kordoc 4.24.7 → 4.24.9

package/dist/{chunk-EIGCAAVM.js → chunk-NDC4AGLX.js}

@@ -6,11 +6,11 @@ import {
   HEADING_RATIO_H3,
   tidyScriptTags,
   toTex
-} from "./chunk-BLWDQEBY.js";
+} from "./chunk-GINSLCL7.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-E35YPWOX.js";
+} from "./chunk-C6QK3PY2.js";
 
 // src/ocr/line-split.ts
 function grayCrop(rgba, pageW, box) {
@@ -482,6 +482,7 @@ function gapGlyphs(gray, w, h, ink) {
     if (ch >= bh * 0.9 && cw <= bh * 0.15) continue;
     const others = comps.filter((o) => o !== c && inBand(o) && !(o.x0 >= c.x0 && o.x1 <= c.x1 && o.y0 >= c.y0 && o.y1 <= c.y1));
     const apart = !others.some((o) => o.x0 < c.x1 + 1 && o.x1 > c.x0 - 1);
+    const abut = !apart && ch >= bh * 0.3 && ch <= bh * 0.65 && !others.some((o) => o.x0 < c.x1 && o.x1 > c.x0);
     const nearlyApart = !others.some((o) => Math.min(o.x1, c.x1) - Math.max(o.x0, c.x0) > 2);
     const spans = [], runs = [];
     let n = 0;
@@ -506,18 +507,11 @@ function gapGlyphs(gray, w, h, ink) {
     const speck = Math.max(3, bh * 0.08);
     const dot = ch <= bh * 0.25 && cw <= bh * 0.25 && ch >= speck && cw >= speck && cw >= ch * 0.6 && cw <= ch * 1.6;
     const small = ch <= bh * 0.25 && cw <= bh * 0.25 && ch >= 2 && cw >= 2 && cw >= ch * 0.6 && cw <= ch * 1.6 && fill >= 0.6 && cy >= 0.3 && cy <= 0.7;
-    if (!apart && !(nearlyApart && ch >= bh * 0.6)) {
+    if (!apart && !abut && !(nearlyApart && ch >= bh * 0.6)) {
       if (small) out.push({ x0: c.x0, x1: c.x1, mark: "\xB7
" });
       continue;
     }
-    if (!apart) {
-    } else if (dot && fill >= 0.6 && cy >= 0.3 && cy <= 0.7) mark = "\xB7";
-    else if (small) mark = "\xB7
";
-    else if (dot && fill >= 0.6 && cy >= 0.75 && cw >= ch * 0.8) mark = ".";
-    else if (ch <= bh * 0.4 && ch >= speck && cw >= ch * 0.75 && cw <= ch * 1.3 && fill < 0.6 && runs[ch >> 1] === 2 && cy >= 0.6) mark = "\u3002";
-    else if (ch <= bh * 0.15 && cw >= ch * 2 && cw <= bh * 0.8 && fill >= 0.7 && cy >= 0.35 && cy <= 0.75) mark = "-";
-    else if (ch >= bh * 0.5 && cw >= ch * 0.9 && cw <= ch * 1.4 && fill >= 0.4 && runs.every((r) => r <= 1) && top <= 0.4 && bottom >= 0.8) mark = "\u25B2";
-    else if (ch >= bh * 0.3 && ch <= bh * 0.65 && fill < 0.5) {
+    const bracket = () => {
       const stem = (from, to) => {
         let sx = 0, k = 0;
         for (let y = from; y < to; y++) for (let x = c.x0; x < c.x1; x++) if (label[y * w + x] === c.id) {
@@ -526,9 +520,21 @@ function gapGlyphs(gray, w, h, ink) {
         }
         return k ? sx / k / cw : 0.5;
       };
-      if (c.y0 <= bandT + bh * 0.15 && top >= 0.7 && bottom <= 0.5 && stem(c.y0 + band, c.y1) <= 0.35) mark = "\u300C";
-      else if (c.y1 >= bandB - bh * 0.15 && bottom >= 0.7 && top <= 0.5 && stem(c.y0, c.y1 - band) >= 0.65) mark = "\u300D";
