kordoc 2.7.2 → 2.9.0

package/dist/index.d.cts

@@ -164,6 +164,34 @@ interface ParseSuccess extends ParseResultBase {
     warnings?: ParseWarning[];
     /** 추출된 이미지 목록 — 마크다운에서 파일명으로 참조됨 */
     images?: ExtractedImage[];
+    /** 페이지별 텍스트 품질 신호 — PDF에서만 제공 */
+    pageQuality?: PageQuality[];
+    /** 문서 단위 품질 요약 — PDF에서만 제공 */
+    qualitySummary?: DocumentQualitySummary;
+}
+/** 페이지별 텍스트 품질 신호 (PDF 전용). 자세한 설명은 src/pdf/quality.ts */
+interface PageQuality {
+    page: number;
+    textChars: number;
+    hangulRatio: number;
+    controlCharRatio: number;
+    replacementCharRatio: number;
+    puaRatio: number;
+    needsOcr: boolean;
+    ocrReason?: "low_text" | "high_pua" | "high_control" | "high_replacement";
+}
+/** 문서 단위 품질 요약 (PDF 전용). */
+interface DocumentQualitySummary {
+    totalPages: number;
+    totalTextChars: number;
+    avgHangulRatio: number;
+    avgControlCharRatio: number;
+    avgReplacementCharRatio: number;
+    avgPuaRatio: number;
+    lowTextPageCount: number;
+    highPuaPageCount: number;
+    needsOcr: boolean;
+    ocrCandidatePages: number[];
 }
 /** 추출된 이미지 — ParseSuccess.images에 포함 */
 interface ExtractedImage {
@@ -315,10 +343,36 @@ declare function fillHwpx(hwpxBuffer: ArrayBuffer, values: Record<string, string
  *       순서/비순서 리스트, 수평선, 인용문, 테이블
  * jszip으로 HWPX ZIP 패키징.
  */
+/** HWPX 생성 시 적용할 시각 테마 (모두 선택) */
+interface HwpxTheme {
+    /**
+     * 헤딩 레벨별 텍스트 색상. 미지정 시 검정.
+     * 현재 charPr 매핑은 h1/h2/h3/h4 4단계 (h5, h6은 h4와 같은 charPr 공유)이므로
+     * 키는 1~4만 받는다.
+     */
+    headingColors?: Partial<Record<1 | 2 | 3 | 4, string>>;
+    /** 본문 단락 텍스트 색상. 미지정 시 검정 */
+    bodyColor?: string;
+    /**
+     * 인용문 텍스트 색상. 미지정 시 검정.
+     *
+     * 주의: 이 옵션을 지정하면 인용문이 별도 charPr(이탤릭)로 렌더링된다.
+     * 미지정 시 기존 동작 그대로 본문 charPr로 렌더링 (이탤릭 아님).
+     */
+    quoteColor?: string;
+    /** 표 첫 행 텍스트 색상. 미지정 시 본문과 동일 */
+    tableHeaderColor?: string;
+    /** 표 첫 행 텍스트를 굵게 표시 (기본 false) */
+    tableHeaderBold?: boolean;
+}
+/** markdownToHwpx 옵션 */
+interface MarkdownToHwpxOptions {
+    theme?: HwpxTheme;
+}
 /**
  * 마크다운 텍스트를 HWPX (ArrayBuffer)로 변환.
  */
-declare function markdownToHwpx(markdown: string): Promise<ArrayBuffer>;
+declare function markdownToHwpx(markdown: string, options?: MarkdownToHwpxOptions): Promise<ArrayBuffer>;
 
 /**
  * Print Renderer — Markdown / IRBlock[] → PDF (puppeteer-core 기반).
@@ -465,4 +519,4 @@ interface FillFormOutput {
  */
 declare function fillForm(input: string | ArrayBuffer | Buffer, values: Record<string, string>, outputFormat?: FillOutputFormat): Promise<FillFormOutput>;
 
-export { type BlockDiff, type BoundingBox, type CellContext, type CellDiff, type DiffChangeType, type DiffResult, type DocumentMetadata, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillOutputFormat, type FillResult, type FormField, type FormResult, type HwpxFillResult, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type OcrProvider, type OutlineItem, type PageMargin, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PrintOptions, type PrintPreset, VERSION, type WarningCode, type WatchOptions, blocksToMarkdown, blocksToPdf, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, fillForm, fillFormFields, fillHwpx, isHwpxFile, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, markdownToHwpx, markdownToPdf, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, renderHtml };
+export { type BlockDiff, type BoundingBox, type CellContext, type CellDiff, type DiffChangeType, type DiffResult, type DocumentMetadata, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillOutputFormat, type FillResult, type FormField, type FormResult, type HwpxFillResult, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type OcrProvider, type OutlineItem, type PageMargin, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PrintOptions, type PrintPreset, VERSION, type WarningCode, type WatchOptions, blocksToMarkdown, blocksToPdf, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, fillForm, fillFormFields, fillHwpx, isHwpxFile, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, markdownToHwpx, markdownToPdf, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, renderHtml };

