teamshare-bridge 0.21.13 → 0.21.14

package/dist/lib/documents/docx-template.js

@@ -0,0 +1,851 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.buildDocxFromTemplate = buildDocxFromTemplate;
+/**
+ * Template merge for Word commits (2026-09).
+ *
+ * The browser cannot rewrite a `.docx` losslessly, and rebuilding one from
+ * HTML throws away the document's identity: its theme, named styles, styles
+ * inheritance, numbering definitions, headers/footers and section properties
+ * (page size, margins, columns). So we do not rebuild - we take the ORIGINAL
+ * package, replace the children of `<w:body>`, and leave every other part
+ * byte-identical.
+ *
+ *   original.docx (zip)
+ *   ├── word/document.xml        <- ONLY the body children are rewritten
+ *   ├── word/theme/theme1.xml    untouched -> theme survives
+ *   ├── word/styles.xml          untouched -> named styles + inheritance survive
+ *   ├── word/numbering.xml       untouched -> list definitions survive
+ *   ├── word/header1.xml, footer1.xml  untouched -> letterhead survives
+ *   ├── word/settings.xml, fontTable.xml, media/*  untouched
+ *   └── w:sectPr (tail of document.xml)            preserved -> page setup survives
+ *
+ * The generated paragraphs reference the DOCUMENT'S OWN styleIds (discovered
+ * from styles.xml by style name), so "Heading 1" stays the document's
+ * Heading 1 and keeps its theme fonts and colours.
+ *
+ * Anything the merge would have to drop is detected beforehand
+ * (`docx-inspect.ts`) and the document is refused instead - we never edit a
+ * file we would silently damage.
+ */
+const xml_escape_1 = require("./xml-escape");
+const MAIN_PART = 'word/document.xml';
+const STYLES_PART = 'word/styles.xml';
+const NUMBERING_PART = 'word/numbering.xml';
+const RELS_PART = 'word/_rels/document.xml.rels';
+const CONTENT_TYPES_PART = '[Content_Types].xml';
+const EMU_PER_PX = 9525;
+// ===========================================================================
+// Minimal HTML parsing (the TipTap / mammoth subset)
+// ===========================================================================
+const VOID_TAGS = new Set([
+    'br',
+    'img',
+    'hr',
+    'meta',
+    'link',
+    'input',
+    'col',
+    'source',
+    'area',
+    'base',
+]);
+function parseAttrs(raw) {
+    const attrs = {};
+    const re = /([a-zA-Z_:][-a-zA-Z0-9_:.]*)\s*=\s*("([^"]*)"|'([^']*)'|([^\s"'>]+))/g;
+    let m;
+    while ((m = re.exec(raw))) {
+        attrs[m[1].toLowerCase()] = m[3] ?? m[4] ?? m[5] ?? '';
+    }
+    return attrs;
+}
+function unescapeHtml(text) {
+    return text
+        .replace(/&nbsp;/g, ' ')
+        .replace(/&lt;/g, '<')
+        .replace(/&gt;/g, '>')
+        .replace(/&quot;/g, '"')
+        .replace(/&#39;/g, "'")
+        .replace(/&amp;/g, '&');
+}
+/**
+ * Tiny stack-based parser for well-formed HTML. TipTap serializes a
+ * ProseMirror document (always balanced) and mammoth emits balanced HTML, so
+ * a full parser is unnecessary.
+ */
+function parseHtml(html) {
+    const root = { tag: '#root', attrs: {}, children: [] };
+    const stack = [root];
+    const re = /<!--[\s\S]*?-->|<[^>]*>|[^<]+/g;
+    let m;
+    while ((m = re.exec(html))) {
+        const token = m[0];
+        if (token.startsWith('<!--'))
+            continue;
+        if (token.startsWith('</')) {
+            const tag = token.slice(2, -1).trim().toLowerCase();
+            for (let i = stack.length - 1; i > 0; i -= 1) {
+                if (stack[i].tag === tag) {
+                    stack.length = i;
+                    break;
+                }
+            }
+            continue;
+        }
+        if (token.startsWith('<')) {
+            const inner = token.slice(1, -1);
+            const selfClosing = inner.endsWith('/');
+            const body = selfClosing ? inner.slice(0, -1) : inner;
+            const split = body.search(/\s/);
+            const tag = (split === -1 ? body : body.slice(0, split)).toLowerCase();
+            if (!tag || tag === '!' || tag === '?')
