kordoc 4.24.13 → 4.24.15

package/dist/{chunk-WAWREGAC.js → chunk-UV2ZS65C.js}

@@ -2,7 +2,7 @@
 
 // src/utils.ts
 import { format } from "util";
-var VERSION = true ? "4.24.13" : "0.0.0-dev";
+var VERSION = true ? "4.24.15" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -172,4 +172,4 @@ export {
   classifyError,
   partExtension
 };
-//# sourceMappingURL=chunk-WAWREGAC.js.map
+//# sourceMappingURL=chunk-UV2ZS65C.js.map

package/dist/{chunk-4I2CKMKU.js → chunk-VNB2IKVA.js}

@@ -1,12 +1,12 @@
 #!/usr/bin/env node
 import {
   parse
-} from "./chunk-PUHBCAXI.js";
+} from "./chunk-7Y4MVNYA.js";
 import {
   classifyError,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-WAWREGAC.js";
+} from "./chunk-UV2ZS65C.js";
 
 // src/cli/worker-parse.ts
 import { readFileSync, statSync } from "fs";
@@ -26,4 +26,4 @@ async function parseWorkerFile(absPath, options) {
 export {
   parseWorkerFile
 };
-//# sourceMappingURL=chunk-4I2CKMKU.js.map
+//# sourceMappingURL=chunk-VNB2IKVA.js.map

package/dist/{chunk-IQGXY63Y.js → chunk-VSYLST4K.js}

@@ -5,15 +5,15 @@ import {
   HEADING_RATIO_H3,
   tidyScriptTags,
   toTex
-} from "./chunk-BJBPBU5B.js";
+} from "./chunk-Y3VNNGB2.js";
 import {
   ensureModelsIn,
   getModelsDir
-} from "./chunk-QAGPSWGH.js";
+} from "./chunk-PCHGHLUP.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-OVQWOOS3.js";
+} from "./chunk-IGLRJCO6.js";
 
