pdfjs-reader-core 0.2.0 → 0.2.1

package/dist/index.cjs

@@ -1331,37 +1331,54 @@ async function extractPageText(document2, pageNumber) {
   const textContent = await page.getTextContent();
   const viewport = page.getViewport({ scale: 1 });
   let fullText = "";
-  const charPositions = [];
+  const textItems = [];
   for (const item of textContent.items) {
     if ("str" in item && item.str) {
-      const tx = item.transform;
-      const x = tx[4];
-      const y = viewport.height - tx[5];
-      const width = item.width ?? 0;
-      const height = item.height ?? 12;
-      const charWidth = item.str.length > 0 ? width / item.str.length : width;
-      for (let i = 0; i < item.str.length; i++) {
-        charPositions.push({
-          char: item.str[i],
-          rect: {
-            x: x + i * charWidth,
-            y: y - height,
-            width: charWidth,
-            height
-          }
-        });
-      }
+      textItems.push({
+        text: item.str,
+        transform: item.transform,
+        width: item.width ?? 0,
+        height: item.height ?? 12
+      });
       fullText += item.str;
     }
   }
-  return { fullText, charPositions };
+  return { fullText, textItems, viewport };
+}
+function calculateMatchRects(textItems, startOffset, length, viewport) {
+  const rects = [];
+  let currentOffset = 0;
+  for (const item of textItems) {
+    const itemStart = currentOffset;
+    const itemEnd = currentOffset + item.text.length;
+    if (itemEnd > startOffset && itemStart < startOffset + length) {
+      const [, , c, d, tx, ty] = item.transform;
+      const x = tx;
+      const y = viewport.height - ty;
+      const height = Math.sqrt(c * c + d * d);
+      const matchStartInItem = Math.max(0, startOffset - itemStart);
+      const matchEndInItem = Math.min(item.text.length, startOffset + length - itemStart);
+      const charWidth = item.text.length > 0 ? item.width / item.text.length : item.width;
+      const matchWidth = charWidth * (matchEndInItem - matchStartInItem);
+      const matchX = x + charWidth * matchStartInItem;
+      const yOffset = height * 0.15;
+      rects.push({
+        x: matchX,
+        y: y - height + yOffset,
+        width: matchWidth,
+        height
+      });
+    }
+    currentOffset = itemEnd;
+  }
+  return rects;
 }
 async function findTextOnPage(document2, pageNumber, query, options = {}) {
   const { caseSensitive = false, wholeWord = false } = options;
   if (!query || pageNumber < 1 || pageNumber > document2.numPages) {
     return [];
   }
-  const { fullText, charPositions } = await extractPageText(document2, pageNumber);
+  const { fullText, textItems, viewport } = await extractPageText(document2, pageNumber);
   const matches = [];
   const searchText = caseSensitive ? query : query.toLowerCase();
   const textToSearch = caseSensitive ? fullText : fullText.toLowerCase();
@@ -1377,17 +1394,15 @@ async function findTextOnPage(document2, pageNumber, query, options = {}) {
         continue;
       }
     }
-    const matchRects = [];
-    for (let i = matchIndex; i < matchIndex + query.length && i < charPositions.length; i++) {
-      matchRects.push(charPositions[i].rect);
+    const matchRects = calculateMatchRects(textItems, matchIndex, query.length, viewport);
+    if (matchRects.length > 0) {
+      matches.push({
+        text: fullText.substring(matchIndex, matchIndex + query.length),
+        rects: matchRects,
+        pageNumber,
+        startIndex: matchIndex
+      });
     }
-    const mergedRects = mergeAdjacentRects(matchRects);
-    matches.push({
-      text: fullText.substring(matchIndex, matchIndex + query.length),
-      rects: mergedRects,
-      pageNumber,
-      startIndex: matchIndex
-    });
     startIndex = matchIndex + 1;
   }
   return matches;
@@ -1640,7 +1655,7 @@ function createSearchStore(initialOverrides = {}) {
               }
             }
             const matchText = pageText.substring(startIndex, startIndex + query.length);
-            const rects = calculateMatchRects(textItems, startIndex, query.length, viewport);
+            const rects = calculateMatchRects2(textItems, startIndex, query.length, viewport);
             results.push({
               pageNumber: pageNum,
               matchIndex: matchIndex++,
@@ -1701,7 +1716,7 @@ function createSearchStore(initialOverrides = {}) {
     }
   }));
 }
-function calculateMatchRects(textItems, startOffset, length, viewport) {
+function calculateMatchRects2(textItems, startOffset, length, viewport) {
   const rects = [];
   let currentOffset = 0;
   for (const item of textItems) {
@@ -1716,9 +1731,10 @@ function calculateMatchRects(textItems, startOffset, length, viewport) {
       const matchEndInItem = Math.min(item.text.length, startOffset + length - itemStart);
