kordoc 4.2.1 → 4.2.3

package/dist/index.d.cts

@@ -15,6 +15,7 @@ interface IRSpan {
     text: string;
     bold?: boolean;
     italic?: boolean;
+    strike?: boolean;
     code?: boolean;
 }
 interface IRBlock {
@@ -78,6 +79,8 @@ interface BoundingBox {
 interface InlineStyle {
     bold?: boolean;
     italic?: boolean;
+    /** 취소선 — 법령 개정문 등의 삭제 표시. 판정은 취소선 모양 whitelist (비트만 믿으면 오탐) */
+    strike?: boolean;
     fontSize?: number;
     fontName?: string;
 }
@@ -96,7 +99,8 @@ interface IRCell {
     rowSpan: number;
     /**
      * 셀 내부 블록 콘텐츠 — v3.0.
-     * 중첩 표·이미지·다중 문단을 구조 그대로 보존한다.
+     * 중첩 표·이미지·다중 문단을 구조 그대로, 문서(원문) 순서대로 보존한다.
+     * 표와 텍스트가 한 줄에 번갈아 놓인 셀도 배치 순서를 따른다 (v4.2.3, #49).
      * blocks가 있으면 text는 blocks의 평탄화 텍스트(하위 호환용)다.
      */
     blocks?: IRBlock[];
@@ -215,7 +219,11 @@ interface ParseSuccess extends ParseResultBase {
     success: true;
     /** 추출된 마크다운 텍스트 */
     markdown: string;
-    /** 중간 표현 블록 (구조화된 데이터 접근용) */
+    /**
+     * 중간 표현 블록 (구조화된 데이터 접근용).
+     * 블록은 문서(원문) 읽기 순서를 따른다 — 한 문단 안에 글자취급(treatAsChar)
+     * 표와 텍스트가 섞여 있어도 배치 순서대로 방출된다 (v4.2.3, #50).
+     */
     blocks: IRBlock[];
     /** 문서 메타데이터 */
     metadata?: DocumentMetadata;

package/dist/index.d.ts

@@ -15,6 +15,7 @@ interface IRSpan {
     text: string;
     bold?: boolean;
     italic?: boolean;
+    strike?: boolean;
     code?: boolean;
 }
 interface IRBlock {
@@ -78,6 +79,8 @@ interface BoundingBox {
 interface InlineStyle {
     bold?: boolean;
     italic?: boolean;
+    /** 취소선 — 법령 개정문 등의 삭제 표시. 판정은 취소선 모양 whitelist (비트만 믿으면 오탐) */
+    strike?: boolean;
     fontSize?: number;
     fontName?: string;
 }
@@ -96,7 +99,8 @@ interface IRCell {
     rowSpan: number;
     /**
      * 셀 내부 블록 콘텐츠 — v3.0.
-     * 중첩 표·이미지·다중 문단을 구조 그대로 보존한다.
+     * 중첩 표·이미지·다중 문단을 구조 그대로, 문서(원문) 순서대로 보존한다.
+     * 표와 텍스트가 한 줄에 번갈아 놓인 셀도 배치 순서를 따른다 (v4.2.3, #49).
      * blocks가 있으면 text는 blocks의 평탄화 텍스트(하위 호환용)다.
      */
     blocks?: IRBlock[];
@@ -215,7 +219,11 @@ interface ParseSuccess extends ParseResultBase {
     success: true;
     /** 추출된 마크다운 텍스트 */
     markdown: string;
-    /** 중간 표현 블록 (구조화된 데이터 접근용) */
+    /**
+     * 중간 표현 블록 (구조화된 데이터 접근용).
+     * 블록은 문서(원문) 읽기 순서를 따른다 — 한 문단 안에 글자취급(treatAsChar)
+     * 표와 텍스트가 섞여 있어도 배치 순서대로 방출된다 (v4.2.3, #50).
+     */
     blocks: IRBlock[];
     /** 문서 메타데이터 */
     metadata?: DocumentMetadata;

package/dist/index.js

@@ -8,7 +8,7 @@ import {
   flattenLayoutTables,
   inlineImagesIntoMarkdown,
   mapPuaText
-} from "./chunk-7IMNZ2BL.js";
+} from "./chunk-K5LGSH5T.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -27,7 +27,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-HTCKGSSW.js";
+} from "./chunk-QNVZAKUQ.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -408,7 +408,7 @@ function comResultToParseResult(pages, pageCount, warnings) {
 