 // src/ocr/models.ts
 import { join } from "path";
@@ -696,8 +696,8 @@ 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) 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";
+      if (c.y0 <= bandT + bh * 0.2 && top >= 0.7 && bottom <= 0.5 && stem(c.y0 + band, c.y1) <= 0.35) return "\u300C";
+      if (c.y1 >= bandB - bh * 0.2 && bottom >= 0.7 && top <= 0.5 && stem(c.y0, c.y1 - band) >= 0.65) return "\u300D";
       return null;
     };
     if (flushDot) mark = ".";
@@ -972,12 +972,30 @@ function leadingBullet(gray, w, h, ink, firstX, squareMax = 0.92) {
   const boxed = cl.length === 1 && midRuns === 2 && edgeFull(y0, 1) && edgeFull(y1 - 1, -1) && thin(run(x0, my0, 1, 0), run(x1 - 1, my0, -1, 0));
   const top = run(mx0, y0, 0, 1), bottom = run(mx0, y1 - 1, 0, -1);
   const shadow = !boxed ? null : thin(bottom, top) ? "\u2750" : thin(top, bottom) ? "\u274F" : null;
+  const diamond = (() => {
+    if (rel < 0.5 || rel > 1) return false;
+    const half = ch / 2, cyc = y0 + half;
+    let ok = 0;
+    for (let y = y0; y < y1; y++) {
+      let l = -1, r = -1;
+      for (let x = x0; x < x1; x++) if (cl.some((c) => label[y * w + x] === c.id)) {
+        if (l < 0) l = x;
+        r = x;
+      }
+      const want = cw * (1 - Math.abs(y + 0.5 - cyc) / half);
+      if (l >= 0 && Math.abs(r - l + 1 - want) <= Math.max(2, cw * 0.15)) ok++;
+    }
+    return ok >= ch * 0.85;
+  })();
   let mark = null;
+  if (diamond) mark = cl.length >= 2 && core ? "\u25C8" : cl.length === 1 && fill >= 0.4 ? "\u25C6" : cl.length === 1 ? "\u25C7" : null;
+  if (mark) return { mark, covers: firstX <= x1 + ch * 0.3 };
   if (shadow && rel >= 0.6 && rel <= 0.92) mark = shadow;
   else if (hollowSquare && rel >= 0.6 && rel <= squareMax) mark = "\u25A1";
   else if (filledSquare && rel >= 0.6 && rel <= 0.92) mark = "\u25A3";
   else if (nested && fill < 0.45 && rel >= 0.6) mark = "\u25CE";
   else if (cl.length === 1 && fill >= 0.9 && rel >= 0.28 && rel <= 0.6) mark = "\u25AA";
+  else if (cl.length === 1 && fill >= 0.9 && rel > 0.6 && rel <= 1) mark = "\u25A0";
   else if (cl.length === 1 && fill >= 0.65 && fill <= 0.88 && rel >= 0.5 && rel <= 0.8) mark = "\u25CF";
   if (!mark) return null;
   return { mark, covers: firstX <= x1 + ch * 0.3 };
@@ -3298,6 +3316,17 @@ var WrapLexicon = class {
   heads = /* @__PURE__ */ new Set();
   /** 줄 안 이웃 어절의 앞 어절 한글 꼬리 + 뒤 어절 첫 음절("지식재산 보") — 어절 머리 증거의 거부권 */
   spacedHeads = /* @__PURE__ */ new Set();
+  /** 쪽 줄 흐름에서 넓은 줄 사이 기준선 간격(글자 크기별) — 줄 둘뿐인 묶음은 묶음 안에 견줄 간격이 없어 넓은 줄 간격 본문을 문단 경계로 봤다 */
+  pitches = [];
+  /** 줄 흐름의 이웃한 넓은 두 줄(같은 글자 크기) 사이 간격 하나를 더한다 */
+  addPitch(fontSize, pitch) {
+    if (fontSize > 0 && pitch > 0 && pitch < 4 * fontSize) this.pitches.push({ fs: fontSize, pitch });
+  }
+  /** 이 글자 크기(±15%) 본문의 줄 간격 — 표본 넷 이상의 중앙값, 없으면 undefined */
+  bodyPitch(fontSize) {
+    const ps = this.pitches.filter((p) => Math.abs(p.fs - fontSize) <= 0.15 * fontSize).map((p) => p.pitch).sort((a, b) => a - b);
+    return ps.length >= 4 ? ps[ps.length >> 1] : void 0;
+  }
   /** 줄 글 한 줄을 증거로 더한다 — 탭(큰 갭)으로 나뉜 조각은 다른 칸·단이라 조각 사이는 어절 경계 증거로 쓰지 않는다 */
   addLine(text) {
     for (const seg of text.replace(MARKUP, "").split("	")) {
@@ -3558,7 +3587,8 @@ function bodyLineJoins(lines, lex) {
   for (let i = 0; i + 1 < lines.length; i++) {
     const a = lines[i], b = lines[i + 1];
     const fs = a.fontSize;
-    const others = i === i1 ? p2 : p1;
+    const own = i === i1 ? p2 : p1;
+    const others = Number.isFinite(own) ? own : lex?.bodyPitch(fs) ?? Infinity;
     const maxPitch = Number.isFinite(others) ? Math.min(BODY_MAX_PITCH_ABS_EM * fs, Math.max(BODY_MAX_PITCH_EM * fs, others * BODY_PITCH_REL)) : BODY_MAX_PITCH_EM * fs;
     const wordWrap = ragged && !full(a) && nextWordNoRoom(a, b) && !MULTI_LEVEL_NUMBER.test(b.text.trimStart());
     const wraps = fs > 0 && (full(a) || wordWrap) && a.right - a.left >= BODY_MIN_WIDTH_EM * fs && a.y - b.y > 0 && a.y - b.y < maxPitch && Math.abs(b.fontSize - fs) <= 0.15 * fs && !startsNewItem(a.text, b.text);
@@ -9357,7 +9387,18 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
     return attachSideTabBlocks(tab, body, pageNum);
   }
   const lex = lexicon ?? new WrapLexicon();
-  for (const line of streamLines(items)) lex.addLine(mergeLineSimple(line));
+  {
+    let prev = null;
+    for (const line of streamLines(items)) {
+      lex.addLine(mergeLineSimple(line));
+      const fs = dominantStyle(line)?.fontSize ?? 0;
+      const left = Math.min(...line.map((i) => i.x)), right = Math.max(...line.map((i) => i.x + i.w));
+      const y = line.reduce((n, i) => n + i.y, 0) / line.length;
+      const wide = fs > 0 && right - left >= 12 * fs;
+      if (wide && prev && Math.abs(prev.fs - fs) <= 0.15 * fs) lex.addPitch(fs, prev.y - y);
+      prev = wide ? { y, fs } : null;
+    }
+  }
   const extracted = extractLines(opList.fnArray, opList.argsArray);
   let { horizontals, verticals } = extracted;
   const rawRules = extracted.horizontals.concat(extracted.shortH);
@@ -10417,4 +10458,4 @@ export {
   UNRULED_REGION_TABLES,
   mergeOcrImageRegions
 };
-//# sourceMappingURL=chunk-IQGXY63Y.js.map
+//# sourceMappingURL=chunk-VSYLST4K.js.map