       const matchWidth = item.width / item.text.length * (matchEndInItem - matchStartInItem);
       const matchX = x + item.width / item.text.length * matchStartInItem;
+      const yOffset = height * 0.15;
       rects.push({
         x: matchX,
-        y: y - height,
+        y: y - height + yOffset,
         width: matchWidth,
         height
       });
@@ -9295,24 +9311,33 @@ function getSrcIdentifier(src) {
   const last = Array.from(data.slice(-4)).map((b) => b.toString(16).padStart(2, "0")).join("");
   return `binary:${len}:${first}:${last}`;
 }
-function mergeRects2(rects) {
-  if (rects.length === 0) return [];
-  const sorted = [...rects].sort((a, b) => a.y - b.y || a.x - b.x);
-  const merged = [];
-  let current = { ...sorted[0] };
-  for (let i = 1; i < sorted.length; i++) {
-    const rect = sorted[i];
-    if (Math.abs(rect.y - current.y) < 2 && rect.x <= current.x + current.width + 2) {
-      const newRight = Math.max(current.x + current.width, rect.x + rect.width);
-      current.width = newRight - current.x;
-      current.height = Math.max(current.height, rect.height);
-    } else {
-      merged.push(current);
-      current = { ...rect };
+function calculateMatchRects3(textItems, startOffset, length, viewport) {
+  const rects = [];
+  let currentOffset = 0;
+  for (const item of textItems) {
+    const itemStart = currentOffset;
+    const itemEnd = currentOffset + item.text.length;
+    if (itemEnd > startOffset && itemStart < startOffset + length) {
+      const [, , c, d, tx, ty] = item.transform;
+      const x = tx;
+      const y = viewport.height - ty;
+      const height = Math.sqrt(c * c + d * d);
+      const matchStartInItem = Math.max(0, startOffset - itemStart);
+      const matchEndInItem = Math.min(item.text.length, startOffset + length - itemStart);
+      const charWidth = item.text.length > 0 ? item.width / item.text.length : item.width;
+      const matchWidth = charWidth * (matchEndInItem - matchStartInItem);
+      const matchX = x + charWidth * matchStartInItem;
+      const yOffset = height * 0.15;
+      rects.push({
+        x: matchX,
+        y: y - height + yOffset,
+        width: matchWidth,
+        height
+      });
     }
+    currentOffset = itemEnd;
   }
-  merged.push(current);
-  return merged;
+  return rects;
 }
 var import_react40, import_jsx_runtime26, PDFViewerInner, PDFViewerInnerWithRef, PDFViewerClient;
 var init_PDFViewerClient = __esm({
@@ -9411,26 +9436,15 @@ var init_PDFViewerClient = __esm({
                 const textContent = await page.getTextContent();
                 const viewport = page.getViewport({ scale: 1 });
                 let fullText = "";
-                const charPositions = [];
+                const textItems = [];
                 for (const item of textContent.items) {
                   if ("str" in item && item.str) {
-                    const tx = item.transform;
-                    const x = tx[4];
-                    const y = viewport.height - tx[5];
-                    const width = item.width ?? 0;
-                    const height = item.height ?? 12;
-                    const charWidth = item.str.length > 0 ? width / item.str.length : width;
-                    for (let i = 0; i < item.str.length; i++) {
-                      charPositions.push({
-                        char: item.str[i],
-                        rect: {
-                          x: x + i * charWidth,
-                          y: y - height,
-                          width: charWidth,
-                          height
-                        }
-                      });
-                    }
+                    textItems.push({
+                      text: item.str,
+                      transform: item.transform,
+                      width: item.width ?? 0,
+                      height: item.height ?? 12
+                    });
                     fullText += item.str;
                   }
                 }
@@ -9439,18 +9453,16 @@ var init_PDFViewerClient = __esm({
                 while (true) {
                   const matchIndex = textToSearch.indexOf(searchText, startIndex);
                   if (matchIndex === -1) break;
-                  const matchRects = [];
-                  for (let i = matchIndex; i < matchIndex + text.length && i < charPositions.length; i++) {
-                    matchRects.push(charPositions[i].rect);
+                  const matchRects = calculateMatchRects3(textItems, matchIndex, text.length, viewport);
+                  if (matchRects.length > 0) {
+                    const highlight = annotationStore.getState().addHighlight({
+                      pageNumber: pageNum,
+                      rects: matchRects,
+                      color,
+                      text: fullText.substring(matchIndex, matchIndex + text.length)
