@signiphi/page-assembly 0.2.0-beta.13 → 0.2.0-beta.14

package/dist/index.mjs

@@ -14,7 +14,7 @@ import * as AlertDialogPrimitive from '@radix-ui/react-alert-dialog';
 import * as DropdownMenuPrimitive from '@radix-ui/react-dropdown-menu';
 import * as TooltipPrimitive from '@radix-ui/react-tooltip';
 import * as pdfjsLib from 'pdfjs-dist';
-import { PDFDocument, PDFTextField, PDFCheckBox, PDFDropdown, PDFRadioGroup, StandardFonts, rgb, PDFName, PDFString, PDFDict, PDFArray, PDFHexString, PDFInvalidObject, PDFObjectParser, PDFRawStream, PDFObjectStreamParser, PDFRef, PDFBool, PDFNumber } from 'pdf-lib';
+import { PDFDocument, PDFName, PDFTextField, PDFCheckBox, PDFDropdown, PDFRadioGroup, PDFHexString, StandardFonts, rgb, PDFString, PDFDict, PDFArray, PDFInvalidObject, PDFObjectParser, PDFRawStream, PDFObjectStreamParser, PDFRef, PDFBool, PDFNumber } from 'pdf-lib';
 
 // src/components/PageAssembler.tsx
 function cn(...inputs) {
@@ -1372,6 +1372,51 @@ function downloadPdf(pdfBytes, filename) {
   a.click();
   URL.revokeObjectURL(url);
 }
+function toCombinedPage(entry, page) {
+  if (!entry.position && !entry.radioPositions && !entry.radioOptions) return entry;
+  const carried = { ...entry };
+  if (entry.position) carried.position = { ...entry.position, page };
+  if (entry.radioPositions) carried.radioPositions = entry.radioPositions.map((p) => ({ ...p, page }));
+  if (entry.radioOptions) {
+    carried.radioOptions = entry.radioOptions.map((o) => ({ ...o, position: { ...o.position, page } }));
+  }
+  return carried;
+}
+function mergeSourceMetadata(sources) {
+  const merged = { version: "1.0", fields: {}, fieldIdIndex: {} };
+  let radioLabelsBaked;
+  let textLabelsBaked;
+  for (const source of sources) {
+    const meta = source.metadata;
+    if (!meta || source.fields.length === 0) continue;
+    if (meta.radioLabelsBaked === false) radioLabelsBaked = false;
+    else if (meta.radioLabelsBaked === true && radioLabelsBaked === void 0) radioLabelsBaked = true;
+    if (meta.textLabelsBaked === false) textLabelsBaked = false;
+    else if (meta.textLabelsBaked === true && textLabelsBaked === void 0) textLabelsBaked = true;
+  }
+  for (const source of sources) {
+    const meta = source.metadata;
+    if (!meta) continue;
+    const sourceTextLabelsBaked = meta.textLabelsBaked !== false;
+    for (const { name, page } of source.fields) {
+      const key = `${name}_p${page}`;
+      const entry = meta.fields?.[name];
+      if (entry) {
+        const carried = toCombinedPage(entry, page);
+        merged.fields[key] = textLabelsBaked === false && sourceTextLabelsBaked ? { ...carried, textLabelBaked: true } : carried;
+      }
+      const fieldId = meta.fieldIdIndex?.[name];
+      if (fieldId) merged.fieldIdIndex[key] = fieldId;
+    }
+  }
+  if (radioLabelsBaked !== void 0) merged.radioLabelsBaked = radioLabelsBaked;
+  if (textLabelsBaked !== void 0) merged.textLabelsBaked = textLabelsBaked;
+  const hasEntries = Object.keys(merged.fields).length > 0;
+  const hasMarkers = radioLabelsBaked !== void 0 || textLabelsBaked !== void 0;
+  if (!hasEntries && !hasMarkers) return null;
+  if (Object.keys(merged.fieldIdIndex).length === 0) delete merged.fieldIdIndex;
+  return merged;
+}
 function getSigniphiMetadata(pdfDoc) {
   try {
     const infoRef = pdfDoc.context.trailerInfo.Info;
@@ -1387,12 +1432,17 @@ function getSigniphiMetadata(pdfDoc) {
     if (!metadataObj) {
       return null;
     }
-    const metadataStr = metadataObj.toString();
-    let jsonStr = metadataStr;
-    if (metadataStr.startsWith("(") && metadataStr.endsWith(")")) {
-      jsonStr = metadataStr.slice(1, -1);
-    } else if (metadataStr.startsWith("<") && metadataStr.endsWith(">")) {
-      jsonStr = metadataStr.slice(1, -1);
+    let jsonStr;
+    if (typeof metadataObj.decodeText === "function") {
+      jsonStr = metadataObj.decodeText();
+    } else {
+      const metadataStr = metadataObj.toString();
+      jsonStr = metadataStr;
+      if (metadataStr.startsWith("(") && metadataStr.endsWith(")")) {
+        jsonStr = metadataStr.slice(1, -1);
+      } else if (metadataStr.startsWith("<") && metadataStr.endsWith(">")) {
+        jsonStr = metadataStr.slice(1, -1);
+      }
     }
     const metadata = JSON.parse(jsonStr);
     return metadata;
@@ -1585,7 +1635,7 @@ function setSigniphiMetadata(pdfDoc, metadata) {
     }
     const infoDict = infoObj;
     const metadataString = JSON.stringify(metadata);
-    infoDict.set(PDFName.of("SigniphiMetadata"), PDFString.of(metadataString));
+    infoDict.set(PDFName.of("SigniphiMetadata"), PDFHexString.fromText(metadataString));
   } catch (error) {
     console.error("Failed to set Signiphi metadata:", error);
     throw error;
@@ -2898,7 +2948,7 @@ var PageAssembler = forwardRef(
         setIsLoading(true);
         setProcessingStatus("Assembling PDF...");
         try {