+                continue;
+            const node = {
+                tag,
+                attrs: parseAttrs(split === -1 ? '' : body.slice(split)),
+                children: [],
+            };
+            stack[stack.length - 1].children.push(node);
+            if (!selfClosing && !VOID_TAGS.has(tag))
+                stack.push(node);
+            continue;
+        }
+        const text = unescapeHtml(token);
+        stack[stack.length - 1].children.push({
+            tag: '#text',
+            attrs: {},
+            children: [],
+            text,
+        });
+    }
+    return root;
+}
+function parseStyleManifest(stylesXml) {
+    const byName = new Map();
+    let paragraphDefault = null;
+    let tableStyleId = null;
+    if (!stylesXml)
+        return { byName, paragraphDefault, tableStyleId };
+    const styleRe = /<w:style\b([^>]*)>([\s\S]*?)<\/w:style>/g;
+    let m;
+    while ((m = styleRe.exec(stylesXml))) {
+        const head = m[1];
+        const inner = m[2];
+        const type = /w:type="([^"]+)"/.exec(head)?.[1];
+        const styleId = /w:styleId="([^"]+)"/.exec(head)?.[1];
+        if (!styleId)
+            continue;
+        const name = /<w:name\s+w:val="([^"]*)"/.exec(inner)?.[1];
+        if (name)
+            byName.set(name.trim().toLowerCase(), styleId);
+        if (type === 'paragraph' && /w:default="1"/.test(head)) {
+            paragraphDefault = styleId;
+        }
+        if (type === 'table' && !tableStyleId)
+            tableStyleId = styleId;
+        if (type === 'table' && styleId.toLowerCase() === 'tablegrid') {
+            tableStyleId = styleId;
+        }
+    }
+    return { byName, paragraphDefault, tableStyleId };
+}
+/**
+ * A regenerable table-of-contents field. It replaces the original TOC (which
+ * lives in the body and would otherwise be dropped). `w:dirty="true"` makes
+ * Word offer to refresh it on open, so nothing is actually lost - the TOC is
+ * rebuilt from the edited headings.
+ */
+function tocField() {
+    return ('<w:p><w:r><w:fldChar w:fldCharType="begin" w:dirty="true"/></w:r>' +
+        '<w:r><w:instrText xml:space="preserve"> TOC \\o "1-3" \\h \\z \\u </w:instrText></w:r>' +
+        '<w:r><w:fldChar w:fldCharType="separate"/></w:r>' +
+        '<w:r><w:t>Right-click and choose &quot;Update Field&quot; to build the table of contents.</w:t></w:r>' +
+        '<w:r><w:fldChar w:fldCharType="end"/></w:r></w:p>');
+}
+/** True when the body has a table-of-contents field. */
+function hasTocField(documentXml) {
+    const instrs = [
+        ...documentXml.matchAll(/<w:instrText[^>]*>([\s\S]*?)<\/w:instrText>/g),
+    ].map((m) => m[1]);
+    const simple = [
+        ...documentXml.matchAll(/<w:fldSimple[^>]*w:instr="([^"]*)"/g),
+    ].map((m) => m[1]);
+    return [...instrs, ...simple].some((instr) => /^\s*TOC\b/i.test(instr));
+}
+/** Extract the trailing `<w:sectPr>` (page setup + header/footer references). */
+function extractSectPrTail(documentXml) {
+    const bodyStart = documentXml.indexOf('<w:body');
+    if (bodyStart === -1)
+        return '';
+    const body = documentXml.slice(bodyStart);
+    const lastOpen = body.lastIndexOf('<w:sectPr');
+    if (lastOpen === -1)
+        return '';
+    const close = body.indexOf('</w:sectPr>', lastOpen);
+    if (close === -1)
+        return '';
+    return body.slice(lastOpen, close + '</w:sectPr>'.length);
+}
+function parseNumbering(numberingXml) {
+    const info = {
+        bulletNumId: null,
+        orderedNumId: null,
+        abstractNumIds: [],
+        numIds: [],
+    };
+    if (!numberingXml)
+        return info;
+    const abstractRe = /<w:abstractNum\b[^>]*w:abstractNumId="([^"]+)"[^>]*>([\s\S]*?)<\/w:abstractNum>/g;
+    const fmtById = new Map();
+    let m;
+    while ((m = abstractRe.exec(numberingXml))) {
+        info.abstractNumIds.push(m[1]);
+        const fmt = /<w:numFmt\s+w:val="([^"]+)"/.exec(m[2])?.[1];
+        if (fmt)
+            fmtById.set(m[1], fmt);
+    }
+    const numRe = /<w:num\b[^>]*w:numId="([^"]+)"[^>]*>([\s\S]*?)<\/w:num>/g;
+    while ((m = numRe.exec(numberingXml))) {
+        info.numIds.push(m[1]);
+        const abstractId = /<w:abstractNumId\s+w:val="([^"]+)"/.exec(m[2])?.[1];
+        const fmt = abstractId ? fmtById.get(abstractId) : undefined;
+        if (fmt === 'bullet' && !info.bulletNumId)
+            info.bulletNumId = m[1];
+        if (fmt === 'decimal' && !info.orderedNumId)
+            info.orderedNumId = m[1];
+    }
+    return info;
+}
+/** Inject a bullet + decimal list definition when the document has none. */
+function injectListNumbering(numberingXml) {
+    const bulletId = '900001';
+    const orderedId = '900002';
+    const bulletAbstract = '900001';