 // src/hwpx/parser-shared.ts
 import { DOMParser } from "@xmldom/xmldom";
-var MAX_DECOMPRESS_SIZE = 100 * 1024 * 1024;
+var MAX_DECOMPRESS_SIZE = 256 * 1024 * 1024;
 var MAX_ZIP_ENTRIES = 500;
 var ZipBombError = class extends KordocError {
   constructor(message) {
@@ -465,6 +465,26 @@ function extractTextFromNode(node, depth = 0) {
 }
 
 // src/hwpx/styles.ts
+function isRealStrikeShape(shape) {
+  switch (shape) {
+    case "SOLID":
+    case "DASH":
+    case "DOT":
+    case "DASH_DOT":
+    case "DASH_DOT_DOT":
+    case "LONG_DASH":
+    case "CIRCLE":
+    case "DOUBLE_SLIM":
+    case "SLIM_THICK":
+    case "THICK_SLIM":
+    case "SLIM_THICK_SLIM":
+    case "WAVE":
+    case "DOUBLE_WAVE":
+      return true;
+    default:
+      return false;
+  }
+}
 async function extractHwpxStyles(zip, decompressed) {
   const result = {
     charProperties: /* @__PURE__ */ new Map(),
@@ -529,6 +549,10 @@ function parseCharProperties(doc, map) {
         const localTag = (k.tagName || "").replace(/^[^:]+:/, "");
         if (localTag === "bold") prop.bold = true;
         else if (localTag === "italic") prop.italic = true;
+        else if (localTag === "strikeout") {
+          const shape = k.getAttribute("shape") || "";
+          if (isRealStrikeShape(shape)) prop.strike = true;
+        }
       }
       const fontFaces = el.getElementsByTagName("*");
       for (let j = 0; j < fontFaces.length; j++) {
@@ -2081,7 +2105,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
         }
         break;
       case "p": {
-        const { text: rawText, href, footnote, style } = extractParagraphInfo(el, ctx.styleMap, ctx);
+        const { text: rawText, href, footnote, style, segments } = extractParagraphInfo(el, ctx.styleMap, ctx);
         let text = rawText;
         let headingLevel;
         const ph = resolveParaHeading(el, ctx);
@@ -2089,6 +2113,33 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
           if (ph?.prefix) text = ph.prefix + " " + text;
           headingLevel = ph?.headingLevel;
         }
+        if (segments) {
+          const cell2 = tableCtx?.cell ?? null;
+          if (text && cell2) cell2.text += (cell2.text ? "\n" : "") + (footnote ? `${text} (\uC8FC: ${footnote})` : text);
+          const segs = [...segments];
+          if (ph?.prefix) {
+            const fi = segs.findIndex((s) => s);
+            if (fi >= 0) segs[fi] = ph.prefix + " " + segs[fi];
+          }
+          let segIdx = 0;
+          let first = true;
+          const flush = () => {
+            const s = segs[segIdx++];
+            if (!s) return;
+            const block = { type: "paragraph", text: s, pageNumber: ctx.sectionNum };
+            if (first && !cell2) {
+              first = false;
+              if (style) block.style = style;
+              if (href) block.href = href;
+              if (footnote) block.footnoteText = footnote;
+            }
+            if (cell2) (cell2.blocks ??= []).push(block);
+            else blocks.push(block);
+          };
+          tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, ctx, depth + 1, flush);
+          while (segIdx < segs.length) flush();
+          break;
+        }
