kordoc 4.7.0 → 4.7.2

package/dist/index.d.cts

@@ -16,6 +16,8 @@ interface IRSpan {
     bold?: boolean;
     italic?: boolean;
     strike?: boolean;
+    /** 밑줄 — 개정문 추가·변경 표시 등. GFM 문법이 없어 <u>…</u> 인라인 HTML 로 방출 */
+    underline?: boolean;
     code?: boolean;
 }
 interface IRBlock {
@@ -81,6 +83,8 @@ interface InlineStyle {
     italic?: boolean;
     /** 취소선 — 법령 개정문 등의 삭제 표시. 판정은 취소선 모양 whitelist (비트만 믿으면 오탐) */
     strike?: boolean;
+    /** 밑줄 — 판정은 밑줄 종류 BOTTOM 한정 (NONE 이 기본 잡음, 코퍼스 실측) */
+    underline?: boolean;
     fontSize?: number;
     fontName?: string;
 }

package/dist/index.d.ts

@@ -16,6 +16,8 @@ interface IRSpan {
     bold?: boolean;
     italic?: boolean;
     strike?: boolean;
+    /** 밑줄 — 개정문 추가·변경 표시 등. GFM 문법이 없어 <u>…</u> 인라인 HTML 로 방출 */
+    underline?: boolean;
     code?: boolean;
 }
 interface IRBlock {
@@ -81,6 +83,8 @@ interface InlineStyle {
     italic?: boolean;
     /** 취소선 — 법령 개정문 등의 삭제 표시. 판정은 취소선 모양 whitelist (비트만 믿으면 오탐) */
     strike?: boolean;
+    /** 밑줄 — 판정은 밑줄 종류 BOTTOM 한정 (NONE 이 기본 잡음, 코퍼스 실측) */
+    underline?: boolean;
     fontSize?: number;
     fontName?: string;
 }

package/dist/index.js

@@ -8,14 +8,16 @@ import {
   flattenLayoutTables,
   inlineImagesIntoMarkdown,
   mapPuaText
-} from "./chunk-BCG2PKQU.js";
+} from "./chunk-PGDJUJ4L.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
 import {
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
-  HEADING_RATIO_H3,
+  HEADING_RATIO_H3
+} from "./chunk-I4TP7HYB.js";
+import {
   KordocError,
   VERSION,
   classifyError,
@@ -27,7 +29,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-YUVKU5CW.js";
+} from "./chunk-JMWBMGNH.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -680,6 +682,8 @@ function parseCharProperties(doc, map) {
         else if (localTag === "strikeout") {
           const shape = k.getAttribute("shape") || "";
           if (isRealStrikeShape(shape)) prop.strike = true;
+        } else if (localTag === "underline") {
+          if (k.getAttribute("type") === "BOTTOM") prop.underline = true;
         }
       }
       const fontFaces = el.getElementsByTagName("*");
@@ -2378,21 +2382,22 @@ function extractRunSpans(para2, ctx, mode, requireMixed) {
       if (cp?.bold) span.bold = true;
       if (cp?.italic) span.italic = true;
       if (cp?.strike) span.strike = true;
+      if (cp?.underline) span.underline = true;
     }
-    if (span.bold || span.italic || span.strike || span.code) styled = true;
+    if (span.bold || span.italic || span.strike || span.underline || span.code) styled = true;
     spans.push(span);
   }
   if (!styled || spans.length === 0) return null;
   const merged = [];
   for (const s of spans) {
     const last = merged[merged.length - 1];
-    if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.strike === !!s.strike && !!last.code === !!s.code) {
+    if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.strike === !!s.strike && !!last.underline === !!s.underline && !!last.code === !!s.code) {
       last.text += s.text;
     } else {
       merged.push(s);
     }
   }
-  if (requireMixed && !merged.some((s) => s.strike) && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
+  if (requireMixed && !merged.some((s) => s.strike || s.underline) && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
   return merged;
 }
 function gongmunDepthFromIndent(para2, left, ctx) {
@@ -4744,7 +4749,11 @@ function parseParagraph(records, start, end, ctx) {
       const style = resolveCharStyle(charShapeIds, ctx.docInfo);
       if (style) {
         block.style = style;
-        if (style.strike) block.text = headMarker ? `${headMarker} ~~${trimmed}~~` : `~~${trimmed}~~`;
+        if (style.strike || style.underline) {
+          let deco = style.strike ? `~~${trimmed}~~` : trimmed;
+          if (style.underline) deco = `<u>${deco}</u>`;
+          block.text = headMarker ? `${headMarker} ${deco}` : deco;
+        }
       }
     }
     if (footnotes.length > 0) block.footnoteText = footnotes.join("; ");
@@ -5137,13 +5146,17 @@ function resolveCharStyle(charShapeIds, docInfo) {
   if (cs.attrFlags & 1) style.italic = true;
   if (cs.attrFlags & 2) style.bold = true;
   if (hasRealStrike(cs.attrFlags)) style.strike = true;
-  return style.fontSize || style.bold || style.italic || style.strike ? style : void 0;
+  if (hasRealUnderline(cs.attrFlags)) style.underline = true;
+  return style.fontSize || style.bold || style.italic || style.strike || style.underline ? style : void 0;
 }
 function hasRealStrike(attrFlags) {
   const strikeBits = attrFlags >> 18 & 7;
   const shapeId = attrFlags >> 26 & 15;
   return strikeBits !== 0 && shapeId <= 12;
 }
+function hasRealUnderline(attrFlags) {
+  return (attrFlags >> 2 & 3) === 1;
+}
 
 // src/hwp3/parser.ts
 import { inflateRawSync as inflateRawSync5 } from "zlib";
@@ -22878,6 +22891,7 @@ function parseInlineMarkdown(text) {
   });
   text = text.replace(/!\[([^\]]*)\]\([^)]*\)/g, "$1");
   text = text.replace(/~~([^~]+)~~/g, "$1");
+  text = text.replace(/<u>([^<]*)<\/u>/g, "$1");
   let spans;
   if (inlineDoc) {
     spans = [];
@@ -31312,7 +31326,7 @@ async function parse(input, options) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-3A3HZX6O.js");
+    const { parseImageDocument } = await import("./image-ocr-TQ5THN35.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     return {
       success: true,
@@ -31371,7 +31385,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-EQN4YKU4.js");
+    const mod = await import("./parser-FXEIK3VJ.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {