@unisim/doc 0.1.0

package/dist/read/odt.d.ts

@@ -0,0 +1,3 @@
+import { type RichDoc } from '../model.js';
+export declare function readOdt(file: File): Promise<RichDoc>;
+//# sourceMappingURL=odt.d.ts.map

package/dist/read/odt.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"odt.d.ts","sourceRoot":"","sources":["../../src/read/odt.ts"],"names":[],"mappings":"AAoBA,OAAO,EAEsB,KAAK,OAAO,EACxC,MAAM,aAAa,CAAA;AAmBpB,wBAAsB,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAyB1D"}

package/dist/read/odt.js

@@ -0,0 +1,359 @@
+// ---------------------------------------------------------------------------
+// ODT (OpenDocument Text) → RichDoc.
+//
+// LibreOffice's and OpenOffice's native format, and what Google Docs gives you
+// if you ask for anything other than .docx. Structurally it is the same bargain
+// as DOCX — a ZIP of XML — which is why it cost one file rather than a rewrite:
+// the reader below is the same shape as `docx.ts`, over a vocabulary that is
+// mostly nicer.
+//
+// TWO PLACES IT IS GENUINELY EASIER THAN DOCX
+// -------------------------------------------
+//   * `<text:h text:outline-level="2">` says it is a heading and how deep,
+//     in the element itself. No style table, no three-way heuristic.
+//   * `<text:list>` is a real element that CONTAINS its items, rather than
+//     DOCX's run of sibling paragraphs that happen to share a numbering id.
+//
+// And one where it is worse: whether a list is bulleted or numbered lives in
+// the list's STYLE, in a different part of the file, so that lookup comes back.
+// ---------------------------------------------------------------------------
+import { addNotice, mergeRuns, tidy, } from '../model.js';
+import { ZipArchive } from '../unzip.js';
+import { attr, child, children, descendant, descendants, parseXml } from '../xml.js';
+const OFFICE = 'urn:oasis:names:tc:opendocument:xmlns:office:1.0';
+const TEXT = 'urn:oasis:names:tc:opendocument:xmlns:text:1.0';
+const TABLE = 'urn:oasis:names:tc:opendocument:xmlns:table:1.0';
+const DRAW = 'urn:oasis:names:tc:opendocument:xmlns:drawing:1.0';
+const STYLE = 'urn:oasis:names:tc:opendocument:xmlns:style:1.0';
+const FO = 'urn:oasis:names:tc:opendocument:xmlns:xsl-fo-compatible:1.0';
+const XLINK = 'http://www.w3.org/1999/xlink';
+const DC = 'http://purl.org/dc/elements/1.1/';
+const META = 'urn:oasis:names:tc:opendocument:xmlns:meta:1.0';
+const IMAGE_MIME = {
+    png: 'image/png', jpg: 'image/jpeg', jpeg: 'image/jpeg', gif: 'image/gif',
+    bmp: 'image/bmp', webp: 'image/webp', svg: 'image/svg+xml',
+};
+export async function readOdt(file) {
+    const zip = await ZipArchive.open(file);
+    const contentXml = await zip.textOf('content.xml');
+    if (!contentXml)
+        throw new Error('This isn’t an OpenDocument file inside — `content.xml` is missing.');
+    const doc = { blocks: [], notices: [] };
+    const content = parseXml(contentXml, 'content');
+    const styles = readTextStyles(content, await zip.textOf('styles.xml'));
+    await readMeta(zip, doc);
+    const body = descendant(content, OFFICE, 'text');
+    if (!body)
+        throw new Error('This OpenDocument file has no text body.');
+    const ctx = { zip, styles, breakBefore: pageBreakStyles.get(content) ?? new Set(), doc };
+    await readBlocks(body, ctx, doc.blocks);
+    if (descendant(content, TEXT, 'note')) {
+        addNotice(doc, 'Footnotes were left out — this repaginates, so their page anchors no longer apply.');
+    }
+    if (descendant(content, DRAW, 'object')) {
+        addNotice(doc, 'A chart or embedded object was left out — those aren’t pictures, and can’t be drawn here.');
+    }
+    return tidy(doc);
+}
+/**
+ * Paragraph style names carrying `fo:break-before="page"`.
+ *
+ * Kept in a module-level map keyed by the content document for the same reason
+ * `spanFormats` is: `readTextStyles` scans both style parts in one pass and can
+ * only return one thing.