+    const orderedAbstract = '900002';
+    const lvl = (fmt, text, font) => `<w:lvl w:ilvl="0"><w:start w:val="1"/><w:numFmt w:val="${fmt}"/><w:lvlText w:val="${text}"/><w:lvlJc w:val="left"/><w:pPr><w:ind w:left="720" w:hanging="360"/></w:pPr>${font ? `<w:rPr><w:rFonts w:ascii="${font}" w:hAnsi="${font}"/></w:rPr>` : ''}</w:lvl>`;
+    const abstracts = `<w:abstractNum w:abstractNumId="${bulletAbstract}"><w:multiLevelType w:val="hybridMultilevel"/>${lvl('bullet', '\u2022', 'Symbol')}</w:abstractNum>` +
+        `<w:abstractNum w:abstractNumId="${orderedAbstract}"><w:multiLevelType w:val="hybridMultilevel"/>${lvl('decimal', '%1.')}</w:abstractNum>`;
+    const nums = `<w:num w:numId="${bulletId}"><w:abstractNumId w:val="${bulletAbstract}"/></w:num>` +
+        `<w:num w:numId="${orderedId}"><w:abstractNumId w:val="${orderedAbstract}"/></w:num>`;
+    if (!numberingXml) {
+        return {
+            xml: `<?xml version="1.0" encoding="UTF-8" standalone="yes"?><w:numbering xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">${abstracts}${nums}</w:numbering>`,
+            bulletNumId: bulletId,
+            orderedNumId: orderedId,
+        };
+    }
+    // Schema order: every w:abstractNum must precede the first w:num.
+    const withAbstracts = numberingXml.replace('</w:numbering>', `${abstracts}</w:numbering>`);
+    const numIndex = withAbstracts.indexOf('<w:num ');
+    const out = numIndex === -1
+        ? withAbstracts.replace('</w:numbering>', `${nums}</w:numbering>`)
+        : withAbstracts.slice(0, numIndex) + nums + withAbstracts.slice(numIndex);
+    return { xml: out, bulletNumId: bulletId, orderedNumId: orderedId };
+}
+// ===========================================================================
+// Inline + block rendering
+// ===========================================================================
+function styleOf(node) {
+    const out = {};
+    for (const part of (node.attrs.style ?? '').split(';')) {
+        const [k, v] = part.split(':');
+        if (k && v)
+            out[k.trim().toLowerCase()] = v.trim();
+    }
+    return out;
+}
+function normalizeColor(value) {
+    const hex = value.trim().replace('#', '');
+    if (/^[0-9a-f]{6}$/i.test(hex))
+        return hex.toUpperCase();
+    if (/^[0-9a-f]{3}$/i.test(hex)) {
+        return hex
+            .split('')
+            .map((c) => c + c)
+            .join('')
+            .toUpperCase();
+    }
+    return null;
+}
+function runProperties(fmt) {
+    const parts = [];
+    if (fmt.code) {
+        parts.push('<w:rFonts w:ascii="Consolas" w:hAnsi="Consolas"/>');
+    }
+    if (fmt.bold)
+        parts.push('<w:b/>');
+    if (fmt.italic)
+        parts.push('<w:i/>');
+    if (fmt.underline)
+        parts.push('<w:u w:val="single"/>');
+    if (fmt.strike)
+        parts.push('<w:strike/>');
+    if (fmt.color)
+        parts.push(`<w:color w:val="${fmt.color}"/>`);
+    if (fmt.highlight)
+        parts.push(`<w:highlight w:val="${fmt.highlight}"/>`);
+    return parts.length ? `<w:rPr>${parts.join('')}</w:rPr>` : '';
+}
+function textRun(text, fmt) {
+    if (!text)
+        return '';
+    const preserve = /^\s|\s$/.test(text) ? ' xml:space="preserve"' : '';
+    return `<w:r>${runProperties(fmt)}<w:t${preserve}>${(0, xml_escape_1.escapeXml)(text)}</w:t></w:r>`;
+}
+const HEADING_TAGS = {
+    h1: 1,
+    h2: 2,
+    h3: 3,
+    h4: 4,
+    h5: 5,
+    h6: 6,
+};
+/** Walk inline content, emitting `<w:r>` (+ hyperlinks and drawings). */
+function renderInline(nodes, fmt, ctx) {
+    let out = '';
+    for (const node of nodes) {
+        if (node.tag === '#text') {
+            out += textRun(node.text ?? '', fmt);
+            continue;
+        }
+        const style = styleOf(node);
+        const next = { ...fmt };
+        switch (node.tag) {
+            case 'br':
+                out += `<w:r>${runProperties(fmt)}<w:br/></w:r>`;
+                continue;
+            case 'img':
+                out += renderImage(node, ctx);
+                continue;
+            case 'strong':
+            case 'b':
+                next.bold = true;
+                break;
+            case 'em':
+            case 'i':
+                next.italic = true;
+                break;
+            case 'u':
+                next.underline = true;
+                break;
+            case 's':
+            case 'del':
+            case 'strike':
+                next.strike = true;
+                break;
+            case 'code':
+                next.code = true;
+                break;
+            case 'mark':
+                next.highlight = 'yellow';
+                break;
+            case 'sub':
+            case 'sup':
+                // No editor-level distinction worth regenerating; keep the text.
