formatferry 1.0.14 → 1.0.15

package/dist/index.js

@@ -22397,7 +22397,8 @@ function renderTextBlockEnhanced(items, medianPageFontSize, outlineTitles) {
       htmlBlocks.push(`<h${headingLevel}>${inlineHtml}</h${headingLevel}>`);
     } else {
       const hasWideGap = /( {3,}|\u00A0\u00A0|\t)/.test(normalizedPlain);
-      const isShort = normalizedPlain.length < 100 && !/[.,:;!?]$/.test(normalizedPlain) && !/\n/.test(normalizedPlain) && !hasWideGap;
+      const looksLikeSentence = normalizedPlain.length > 40 && /^[A-Z]/.test(normalizedPlain);
+      const isShort = normalizedPlain.length < 80 && !/[.,:;!?]$/.test(normalizedPlain) && !/\n/.test(normalizedPlain) && !hasWideGap && !looksLikeSentence;
       if (isShort && !hasAnyStyle) {
         htmlBlocks.push(`<p><strong>${inlineHtml}</strong></p>`);
       } else {
@@ -22528,7 +22529,15 @@ function isHeaderFooterItem(it, suppressSet, Y_TOLERANCE = 5) {
   const yBucket = Math.round(it.y / Y_TOLERANCE) * Y_TOLERANCE;
   return suppressSet.has(`${normalized}@${yBucket}`);
 }
-var PDF_OP = { moveTo: 13, lineTo: 14, closePath: 18, rectangle: 92 };
+var PDF_OP = {
+  moveTo: 13,
+  lineTo: 14,
+  closePath: 18,
+  rectangle: 92,
+  stroke: 19,
+  closeStroke: 20,
+  fillStroke: 37
+};
 function extractVectorLines(fnArray, argsArray, pageWidth, pageHeight) {
   const MIN_H_SPAN = pageWidth * 0.3;
   const MIN_V_SPAN = pageHeight * 0.3;
@@ -22537,22 +22546,28 @@ function extractVectorLines(fnArray, argsArray, pageWidth, pageHeight) {
   let curY = 0;
   let pathStartX = 0;
   let pathStartY = 0;
-  const pushIfSignificant = (x0, y0, x1, y1) => {
-    const dx = Math.abs(x1 - x0);
-    const dy = Math.abs(y1 - y0);
-    if (dy < 2 && dx >= MIN_H_SPAN) {
-      lines.push({
-        orientation: "h",
-        position: (y0 + y1) / 2,
-        span: [Math.min(x0, x1), Math.max(x0, x1)]
-      });
-    } else if (dx < 2 && dy >= MIN_V_SPAN) {
-      lines.push({
-        orientation: "v",
-        position: (x0 + x1) / 2,
-        span: [Math.min(y0, y1), Math.max(y0, y1)]
-      });
+  let pending = [];
+  const isPathOp = (op) => op === PDF_OP.moveTo || op === PDF_OP.lineTo || op === PDF_OP.closePath || op === PDF_OP.rectangle;
+  const isStrokeOp = (op) => op === PDF_OP.stroke || op === PDF_OP.closeStroke || op === PDF_OP.fillStroke;
+  const flushPending = () => {
+    for (const p of pending) {
+      const dx = Math.abs(p.x1 - p.x0);
+      const dy = Math.abs(p.y1 - p.y0);
+      if (dy < 2 && dx >= MIN_H_SPAN) {
+        lines.push({
+          orientation: "h",
+          position: (p.y0 + p.y1) / 2,
+          span: [Math.min(p.x0, p.x1), Math.max(p.x0, p.x1)]
+        });
+      } else if (dx < 2 && dy >= MIN_V_SPAN) {
+        lines.push({
+          orientation: "v",
+          position: (p.x0 + p.x1) / 2,
+          span: [Math.min(p.y0, p.y1), Math.max(p.y0, p.y1)]
+        });
+      }
     }
+    pending = [];
   };
   for (let i = 0; i < fnArray.length; i++) {
     const op = fnArray[i];
@@ -22565,7 +22580,7 @@ function extractVectorLines(fnArray, argsArray, pageWidth, pageHeight) {
     } else if (op === PDF_OP.lineTo) {
       const nx = args[0] ?? 0;
       const ny = args[1] ?? 0;
-      pushIfSignificant(curX, curY, nx, ny);
+      pending.push({ x0: curX, y0: curY, x1: nx, y1: ny });
       curX = nx;
       curY = ny;
     } else if (op === PDF_OP.rectangle) {
@@ -22573,14 +22588,18 @@ function extractVectorLines(fnArray, argsArray, pageWidth, pageHeight) {
       const ry = args[1] ?? 0;
       const rw = args[2] ?? 0;
       const rh = args[3] ?? 0;
-      pushIfSignificant(rx, ry, rx + rw, ry);
-      pushIfSignificant(rx, ry + rh, rx + rw, ry + rh);
-      pushIfSignificant(rx, ry, rx, ry + rh);
-      pushIfSignificant(rx + rw, ry, rx + rw, ry + rh);
+      pending.push({ x0: rx, y0: ry, x1: rx + rw, y1: ry });
+      pending.push({ x0: rx, y0: ry + rh, x1: rx + rw, y1: ry + rh });
+      pending.push({ x0: rx, y0: ry, x1: rx, y1: ry + rh });
+      pending.push({ x0: rx + rw, y0: ry, x1: rx + rw, y1: ry + rh });
     } else if (op === PDF_OP.closePath) {
-      pushIfSignificant(curX, curY, pathStartX, pathStartY);
+      pending.push({ x0: curX, y0: curY, x1: pathStartX, y1: pathStartY });
       curX = pathStartX;
       curY = pathStartY;
+    } else if (isStrokeOp(op)) {
+      flushPending();
+    } else if (!isPathOp(op)) {
+      pending = [];
     }
   }
   return lines;
@@ -22703,7 +22722,8 @@ function renderPage(items, pageNum, nextPageNum, enableTableDetection, translate
           const hasWideGap = /( {3,}|\u00A0\u00A0|\t)/.test(trimmed);
           const isFirstPara = paraIdx === 1;
           const looksTabular = hasWideGap || isFirstPara && topRowLooksTabular;
-          const short = trimmed.length < 100 && !/[.,:;!?]$/.test(trimmed) && !/\n/.test(trimmed) && !looksTabular;
+          const looksLikeSentence = trimmed.length > 40 && /^[A-Z]/.test(trimmed);
+          const short = trimmed.length < 80 && !/[.,:;!?]$/.test(trimmed) && !/\n/.test(trimmed) && !looksTabular && !looksLikeSentence;
           if (short) {
             htmlBlocks.push(`<p><strong>${escapeHtml(trimmed)}</strong></p>`);
           } else {
@@ -25294,8 +25314,8 @@ import { readFileSync as readFileSync4 } from "fs";
 import { resolve, dirname as dirname3 } from "path";
 import { fileURLToPath } from "url";
 var getVersion = () => {
-  if ("1.0.14") {
-    return "1.0.14";
+  if ("1.0.15") {
+    return "1.0.15";
   }
   try {
     const __dirname = dirname3(fileURLToPath(import.meta.url));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "formatferry",
-  "version": "1.0.14",
+  "version": "1.0.15",
   "description": "FormatFerry CLI — Convert HTML, DOCX, PDF, XLSX, CSV, JSON, XML, PPTX to Markdown",
   "keywords": [
     "markdown",