kordoc 2.0.1 → 2.0.2

package/dist/index.js

@@ -138,7 +138,7 @@ import { inflateRawSync } from "zlib";
 import { DOMParser } from "@xmldom/xmldom";
 
 // src/utils.ts
-var VERSION = true ? "2.0.1" : "0.0.0-dev";
+var VERSION = true ? "2.0.2" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -388,8 +388,11 @@ function blocksToMarkdown(blocks) {
       if (lines.length > 0 && lines[lines.length - 1] !== "") {
         lines.push("");
       }
-      lines.push(tableToMarkdown(block.table));
-      lines.push("");
+      const tableMd = tableToMarkdown(block.table);
+      if (tableMd) {
+        lines.push(tableMd);
+        lines.push("");
+      }
     }
   }
   return lines.join("\n").trim();
@@ -399,6 +402,7 @@ function tableToMarkdown(table) {
   const { cells, rows: numRows, cols: numCols } = table;
   if (numRows === 1 && numCols === 1) {
     const content = sanitizeText(cells[0][0].text);
+    if (!content) return "";
     return content.split(/\n/).map((line) => {
       const trimmed = line.trim();
       if (!trimmed) return "";
@@ -435,9 +439,9 @@ function tableToMarkdown(table) {
     const row = display[r];
     const isEmptyPlaceholder = row.every((cell) => cell === "");
     if (isEmptyPlaceholder) continue;
-    const hasSkippedCols = row.some((cell, c) => cell === "" && skip.has(`${r},${c}`));
     const nonEmptyCols = row.filter((cell) => cell !== "");
-    if (!hasSkippedCols && nonEmptyCols.length === 1 && row[0] !== "" && row.slice(1).every((c) => c === "")) {
+    const hasSkipInRow = row.some((_, c) => skip.has(`${r},${c}`));
+    if (!hasSkipInRow && nonEmptyCols.length === 1 && row[0] !== "" && row.slice(1).every((c) => c === "")) {
       pendingFirstCol = row[0];
       continue;
     }
@@ -852,7 +856,8 @@ function detectHwpxHeadings(blocks, styleMap) {
       else if (ratio >= HEADING_RATIO_H2) level = 2;
       else if (ratio >= HEADING_RATIO_H3) level = 3;
     }
-    if (/^제\d+[조장절편]/.test(text) && text.length <= 50) {
+    const compactText = text.replace(/\s+/g, "");
+    if (/^제\d+[조장절편]/.test(compactText) && text.length <= 50) {
       if (level === 0) level = 3;
     }
     if (level > 0) {
@@ -904,9 +909,14 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
         if (newTable.rows.length > 0) {
           if (tableStack.length > 0) {
             const parentTable = tableStack.pop();
-            const nestedText = convertTableToText(newTable.rows);
-            if (parentTable.cell) {
-              parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
+            const nestedCols = Math.max(...newTable.rows.map((r) => r.length));
+            if (newTable.rows.length >= 3 && nestedCols >= 2) {
+              blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
+            } else {
+              const nestedText = convertTableToText(newTable.rows);
+              if (parentTable.cell) {
+                parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
+              }
             }
             tableCtx = parentTable;
           } else {
@@ -1006,9 +1016,14 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
         if (newTable.rows.length > 0) {
           if (tableStack.length > 0) {
             const parentTable = tableStack.pop();
-            const nestedText = convertTableToText(newTable.rows);
-            if (parentTable.cell) {
-              parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
+            const nestedCols = Math.max(...newTable.rows.map((r) => r.length));
+            if (newTable.rows.length >= 3 && nestedCols >= 2) {
+              blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
+            } else {
+              const nestedText = convertTableToText(newTable.rows);
+              if (parentTable.cell) {
+                parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
+              }
             }
             tableCtx = parentTable;
           } else {
@@ -1019,13 +1034,20 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
           tableCtx = tableStack.length > 0 ? tableStack.pop() : null;
         }
       } else if (localTag === "pic" || localTag === "shape" || localTag === "drawingObject") {
-        const imgRef = extractImageRef(el);
-        if (imgRef) {
-          blocks.push({ type: "image", text: imgRef, pageNumber: sectionNum });
-        } else if (warnings && sectionNum) {
-          warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
+        const drawTextChild = findDescendant(el, "drawText");
+        if (drawTextChild) {
+          extractDrawTextBlocks(drawTextChild, blocks, styleMap, sectionNum);
+        } else {
+          const imgRef = extractImageRef(el);
+          if (imgRef) {
+            blocks.push({ type: "image", text: imgRef, pageNumber: sectionNum });
+          } else if (warnings && sectionNum) {
+            warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
+          }
         }
-      } else if (localTag === "r" || localTag === "run" || localTag === "ctrl") {
+      } else if (localTag === "drawText") {
+        extractDrawTextBlocks(el, blocks, styleMap, sectionNum);
+      } else if (localTag === "r" || localTag === "run" || localTag === "ctrl" || localTag === "rect" || localTag === "ellipse" || localTag === "polygon" || localTag === "line" || localTag === "arc" || localTag === "curve" || localTag === "connectLine" || localTag === "container") {
         walkChildren(el, d + 1);
       }
     }
@@ -1033,6 +1055,40 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
   walkChildren(node, depth);
   return tableCtx;
 }
+function findDescendant(node, targetTag, depth = 0) {
+  if (depth > 5) return null;
+  const children = node.childNodes;
+  if (!children) return null;
+  for (let i = 0; i < children.length; i++) {
+    const child = children[i];
+    if (child.nodeType !== 1) continue;
+    const tag = (child.tagName || child.localName || "").replace(/^[^:]+:/, "");
+    if (tag === targetTag) return child;
+    const found = findDescendant(child, targetTag, depth + 1);
+    if (found) return found;
+  }
+  return null;
+}
+function extractDrawTextBlocks(drawTextNode, blocks, styleMap, sectionNum) {
+  const children = drawTextNode.childNodes;
+  if (!children) return;
+  for (let i = 0; i < children.length; i++) {
+    const child = children[i];
+    if (child.nodeType !== 1) continue;
+    const tag = (child.tagName || child.localName || "").replace(/^[^:]+:/, "");
+    if (tag === "subList" || tag === "p" || tag === "para") {
+      if (tag === "subList") {
+        extractDrawTextBlocks(child, blocks, styleMap, sectionNum);
+      } else {
+        const info = extractParagraphInfo(child, styleMap);
+        const text = info.text.trim();
+        if (text) {
+          blocks.push({ type: "paragraph", text, style: info.style ?? void 0, pageNumber: sectionNum });
+        }
+      }
+    }
+  }
+}
 function extractParagraphInfo(para, styleMap) {
   let text = "";
   let href;
@@ -1051,11 +1107,18 @@ function extractParagraphInfo(para, styleMap) {
       const tag = (child.tagName || child.localName || "").replace(/^[^:]+:/, "");
       switch (tag) {
         case "t":
-          text += child.textContent || "";
+          walk(child);
           break;
-        case "tab":
-          text += "	";
+        // 자식 순회 (tab 등 하위 요소 처리)
+        case "tab": {
+          const leader = child.getAttribute("leader");
+          if (leader && leader !== "0") {
+            text += "";
+          } else {
+            text += "	";
+          }
           break;
+        }
         case "br":
           if ((child.getAttribute("type") || "line") === "line") text += "\n";
           break;
@@ -1122,6 +1185,8 @@ function extractParagraphInfo(para, styleMap) {
     }
   };
   walk(para);
+  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();
@@ -2592,7 +2657,10 @@ function parseSection(records, docInfo, warnings, sectionNum) {
         if (binId >= 0) {
           blocks.push({ type: "image", text: String(binId), pageNumber: sectionNum });
         } else {
-          warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC81C\uC5B4 \uC694\uC18C: ${ctrlId.trim()}`, code: "SKIPPED_IMAGE" });
+          const boxText = extractTextBoxText(records, i);
+          if (boxText) {
+            blocks.push({ type: "paragraph", text: boxText, pageNumber: sectionNum });
+          }
         }
       } else if (ctrlId === " elo" || ctrlId === "ole ") {
         warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC81C\uC5B4 \uC694\uC18C: ${ctrlId.trim()}`, code: "SKIPPED_IMAGE" });
@@ -2631,6 +2699,19 @@ function extractNoteText(records, ctrlIdx) {
   }
   return texts.length > 0 ? texts.join(" ") : null;
 }
+function extractTextBoxText(records, ctrlIdx) {
+  const ctrlLevel = records[ctrlIdx].level;
+  const texts = [];
+  for (let j = ctrlIdx + 1; j < records.length && j < ctrlIdx + 200; j++) {
+    const r = records[j];
+    if (r.level <= ctrlLevel) break;
+    if (r.tagId === TAG_PARA_TEXT) {
+      const t = extractText(r.data).trim();
+      if (t) texts.push(t);
+    }
+  }
+  return texts.length > 0 ? texts.join("\n") : null;
+}
 function extractHyperlinkUrl(data) {
   try {
     const httpSig = Buffer.from("http", "utf16le");