+                break;
+            case 'span':
+            case 'font': {
+                const color = style['color'] ? normalizeColor(style['color']) : null;
+                if (color)
+                    next.color = color;
+                break;
+            }
+            case 'a': {
+                const href = node.attrs.href;
+                if (href && !href.startsWith('#')) {
+                    const relId = ctx.nextRelId();
+                    ctx.newRels.push({
+                        id: relId,
+                        type: 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink',
+                        target: href,
+                        external: true,
+                    });
+                    out += `<w:hyperlink r:id="${relId}">${renderInline(node.children, next, ctx)}</w:hyperlink>`;
+                    continue;
+                }
+                if (href && href.startsWith('#')) {
+                    ctx.notes.push('internal anchor link flattened to plain text');
+                }
+                break;
+            }
+            default:
+                break;
+        }
+        out += renderInline(node.children, next, ctx);
+    }
+    return out;
+}
+/** Convert an `<img>` into an inline picture (data URIs and http(s) URLs). */
+function renderImage(node, ctx) {
+    const src = node.attrs.src ?? '';
+    if (!src)
+        return '';
+    // Images are embedded by the async pass BEFORE rendering; the placeholder is
+    // resolved there. See `embeddedImages` in buildDocxFromTemplate.
+    const ref = ctx.imageRefs?.get(src);
+    if (!ref) {
+        ctx.notes.push('an image could not be embedded and was omitted');
+        return '';
+    }
+    const cx = Math.round(ref.widthPx * EMU_PER_PX);
+    const cy = Math.round(ref.heightPx * EMU_PER_PX);
+    const id = ref.docPrId;
+    return (`<w:r><w:drawing><wp:inline xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" distT="0" distB="0" distL="0" distR="0">` +
+        `<wp:extent cx="${cx}" cy="${cy}"/><wp:effectExtent l="0" t="0" r="0" b="0"/>` +
+        `<wp:docPr id="${id}" name="Picture ${id}"/>` +
+        `<a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main">` +
+        `<a:graphicData uri="http://schemas.openxmlformats.org/drawingml/2006/picture">` +
+        `<pic:pic xmlns:pic="http://schemas.openxmlformats.org/drawingml/2006/picture">` +
+        `<pic:nvPicPr><pic:cNvPr id="${id}" name="Picture ${id}"/><pic:cNvPicPr/></pic:nvPicPr>` +
+        `<pic:blipFill><a:blip r:embed="${ref.relId}"/><a:stretch><a:fillRect/></a:stretch></pic:blipFill>` +
+        `<pic:spPr><a:xfrm><a:off x="0" y="0"/><a:ext cx="${cx}" cy="${cy}"/></a:xfrm>` +
+        `<a:prstGeom prst="rect"><a:avLst/></a:prstGeom></pic:spPr>` +
+        `</pic:pic></a:graphicData></a:graphic></wp:inline></w:drawing></w:r>`);
+}
+function alignmentProps(node) {
+    const align = styleOf(node)['text-align'];
+    if (!align)
+        return '';
+    const map = {
+        left: 'left',
+        right: 'right',
+        center: 'center',
+        justify: 'both',
+    };
+    const jc = map[align];
+    return jc ? `<w:jc w:val="${jc}"/>` : '';
+}
+function paragraph(styleId, extraPr, runs) {
+    const pr = `${styleId ? `<w:pStyle w:val="${(0, xml_escape_1.escapeXml)(styleId)}"/>` : ''}${extraPr}`;
+    return `<w:p>${pr ? `<w:pPr>${pr}</w:pPr>` : ''}${runs}</w:p>`;
+}
+function listParagraph(listStyleId, numId, runs) {
+    const numPr = numId
+        ? `<w:numPr><w:ilvl w:val="0"/><w:numId w:val="${numId}"/></w:numPr>`
+        : '';
+    const pr = `${listStyleId ? `<w:pStyle w:val="${(0, xml_escape_1.escapeXml)(listStyleId)}"/>` : ''}${numPr}`;
+    return `<w:p>${pr ? `<w:pPr>${pr}</w:pPr>` : ''}${runs}</w:p>`;
+}
+function renderTable(node, ctx) {
+    const rows = [];
+    const collectRows = (n) => {
+        for (const child of n.children) {
+            if (child.tag === 'tr')
+                rows.push(renderRow(child, ctx));
+            else if (child.tag !== '#text')
+                collectRows(child);
+        }
+    };
+    collectRows(node);