+ */
+const pageBreakStyles = new WeakMap();
+async function readBlocks(root, ctx, out) {
+    for (const node of Array.from(root.childNodes)) {
+        if (node.nodeType !== 1)
+            continue;
+        const el = node;
+        if (el.namespaceURI === TEXT) {
+            switch (el.localName) {
+                case 'h': {
+                    const raw = Number(attr(el, TEXT, 'outline-level') ?? '1');
+                    const level = Math.min(4, Math.max(1, Number.isFinite(raw) ? raw : 1));
+                    out.push({ kind: 'heading', level, runs: mergeRuns(await readRuns(el, ctx)) });
+                    continue;
+                }
+                case 'p': {
+                    const runs = mergeRuns(await readRuns(el, ctx));
+                    const images = await readImages(el, ctx);
+                    const styleName = attr(el, TEXT, 'style-name') ?? '';
+                    // ⚠️ A HARD page break in ODF is a PARAGRAPH PROPERTY, not an element:
+                    // `fo:break-before="page"` on the style this paragraph uses. Only the
+                    // *soft* break — where LibreOffice's own layout happened to land last
+                    // time — appears in the body, and that one is ignored. Reading only
+                    // the body therefore loses every deliberate page break in the file.
+                    if (ctx.breakBefore.has(styleName))
+                        out.push({ kind: 'pagebreak' });
+                    if (runs.length) {
+                        out.push(/quotation|blockquote/i.test(styleName)
+                            ? { kind: 'quote', runs }
+                            : { kind: 'paragraph', runs });
+                    }
+                    else if (!images.length) {
+                        out.push({ kind: 'paragraph', runs: [{ text: ' ' }] });
+                    }
+                    out.push(...images);
+                    continue;
+                }
+                case 'list':
+                case 'numbered-paragraph': {
+                    const items = [];
+                    await readListItems(el, ctx, items, 0);
+                    if (items.length) {
+                        const styleName = attr(el, TEXT, 'style-name') ?? '';
+                        out.push({ kind: 'list', ordered: ctx.styles.get(styleName) ?? false, items });
+                    }
+                    continue;
+                }
+                case 'section':
+                case 'list-header':
+                    await readBlocks(el, ctx, out);
+                    continue;
+                case 'soft-page-break':
+                    // A RENDERING artefact — where LibreOffice's own layout happened to
+                    // break the page last time it was opened, not an instruction. Honouring
+                    // it would scatter page breaks through a document that has none.
+                    continue;
+                default:
+                    continue;
+            }
+        }
+        if (el.namespaceURI === TABLE && el.localName === 'table') {
+            out.push(await readTable(el, ctx));
+            continue;
+        }
+        if (el.namespaceURI === TEXT && el.localName === 'table-of-content')
+            continue;
+    }
+}
+/** Items, recursing into nested lists so indentation survives. */
+async function readListItems(list, ctx, out, level) {
+    for (const item of children(list, TEXT, 'list-item')) {
+        for (const node of Array.from(item.childNodes)) {
+            if (node.nodeType !== 1)
+                continue;
+            const el = node;
+            if (el.namespaceURI !== TEXT)
+                continue;
+            if (el.localName === 'p' || el.localName === 'h') {
+                out.push({ runs: mergeRuns(await readRuns(el, ctx)), level });
+            }
+            else if (el.localName === 'list') {
+                await readListItems(el, ctx, out, Math.min(4, level + 1));
+            }
+        }
+    }
+}
+/**
+ * Inline content.
+ *
+ * The one thing worth knowing: ODF does not put runs of spaces in the text.
+ * `<text:s text:c="4"/>` means four of them, because XML would collapse the
+ * literal characters. Miss it and indented code inside a document loses its
+ * shape entirely.
+ */
+async function readRuns(parent, ctx, inherited = {}) {
+    const out = [];
+    for (const node of Array.from(parent.childNodes)) {
+        if (node.nodeType === 3) {
+            out.push({ ...inherited, text: node.nodeValue ?? '' });
+            continue;
+        }
+        if (node.nodeType !== 1)
+            continue;
+        const el = node;
+        if (el.namespaceURI !== TEXT)
+            continue;
+        switch (el.localName) {
+            case 'span':
+                out.push(...(await readRuns(el, ctx, { ...inherited, ...spanFormat(el) })));
+                break;
+            case 'a': {
+                const href = attr(el, XLINK, 'href') ?? undefined;
+                // Same rule as DOCX: only a link that leaves the document survives.
+                const external = href && !href.startsWith('#') ? href : undefined;
+                out.push(...(await readRuns(el, ctx, { ...inherited, link: external })));
+                break;
+            }
+            case 's': {
+                const count = Number(attr(el, TEXT, 'c') ?? '1');
+                out.push({ ...inherited, text: ' '.repeat(Number.isFinite(count) ? Math.min(count, 200) : 1) });
+                break;
+            }
+            case 'tab':
+                out.push({ ...inherited, text: '  ' });
+                break;
+            case 'line-break':
+                out.push({ ...inherited, text: '\n' });
+                break;
+            case 'note':
+                // Skipped deliberately — see the notice raised in `readOdt`.
