dsh-cait 1.1.3 → 1.1.7

package/lib/docx.js

@@ -239,6 +239,88 @@ function paragraphText(paragraph) {
   return out.replace(/\n{3,}/g, '\n\n').trim()
 }
 
+function contentPartRank(name) {
+  if (name === 'word/document.xml') return 0
+  if (/^word\/header\d*\.xml$/i.test(name)) return 1
+  if (/^word\/footer\d*\.xml$/i.test(name)) return 2
+  if (name === 'word/footnotes.xml') return 3
+  if (name === 'word/endnotes.xml') return 4
+  return 9
+}
+
+function contentPartNames(entries) {
+  return entries
+    .map((entry) => entry.name)
+    .filter((name) => contentPartRank(name) < 9)
+    .sort((left, right) => {
+      const rank = contentPartRank(left) - contentPartRank(right)
+      return rank || left.localeCompare(right)
+    })
+}
+
+/**
+ * Collect every paragraph in one document part, including table cells, in
+ * document order. `docxParagraph` is the ordinal within that part, which is
+ * what the exporter patches; table/row/cell are informative metadata.
+ */
+function collectPartParagraphs(xml, partName) {
+  const pattern = /<w:tbl\b[^>]*>|<\/w:tbl>|<w:tr\b[^>]*>|<\/w:tr>|<w:tc\b[^>]*>|<\/w:tc>|<w:p\b[^>]*>[\s\S]*?<\/w:p>/gi
+  const paragraphs = []
+  let tableDepth = 0
+  let tableIndex = -1
+  let tableCount = 0
+  let rowIndex = -1
+  let cellIndex = -1
+  let ordinal = 0
+  let match
+
+  while ((match = pattern.exec(xml)) !== null) {
+    const tag = match[0]
+    if (/^<w:tbl\b/i.test(tag)) {
+      if (tableDepth === 0) {
+        tableIndex = tableCount
+        tableCount += 1
+        rowIndex = -1
+        cellIndex = -1
+      }
+      tableDepth += 1
+      continue
+    }
+    if (/^<\/w:tbl>/i.test(tag)) {
+      tableDepth = Math.max(0, tableDepth - 1)
+      continue
+    }
+    if (/^<w:tr\b/i.test(tag)) {
+      if (tableDepth > 0) {
+        rowIndex += 1
+        cellIndex = -1
+      }
+      continue
+    }
+    if (/^<\/w:tr>/i.test(tag)) continue
+    if (/^<w:tc\b/i.test(tag)) {
+      if (tableDepth > 0) cellIndex += 1
+      continue
+    }
+    if (/^<\/w:tc>/i.test(tag)) continue
+
+    const inner = tag.replace(/^<w:p\b[^>]*>/i, '').replace(/<\/w:p>\s*$/i, '')
+    const text = paragraphText(inner)
+    if (text) {
+      const meta = { docxPart: partName, docxParagraph: ordinal }
+      if (tableDepth > 0) {
+        meta.table = tableIndex
+        meta.row = Math.max(0, rowIndex)
+        meta.cell = Math.max(0, cellIndex)
+      }
+      paragraphs.push({ text, meta })
+    }
+    ordinal += 1
+  }
+
+  return paragraphs
+}
+
 /**
  * Like parseDocx, but also returns the zero-based body paragraph ordinal for
  * every non-empty paragraph. The ordinal counts all `<w:p>` elements outside
@@ -248,33 +330,35 @@ export function parseDocxDetailed(buffer) {
   if (!Buffer.isBuffer(buffer) || buffer.length < 4) throw new Error('DOCX 内容为空')
   if (buffer[0] !== 0x50 || buffer[1] !== 0x4b) throw new Error('DOCX 文件头不是 ZIP')
 
-  const xml = readZipEntry(buffer, 'word/document.xml')
-  if (!xml) throw new Error('DOCX 中缺少 word/document.xml')
+  const entries = readZipEntries(buffer)
+  const documentEntry = entries.find((entry) => entry.name === 'word/document.xml')
+  if (!documentEntry) throw new Error('DOCX 中缺少 word/document.xml')
 