+    if (rows.length === 0)
+        return '';
+    const style = ctx.tableStyleId
+        ? `<w:tblStyle w:val="${(0, xml_escape_1.escapeXml)(ctx.tableStyleId)}"/>`
+        : '';
+    return (`<w:tbl><w:tblPr>${style}<w:tblW w:w="0" w:type="auto"/><w:tblBorders>` +
+        `<w:top w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `<w:left w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `<w:bottom w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `<w:right w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `<w:insideH w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `<w:insideV w:val="single" w:sz="4" w:space="0" w:color="auto"/>` +
+        `</w:tblBorders></w:tblPr><w:tblGrid/>${rows.join('')}</w:tbl>`);
+}
+function renderRow(node, ctx) {
+    const cells = [];
+    for (const cell of node.children) {
+        if (cell.tag !== 'td' && cell.tag !== 'th')
+            continue;
+        const bold = cell.tag === 'th' ? { bold: true } : {};
+        const runs = renderInline(cell.children, bold, ctx);
+        const shade = cell.tag === 'th'
+            ? `<w:shd w:val="clear" w:color="auto" w:fill="F1F5F9"/>`
+            : '';
+        cells.push(`<w:tc><w:tcPr><w:tcW w:w="0" w:type="auto"/>${shade}</w:tcPr><w:p>${runs}</w:p></w:tc>`);
+    }
+    return `<w:tr>${cells.join('')}</w:tr>`;
+}
+/**
+ * Render block content to a flat list of top-level OOXML elements (paragraphs
+ * and tables). Returning a list - rather than one joined string - lets the
+ * caller re-insert section breaks at their matching positions.
+ */
+function renderBlocks(nodes, ctx, listDepth = 0) {
+    const out = [];
+    for (const node of nodes) {
+        if (node.tag === '#text') {
+            const text = (node.text ?? '').trim();
+            if (text)
+                out.push(paragraph(ctx.bodyStyleId, '', textRun(node.text ?? '', {})));
+            continue;
+        }
+        const heading = HEADING_TAGS[node.tag];
+        if (heading) {
+            // Prefer the document's own heading style so the theme's fonts and
+            // colours apply automatically.
+            const styleId = ctx.paragraphStyle(`heading ${heading}`) ??
+                ctx.paragraphStyle(`heading${heading}`) ??
+                null;
+            out.push(paragraph(styleId, '', renderInline(node.children, {}, ctx)));
+            continue;
+        }
+        switch (node.tag) {
+            case 'p':
+            case 'div':
+                out.push(paragraph(ctx.bodyStyleId, alignmentProps(node), renderInline(node.children, {}, ctx)));
+                break;
+            case 'blockquote':
+                out.push(paragraph(ctx.quoteStyleId ?? ctx.bodyStyleId, ctx.quoteStyleId ? '' : '<w:ind w:left="720"/>', renderInline(node.children, {}, ctx)));
+                break;
+            case 'pre': {
+                const text = collectText(node);
+                for (const line of text.split('\n')) {
+                    out.push(paragraph(null, '<w:shd w:val="clear" w:color="auto" w:fill="F1F5F9"/>', textRun(line, { code: true })));
+                }
+                break;
+            }
+            case 'ul':
+            case 'ol': {
+                const numId = node.tag === 'ol' ? ctx.orderedNumId : ctx.bulletNumId;
+                const items = node.children.filter((c) => c.tag === 'li');
+                for (const li of items) {
+                    out.push(listParagraph(ctx.listStyleId, numId, renderInline(li.children, {}, ctx)));
+                }
+                break;
+            }
+            case 'table':
+                out.push(renderTable(node, ctx));
+                break;
+            case 'hr':
+                out.push(paragraph(null, '<w:pBdr><w:bottom w:val="single" w:sz="6" w:space="1" w:color="auto"/></w:pBdr>', ''));
+                break;
+            case 'li':
+                out.push(listParagraph(ctx.listStyleId, ctx.bulletNumId, renderInline(node.children, {}, ctx)));
+                break;
+            default:
+                // Unknown wrapper: recurse so nested content survives.