+                break;
+            case 'bookmark':
+            case 'bookmark-start':
+            case 'bookmark-end':
+            case 'sequence-decls':
+                break;
+            default:
+                out.push(...(await readRuns(el, ctx, inherited)));
+                break;
+        }
+    }
+    return out;
+}
+/**
+ * The formatting a `<text:span>` carries.
+ *
+ * ODF holds it in an automatic style in the same file, so the span names a
+ * style and the style names the weight. `styleFormats` is built once up front
+ * because a document of any size names the same handful over and over.
+ */
+const spanFormats = new WeakMap();
+function spanFormat(el) {
+    const name = attr(el, TEXT, 'style-name');
+    if (!name)
+        return {};
+    return spanFormats.get(el.ownerDocument)?.get(name) ?? {};
+}
+/**
+ * Read both style parts.
+ *
+ * Two maps in one pass, keyed differently so they can share a return: list
+ * style name → is-numbered, and (in the WeakMap above) text style name → run
+ * formatting. Automatic styles live in `content.xml`, named ones in
+ * `styles.xml`, and a document uses both.
+ */
+function readTextStyles(content, stylesXml) {
+    const ordered = new Map();
+    const runs = new Map();
+    const breaks = new Set();
+    const scan = (doc) => {
+        for (const style of descendants(doc, TEXT, 'list-style')) {
+            const name = attr(style, STYLE, 'name');
+            if (!name)
+                continue;
+            // A numbered list declares `<text:list-level-style-number>`; a bulleted
+            // one declares `…-bullet`. Level 1 decides for the whole list.
+            ordered.set(name, descendant(style, TEXT, 'list-level-style-number') !== null);
+        }
+        for (const style of descendants(doc, STYLE, 'style')) {
+            const name = attr(style, STYLE, 'name');
+            if (!name)
+                continue;
+            if (attr(style, STYLE, 'family') === 'paragraph') {
+                const paragraphProps = child(style, STYLE, 'paragraph-properties');
+                // 'page' and 'column' are the two values; only 'page' is a page break.
+                if (paragraphProps && attr(paragraphProps, FO, 'break-before') === 'page') {
+                    breaks.add(name);
+                }
+            }
+            if (attr(style, STYLE, 'family') !== 'text')
+                continue;
+            const props = child(style, STYLE, 'text-properties');
+            if (!props)
+                continue;
+            const format = {};
+            if ((attr(props, FO, 'font-weight') ?? '') === 'bold')
+                format.bold = true;
+            if (/italic|oblique/.test(attr(props, FO, 'font-style') ?? ''))
+                format.italic = true;
+            const underline = attr(props, STYLE, 'text-underline-style') ?? 'none';
+            if (underline !== 'none')
+                format.underline = true;
+            const strike = attr(props, STYLE, 'text-line-through-style') ?? 'none';
+            if (strike !== 'none')
+                format.strike = true;
+            const family = attr(props, STYLE, 'font-name') ?? attr(props, FO, 'font-family') ?? '';
+            if (/mono|courier|consolas|menlo/i.test(family))
+                format.code = true;
+            if (Object.keys(format).length)
+                runs.set(name, format);
+        }
+    };
+    scan(content);
+    // `styles.xml` holds the NAMED styles a document defines; `content.xml` holds
+    // the AUTOMATIC ones LibreOffice generates per paragraph. A file uses both,
+    // and a page break almost always lands on an automatic style — so scanning
+    // only the named part finds nothing.
+    if (stylesXml)
+        scan(parseXml(stylesXml, 'styles'));
+    // Both parts feed the same maps, keyed off `content`'s document, because that
+    // is the one every element the readers touch actually belongs to.
+    spanFormats.set(content, runs);
+    pageBreakStyles.set(content, breaks);
+    return ordered;
+}
+async function readTable(el, ctx) {
+    const rows = [];
+    // `table:table-header-rows` wraps the header when there is one; its rows are
+    // ordinary `table-row`s one level down, so both are collected.
+    const rowElements = [
+        ...descendants(el, TABLE, 'table-row'),
+    ];
+    const headerRows = new Set(children(el, TABLE, 'table-header-rows').flatMap((h) => descendants(h, TABLE, 'table-row')));
+    let headerCount = 0;
+    for (const tr of rowElements) {
+        const cells = [];
+        for (const tc of children(tr, TABLE, 'table-cell')) {
+            const runs = [];
+            for (const p of [...children(tc, TEXT, 'p'), ...children(tc, TEXT, 'h')]) {
+                if (runs.length)
+                    runs.push({ text: '\n' });
+                runs.push(...(await readRuns(p, ctx)));
+            }
+            // A repeated cell is how ODF stores a run of identical ones — usually the
+            // empty tail of a row. Expanded, but capped: a spreadsheet-shaped ODT can
+            // claim 16,384 repeats of one blank cell, and honouring that literally
+            // builds a table nothing can lay out.