-    } else if (ch >= charH * 0.6 && cw >= ch * 0.8 && cw <= ch * 1.25 && fill < 0.5 && top >= 0.8 && bottom >= 0.8 && runs[ch >> 1] === 2) mark = ch <= charH * 0.92 ? "\u25A1" : "\u25A1\0";
+      if (c.y0 <= bandT + bh * 0.15 && top >= 0.7 && bottom <= 0.5 && stem(c.y0 + band, c.y1) <= 0.35) return "\u300C";
+      if (c.y1 >= bandB - bh * 0.15 && bottom >= 0.7 && top <= 0.5 && stem(c.y0, c.y1 - band) >= 0.65) return "\u300D";
+      return null;
+    };
+    if (!apart && !abut) {
+    } else if (abut) {
+      if (fill < 0.5) mark = bracket();
+    } else if (dot && fill >= 0.6 && cy >= 0.3 && cy <= 0.7) mark = "\xB7";
+    else if (small) mark = "\xB7
";
+    else if (dot && fill >= 0.6 && cy >= 0.75 && cw >= ch * 0.8) mark = ".";
+    else if (ch <= bh * 0.4 && ch >= speck && cw >= ch * 0.75 && cw <= ch * 1.3 && fill < 0.6 && runs[ch >> 1] === 2 && cy >= 0.6) mark = "\u3002";
+    else if (ch <= bh * 0.15 && cw >= ch * 2 && cw <= bh * 0.8 && fill >= 0.7 && cy >= 0.35 && cy <= 0.75) mark = "-";
+    else if (ch >= bh * 0.5 && cw >= ch * 0.9 && cw <= ch * 1.4 && fill >= 0.4 && runs.every((r) => r <= 1) && top <= 0.4 && bottom >= 0.8) mark = "\u25B2";
+    else if (ch >= bh * 0.3 && ch <= bh * 0.65 && fill < 0.5) mark = bracket();
+    else if (ch >= charH * 0.6 && cw >= ch * 0.8 && cw <= ch * 1.25 && fill < 0.5 && top >= 0.8 && bottom >= 0.8 && runs[ch >> 1] === 2) mark = ch <= charH * 0.92 ? "\u25A1" : "\u25A1\0";
     if (!mark && ch >= bh * 0.6 && cw >= ch * 0.25 && cw <= ch * 0.7 && fill <= 0.4 && runs.every((r) => r <= 1)) {
       const mx = (y) => {
         let sx = 0, k = 0;
@@ -598,6 +604,34 @@ function dashRatio(gray, w, h, ink, cx) {
   const dash = comps.filter((c) => c.y1 - c.y0 <= bh * 0.25 && c.x1 - c.x0 >= (c.y1 - c.y0) * 2 && Math.abs((c.x0 + c.x1) / 2 - cx) <= bh * 0.6).sort((a, b) => Math.abs((a.x0 + a.x1) / 2 - cx) - Math.abs((b.x0 + b.x1) / 2 - cx))[0];
   return dash ? (dash.x1 - dash.x0) / bh : null;
 }
+function leadingRing(gray, w, h, ink, firstX) {
+  if (firstX < h * 0.8) return false;
+  const isInk = (v) => ink.darkInk ? v <= ink.threshold : v > ink.threshold;
+  let left = -1;
+  for (let x = 0; x < w && left < 0; x++) for (let y = 0; y < h; y++) if (isInk(gray[y * w + x])) {
+    left = x;
+    break;
+  }
+  if (left < 0 || ringBullet(gray, w, h, ink, left + h * 0.25) !== "\u25CB") return false;
+  const colInk = (x) => {
+    for (let y = 0; y < h; y++) if (isInk(gray[y * w + x])) return true;
+    return false;
+  };
+  let right = left;
+  while (right + 1 < w && colInk(right + 1)) right++;
+  let top = h, bottom = -1;
+  for (let y = 0; y < h; y++) for (let x = left; x <= right; x++) if (isInk(gray[y * w + x])) {
+    top = Math.min(top, y);
+    bottom = Math.max(bottom, y);
+    break;
+  }