+                if (node.children.length) {
+                    out.push(...renderBlocks(node.children, ctx, listDepth));
+                }
+                break;
+        }
+    }
+    return out;
+}
+/**
+ * Re-insert the original section breaks (`w:sectPr` blocks that lived inside
+ * the body) so a multi-section document keeps its section count, per-section
+ * page setup and per-section header/footer references. Positions are scaled to
+ * the new paragraph count, since the body itself was rewritten.
+ */
+function placeSectionBreaks(elements, breaks) {
+    if (breaks.length === 0)
+        return elements;
+    const total = elements.length;
+    const out = [];
+    let next = 0;
+    for (let i = 0; i < total; i += 1) {
+        while (next < breaks.length &&
+            Math.round(((next + 1) / (breaks.length + 1)) * total) <= i) {
+            out.push(`<w:p><w:pPr>${breaks[next]}</w:pPr></w:p>`);
+            next += 1;
+        }
+        out.push(elements[i]);
+    }
+    while (next < breaks.length) {
+        out.push(`<w:p><w:pPr>${breaks[next]}</w:pPr></w:p>`);
+        next += 1;
+    }
+    return out;
+}
+/** Ordered `w:sectPr` blocks in the body; the last one is body-level. */
+function extractSectionBreaks(documentXml) {
+    const all = documentXml.match(/<w:sectPr[\s\S]*?<\/w:sectPr>/g) ?? [];
+    if (all.length === 0)
+        return { breaks: [], trailing: '' };
+    return { breaks: all.slice(0, -1), trailing: all[all.length - 1] };
+}
+function collectText(node) {
+    if (node.tag === '#text')
+        return node.text ?? '';
+    if (node.tag === 'br')
+        return '\n';
+    return node.children.map(collectText).join('');
+}
+function imageDimensions(buf) {
+    // PNG
+    if (buf.length > 24 && buf.readUInt32BE(0) === 0x89504e47) {
+        return { widthPx: buf.readUInt32BE(16), heightPx: buf.readUInt32BE(20) };
+    }
+    // GIF
+    if (buf.length > 10 && buf.slice(0, 3).toString('ascii') === 'GIF') {
+        return { widthPx: buf.readUInt16LE(6), heightPx: buf.readUInt16LE(8) };
+    }
+    // JPEG
+    if (buf.length > 4 && buf[0] === 0xff && buf[1] === 0xd8) {
+        let offset = 2;
+        while (offset + 9 < buf.length) {
+            if (buf[offset] !== 0xff) {
+                offset += 1;
+                continue;
+            }
+            const marker = buf[offset + 1];
+            const size = buf.readUInt16BE(offset + 2);
+            if ((marker >= 0xc0 && marker <= 0xc3) ||
+                (marker >= 0xc5 && marker <= 0xc7) ||
+                (marker >= 0xc9 && marker <= 0xcb)) {
+                return {
+                    heightPx: buf.readUInt16BE(offset + 5),
+                    widthPx: buf.readUInt16BE(offset + 7),
+                };
+            }
+            offset += 2 + size;
+        }
+    }
+    return { widthPx: 600, heightPx: 400 };
+}
+function extensionForMime(mime, fallback = 'png') {
+    if (mime.includes('png'))
+        return 'png';
+    if (mime.includes('jpeg') || mime.includes('jpg'))
+        return 'jpeg';
+    if (mime.includes('gif'))
+        return 'gif';
+    if (mime.includes('webp'))
+        return 'png';
+    if (mime.includes('svg'))
+        return 'png';
+    if (mime.includes('bmp'))
+        return 'bmp';
+    return fallback;
+}
+const MIME_BY_EXT = {
+    png: 'image/png',
+    jpeg: 'image/jpeg',
+    jpg: 'image/jpeg',
+    gif: 'image/gif',
+    bmp: 'image/bmp',
+    tiff: 'image/tiff',
+};
+/** Resolve every `<img src>` to bytes, ready to be embedded. */
+async function collectImages(html, ctx) {
+    const refs = new Map();
+    const sources = new Set();
+    const imgRe = /<img\b[^>]*\bsrc\s*=\s*("([^"]*)"|'([^']*)')/gi;
+    let m;
+    while ((m = imgRe.exec(html))) {
+        const src = m[2] ?? m[3] ?? '';
+        if (src)
+            sources.add(src);
+    }
+    let index = 0;
+    for (const src of sources) {
+        try {
+            let bytes;
+            let mime = 'image/png';
+            const dataUri = /^data:([^;,]+);base64,(.*)$/i.exec(src);
+            if (dataUri) {
+                mime = dataUri[1];
+                bytes = Buffer.from(dataUri[2], 'base64');
+            }
+            else if (/^https?:\/\//i.test(src)) {
+                const res = await fetch(src, { signal: AbortSignal.timeout(30_000) });