+            const repeat = Math.min(Number(attr(tc, TABLE, 'number-columns-repeated') ?? '1') || 1, 64);
+            for (let i = 0; i < repeat; i += 1)
+                cells.push(mergeRuns(runs));
+        }
+        if (cells.length) {
+            rows.push(cells);
+            if (headerRows.has(tr))
+                headerCount += 1;
+        }
+    }
+    if (!rows.length)
+        return { kind: 'paragraph', runs: [{ text: ' ' }] };
+    const useHeader = headerCount > 0 || rows.length > 1;
+    return {
+        kind: 'table',
+        header: useHeader ? rows[0] : null,
+        rows: useHeader ? rows.slice(1) : rows,
+    };
+}
+async function readImages(el, ctx) {
+    const out = [];
+    for (const image of descendants(el, DRAW, 'image')) {
+        const href = attr(image, XLINK, 'href');
+        if (!href || href.startsWith('http'))
+            continue;
+        const path = href.replace(/^\.\//, '');
+        const ext = path.split('.').pop()?.toLowerCase() ?? '';
+        const blob = await ctx.zip.blobOf(path, IMAGE_MIME[ext] ?? 'application/octet-stream');
+        if (blob) {
+            const frame = image.parentElement;
+            const alt = frame ? descendant(frame, SVG_NS, 'title')?.textContent ?? '' : '';
+            out.push({ kind: 'image', file: blob, alt });
+        }
+    }
+    return out;
+}
+const SVG_NS = 'urn:oasis:names:tc:opendocument:xmlns:svg-compatible:1.0';
+async function readMeta(zip, doc) {
+    const xml = await zip.textOf('meta.xml');
+    if (!xml)
+        return;
+    const meta = parseXml(xml, 'metadata');
+    doc.title = descendant(meta, DC, 'title')?.textContent?.trim() || undefined;
+    doc.author =
+        descendant(meta, META, 'initial-creator')?.textContent?.trim() ||
+            descendant(meta, DC, 'creator')?.textContent?.trim() ||
+            undefined;
+}
+//# sourceMappingURL=odt.js.map

package/dist/read/odt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"odt.js","sourceRoot":"","sources":["../../src/read/odt.ts"],"names":[],"mappings":"AAAA,8EAA8E;AAC9E,qCAAqC;AACrC,EAAE;AACF,+EAA+E;AAC/E,gFAAgF;AAChF,gFAAgF;AAChF,6EAA6E;AAC7E,gBAAgB;AAChB,EAAE;AACF,8CAA8C;AAC9C,8CAA8C;AAC9C,2EAA2E;AAC3E,qEAAqE;AACrE,2EAA2E;AAC3E,4EAA4E;AAC5E,EAAE;AACF,6EAA6E;AAC7E,gFAAgF;AAChF,8EAA8E;AAE9E,OAAO,EACL,SAAS,EAAE,SAAS,EAAE,IAAI,GAE3B,MAAM,aAAa,CAAA;AACpB,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAA;AACxC,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,EAAE,UAAU,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AAEpF,MAAM,MAAM,GAAG,kDAAkD,CAAA;AACjE,MAAM,IAAI,GAAG,gDAAgD,CAAA;AAC7D,MAAM,KAAK,GAAG,iDAAiD,CAAA;AAC/D,MAAM,IAAI,GAAG,mDAAmD,CAAA;AAChE,MAAM,KAAK,GAAG,iDAAiD,CAAA;AAC/D,MAAM,EAAE,GAAG,6DAA6D,CAAA;AACxE,MAAM,KAAK,GAAG,8BAA8B,CAAA;AAC5C,MAAM,EAAE,GAAG,kCAAkC,CAAA;AAC7C,MAAM,IAAI,GAAG,gDAAgD,CAAA;AAE7D,MAAM,UAAU,GAA2B;IACzC,GAAG,EAAE,WAAW,EAAE,GAAG,EAAE,YAAY,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,WAAW;IACzE,GAAG,EAAE,WAAW,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,eAAe;CAC3D,CAAA;AAED,MAAM,CAAC,KAAK,UAAU,OAAO,CAAC,IAAU;IACtC,MAAM,GAAG,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEvC,MAAM,UAAU,GAAG,MAAM,GAAG,CAAC,MAAM,CAAC,aAAa,CAAC,CAAA;IAClD,IAAI,CAAC,UAAU;QAAE,MAAM,