+  const cw = right - left + 1, ch = bottom - top + 1, kx = Math.max(1, Math.round(cw * 0.1)), ky = Math.max(1, Math.round(ch * 0.1));
+  const corner = (x0, y0) => {
+    for (let y = y0; y < y0 + ky; y++) for (let x = x0; x < x0 + kx; x++) if (isInk(gray[y * w + x])) return true;
+    return false;
+  };
+  return !corner(left, top) && !corner(right + 1 - kx, top) && !corner(left, bottom + 1 - ky) && !corner(right + 1 - kx, bottom + 1 - ky);
+}
 function discAt(gray, w, h, ink, cx) {
   const { comps, label } = components(gray, w, h, ink);
   let charH = 0;
@@ -1873,7 +1907,17 @@ function dropHeadBandClipGrids(clipGrids, lineGrids) {
   if (clipGrids.length === 0 || lineGrids.length === 0) return clipGrids;
   const near2 = (a, b) => Math.abs(a - b) <= SHADE_CLIP_TOL;
   const overlaps = (a, b) => Math.min(a.x2, b.x2) > Math.max(a.x1, b.x1) && Math.min(a.y2, b.y2) > Math.max(a.y1, b.y1);
-  return clipGrids.filter((c) => c.clipParent || !c.cells?.length || !lineGrids.some((l) => l.rowYs.length > c.rowYs.length && l.colXs.length === c.colXs.length && c.colXs.every((x, k) => near2(x, l.colXs[k])) && c.rowYs.every((y, k) => near2(y, l.rowYs[k])) && !clipGrids.some((o) => o !== c && overlaps(o.bbox, l.bbox))));
+  const kept = clipGrids.filter((c) => c.clipParent || !c.cells?.length || !lineGrids.some((l) => l.rowYs.length > c.rowYs.length && l.colXs.length === c.colXs.length && c.colXs.every((x, k) => near2(x, l.colXs[k])) && c.rowYs.every((y, k) => near2(y, l.rowYs[k])) && !clipGrids.some((o) => o !== c && overlaps(o.bbox, l.bbox))));
+  return kept.filter((c) => !lineGrids.some((l) => headBandTiles(kept, l, near2, overlaps).includes(c)));
+}
+function headBandTiles(clips, l, near2, overlaps) {
+  const onRow = (y) => l.rowYs.some((r) => near2(r, y));
+  const band = clips.filter((c) => !c.clipParent && c.cells?.length && near2(c.bbox.y2, l.bbox.y2) && c.bbox.y1 > l.bbox.y1 + SHADE_CLIP_TOL && c.bbox.x1 >= l.bbox.x1 - SHADE_CLIP_TOL && c.bbox.x2 <= l.bbox.x2 + SHADE_CLIP_TOL && c.rowYs.every(onRow)).sort((a, b) => a.bbox.x1 - b.bbox.x1);
+  if (band.length < 2 || !near2(band[0].bbox.x1, l.bbox.x1) || !near2(band[band.length - 1].bbox.x2, l.bbox.x2)) return [];
+  if (band.some((c, k) => k > 0 && c.bbox.x1 > band[k - 1].bbox.x2 + SHADE_CLIP_TOL)) return [];
+  const bottom = Math.min(...band.map((c) => c.bbox.y1));
+  const body = { ...l.bbox, y2: bottom };
+  return clips.some((o) => !band.includes(o) && overlaps(o.bbox, body)) ? [] : band;
 }
 function enforceMinWidth(colXs, minWidth, narrowRuns = true) {
   if (colXs.length <= 2) return colXs;
@@ -9988,6 +10032,7 @@ export {
   gapGlyphs,
   ringBullet,
   dashRatio,
+  leadingRing,
   discAt,
   serifOne,
   starRun,
@@ -10050,4 +10095,4 @@ export {
   UNRULED_REGION_TABLES,
   mergeOcrImageRegions
 };
-//# sourceMappingURL=chunk-EIGCAAVM.js.map
+//# sourceMappingURL=chunk-NDC4AGLX.js.map