@absolutejs/absolute 0.19.0-beta.473 → 0.19.0-beta.474

package/dist/ai/index.js

@@ -2455,6 +2455,11 @@ var extractTextFromPDFBytes = (data) => {
 `);
   return normalizeWhitespace(combined);
 };
+var estimatePDFPageCount = (data) => {
+  const raw = Buffer.from(data).toString("latin1");
+  const count = [...raw.matchAll(/\/Type\s*\/Page\b/g)].length;
+  return count > 0 ? count : 1;
+};
 var readUInt16LE = (data, offset) => data[offset] | data[offset + 1] << 8;
 var readUInt32LE = (data, offset) => (data[offset] | data[offset + 1] << 8 | data[offset + 2] << 16 | data[offset + 3] << 24) >>> 0;
 var decodeUtf8 = (data) => Buffer.from(data).toString("utf8");
@@ -2545,6 +2550,14 @@ var officeDocumentText = (entries) => {
   }
   return extractXmlText(decodeUtf8(documentEntry.data));
 };
+var officeDocumentSectionCount = (entries) => {
+  const documentEntry = entries.find((entry) => entry.path === "word/document.xml");
+  if (!documentEntry) {
+    return;
+  }
+  const count = [...decodeUtf8(documentEntry.data).matchAll(/<w:p\b/g)].length;
+  return count > 0 ? count : undefined;
+};
 var spreadsheetText = (entries) => {
   const sharedStrings = entries.filter((entry) => entry.path === "xl/sharedStrings.xml").flatMap((entry) => [
     ...decodeUtf8(entry.data).matchAll(/<t[^>]*>([\s\S]*?)<\/t>/g)
@@ -2558,12 +2571,16 @@ var spreadsheetText = (entries) => {
   return normalizeWhitespace(sheetValues.join(`
 `));
 };
+var spreadsheetSheetNames = (entries) => entries.filter((entry) => entry.path === "xl/workbook.xml").flatMap((entry) => [
+  ...decodeUtf8(entry.data).matchAll(/<sheet[^>]*name="([^"]+)"/g)
+].map((match) => match[1] ?? "")).filter(Boolean);
 var presentationText = (entries) => {
   const slides = entries.filter((entry) => entry.path.startsWith("ppt/slides/") && entry.path.endsWith(".xml")).map((entry) => extractXmlText(decodeUtf8(entry.data)));
   return normalizeWhitespace(slides.join(`
 
 `));
 };
+var presentationSlideCount = (entries) => entries.filter((entry) => entry.path.startsWith("ppt/slides/") && entry.path.endsWith(".xml")).length;
 var epubText = (entries) => {
   const htmlEntries = entries.filter((entry) => /\.(xhtml|html|htm)$/i.test(entry.path));
   return normalizeWhitespace(htmlEntries.map((entry) => stripHtml(decodeUtf8(entry.data))).join(`
@@ -2588,6 +2605,23 @@ var extractEmailText = (raw) => {
   }
   return normalizeWhitespace(body);
 };
+var parseEmailHeaders = (raw) => {
+  const normalized = raw.replace(/\r\n?/g, `
+`);
+  const [headerBlock = ""] = normalized.split(`
+
+`);
+  const getHeader = (name) => {
+    const match = headerBlock.match(new RegExp(`^${name}:\\s*(.+)$`, "im"));
+    return match?.[1]?.trim();
+  };
+  return {
+    from: getHeader("From"),
+    subject: getHeader("Subject"),
+    threadTopic: getHeader("Thread-Topic") ?? getHeader("Subject"),
+    to: getHeader("To")
+  };
+};
 var stripRTF = (value) => {
   const withoutBinary = value.replace(/\\bin\d+ [\s\S]*?(?=[\\}])/g, " ");
   const withoutControls = withoutBinary.replace(/\\par[d]?/g, `
@@ -2750,12 +2784,22 @@ var createOfficeDocumentExtractor = () => ({
     const extension = inferExtensionFromInput(input);
     const entries = unzipEntries(input.data);
     let text = "";
+    let officeMetadata = {};
     if (extension === ".docx" || extension === ".odt") {
       text = officeDocumentText(entries);
+      officeMetadata = {
+        sectionCount: officeDocumentSectionCount(entries)
+      };
     } else if (extension === ".xlsx" || extension === ".ods") {
       text = spreadsheetText(entries);
+      officeMetadata = {
+        sheetNames: spreadsheetSheetNames(entries)
+      };
     } else if (extension === ".pptx" || extension === ".odp") {
       text = presentationText(entries);
+      officeMetadata = {
+        slideCount: presentationSlideCount(entries)
+      };
     }
     if (!text) {
       throw new Error(`AbsoluteJS could not extract readable text from ${inferNameFromInput(input)}`);
@@ -2766,7 +2810,8 @@ var createOfficeDocumentExtractor = () => ({
       format: "text",
       metadata: {
         ...input.metadata ?? {},
-        fileKind: "office"
+        fileKind: "office",
+        ...officeMetadata
       },
       source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}${extension || ".office"}`,
       text,
@@ -2801,18 +2846,21 @@ var createEmailExtractor = () => ({
   supports: emailExtractorSupports,
   extract: (input) => {
     const raw = decodeUtf8(input.data);
-    const subjectMatch = raw.match(/^Subject:\s*(.+)$/im);
+    const headers = parseEmailHeaders(raw);
     return {
       chunking: input.chunking,
       contentType: input.contentType,
       format: "text",
       metadata: {
         ...input.metadata ?? {},
-        fileKind: "email"
+        fileKind: "email",
+        from: headers.from,
+        threadTopic: headers.subject,
+        to: headers.to
       },
       source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.eml`,
       text: extractEmailText(raw),
-      title: input.title ?? subjectMatch?.[1]?.trim()
+      title: input.title ?? headers.subject
     };
   }
 });
@@ -2889,7 +2937,8 @@ var createPDFFileExtractor = () => ({
       format: "text",
       metadata: {
         ...input.metadata ?? {},
-        fileKind: "pdf"
+        fileKind: "pdf",
+        pageCount: estimatePDFPageCount(input.data)
       },
       source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.pdf`,
       text,
@@ -2912,6 +2961,7 @@ var createRAGPDFOCRExtractor = (options) => ({
         metadata: {
           ...input.metadata ?? {},
           fileKind: "pdf",
+          pageCount: estimatePDFPageCount(input.data),
           pdfTextMode: "native"
         },
         source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.pdf`,
@@ -2931,6 +2981,7 @@ var createRAGPDFOCRExtractor = (options) => ({
         ...input.metadata ?? {},
         ...ocr.metadata ?? {},
         fileKind: "pdf",
+        pageCount: estimatePDFPageCount(input.data),
         pdfTextMode: "ocr"
       },
       source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.pdf`,
@@ -6437,5 +6488,5 @@ export {
   aiChat
 };
 
-//# debugId=C9D2C43A1F5E327064756E2164756E21
+//# debugId=4C8CFD94D17CA32664756E2164756E21
 //# sourceMappingURL=index.js.map