package/dist/index.d.ts

@@ -164,6 +164,34 @@ interface ParseSuccess extends ParseResultBase {
     warnings?: ParseWarning[];
     /** 추출된 이미지 목록 — 마크다운에서 파일명으로 참조됨 */
     images?: ExtractedImage[];
+    /** 페이지별 텍스트 품질 신호 — PDF에서만 제공 */
+    pageQuality?: PageQuality[];
+    /** 문서 단위 품질 요약 — PDF에서만 제공 */
+    qualitySummary?: DocumentQualitySummary;
+}
+/** 페이지별 텍스트 품질 신호 (PDF 전용). 자세한 설명은 src/pdf/quality.ts */
+interface PageQuality {
+    page: number;
+    textChars: number;
+    hangulRatio: number;
+    controlCharRatio: number;
+    replacementCharRatio: number;
+    puaRatio: number;
+    needsOcr: boolean;
+    ocrReason?: "low_text" | "high_pua" | "high_control" | "high_replacement";
+}
+/** 문서 단위 품질 요약 (PDF 전용). */
+interface DocumentQualitySummary {
+    totalPages: number;
+    totalTextChars: number;
+    avgHangulRatio: number;
+    avgControlCharRatio: number;
+    avgReplacementCharRatio: number;
+    avgPuaRatio: number;
+    lowTextPageCount: number;
+    highPuaPageCount: number;
+    needsOcr: boolean;
+    ocrCandidatePages: number[];
 }
 /** 추출된 이미지 — ParseSuccess.images에 포함 */
 interface ExtractedImage {
@@ -315,10 +343,36 @@ declare function fillHwpx(hwpxBuffer: ArrayBuffer, values: Record<string, string
  *       순서/비순서 리스트, 수평선, 인용문, 테이블
  * jszip으로 HWPX ZIP 패키징.
  */
+/** HWPX 생성 시 적용할 시각 테마 (모두 선택) */
+interface HwpxTheme {
+    /**
+     * 헤딩 레벨별 텍스트 색상. 미지정 시 검정.
+     * 현재 charPr 매핑은 h1/h2/h3/h4 4단계 (h5, h6은 h4와 같은 charPr 공유)이므로
+     * 키는 1~4만 받는다.
+     */
+    headingColors?: Partial<Record<1 | 2 | 3 | 4, string>>;
+    /** 본문 단락 텍스트 색상. 미지정 시 검정 */
+    bodyColor?: string;
+    /**
+     * 인용문 텍스트 색상. 미지정 시 검정.
+     *
+     * 주의: 이 옵션을 지정하면 인용문이 별도 charPr(이탤릭)로 렌더링된다.
+     * 미지정 시 기존 동작 그대로 본문 charPr로 렌더링 (이탤릭 아님).
+     */
+    quoteColor?: string;
+    /** 표 첫 행 텍스트 색상. 미지정 시 본문과 동일 */
+    tableHeaderColor?: string;
+    /** 표 첫 행 텍스트를 굵게 표시 (기본 false) */
+    tableHeaderBold?: boolean;
+}
+/** markdownToHwpx 옵션 */
+interface MarkdownToHwpxOptions {
+    theme?: HwpxTheme;
+}
 /**
  * 마크다운 텍스트를 HWPX (ArrayBuffer)로 변환.
  */
-declare function markdownToHwpx(markdown: string): Promise<ArrayBuffer>;
+declare function markdownToHwpx(markdown: string, options?: MarkdownToHwpxOptions): Promise<ArrayBuffer>;
 
 /**
  * Print Renderer — Markdown / IRBlock[] → PDF (puppeteer-core 기반).
@@ -465,4 +519,4 @@ interface FillFormOutput {
  */
 declare function fillForm(input: string | ArrayBuffer | Buffer, values: Record<string, string>, outputFormat?: FillOutputFormat): Promise<FillFormOutput>;
 