-  const document = xml.toString('utf8')
-  const body = document.match(/<w:body\b[^>]*>([\s\S]*?)<\/w:body>/i)?.[1] ?? document
-  const content = withoutTables(body)
+  const parts = contentPartNames(entries)
   const paragraphs = []
-  const pattern = /<w:p\b[^>]*>([\s\S]*?)<\/w:p>/gi
-  let match
-  let paragraphIndex = 0
-
-  while ((match = pattern.exec(content)) !== null) {
-    const text = paragraphText(match[1])
-    if (text) paragraphs.push({ text, index: paragraphIndex })
-    paragraphIndex += 1
+  for (const name of parts) {
+    const entry = entries.find((item) => item.name === name)
+    if (!entry) continue
+    const found = collectPartParagraphs(entry.data.toString('utf8'), name)
+    for (const paragraph of found) {
+      paragraphs.push({ ...paragraph, index: paragraphs.length })
+    }
   }
 
   if (!paragraphs.length) {
-    const fallback = content.replace(/<w:tab\b[^>]*\/>/g, '\t').replace(/<w:br\b[^>]*\/>/g, '\n')
+    const fallback = documentEntry.data
+      .toString('utf8')
+      .replace(/<w:tab\b[^>]*\/>/g, '\t')
+      .replace(/<w:br\b[^>]*\/>/g, '\n')
     const text = decodeXml(fallback.replace(/<[^>]*>/g, '')).replace(/\n{3,}/g, '\n\n').trim()
-    if (text) paragraphs.push({ text, index: 0 })
+    if (text) paragraphs.push({ text, index: 0, meta: { docxPart: 'word/document.xml', docxParagraph: 0 } })
   }
 
   if (!paragraphs.length) throw new Error('DOCX 中未找到可翻译的文本')
   return {
     text: paragraphs.map((paragraph) => paragraph.text).join('\n\n'),
     paragraphs,
+    formatMeta: { docxParts: parts },
   }
 }
 
@@ -282,9 +366,9 @@ export function parseDocx(buffer) {
   return parseDocxDetailed(buffer).text
 }
 
-/** Read `word/document.xml` as text; used by the export tests. */
-export function readDocxXml(buffer) {
-  const xml = readZipEntry(buffer, 'word/document.xml')
+/** Read one DOCX part as text; used by the export tests. */
+export function readDocxXml(buffer, partName = 'word/document.xml') {
+  const xml = readZipEntry(buffer, partName)
   return xml ? xml.toString('utf8') : null
 }
 
@@ -355,19 +439,33 @@ function patchDocumentXml(xml, replacements) {
   return `${xml.slice(0, start)}${patchBody(body, replacements)}${xml.slice(start + body.length)}`
 }
 
+function patchPartXml(xml, replacements) {
+  let ordinal = 0
+  return xml.replace(/<w:p\b[^>]*>[\s\S]*?<\/w:p>/gi, (paragraph) => {
+    const current = ordinal
+    ordinal += 1
+    if (!replacements.has(current)) return paragraph
+    return patchParagraph(paragraph, replacements.get(current))
+  })
+}
+
 /** Patch translated text into the source package, preserving its formatting. */
-export function buildDocx(template, replacements) {
+export function buildDocx(template, replacements, partReplacements = null) {
   const entries = readZipEntries(template)
   const documentEntry = entries.find((entry) => entry.name === 'word/document.xml')
   if (!documentEntry) throw new Error('DOCX 中缺少 word/document.xml')
-  const xml = documentEntry.data.toString('utf8')
-  const patched = patchDocumentXml(xml, replacements)
+  const byPart = partReplacements instanceof Map
+    ? partReplacements
+    : new Map([['word/document.xml', replacements]])
   return writeZip(
-    entries.map((entry) =>
-      entry.name === 'word/document.xml'
-        ? { name: entry.name, data: Buffer.from(patched, 'utf8') }
-        : entry,
-    ),
+    entries.map((entry) => {
+      if (!byPart.has(entry.name)) return entry
+      const xml = entry.data.toString('utf8')
+      const patched = partReplacements instanceof Map
+        ? patchPartXml(xml, byPart.get(entry.name))
+        : patchDocumentXml(xml, replacements)
+      return { name: entry.name, data: Buffer.from(patched, 'utf8') }
+    }),
   )
 }
 
@@ -389,4 +487,35 @@ export function buildDocxFromText(paragraphs) {
   ])
 }
 
+export function buildBilingualDocx(segments, { includeStatus = false, title = '' } = {}) {
+  const columns = includeStatus ? 3 : 2
+  const width = Math.floor(9000 / columns)
+  const cell = (text, shaded = false) => {
+    const paragraphs = String(text ?? '')
+      .split(/\r?\n/)
+      .map((line) => `<w:p><w:r><w:t xml:space="preserve">${escapeXml(line)}</w:t></w:r></w:p>`)
+      .join('')
+    const shade = shaded ? '<w:shd w:val="clear" w:color="auto" w:fill="F2F2F2"/>' : ''
+    return `<w:tc><w:tcPr><w:tcW w:w="${width}" w:type="dxa"/>${shade}</w:tcPr>${paragraphs}</w:tc>`
+  }
+  const header = `<w:tr><w:trPr><w:tblHeader/></w:trPr>${cell('Source', true)}${cell('Target', true)}${includeStatus ? cell('Status', true) : ''}</w:tr>`
+  const body = segments
+    .map(
+      (segment) =>
+        `<w:tr>${cell(segment.source)}${cell(segment.target || '')}${includeStatus ? cell(segment.status || '') : ''}</w:tr>`,
+    )
+    .join('')
+  const titleXml = title
+    ? `<w:p><w:r><w:rPr><w:b/><w:sz w:val="28"/></w:rPr><w:t xml:space="preserve">${escapeXml(title)}</w:t></w:r></w:p>`
+    : ''
+  const document = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?><w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"><w:body>${titleXml}<w:tbl><w:tblPr><w:tblW w:w="0" w:type="auto"/><w:tblBorders><w:top w:val="single" w:sz="4"/><w:left w:val="single" w:sz="4"/><w:bottom w:val="single" w:sz="4"/><w:right w:val="single" w:sz="4"/><w:insideH w:val="single" w:sz="4"/><w:insideV w:val="single" w:sz="4"/></w:tblBorders></w:tblPr><w:tblGrid>${Array.from({ length: columns }, () => `<w:gridCol w:w="${width}"/>`).join('')}</w:tblGrid>${header}${body}</w:tbl><w:sectPr><w:pgSz w:w="11906" w:h="16838"/><w:pgMar w:top="1440" w:right="1440" w:bottom="1440" w:left="1440"/></w:sectPr></w:body></w:document>`
+  const contentTypes = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?><Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"><Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/><Default Extension="xml" ContentType="application/xml"/><Override PartName="/word/document.xml" ContentType="application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml"/></Types>`
+  const rels = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?><Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument" Target="word/document.xml"/></Relationships>`
+  return writeZip([
+    { name: '[Content_Types].xml', data: Buffer.from(contentTypes, 'utf8') },
+    { name: '_rels/.rels', data: Buffer.from(rels, 'utf8') },
+    { name: 'word/document.xml', data: Buffer.from(document, 'utf8') },
+  ])
+}
+
 export { readZipEntries, writeZip }

package/lib/formats.js

@@ -577,6 +577,9 @@ function parseXliffUnit(body, attrs, format, tokens) {
   const targetMatch = body.match(/<(?:[\w.-]+:)?target\b[^>]*>([\s\S]*?)<\/(?:[\w.-]+:)?target>/i)
     ?? body.match(/<(?:[\w.-]+:)?target\b[^>]*\/>/i)
   const target = targetMatch ? extractInline(targetMatch[1] ?? '').text : ''
+  const targetTag = body.match(/<(?:[\w.-]+:)?target\b([^>]*)>/i)
+  const targetState = targetTag?.[1] ? readXmlAttr(targetTag[1], 'state') : ''
+  const caitUid = readXmlAttr(attrs, 'x-cait-uid')
   tokens.push(...source.tokens)
   return {
     source: source.text,
@@ -585,6 +588,8 @@ function parseXliffUnit(body, attrs, format, tokens) {
     meta: {
       format,
       xliffId: readXmlAttr(attrs, 'id'),
+      ...(caitUid ? { caitUid } : {}),
+      ...(targetState ? { targetState } : {}),
       ...(source.tokens.length ? { inline: source.tokens } : {}),
     },
   }
@@ -661,6 +666,38 @@ export function buildXliff(doc) {
   ].join('\n')
 }
 
+export function buildXliffReview(doc) {
+  const units = doc.segments.map((segment, index) => {
+    const id = String(segment.uid || segment.meta?.xliffId || index + 1)
+    const tokens = Array.isArray(segment.meta?.inline) ? segment.meta.inline : []
+    const target = completeInline(segment.target || '', tokens)
+    const state = segment.confirmedAt ? 'final' : segment.target ? 'translated' : 'new'
+    const note = segment.confirmedAt
+      ? 'confirmed'
+      : segment.origin
+        ? `origin: ${segment.origin}`
+        : ''
+    return [
+      `      <trans-unit id="${escapeXml(id)}" x-cait-uid="${escapeXml(segment.uid || id)}">`,
+      `        <source xml:lang="${escapeXml(doc.sourceLang)}">${renderInline(segment.source, tokens)}</source>`,
+      `        <target xml:lang="${escapeXml(doc.targetLang)}" state="${state}">${renderInline(target, tokens)}</target>`,
+      ...(note ? [`        <note from="cait">${escapeXml(note)}</note>`] : []),
+      '      </trans-unit>',
+    ].join('\n')
+  })
+  return [
+    '<?xml version="1.0" encoding="UTF-8"?>',
+    '<xliff version="1.2">',
+    `  <file source-language="${escapeXml(doc.sourceLang)}" target-language="${escapeXml(doc.targetLang)}" datatype="plaintext" original="${escapeXml(doc.name)}">`,
+    '    <body>',
+    ...units,
+    '    </body>',
+    '  </file>',
+    '</xliff>',
+    '',
+  ].join('\n')
+}
+
 export function buildXliff2(doc) {
   const units = doc.segments.map((segment, index) => {
     const id = String(segment.meta?.xliffId || index + 1)