-          const { PDFDocument: PDFDocument5, PDFName: PDFName4, PDFString: PDFString4 } = await import('pdf-lib');
+          const { PDFDocument: PDFDocument5, PDFName: PDFName4 } = await import('pdf-lib');
           const loadedPdfs = /* @__PURE__ */ new Map();
           const fileMap = /* @__PURE__ */ new Map();
           const pdfFiles = state.files.filter((f) => f.type === "pdf" && f.originalBytes?.length > 0);
@@ -2977,11 +3027,7 @@ var PageAssembler = forwardRef(
           }
           console.log("[PageAssembler] Starting multi-PDF assembly with field preservation");
           const allExtractedFields = [];
-          const mergedMetadata = {
-            version: "1.0",
-            fields: {},
-            fieldIdIndex: {}
-          };
+          const metadataSources = [];
           const filePageUsage = /* @__PURE__ */ new Map();
           for (const file of state.files) {
             for (const page of file.pages) {
@@ -3006,28 +3052,12 @@ var PageAssembler = forwardRef(
           }
           console.log("[PageAssembler] Page mapping:", Object.fromEntries(pageIdToNewPageNumber));
           const extractAndMapFields = async (fileId, fileName, pdfBytes2, sourcePdf) => {
-            let sourceMetadata = null;
-            if (sourcePdf) {
-              try {
-                const sourceInfoRef = sourcePdf.context.trailerInfo.Info;
-                if (sourceInfoRef) {
-                  const sourceInfoDict = sourcePdf.context.lookup(sourceInfoRef);
-                  if (sourceInfoDict && typeof sourceInfoDict.get === "function") {
-                    const metadataObj = sourceInfoDict.get(PDFName4.of("SigniphiMetadata"));
-                    if (metadataObj) {
-                      const metadataStr = metadataObj.toString();
-                      let jsonStr = metadataStr;
-                      if (metadataStr.startsWith("(") && metadataStr.endsWith(")")) {
-                        jsonStr = metadataStr.slice(1, -1);
-                      }
-                      sourceMetadata = JSON.parse(jsonStr);
-                    }
-                  }
-                }
-              } catch (metadataError) {
-                console.warn(`Could not read metadata from PDF ${fileName}:`, metadataError);
-              }
+            const sourceMetadata = sourcePdf ? getSigniphiMetadata(sourcePdf) : null;
+            if (sourcePdf && !sourceMetadata) {
+              console.warn(`[PageAssembler] No readable SigniphiMetadata in "${fileName}"`);
             }
+            const placements = [];
+            metadataSources.push({ metadata: sourceMetadata, fields: placements });
             console.log(`[PageAssembler] Extracting fields from "${fileName}"`);
             const extractedFields = await extractFieldsFromPdf(pdfBytes2);
             console.log(`[PageAssembler] Found ${extractedFields.length} fields in "${fileName}"`);
@@ -3061,13 +3091,7 @@ var PageAssembler = forwardRef(
                   };
                   allExtractedFields.push(trackedField);
                   console.log(`[PageAssembler] Mapped field "${field.name}" from page ${originalPage} to page ${newPageNumber}`);
-                  const metadataKey = `${field.name}_p${newPageNumber}`;
-                  if (sourceMetadata?.fields?.[field.name]) {
-                    mergedMetadata.fields[metadataKey] = sourceMetadata.fields[field.name];
-                  }
-                  if (sourceMetadata?.fieldIdIndex?.[field.name]) {
-                    mergedMetadata.fieldIdIndex[metadataKey] = sourceMetadata.fieldIdIndex[field.name];
-                  }
+                  placements.push({ name: field.name, page: newPageNumber });
                 }
               }
             }
@@ -3145,21 +3169,15 @@ var PageAssembler = forwardRef(
               }
             }
           }
-          if (Object.keys(mergedMetadata.fields).length > 0) {
+          const mergedMetadata = mergeSourceMetadata(metadataSources);
+          if (mergedMetadata) {
             try {
-              const newInfoRef = combinedDoc.context.trailerInfo.Info;
-              if (newInfoRef) {
-                const newInfoDict = combinedDoc.context.lookup(newInfoRef);
-                if (newInfoDict && typeof newInfoDict.set === "function") {
-                  newInfoDict.set(
-                    PDFName4.of("SigniphiMetadata"),
-                    PDFString4.of(JSON.stringify(mergedMetadata))
-                  );
-                }
-              }
+              setSigniphiMetadata(combinedDoc, mergedMetadata);
             } catch (metadataError) {
               console.warn("Could not write merged metadata:", metadataError);
             }
+          } else {
+            console.warn("[PageAssembler] No source metadata survived the merge; the assembled document will read as unprepared downstream");
           }
           let pdfBytes = await combinedDoc.save();
           if (allExtractedFields.length > 0) {