-export { type BlockDiff, type BoundingBox, type CellContext, type CellDiff, type DiffChangeType, type DiffResult, type DocumentMetadata, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillOutputFormat, type FillResult, type FormField, type FormResult, type HwpxFillResult, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type OcrProvider, type OutlineItem, type PageMargin, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PrintOptions, type PrintPreset, VERSION, type WarningCode, type WatchOptions, blocksToMarkdown, blocksToPdf, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, fillForm, fillFormFields, fillHwpx, isHwpxFile, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, markdownToHwpx, markdownToPdf, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, renderHtml };
+export { type BlockDiff, type BoundingBox, type CellContext, type CellDiff, type DiffChangeType, type DiffResult, type DocumentMetadata, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillOutputFormat, type FillResult, type FormField, type FormResult, type HwpxFillResult, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type OcrProvider, type OutlineItem, type PageMargin, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PrintOptions, type PrintPreset, VERSION, type WarningCode, type WatchOptions, blocksToMarkdown, blocksToPdf, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, fillForm, fillFormFields, fillHwpx, isHwpxFile, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, markdownToHwpx, markdownToPdf, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, renderHtml };

package/dist/index.js

@@ -16,7 +16,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-LB7E2KDF.js";
+} from "./chunk-RXZLTACX.js";
 import {
   parsePageRange
 } from "./chunk-SBVRCJFH.js";
@@ -18397,6 +18397,8 @@ var CHAR_H1 = 5;
 var CHAR_H2 = 6;
 var CHAR_H3 = 7;
 var CHAR_H4 = 8;
+var CHAR_TABLE_HEADER = 9;
+var CHAR_QUOTE = 10;
 var PARA_NORMAL = 0;
 var PARA_H1 = 1;
 var PARA_H2 = 2;
@@ -18405,14 +18407,30 @@ var PARA_H4 = 4;
 var PARA_CODE = 5;
 var PARA_QUOTE = 6;
 var PARA_LIST = 7;
-async function markdownToHwpx(markdown) {
+var DEFAULT_TEXT_COLOR = "#000000";
+function resolveTheme(theme) {
+  return {
+    h1: theme?.headingColors?.[1] ?? DEFAULT_TEXT_COLOR,
+    h2: theme?.headingColors?.[2] ?? DEFAULT_TEXT_COLOR,
+    h3: theme?.headingColors?.[3] ?? DEFAULT_TEXT_COLOR,
+    h4: theme?.headingColors?.[4] ?? theme?.headingColors?.[3] ?? DEFAULT_TEXT_COLOR,
+    body: theme?.bodyColor ?? DEFAULT_TEXT_COLOR,
+    quote: theme?.quoteColor ?? DEFAULT_TEXT_COLOR,
+    /** quoteColor가 명시되었는지 — blockquote charPr 분기에 사용 (baseline 호환) */
+    hasQuoteOption: theme?.quoteColor !== void 0,
+    tableHeader: theme?.tableHeaderColor ?? theme?.bodyColor ?? DEFAULT_TEXT_COLOR,
+    tableHeaderBold: !!theme?.tableHeaderBold
+  };
+}
+async function markdownToHwpx(markdown, options) {
+  const theme = resolveTheme(options?.theme);
   const blocks = parseMarkdownToBlocks(markdown);
-  const sectionXml = blocksToSectionXml(blocks);
+  const sectionXml = blocksToSectionXml(blocks, theme);
   const zip = new JSZip6();
   zip.file("mimetype", "application/hwp+zip", { compression: "STORE" });
   zip.file("META-INF/container.xml", generateContainerXml());
   zip.file("Contents/content.hpf", generateManifest());
-  zip.file("Contents/header.xml", generateHeaderXml());
+  zip.file("Contents/header.xml", generateHeaderXml(theme));
   zip.file("Contents/section0.xml", sectionXml);
   zip.file("Preview/PrvText.txt", buildPrvText(blocks));
   return await zip.generateAsync({ type: "arraybuffer" });
@@ -18592,11 +18610,11 @@ function generateManifest() {
   </opf:spine>
 </opf:package>`;
 }
-function charPr(id, height, bold, italic, fontId = 0) {
+function charPr(id, height, bold, italic, fontId = 0, textColor = DEFAULT_TEXT_COLOR) {
   const boldAttr = bold ? ` bold="1"` : "";
   const italicAttr = italic ? ` italic="1"` : "";
   const effFont = bold ? 2 : fontId;
-  return `      <hh:charPr id="${id}" height="${height}" textColor="#000000" shadeColor="none" useFontSpace="0" useKerning="0" symMark="NONE" borderFillIDRef="0"${boldAttr}${italicAttr}>