+                if (!res.ok)
+                    throw new Error(`HTTP ${res.status}`);
+                mime = res.headers.get('content-type') ?? 'image/png';
+                bytes = Buffer.from(await res.arrayBuffer());
+            }
+            else {
+                throw new Error('unsupported image source');
+            }
+            index += 1;
+            const ext = extensionForMime(mime);
+            const relId = ctx.nextRelId();
+            const path = `word/media/ts-image${index}.${ext}`;
+            ctx.newMedia.push({ path, bytes });
+            ctx.newRels.push({
+                id: relId,
+                type: 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/image',
+                target: `media/ts-image${index}.${ext}`,
+                external: false,
+            });
+            const { widthPx, heightPx } = imageDimensions(bytes);
+            refs.set(src, { relId, docPrId: index, widthPx, heightPx });
+        }
+        catch {
+            ctx.notes.push('an image could not be embedded and was omitted');
+        }
+    }
+    return refs;
+}
+// ===========================================================================
+// Package assembly
+// ===========================================================================
+/** Ensure `[Content_Types].xml` declares the image extensions we added. */
+function ensureContentTypes(xml, extensions) {
+    if (extensions.length === 0)
+        return xml;
+    const base = xml ??
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?><Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"></Types>';
+    let out = base;
+    for (const ext of extensions) {
+        const re = new RegExp(`<Default\\s+Extension="${ext}"`, 'i');
+        if (re.test(out))
+            continue;
+        const mime = MIME_BY_EXT[ext] ?? 'application/octet-stream';
+        out = out.replace('</Types>', `<Default Extension="${ext}" ContentType="${mime}"/></Types>`);
+    }
+    return out;
+}
+function parseRels(xml) {
+    const base = xml ??
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?><Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"></Relationships>';
+    let maxId = 0;
+    const re = /Id="rId(\d+)"/g;
+    let m;
+    while ((m = re.exec(base))) {
+        maxId = Math.max(maxId, Number(m[1]));
+    }
+    const prefix = xml ? 'rId' : 'rId';
+    return { xml: base, maxId, prefix };
+}
+/**
+ * Merge the edited HTML body into the original package.
+ *
+ * Throws on any structural problem - the caller fails the job and keeps the
+ * draft rather than writing a degraded file.
+ */
+async function buildDocxFromTemplate(original, html) {
+    // eslint-disable-next-line @typescript-eslint/no-require-imports
+    const JSZipCtor = require('jszip');
+    const zip = await JSZipCtor.loadAsync(original);
+    const mainEntry = zip.file(MAIN_PART);
+    if (!mainEntry) {
+        throw new Error('The document has no word/document.xml (not a valid .docx)');
+    }
+    const documentXml = await mainEntry.async('string');
+    const stylesEntry = zip.file(STYLES_PART);
+    const stylesXml = stylesEntry ? await stylesEntry.async('string') : null;
+    const numberingEntry = zip.file(NUMBERING_PART);
+    const numberingXml = numberingEntry ? await numberingEntry.async('string') : null;
+    const relsEntry = zip.file(RELS_PART);
+    const relsXml = relsEntry ? await relsEntry.async('string') : null;
+    const contentTypesEntry = zip.file(CONTENT_TYPES_PART);
+    const contentTypesXml = contentTypesEntry
+        ? await contentTypesEntry.async('string')
+        : null;
+    const manifest = parseStyleManifest(stylesXml);
+    const numbering = parseNumbering(numberingXml);
+    let finalNumberingXml = numberingXml;
+    let bulletNumId = numbering.bulletNumId;
+    let orderedNumId = numbering.orderedNumId;
+    if ((!bulletNumId || !orderedNumId) && /<(ul|ol)[\s>]/i.test(html)) {
+        const injected = injectListNumbering(numberingXml);
+        finalNumberingXml = injected.xml;
+        bulletNumId = bulletNumId ?? injected.bulletNumId;
+        orderedNumId = orderedNumId ?? injected.orderedNumId;
+    }
+    const rels = parseRels(relsXml);
+    let relCounter = rels.maxId;
+    const ctx = {
+        paragraphStyle: (name) => manifest.byName.get(name.toLowerCase()) ?? null,