+                    });
+                    highlightIds.push(highlight.id);
                   }
-                  const mergedRects = mergeRects2(matchRects);
-                  const highlight = annotationStore.getState().addHighlight({
-                    pageNumber: pageNum,
-                    rects: mergedRects,
-                    color,
-                    text: fullText.substring(matchIndex, matchIndex + text.length)
-                  });
-                  highlightIds.push(highlight.id);
                   startIndex = matchIndex + 1;
                 }
               } catch {
@@ -9628,33 +9640,22 @@ var init_PDFViewerClient = __esm({
                 const textContent = await page.getTextContent();
                 const viewport = page.getViewport({ scale: 1 });
                 let fullText = "";
-                const charPositions = [];
+                const textItems = [];
                 for (const item of textContent.items) {
                   if ("str" in item && item.str) {
-                    const tx = item.transform;
-                    const x = tx[4];
-                    const y = viewport.height - tx[5];
-                    const width = item.width ?? 0;
-                    const height = item.height ?? 12;
-                    const charWidth = item.str.length > 0 ? width / item.str.length : width;
-                    for (let i = 0; i < item.str.length; i++) {
-                      charPositions.push({
-                        char: item.str[i],
-                        rect: {
-                          x: x + i * charWidth,
-                          y: y - height,
-                          width: charWidth,
-                          height
-                        }
-                      });
-                    }
+                    textItems.push({
+                      text: item.str,
+                      transform: item.transform,
+                      width: item.width ?? 0,
+                      height: item.height ?? 12
+                    });
                     fullText += item.str;
                   }
                 }
                 const textToSearch = caseSensitive ? fullText : fullText.toLowerCase();
                 let startIndex = 0;
                 while (true) {
-                  let matchIndex = textToSearch.indexOf(searchText, startIndex);
+                  const matchIndex = textToSearch.indexOf(searchText, startIndex);
                   if (matchIndex === -1) break;
                   if (wholeWord) {
                     const beforeChar = matchIndex > 0 ? textToSearch[matchIndex - 1] : " ";
@@ -9664,26 +9665,24 @@ var init_PDFViewerClient = __esm({
                       continue;
                     }
                   }
-                  const matchRects = [];
-                  for (let i = matchIndex; i < matchIndex + query.length && i < charPositions.length; i++) {
-                    matchRects.push(charPositions[i].rect);
+                  const matchRects = calculateMatchRects3(textItems, matchIndex, query.length, viewport);
+                  if (matchRects.length > 0) {
+                    const highlight = annotationStore.getState().addHighlight({
+                      pageNumber: pageNum,
+                      rects: matchRects,
+                      color,
+                      text: fullText.substring(matchIndex, matchIndex + query.length),
+                      source: "search"
+                    });
+                    result.matchCount++;
+                    result.highlightIds.push(highlight.id);
+                    result.matches.push({
+                      pageNumber: pageNum,
+                      text: fullText.substring(matchIndex, matchIndex + query.length),
+                      highlightId: highlight.id,
+                      rects: matchRects
+                    });
                   }
-                  const mergedRects = mergeRects2(matchRects);
-                  const highlight = annotationStore.getState().addHighlight({
-                    pageNumber: pageNum,
-                    rects: mergedRects,
-                    color,
-                    text: fullText.substring(matchIndex, matchIndex + query.length),
-                    source: "search"
-                  });
-                  result.matchCount++;
-                  result.highlightIds.push(highlight.id);
-                  result.matches.push({
-                    pageNumber: pageNum,
-                    text: fullText.substring(matchIndex, matchIndex + query.length),
-                    highlightId: highlight.id,
-                    rects: mergedRects
-                  });
                   startIndex = matchIndex + 1;
                 }
               } catch {