package/lib/glossary.js

@@ -3,13 +3,16 @@ import { DEFAULT_GLOSSARY_ID, DEFAULT_GLOSSARY_NAME, MAX_SEGMENTS, canonicalLang
 import { normalizeText } from './similarity.js'
 
 export const glossaryMethods = {
-  findGlossaryHits(text, sourceLang) {
+  findGlossaryHits(text, sourceLang, documentId) {
     const haystack = String(text ?? '')
     if (!haystack) return []
     const lower = haystack.toLowerCase()
     const lang = sourceLang ? canonicalLanguage(sourceLang) : ''
     const hits = []
-    const mountedGlossary = new Set(this.mounts.glossaryIds)
+    const scopedDoc = documentId ? this.requireDocument(documentId) : null
+    const mountedGlossary = new Set(
+      scopedDoc ? this.effectiveGlossaryIds(scopedDoc) : this.mounts.glossaryIds,
+    )
 
     for (const entry of this.glossary) {
       if (!mountedGlossary.has(entry.glossaryId || DEFAULT_GLOSSARY_ID)) continue
@@ -29,6 +32,8 @@ export const glossaryMethods = {
         term: entry.source,
         target: entry.target,
         note: entry.note || '',
+        forbidden: Boolean(entry.forbidden),
+        preferred: Boolean(entry.preferred),
         count,
       })
     }
@@ -36,7 +41,7 @@ export const glossaryMethods = {
     return hits.sort((a, b) => b.term.length - a.term.length)
   }
 ,
-  addGlossary({ source, target, note, sourceLang, caseSensitive, glossaryId } = {}) {
+  addGlossary({ source, target, note, sourceLang, caseSensitive, forbidden, preferred, glossaryId } = {}) {
     const term = cleanText(source, '术语')
     const translation = cleanText(target, '术语译文')
     if (!term || !translation) throw new CatError('EMPTY_TERM', '术语和译文都不能为空。')
@@ -53,6 +58,8 @@ export const glossaryMethods = {
       existing.target = translation
       existing.note = String(note ?? '').trim()
       existing.caseSensitive = Boolean(caseSensitive)
+      existing.forbidden = Boolean(forbidden)
+      existing.preferred = Boolean(preferred)
       existing.updatedAt = now()
       this.touch()
       this.persist()
@@ -66,6 +73,8 @@ export const glossaryMethods = {
       note: String(note ?? '').trim(),
       sourceLang: lang,
       caseSensitive: Boolean(caseSensitive),
+      forbidden: Boolean(forbidden),
+      preferred: Boolean(preferred),
       createdAt: now(),
       updatedAt: now(),
     }
@@ -75,7 +84,7 @@ export const glossaryMethods = {
     return { entry: { ...entry }, created: true }
   }
 ,
-  updateGlossary({ id, source, target, note, caseSensitive } = {}) {
+  updateGlossary({ id, source, target, note, caseSensitive, forbidden, preferred } = {}) {
     const entry = this.glossary.find((item) => item.id === id)
     if (!entry) throw new CatError('TERM_NOT_FOUND', '术语不存在。', 404)
     // The library page manages every library, mounted or not; only the target
@@ -93,6 +102,9 @@ export const glossaryMethods = {
     }
     if (note !== undefined) entry.note = String(note ?? '').trim()
     if (caseSensitive !== undefined) entry.caseSensitive = Boolean(caseSensitive)
+    if (forbidden !== undefined) entry.forbidden = Boolean(forbidden)
+    if (preferred !== undefined) entry.preferred = Boolean(preferred)
+    if (entry.forbidden) entry.preferred = false
     entry.updatedAt = now()
     this.touch()
     this.persist()
@@ -129,6 +141,8 @@ export const glossaryMethods = {
       if (existing) {
         existing.target = String(item.target).trim()
         existing.note = String(item.note ?? '').trim()
+        existing.forbidden = Boolean(item.forbidden)
+        existing.preferred = Boolean(item.preferred) && !existing.forbidden
         existing.updatedAt = now()
         updated += 1
       } else {
@@ -140,6 +154,8 @@ export const glossaryMethods = {
           note: String(item.note ?? '').trim(),
           sourceLang: lang,
           caseSensitive: Boolean(item.caseSensitive),
+          forbidden: Boolean(item.forbidden),
+          preferred: Boolean(item.preferred) && !item.forbidden,
           createdAt: now(),
           updatedAt: now(),
         })