAG,CAAC,CAAC,CAAC,iBAAiB,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,YAAY,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE;YACxF,GAAG,EAAE,KAAK,CAAC,IAAI,IAAI,EAAE;SACtB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,SAAS,UAAU,CAAC,QAAgB;IAClC,wEAAwE;IACxE,yEAAyE;IACzE,0EAA0E;IAC1E,uEAAuE;IACvE,2EAA2E;IAC3E,2EAA2E;IAC3E,2CAA2C;IAC3C,MAAM,IAAI,GAAe,EAAE,CAAC;IAC5B,IAAI,YAAY,GAAG,CAAC,CAAC;IACrB,IAAI,aAAa,GAAG,KAAK,CAAC;IAC1B,IAAI,aAAa,GAAG,KAAK,CAAC;IAC1B,sBAAsB,CAAC,QAAQ,EAAE,CAAC,WAAW,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE;QACtD,MAAM,MAAM,GAAG,WAAW,CAAC,eAAe,CAAC,GAAG,CAAC,UAAU,EAAE,sBAAsB,CAAC,CAAC,CAAC;QACpF,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC;QAC1C,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC7B,oEAAoE;YACpE,yEAAyE;YACzE,YAAY,IAAI,MAAM,CAAC;YACvB,OAAO,KAAK,CAAC;QACf,CAAC;QACD,IAAI,KAAK,CAAC,SAAS;YAAE,aAAa,GAAG,IAAI,CAAC;QAC1C,OAAO,YAAY,GAAG,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,kBAAkB,EAAE,YAAY,EAAE;YAAE,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAC3F,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,OAAO,CAAC,GAAG,MAAM,IAAI,IAAI,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC,EAAE;YAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QACnF,IAAI,YAAY,GAAG,CAAC,IAAI,CAAC,GAAG,MAAM,EAAE,CAAC;YACnC,4EAA4E;YAC5E,aAAa,GAAG,IAAI,CAAC;YACrB,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC,CAAC,CAAC;IACH,MAAM,SAAS,GAAa,EAAE,CAAC;IAC/B,IAAI,aAAa;QAAE,SAAS,CAAC,IAAI,CAAC,SAAS,kBAAkB,OAAO,CAAC,CAAC;IACtE,IAAI,aAAa;QAAE,SAAS,CAAC,IAAI,CAAC,SAAS,kBAAkB,UAAU,CAAC,CAAC;IACzE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC;AAC7B,CAAC;AAED;;;;;GAKG;AACH,SAAS,WAAW,CAAC,MAAc;IACjC,oEAAoE;IACpE,sDAAsD;IACtD,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,YAAY,GAAG,CAAC,CAAC;IACrB,IAAI,SAAS,GAAG,KAAK,CAAC;IACtB,sBAAsB,CAAC,MAAM,EAAE,CAAC,YAAY,EAAE,oBAAoB,CAAC,EAAE,CAAC,IAAI,EAAE,EAAE;QAC5E,MAAM,MAAM,GAAG,WAAW,CAAC,eAAe,CAAC,IAAI,CAAC,UAAU,EAAE,yBAAyB,CAAC,CAAC,CAAC;QACxF,yEAAyE;QACzE,wEAAwE;QACxE,0EAA0E;QAC1E,yEAAyE;QACzE,yEAAyE;QACzE,0EAA0E;QAC1E,kDAAkD;QAClD,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,oBAAoB;YACxD,CAAC,CAAC,EAAE;YACJ,CAAC,CAAC,kBAAkB,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;iBAChC,GAAG,CAAC,gBAAgB,CAAC;iBACrB,IAAI,CAAC,GAAG,CAAC;iBACT,OAAO,CAAC,YAAY,EAAE,GAAG,CAAC,CAAC;QAClC,IAAI,IAAI,KAAK,EAAE,EAAE,CAAC;YAChB,YAAY,IAAI,MAAM,CAAC;YACvB,OAAO,KAAK,CAAC;QACf,CAAC;QACD,OAAO,YAAY,GAAG,CAAC,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,YAAY,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAC7F,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,OAAO,CAAC,GAAG,MAAM,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC9E,IAAI,YAAY,GAAG,CAAC,IAAI,CAAC,GAAG,MAAM,EAAE,CAAC;YACnC,0EAA0E;YAC1E,SAAS,GAAG,IAAI,CAAC;YACjB,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC,CAAC,CAAC;IACH,OAAO,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;AAC9B,CAAC;AAED,8EAA8E;AAC9E,SAAS,WAAW,CAAC,GAAuB;IAC1C,4EAA4E;IAC5E,yEAAyE;IACzE,mCAAmC;IACnC,MAAM,CAAC,GAAG,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,iBAAiB,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACvE,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAC9C,CAAC"}