         if (text) {
           if (tableCtx?.cell) {
             const cell2 = tableCtx.cell;
@@ -2307,7 +2358,7 @@ function findTopLevelTbls(el, out, depth = 0) {
     findTopLevelTbls(ch, out, depth + 1);
   }
 }
-function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
+function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth = 0, onTbl) {
   if (depth > MAX_XML_DEPTH) return tableCtx;
   const children = node.childNodes;
   if (!children) return tableCtx;
@@ -2321,6 +2372,7 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth =
       const tag = el.tagName || el.localName || "";
       const localTag = tag.replace(/^[^:]+:/, "");
       if (localTag === "tbl") {
+        if (onTbl && isInlineTbl(el)) onTbl();
         if (!tableCtx) {
           const chan = kordocTableChannel(el, ctx);
           if (chan) {
@@ -2367,6 +2419,9 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth =
   walkChildren(node, depth);
   return tableCtx;
 }
+function isInlineTbl(tbl2) {
+  return findChildByLocalName(tbl2, "pos")?.getAttribute("treatAsChar") === "1";
+}
 function findDescendant(node, targetTag, depth = 0) {
   if (depth > 5) return null;
   const children = node.childNodes;
@@ -2540,7 +2595,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
             ctx.warnings?.push({ page: ctx.sectionNum, message: "\uBCC0\uACBD\uCD94\uC801 \uC0AD\uC81C \uD14D\uC2A4\uD2B8 \uCD9C\uB825 \uC81C\uC678", code: "HIDDEN_TEXT_FILTERED" });
           }
         } else {
-          text += t;
+          text += t.replace(/\x1E/g, "");
         }
         continue;
       }
@@ -2571,9 +2626,12 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         case "hwSpace":
           text += " ";
           break;
+        // 테이블 자체는 walkSection에서 처리 — 글자취급(inline) 표만 경계 마커를 남겨
+        // 표 앞뒤 텍스트를 문서 순서대로 분할 방출할 수 있게 한다 (#49/#50).
+        // float·페이지 앵커 표는 텍스트 흐름 불참(reflow 모델 정합) — 종전대로 텍스트 뒤 방출
         case "tbl":
+          if (isInlineTbl(child)) text += "";
           break;
-        // 테이블은 walkSection에서 처리
         // 하이퍼링크
         case "hyperlink": {
           const url = child.getAttribute("url") || child.getAttribute("href") || "";
@@ -2668,7 +2726,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
     for (const r of closed) {
       if (applied.some(([s, e]) => r.start < e && r.end > s)) continue;
       const anchor = text.slice(r.start, r.end);
-      if (!anchor.trim() || /[\n\x1F\[\]]/.test(anchor)) continue;
+      if (!anchor.trim() || /[\n\x1F\x1E\[\]]/.test(anchor)) continue;
       text = text.slice(0, r.start) + `[${anchor}](${r.url})` + text.slice(r.end);
       applied.push([r.start, r.end]);
     }
@@ -2676,10 +2734,18 @@ function extractParagraphInfo(para2, styleMap, ctx) {
   }
   const leaderIdx = text.indexOf("");
   if (leaderIdx >= 0) text = text.substring(0, leaderIdx);
-  let cleanText = text.replace(/[ \t]+/g, " ").trim();
-  if (/^그림입니다\.?\s*원본\s*그림의\s*(이름|크기)/.test(cleanText)) cleanText = "";
-  cleanText = cleanText.replace(/그림입니다\.?\s*원본\s*그림의\s*(이름|크기)[^\n]*(\n[^\n]*원본\s*그림의\s*(이름|크기)[^\n]*)*/g, "").trim();
-  cleanText = cleanText.replace(/^(?:모서리가 둥근 |둥근 )?(?:사각형|직사각형|정사각형|원|타원|삼각형|선|직선|곡선|화살표|오각형|육각형|팔각형|별|십자|구름|마름모|도넛|평행사변형|사다리꼴|개체|그리기\s?개체|묶음\s?개체|글상자|표|그림|OLE\s?개체)\s?입니다\.?$/gm, "").trim();