IAAI,KAAK,CAAC,oEAAoE,CAAC,CAAA;IAEtG,MAAM,GAAG,GAAY,EAAE,MAAM,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,EAAE,CAAA;IAChD,MAAM,OAAO,GAAG,QAAQ,CAAC,UAAU,EAAE,SAAS,CAAC,CAAA;IAC/C,MAAM,MAAM,GAAG,cAAc,CAAC,OAAO,EAAE,MAAM,GAAG,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,CAAA;IACtE,MAAM,QAAQ,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;IAExB,MAAM,IAAI,GAAG,UAAU,CAAC,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,CAAA;IAChD,IAAI,CAAC,IAAI;QAAE,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC,CAAA;IAEtE,MAAM,GAAG,GAAQ,EAAE,GAAG,EAAE,MAAM,EAAE,WAAW,EAAE,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,GAAG,EAAE,EAAE,GAAG,EAAE,CAAA;IAC7F,MAAM,UAAU,CAAC,IAAI,EAAE,GAAG,EAAE,GAAG,CAAC,MAAM,CAAC,CAAA;IAEvC,IAAI,UAAU,CAAC,OAAO,EAAE,IAAI,EAAE,MAAM,CAAC,EAAE,CAAC;QACtC,SAAS,CAAC,GAAG,EAAE,oFAAoF,CAAC,CAAA;IACtG,CAAC;IACD,IAAI,UAAU,CAAC,OAAO,EAAE,IAAI,EAAE,QAAQ,CAAC,EAAE,CAAC;QACxC,SAAS,CAAC,GAAG,EAAE,2FAA2F,CAAC,CAAA;IAC7G,CAAC;IAED,OAAO,IAAI,CAAC,GAAG,CAAC,CAAA;AAClB,CAAC;AAWD;;;;;;GAMG;AACH,MAAM,eAAe,GAAG,IAAI,OAAO,EAAyB,CAAA;AAE5D,KAAK,UAAU,UAAU,CAAC,IAAa,EAAE,GAAQ,EAAE,GAAY;IAC7D,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;QAC/C,IAAI,IAAI,CAAC,QAAQ,KAAK,CAAC;YAAE,SAAQ;QACjC,MAAM,EAAE,GAAG,IAAe,CAAA;QAE1B,IAAI,EAAE,CAAC,YAAY,KAAK,IAAI,EAAE,CAAC;YAC7B,QAAQ,EAAE,CAAC,SAAS,EAAE,CAAC;gBACrB,KAAK,GAAG,CAAC,CAAC,CAAC;oBACT,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,eAAe,CAAC,IAAI,GAAG,CAAC,CAAA;oBAC1D,MAAM,KAAK,GAAI,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAmB,CAAA;oBACzF,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAA;oBAC9E,SAAQ;gBACV,CAAC;gBACD,KAAK,GAAG,CAAC,CAAC,CAAC;oBACT,MAAM,IAAI,GAAG,SAAS,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,CAAA;oBAC/C,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,EAAE,EAAE,GAAG,CAAC,CAAA;oBACxC,MAAM,SAAS,GAAG,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,YAAY,CAAC,IAAI,EAAE,CAAA;oBACpD,uEAAuE;oBACvE,sEAAsE;oBACtE,sEAAsE;oBACtE,oEAAoE;oBACpE,oEAAoE;oBACpE,IAAI,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,SAAS,CAAC;wBAAE,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAA;oBACnE,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;wBAChB,GAAG,CAAC,IAAI,CACN,uBAAuB,CAAC,IAAI,CAAC,SAAS,CAAC;4BACrC,CAAC,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE;4BACzB,CAAC,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAChC,CAAA;oBACH,CAAC;yBAAM,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;wBAC1B,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAA;oBACxD,CAAC;oBACD,GAAG,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,CAAA;oBACnB,SAAQ;gBACV,CAAC;gBACD,KAAK,MAAM,CAAC;gBACZ,KAAK,oBAAoB,CAAC,CAAC,CAAC;oBAC1B,MAAM,KAAK,GAAe,EAAE,CAAA;oBAC5B,MAAM,aAAa,CAAC,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,CAAC,CAAA;oBACtC,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;wBACjB,MAAM,SAAS,GAAG,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,YAAY,CAAC,IAAI,EAAE,CAAA;wBACpD,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,KAAK,EAAE,KAAK,EAAE,CAAC,CAAA;oBAChF,CAAC;oBACD,SAAQ;gBACV,CAAC;gBACD,KAAK,SAAS,CAAC;gBACf,KAAK,aAAa;oBAChB,MAAM,UAAU,CAAC,EAAE,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;oBAC9B,SAAQ;gBACV,KAAK,iBAAiB;oBACpB,oEAAoE;oBACpE,wEAAwE;oBACxE,iEAAiE;oBACjE,SAAQ;gBACV;oBACE,SAAQ;YACZ,CAAC;QACH,CAAC;QAED,IAAI,EAAE,CAAC,YAAY,KAAK,KAAK,IAAI,EAAE,CAAC,SAAS,KAAK,OAAO,EAAE,CAAC;YAC1D,GAAG,CAAC,IAAI,CAAC,MAAM,SAAS,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,CAAA;YAClC,SAAQ;QACV,CAAC;QAED,IAAI,EAAE,CAAC,YAAY,KAAK,IAAI,IAAI,EAAE,CAAC,SAAS,KAAK,kBAAkB;YAAE,SAAQ;IAC/E,CAAC;AACH,CAAC;AAED,kEAAkE;AAClE,KAAK,UAAU,aAAa,CAAC,IAAa,EAAE,GAAQ,EAAE,GAAe,EAAE,KAAa;IAClF,KAAK,MAAM,IAAI,IAAI,QAAQ,CAAC,IAAI,EAAE,IAAI,EAAE,WAAW,CAAC,EAAE,CAAC;QACrD,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;YAC/C,IAAI,IAAI,CAAC,QAAQ,KAAK,CAAC;gBAAE,SAAQ;YACjC,MAAM,EAAE,GAAG,IAAe,CAAA;YAC1B,IAAI,EAAE,CAAC,YAAY,KAAK,IAAI;gBAAE,SAAQ;YACtC,IAAI,EAAE,CAAC,SAAS,KAAK,GAAG,IAAI,EAAE,CAAC,SAAS,KAAK,GAAG,EAAE,CAAC;gBACjD,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,SAAS,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAA;YAC/D,CAAC;iBAAM,IAAI,EAAE,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;gBACnC,MAAM,aAAa,CAAC,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC,CAAA;YAC3D,CAAC;QACH,CAAC;IACH,CAAC;AACH,CAAC;AAED;;;;;;;GAOG;AACH,KAAK,UAAU,QAAQ,CAAC,MAAe,EAAE,GAAQ,EAAE,YAA0B,EAAE;IAC7E,MAAM,GAAG,GAAU,EAAE,CAAA;IAErB,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,CAAC,EAAE,CAAC;QACjD,IAAI,IAAI,CAAC,QAAQ,KAAK,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,SAAS,EAAE,IAAI,EAAE,IAAI,CAAC,SAAS,IAAI,EAAE,EAAE,CAAC,CAAA;YACtD,SAAQ;QACV,CAAC;QACD,IAAI,IAAI,CAAC,QAAQ,KAAK,CAAC;YAAE,SAAQ;QACjC,MAAM,EAAE,GAAG,IAAe,CAAA;QAC1B,IAAI,EAAE,CAAC,YAAY,KAAK,IAAI;YAAE,SAAQ;QAEtC,QAAQ,EAAE,CAAC,SAAS,EAAE,CAAC;YACrB,KAAK,MAAM;gBACT,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,EAAE,EAAE,GAAG,SAAS,EAAE,GAAG,UAAU,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;gBAC3E,MAAK;YACP,KAAK,GAAG,CAAC,CAAC,CAAC;gBACT,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,IAAI,SAAS,CAAA;gBACjD,oEAAoE;gBACpE,MAAM,QAAQ,GAAG,IAAI,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAA;gBACjE,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,EAAE,EAAE,GAAG,SAAS,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,CAAA;gBACxE,MAAK;YACP,CAAC;YACD,KAAK,GAAG,CAAC,CAAC,CAAC;gBACT,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,GAAG,CAAC,CAAA;gBAChD,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,SAAS,EAAE,IAAI,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;gBAC/F,MAAK;YACP,CAAC;YACD,KAAK,KAAK;gBACR,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;gBACtC,MAAK;YACP,KAAK,YAAY;gBACf,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;gBACtC,MAAK;YACP,KAAK,MAAM;gBACT,6DAA6D;gBAC7D,MAAK;YACP,KAAK,UAAU,CAAC;YAChB,KAAK,gBAAgB,CAAC;YACtB,KAAK,cAAc,CAAC;YACpB,KAAK,gBAAgB;gBACnB,MAAK;YACP;gBACE,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,QAAQ,CAAC,EAAE,EAAE,GAAG,EAAE,SAAS,CAAC,CAAC,CAAC,CAAA;gBACjD,MAAK;QACT,CAAC;IACH,CAAC;IAED,OAAO,GAAG,CAAA;AACZ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,WAAW,GAAG,IAAI,OAAO,EAAuC,CAAA;AAEtE,SAAS,UAAU,CAAC,EAAW;IAC7B,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,EAAE,IAAI,EAAE,YAAY,CAAC,CAAA;IACzC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAA;IACpB,OAAO,WAAW,CAAC,GAAG,CAAC,EAAE,CAAC,aAAa,CAAC,EAAE,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA;AAC3D,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,cAAc,CAAC,OAAiB,EAAE,SAAwB;IACjE,MAAM,OAAO,GAAG,IAAI,GAAG,EAAmB,CAAA;IAC1C,MAAM,IAAI,GAAG,IAAI,GAAG,EAAwB,CAAA;IAC5C,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAA;IAEhC,MAAM,IAAI,GAAG,CAAC,GAAa,EAAE,EAAE;QAC7B,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,GAAG,EAAE,IAAI,EAAE,YAAY,CAAC,EAAE,CAAC;YACzD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,CAAC,CAAA;YACvC,IAAI,CAAC,IAAI;gBAAE,SAAQ;YACnB,wEAAwE;YACxE,+DAA+D;YAC/D,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,UAAU,CAAC,KAAK,EAAE,IAAI,EAAE,yBAAyB,CAAC,KAAK,IAAI,CAAC,CAAA;QAChF,CAAC;QAED,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,GAAG,EAAE,KAAK,EAAE,OAAO,CAAC,EAAE,CAAC;YACrD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,CAAC,CAAA;YACvC,IAAI,CAAC,IAAI;gBAAE,SAAQ;YACnB,IAAI,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,WAAW,EAAE,CAAC;gBACjD,MAAM,cAAc,GAAG,KAAK,CAAC,KAAK,EAAE,KAAK,EAAE,sBAAsB,CAAC,CAAA;gBAClE,uEAAuE;gBACvE,IAAI,cAAc,IAAI,IAAI,CAAC,cAAc,EAAE,EAAE,EAAE,cAAc,CAAC,KAAK,MAAM,EAAE,CAAC;oBAC1E,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;gBAClB,CAAC;YACH,CAAC;YACD,IAAI,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,MAAM;gBAAE,SAAQ;YACrD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,EAAE,KAAK,EAAE,iBAAiB,CAAC,CAAA;YACpD,IAAI,CAAC,KAAK;gBAAE,SAAQ;YACpB,MAAM,MAAM,GAAiB,EAAE,CAAA;YAC/B,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE,aAAa,CAAC,IAAI,EAAE,CAAC,KAAK,MAAM;gBAAE,MAAM,CAAC,IAAI,GAAG,IAAI,CAAA;YACzE,IAAI,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE,YAAY,CAAC,IAAI,EAAE,CAAC;gBAAE,MAAM,CAAC,MAAM,GAAG,IAAI,CAAA;YACpF,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,sBAAsB,CAAC,IAAI,MAAM,CAAA;YACtE,IAAI,SAAS,KAAK,MAAM;gBAAE,MAAM,CAAC,SAAS,GAAG,IAAI,CAAA;YACjD,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,yBAAyB,CAAC,IAAI,MAAM,CAAA;YACtE,IAAI,MAAM,KAAK,MAAM;gBAAE,MAAM,CAAC,MAAM,GAAG,IAAI,CAAA;YAC3C,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,WAAW,CAAC,IAAI,IAAI,CAAC,KAAK,EAAE,EAAE,EAAE,aAAa,CAAC,IAAI,EAAE,CAAA;YACtF,IAAI,8BAA8B,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,MAAM,CAAC,IAAI,GAAG,IAAI,CAAA;YACnE,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,MAAM;gBAAE,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;QACxD,CAAC;IACH,CAAC,CAAA;IAED,IAAI,CAAC,OAAO,CAAC,CAAA;IACb,8EAA8E;IAC9E,4EAA4E;IAC5E,2EAA2E;IAC3E,qCAAqC;IACrC,IAAI,SAAS;QAAE,IAAI,CAAC,QAAQ,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC,CAAA;IAElD,8EAA8E;IAC9E,kEAAkE;IAClE,WAAW,CAAC,GAAG,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;IAC9B,eAAe,CAAC,GAAG,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IACpC,OAAO,OAAO,CAAA;AAChB,CAAC;AAED,KAAK,UAAU,SAAS,CAAC,EAAW,EAAE,GAAQ;IAC5C,MAAM,IAAI,GAAc,EAAE,CAAA;IAC1B,6EAA6E;IAC7E,+DAA+D;IAC/D,MAAM,WAAW,GAAG;QAClB,GAAG,WAAW,CAAC,EAAE,EAAE,KAAK,EAAE,WAAW,CAAC;KACvC,CAAA;IACD,MAAM,UAAU,GAAG,IAAI,GAAG,CACxB,QAAQ,CAAC,EAAE,EAAE,KAAK,EAAE,mBAAmB,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,WAAW,CAAC,CAAC,EAAE,KAAK,EAAE,WAAW,CAAC,CAAC,CAC5F,CAAA;IAED,IAAI,WAAW,GAAG,CAAC,CAAA;IACnB,KAAK,MAAM,EAAE,IAAI,WAAW,EAAE,CAAC;QAC7B,MAAM,KAAK,GAAY,EAAE,CAAA;QACzB,KAAK,MAAM,EAAE,IAAI,QAAQ,CAAC,EAAE,EAAE,KAAK,EAAE,YAAY,CAAC,EAAE,CAAC;YACnD,MAAM,IAAI,GAAU,EAAE,CAAA;YACtB,KAAK,MAAM,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,EAAE,GAAG,QAAQ,CAAC,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC;gBACzE,IAAI,IAAI,CAAC,MAAM;oBAAE,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;gBAC1C,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAA;YACxC,CAAC;YACD,0EAA0E;YAC1E,0EAA0E;YAC1E,uEAAuE;YACvE,sCAAsC;YACtC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,EAAE,yBAAyB,CAAC,IAAI,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,CAAA;YAC3F,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,IAAI,CAAC;gBAAE,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;QACjE,CAAC;QACD,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;YACjB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YAChB,IAAI,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBAAE,WAAW,IAAI,CAAC,CAAA;QAC1C,CAAC;IACH,CAAC;IAED,IAAI,CAAC,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,EAAE,CAAA;IACrE,MAAM,SAAS,GAAG,WAAW,GAAG,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAA;IACpD,OAAO;QACL,IAAI,EAAE,OAAO;QACb,MAAM,EAAE,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI;QAClC,IAAI,EAAE,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI;KACvC,CAAA;AACH,CAAC;AAED,KAAK,UAAU,UAAU,CAAC,EAAW,EAAE,GAAQ;IAC7C,MAAM,GAAG,GAAY,EAAE,CAAA;IACvB,KAAK,MAAM,KAAK,IAAI,WAAW,CAAC,EAAE,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,CAAC;QACnD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,MAAM,CAAC,CAAA;QACvC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC;YAAE,SAAQ;QAC9C,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAA;QACtC,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,WAAW,EAAE,IAAI,EAAE,CAAA;QACtD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,UAAU,CAAC,GAAG,CAAC,IAAI,0BAA0B,CAAC,CAAA;QACtF,IAAI,IAAI,EAAE,CAAC;YACT,MAAM,KAAK,GAAG,KAAK,CAAC,aAAa,CAAA;YACjC,MAAM,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,WAAW,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YAC9E,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC,CAAA;QAC9C,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAA;AACZ,CAAC;AAED,MAAM,MAAM,GAAG,0DAA0D,CAAA;AAEzE,KAAK,UAAU,QAAQ,CAAC,GAAe,EAAE,GAAY;IACnD,MAAM,GAAG,GAAG,MAAM,GAAG,CAAC,MAAM,CAAC,UAAU,CAAC,CAAA;IACxC,IAAI,CAAC,GAAG;QAAE,OAAM;IAChB,MAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,EAAE,UAAU,CAAC,CAAA;IACtC,GAAG,CAAC,KAAK,GAAG,UAAU,CAAC,IAAI,EAAE,EAAE,EAAE,OAAO,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,SAAS,CAAA;IAC3E,GAAG,CAAC,MAAM;QACR,UAAU,CAAC,IAAI,EAAE,IAAI,EAAE,iBAAiB,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE;YAC9D,UAAU,CAAC,IAAI,EAAE,EAAE,EAAE,SAAS,CAAC,EAAE,WAAW,EAAE,IAAI,EAAE;YACpD,SAAS,CAAA;AACb,CAAC"}

package/dist/read/rtf.d.ts

@@ -0,0 +1,3 @@
+import { type RichDoc } from '../model.js';
+export declare function readRtf(file: File): Promise<RichDoc>;
+//# sourceMappingURL=rtf.d.ts.map

package/dist/read/rtf.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"rtf.d.ts","sourceRoot":"","sources":["../../src/read/rtf.ts"],"names":[],"mappings":"AAwBA,OAAO,EAAyD,KAAK,OAAO,EAAY,MAAM,aAAa,CAAA;AA0B3G,wBAAsB,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,CAgP1D"}