+        bodyStyleId: manifest.paragraphDefault,
+        listStyleId: manifest.byName.get('list paragraph') ?? null,
+        tableStyleId: manifest.tableStyleId,
+        quoteStyleId: manifest.byName.get('quote') ?? manifest.byName.get('intense quote') ?? null,
+        bulletNumId,
+        orderedNumId,
+        newRels: [],
+        newMedia: [],
+        imageRefs: new Map(),
+        notes: [],
+        nextRelId: () => {
+            relCounter += 1;
+            return `rId${relCounter}`;
+        },
+    };
+    ctx.imageRefs = await collectImages(html, ctx);
+    const tree = parseHtml(html);
+    const toc = hasTocField(documentXml);
+    const { breaks: sectionBreaks, trailing: sectPr } = extractSectionBreaks(documentXml);
+    const elements = placeSectionBreaks(renderBlocks(tree.children, ctx), sectionBreaks);
+    const bodyInner = (toc ? tocField() : '') + elements.join('');
+    const openTagMatch = /^[\s\S]*?<w:body\b[^>]*>/.exec(documentXml);
+    const openTag = openTagMatch
+        ? openTagMatch[0]
+        : documentXml.slice(0, documentXml.indexOf('<w:body') + '<w:body>'.length);
+    const newDocumentXml = `${openTag}${bodyInner}${sectPr}</w:body></w:document>`;
+    // --- assemble the package -------------------------------------------------
+    const out = new JSZipCtor();
+    const preserved = [];
+    const dropped = [];
+    for (const [path, entry] of Object.entries(zip.files)) {
+        if (entry.dir)
+            continue;
+        if (path === MAIN_PART) {
+            out.file(path, newDocumentXml);
+            continue;
+        }
+        if (path === NUMBERING_PART && finalNumberingXml) {
+            out.file(path, finalNumberingXml);
+            continue;
+        }
+        if (path === RELS_PART) {
+            // written after the render, with the new relationships appended
+            continue;
+        }
+        // Every other part is carried over byte-for-byte.
+        if (path === STYLES_PART)
+            preserved.push('named styles');
+        else if (path.startsWith('word/theme/'))
+            preserved.push('theme');
+        else if (path.startsWith('word/header') || path.startsWith('word/footer')) {
+            preserved.push('headers and footers');
+        }
+        else if (path === 'word/settings.xml')
+            preserved.push('document settings');
+        out.file(path, await entry.async('nodebuffer'));
+    }
+    if (sectPr)
+        preserved.push('page setup, margins and columns');
+    if (sectionBreaks.length > 0) {
+        preserved.push(`sections (${sectionBreaks.length + 1})`);
+    }
+    if (toc)
+        preserved.push('table of contents (regenerated field)');
+    // Relationships (original + new link/image rels).
+    const newRelXml = ctx.newRels
+        .map((rel) => `<Relationship Id="${rel.id}" Type="${rel.type}" Target="${(0, xml_escape_1.escapeXml)(rel.target)}"${rel.external ? ' TargetMode="External"' : ''}/>`)
+        .join('');
+    const mergedRels = rels.xml.replace('</Relationships>', `${newRelXml}</Relationships>`);
+    out.file(RELS_PART, mergedRels);
+    // Media parts.
+    for (const media of ctx.newMedia) {
+        out.file(media.path, media.bytes);
+    }
+    // Content types for the image extensions we introduced.
+    const newExtensions = Array.from(new Set(ctx.newMedia.map((m) => m.path.split('.').pop() ?? 'png')));
+    const mergedTypes = ensureContentTypes(contentTypesXml, newExtensions);
+    if (mergedTypes)
+        out.file(CONTENT_TYPES_PART, mergedTypes);
+    // Honest report of body-level details that cannot survive a body rewrite.
+    if (/<w:bookmarkStart/.test(documentXml)) {
+        dropped.push('hidden bookmark anchors');
+    }
+    if (/<w:hyperlink\b[^>]*w:anchor="/.test(documentXml)) {
+        dropped.push('internal links to bookmarks');
+    }
+    const bytes = (await out.generateAsync({
+        type: 'nodebuffer',
+        compression: 'DEFLATE',
+    }));
+    return {
+        bytes,
+        report: {
+            preserved: Array.from(new Set(preserved)),
+            dropped: Array.from(new Set([...dropped, ...ctx.notes])).filter((d) => !d.startsWith('internal anchor link')),
+            notes: ctx.notes,
+        },
+    };
+}
+//# sourceMappingURL=docx-template.js.map