+  const cleanParaText = (raw) => {
+    let t = raw.replace(/[ \t]+/g, " ").trim();
+    if (/^그림입니다\.?\s*원본\s*그림의\s*(이름|크기)/.test(t)) t = "";
+    t = t.replace(/그림입니다\.?\s*원본\s*그림의\s*(이름|크기)[^\n]*(\n[^\n]*원본\s*그림의\s*(이름|크기)[^\n]*)*/g, "").trim();
+    return t.replace(/^(?:모서리가 둥근 |둥근 )?(?:사각형|직사각형|정사각형|원|타원|삼각형|선|직선|곡선|화살표|오각형|육각형|팔각형|별|십자|구름|마름모|도넛|평행사변형|사다리꼴|개체|그리기\s?개체|묶음\s?개체|글상자|표|그림|OLE\s?개체)\s?입니다\.?$/gm, "").trim();
+  };
+  let segments;
+  if (text.includes("")) {
+    segments = text.split("").map(cleanParaText);
+    text = text.replace(/\x1E/g, "");
+  }
+  const cleanText = cleanParaText(text);
   let style;
   if (styleMap && charPrId) {
     const charProp = styleMap.charProperties.get(charPrId);
@@ -2692,7 +2758,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
       if (!style.fontSize && !style.bold && !style.italic) style = void 0;
     }
   }
-  return { text: cleanText, href, footnote, style };
+  return { text: cleanText, href, footnote, style, segments };
 }
 var KORDOC_CHAR_CODE = "4";
 var KORDOC_PARA_QUOTE = "6";
@@ -2743,21 +2809,22 @@ function extractRunSpans(para2, ctx, mode, requireMixed) {
     else {
       if (cp?.bold) span.bold = true;
       if (cp?.italic) span.italic = true;
+      if (cp?.strike) span.strike = true;
     }
-    if (span.bold || span.italic || span.code) styled = true;
+    if (span.bold || span.italic || span.strike || 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.code === !!s.code) {
+    if (last && !!last.bold === !!s.bold && !!last.italic === !!s.italic && !!last.strike === !!s.strike && !!last.code === !!s.code) {
       last.text += s.text;
     } else {
       merged.push(s);
     }
   }
-  if (requireMixed && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
+  if (requireMixed && !merged.some((s) => s.strike) && !merged.some((s) => !(s.bold || s.italic || s.code))) return null;
   return merged;
 }
 function gongmunDepthFromIndent(para2, left, ctx) {
@@ -5149,7 +5216,10 @@ function parseParagraph(records, start, end, ctx) {
     if (headingLevel > 0) block.level = headingLevel;
     if (ctx.docInfo && charShapeIds.length > 0) {
       const style = resolveCharStyle(charShapeIds, ctx.docInfo);
-      if (style) block.style = style;
+      if (style) {
+        block.style = style;
+        if (style.strike) block.text = headMarker ? `${headMarker} ~~${trimmed}~~` : `~~${trimmed}~~`;
+      }
     }
     if (footnotes.length > 0) block.footnoteText = footnotes.join("; ");
     blocks.push(block);
@@ -5540,7 +5610,13 @@ function resolveCharStyle(charShapeIds, docInfo) {
   if (cs.fontSize > 0) style.fontSize = cs.fontSize / 10;
   if (cs.attrFlags & 1) style.italic = true;
   if (cs.attrFlags & 2) style.bold = true;
-  return style.fontSize || style.bold || style.italic ? style : void 0;
+  if (hasRealStrike(cs.attrFlags)) style.strike = true;
+  return style.fontSize || style.bold || style.italic || style.strike ? style : void 0;
+}
+function hasRealStrike(attrFlags) {
+  const strikeBits = attrFlags >> 18 & 7;
+  const shapeId = attrFlags >> 26 & 15;
+  return strikeBits !== 0 && shapeId <= 12;
 }
 
 // src/hwp3/parser.ts
@@ -23635,6 +23711,18 @@ function htmlCellInnerToLines(inner) {
   const lines = work.split(/<br\s*\/?>/gi).map((s) => s.trim()).filter((s) => s.length > 0);
   return { lines, hadNonText, imgSrcs };
 }
+function splitCellByTopLevelTables(html) {
+  const tables = extractTopLevelTables(html);
+  const texts = [];
+  let rest = html;
+  for (const t of tables) {
+    const i = rest.indexOf(t);
+    texts.push(rest.slice(0, i));
+    rest = rest.slice(i + t.length);
+  }
+  texts.push(rest);
+  return { texts, tables };
+}
 function extractTopLevelTables(html) {
   const result = [];
   let depth = 0;
@@ -24313,17 +24401,18 @@ function generateHtmlTableXml(rawHtml, theme, totalWidth = 44e3, style = null, r
     const rowSpan = Math.min(cell2.rowSpan, rowCnt - cell2.r);
     const lines = cellLines[i];
     let nestedH = 0;
-    const nestedXmls = [];
-    for (const nested of extractTopLevelTables(cell2.inner)) {
+    const { texts: segTexts, tables: nestedTables } = splitCellByTopLevelTables(cell2.inner);
+    const nestedXmls = nestedTables.map((nested) => {
       const sw = spanW(cell2);
       const nestedW = Math.max(Math.min(Math.max(sw - 1020, 4e3), sw - 282), 500);
       const nestedXml = generateHtmlTableXml(nested, theme, nestedW, style ? { ...style, totalWidth: nestedW } : null);
       if (nestedXml) {
-        nestedXmls.push(nestedXml);
         const szH = nestedXml.match(/<hp:sz [^>]*height="(\d+)"/)?.[1];
         nestedH += (szH ? Number(szH) : (nested.match(/<tr[\s>]/gi) ?? []).length * cellH) + 300;
       }
-    }
+      return nestedXml;
+    });
+    const segLines = segTexts.map((t) => htmlCellInnerToLines(t).lines);
     const usable = Math.max(spanW(cell2) - CELL_PAD, 1e3);
     let wrapLines = 0;
     for (const line of lines) wrapLines += Math.max(1, Math.ceil(measureTextWidth(unescapeHtml(line).trim(), measureH, 100) / usable));
@@ -24332,9 +24421,9 @@ function generateHtmlTableXml(rawHtml, theme, totalWidth = 44e3, style = null, r
     const cellHeight = Math.max(prof?.cellH.get(`${cell2.r},${cell2.c}`) ?? 0, contentH);
     const perRow = Math.ceil(cellHeight / rowSpan);
     for (let r = cell2.r; r < cell2.r + rowSpan; r++) tableRowHeights[r] = Math.max(tableRowHeights[r], perRow);
-    return { cell: cell2, rowSpan, lines, nestedXmls, imgSrcs: cellParsed[i].imgSrcs };
+    return { cell: cell2, rowSpan, lines, nestedXmls, segLines, imgSrcs: cellParsed[i].imgSrcs };
   });
-  const tcXmls = meta.map(({ cell: cell2, rowSpan, lines, nestedXmls, imgSrcs }) => {
+  const tcXmls = meta.map(({ cell: cell2, rowSpan, nestedXmls, segLines, imgSrcs }) => {
     const k = `${cell2.r},${cell2.c}`;
     const isHeader = cell2.isHeader;
     const baseCharPr = style ? style.charPr : CHAR_NORMAL;
@@ -24345,7 +24434,7 @@ function generateHtmlTableXml(rawHtml, theme, totalWidth = 44e3, style = null, r
     const paraPrId = style ? centered ? reg ? style.tblCenterParaPr ?? style.centerParaPr : style.centerParaPr : reg ? style.tblLeftParaPr ?? PARA_NORMAL : PARA_NORMAL : PARA_NORMAL;
     const picUrls = images ? [...imgSrcs] : [];
     const paras = [];
-    for (const line of lines) {
+    const pushTextLine = (line) => {
       let text = unescapeHtml(line);
       if (images) {
         const { text: rest, urls } = splitImageRefs(text);
@@ -24357,7 +24446,14 @@ function generateHtmlTableXml(rawHtml, theme, totalWidth = 44e3, style = null, r
       if (!images || text.trim() || picUrls.length === 0) {
         paras.push(`<hp:p paraPrIDRef="${paraPrId}" styleIDRef="0"><hp:run charPrIDRef="${charPrId}"><hp:t>${escapeXml(text)}</hp:t></hp:run></hp:p>`);
       }
-    }
+    };
+    segLines.forEach((seg, si) => {
+      for (const line of seg) pushTextLine(line);
+      const nestedXml = nestedXmls[si];
+      if (nestedXml) {
+        paras.push(`<hp:p paraPrIDRef="0" styleIDRef="0"><hp:run charPrIDRef="0">${nestedXml}</hp:run></hp:p>`);
+      }
+    });
     if (images && picUrls.length > 0) {
       const pics = picUrls.map((u) => {
         const part = images.take(u);
@@ -24367,9 +24463,6 @@ function generateHtmlTableXml(rawHtml, theme, totalWidth = 44e3, style = null, r
         paras.push(`<hp:p paraPrIDRef="${paraPrId}" styleIDRef="0"><hp:run charPrIDRef="${charPrId}">${pics.join("")}</hp:run></hp:p>`);
       }
     }
-    for (const nestedXml of nestedXmls) {
-      paras.push(`<hp:p paraPrIDRef="0" styleIDRef="0"><hp:run charPrIDRef="0">${nestedXml}</hp:run></hp:p>`);
-    }
     if (paras.length === 0) {
       paras.push(`<hp:p paraPrIDRef="${paraPrId}" styleIDRef="0"><hp:run charPrIDRef="${charPrId}"><hp:t></hp:t></hp:run></hp:p>`);
     }
@@ -30463,7 +30556,7 @@ async function parse(input, options) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-DALT7PFQ.js");
+    const { parseImageDocument } = await import("./image-ocr-PLTV4NOK.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     return {
       success: true,
@@ -30521,7 +30614,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-33TK5SMH.js");
+    const mod = await import("./parser-Y2X4JMAK.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {