formatferry 1.0.11 → 1.0.12

package/dist/index.js

@@ -21361,16 +21361,45 @@ var DATE_LIKE_REGEX = /^\d{1,4}[-/.]\d{1,2}([-/.]\d{1,4})?$/;
 var SUPERSCRIPT_MERGE_GAP_FACTOR = 0.3;
 var VERTICAL_OVERLAP_THRESHOLD = 0.5;
 var COLUMN_SUPPORT_THRESHOLD = 0.6;
-var COLUMN_GAP_MERGE_FACTOR = 0.5;
+var COLUMN_GAP_MERGE_FACTOR = 0.35;
+var COLUMN_BOUNDARY_VOTE_TOLERANCE_FACTOR = 0.5;
+var MIN_CELL_OVERLAP_SCORE = 0.35;
+var WIDE_WHITESPACE_MIN_RUN = 2;
 var SHORT_CELL_THRESHOLD_LATIN = 40;
 var SHORT_CELL_THRESHOLD_CJK = 20;
 var MIN_CONFIDENCE_EMIT = 0.55;
 var MIN_CONFIDENCE_HIGH = 0.7;
 var PROSE_FLOW_MAX_FRACTION = 0.3;
+function splitWideWhitespaceItems(items) {
+  const out = [];
+  for (const it of items) {
+    const tokens = splitAtWideRuns(it.str, WIDE_WHITESPACE_MIN_RUN);
+    if (tokens.length < 2 || it.str.length < 4 || it.width <= 0) {
+      out.push(it);
+      continue;
+    }
+    const perChar = it.width / it.str.length;
+    for (const t of tokens) {
+      const subX = it.x + t.start * perChar;
+      const subRight = it.x + t.end * perChar;
+      const subWidth = Math.max(subRight - subX, perChar);
+      out.push({
+        ...it,
+        str: t.str,
+        x: subX,
+        width: subWidth,
+        right: subX + subWidth
+      });
+    }
+  }
+  return out;
+}
 function preFilterAndMerge(items) {
-  const alive = items.filter(
+  const filtered = items.filter(
     (it) => it.str.trim().length > 0 && it.width > 0 && it.height > 0
   );
+  if (filtered.length === 0) return [];
+  const alive = splitWideWhitespaceItems(filtered);
   if (alive.length === 0) return [];
   const widths = alive.map((it) => estimateCharWidth(it)).filter((w) => w > 0);
   const medianCharWidth = median(widths) || 1;
@@ -21428,11 +21457,13 @@ function detectColumns(rows) {
   if (allItems.length === 0) return [];
   const medianCharWidth = median(allItems.map((it) => estimateCharWidth(it)).filter((w) => w > 0)) || 1;
   const gapTol = medianCharWidth * COLUMN_GAP_MERGE_FACTOR;
+  const hardBoundaries = voteColumnBoundaries(rows, medianCharWidth);
   const intervals = allItems.map((it) => ({ x0: it.x, x1: it.right })).sort((a, b) => a.x0 - b.x0);
   const merged = [];
   for (const iv of intervals) {
     const last = merged[merged.length - 1];
-    if (last && iv.x0 - last.x1 <= gapTol) {
+    const crossesHardBoundary = last !== void 0 && hardBoundaries.some((b) => b >= last.x1 && b <= iv.x0);
+    if (last && iv.x0 - last.x1 <= gapTol && !crossesHardBoundary) {
       last.x1 = Math.max(last.x1, iv.x1);
     } else {
       merged.push({ ...iv });
@@ -21611,6 +21642,67 @@ function estimateCharWidth(item) {
   const chars = item.str.length;
   return chars > 0 ? item.width / chars : 0;
 }
+function splitAtWideRuns(str, minRun) {
+  const tokens = [];
+  const isWs = (ch) => ch === " " || ch === "	" || ch === "\xA0";
+  let i = 0;
+  while (i < str.length) {
+    while (i < str.length && isWs(str[i])) i++;
+    if (i >= str.length) break;
+    const start = i;
+    let end = i;
+    while (i < str.length) {
+      if (isWs(str[i])) {
+        let j = i;
+        while (j < str.length && isWs(str[j])) j++;
+        if (j - i >= minRun) break;
+        i = j;
+      } else {
+        i++;
+        end = i;
+      }
+    }
+    if (end > start) tokens.push({ str: str.slice(start, end), start, end });
+  }
+  return tokens;
+}
+function voteColumnBoundaries(rows, medianCharWidth) {
+  if (rows.length < 2 || medianCharWidth <= 0) return [];
+  const candidates = [];
+  rows.forEach((row, rIdx) => {
+    const sorted = [...row.items].sort((a, b) => a.x - b.x);
+    for (let i = 1; i < sorted.length; i++) {
+      const gap = sorted[i].x - sorted[i - 1].right;
+      if (gap > 0) {
+        candidates.push({
+          mid: (sorted[i - 1].right + sorted[i].x) / 2,
+          rowIdx: rIdx
+        });
+      }
+    }
+  });
+  if (candidates.length === 0) return [];
+  candidates.sort((a, b) => a.mid - b.mid);
+  const tolerance = medianCharWidth * COLUMN_BOUNDARY_VOTE_TOLERANCE_FACTOR;
+  const clusters = [];
+  for (const c of candidates) {
+    const last = clusters[clusters.length - 1];
+    if (last && c.mid - last.mids[last.mids.length - 1] <= tolerance) {
+      last.mids.push(c.mid);
+      last.rowSet.add(c.rowIdx);
+    } else {
+      clusters.push({ mids: [c.mid], rowSet: /* @__PURE__ */ new Set([c.rowIdx]) });
+    }
+  }
+  const hard = [];
+  for (const cl of clusters) {
+    if (cl.rowSet.size / rows.length >= COLUMN_SUPPORT_THRESHOLD) {
+      hard.push(median(cl.mids));
+    }
+  }
+  hard.sort((a, b) => a - b);
+  return hard;
+}
 function median(xs) {
   if (xs.length === 0) return 0;
   const sorted = [...xs].sort((a, b) => a - b);
@@ -21702,7 +21794,7 @@ function assignRowToColumns(row, columns) {
         best = c;
       }
     }
-    if (best >= 0 && bestScore > 0.2) buckets[best].push(item);
+    if (best >= 0 && bestScore > MIN_CELL_OVERLAP_SCORE) buckets[best].push(item);
   }
   for (let c = 0; c < columns.length; c++) {
     const bucket = buckets[c].sort((a, b) => a.x - b.x);
@@ -23862,8 +23954,8 @@ import { readFileSync as readFileSync4 } from "fs";
 import { resolve, dirname as dirname3 } from "path";
 import { fileURLToPath } from "url";
 var getVersion = () => {
-  if ("1.0.11") {
-    return "1.0.11";
+  if ("1.0.12") {
+    return "1.0.12";
   }
   try {
     const __dirname = dirname3(fileURLToPath(import.meta.url));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "formatferry",
-  "version": "1.0.11",
+  "version": "1.0.12",
   "description": "FormatFerry CLI — Convert HTML, DOCX, PDF, XLSX, CSV to Markdown",
   "keywords": [
     "markdown",