+  return `      <hh:charPr id="${id}" height="${height}" textColor="${textColor}" shadeColor="none" useFontSpace="0" useKerning="0" symMark="NONE" borderFillIDRef="0"${boldAttr}${italicAttr}>
         <hh:fontRef hangul="${effFont}" latin="${effFont}" hanja="${effFont}" japanese="${effFont}" other="${effFont}" symbol="${effFont}" user="${effFont}"/>
         <hh:ratio hangul="100" latin="100" hanja="100" japanese="100" other="100" symbol="100" user="100"/>
         <hh:spacing hangul="0" latin="0" hanja="0" japanese="0" other="0" symbol="0" user="0"/>
@@ -18616,7 +18634,7 @@ function paraPr(id, opts = {}) {
         <hh:border borderFillIDRef="0" offsetLeft="0" offsetRight="0" offsetTop="0" offsetBottom="0" connect="0" ignoreMargin="0"/>
       </hh:paraPr>`;
 }
-function generateHeaderXml() {
+function generateHeaderXml(theme) {
   return `<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
 <hh:head xmlns:hh="${NS_HEAD}" xmlns:hp="${NS_PARA}" version="1.4" secCnt="1">
   <hh:beginNum page="1" footnote="1" endnote="1" pic="1" tbl="1" equation="1"/>
@@ -18692,16 +18710,18 @@ function generateHeaderXml() {
         <hh:fillInfo/>
       </hh:borderFill>
     </hh:borderFills>
-    <hh:charProperties itemCnt="9">
-${charPr(0, 1e3, false, false)}
-${charPr(1, 1e3, true, false)}
-${charPr(2, 1e3, false, true)}
-${charPr(3, 1e3, true, true)}
+    <hh:charProperties itemCnt="11">
+${charPr(0, 1e3, false, false, 0, theme.body)}
+${charPr(1, 1e3, true, false, 0, theme.body)}
+${charPr(2, 1e3, false, true, 0, theme.body)}
+${charPr(3, 1e3, true, true, 0, theme.body)}
 ${charPr(4, 900, false, false, 1)}
-${charPr(5, 1800, true, false, 1)}
-${charPr(6, 1400, true, false, 1)}
-${charPr(7, 1200, true, false, 1)}
-${charPr(8, 1100, true, false, 1)}
+${charPr(5, 1800, true, false, 1, theme.h1)}
+${charPr(6, 1400, true, false, 1, theme.h2)}
+${charPr(7, 1200, true, false, 1, theme.h3)}
+${charPr(8, 1100, true, false, 1, theme.h4)}
+${charPr(CHAR_TABLE_HEADER, 1e3, theme.tableHeaderBold, false, 0, theme.tableHeader)}
+${charPr(CHAR_QUOTE, 1e3, false, true, 0, theme.quote)}
     </hh:charProperties>
     <hh:tabProperties itemCnt="0"/>
     <hh:numberings itemCnt="0"/>
@@ -18731,7 +18751,7 @@ var tableIdCounter = TABLE_ID_BASE;
 function nextTableId() {
   return ++tableIdCounter;
 }
-function generateTable(rows) {
+function generateTable(rows, theme) {
   const rowCnt = rows.length;
   const colCnt = Math.max(...rows.map((r) => r.length), 1);
   const cellW = Math.floor(44e3 / colCnt);
@@ -18739,12 +18759,15 @@ function generateTable(rows) {
   const tblW = cellW * colCnt;
   const tblH = cellH * rowCnt;
   const tblId = nextTableId();
+  const useHeaderStyle = theme.tableHeader !== theme.body || theme.tableHeaderBold;
   const trElements = rows.map((row, rowIdx) => {
     const cells = row.length < colCnt ? [...row, ...Array(colCnt - row.length).fill("")] : row;
+    const isHeaderRow = rowIdx === 0;
+    const headerCharPr = isHeaderRow && useHeaderStyle ? CHAR_TABLE_HEADER : CHAR_NORMAL;
     const tdElements = cells.map((cell, colIdx) => {
-      const runs = generateRuns(cell);
+      const runs = generateRuns(cell, headerCharPr);
       const p = `<hp:p paraPrIDRef="0" styleIDRef="0">${runs}</hp:p>`;
-      return `<hp:tc name="" header="${rowIdx === 0 ? 1 : 0}" hasMargin="0" protect="0" editable="1" dirty="0" borderFillIDRef="1"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="TOP" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0">${p}</hp:subList><hp:cellAddr colAddr="${colIdx}" rowAddr="${rowIdx}"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="${cellW}" height="${cellH}"/><hp:cellMargin left="141" right="141" top="141" bottom="141"/></hp:tc>`;
+      return `<hp:tc name="" header="${isHeaderRow ? 1 : 0}" hasMargin="0" protect="0" editable="1" dirty="0" borderFillIDRef="1"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="TOP" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0">${p}</hp:subList><hp:cellAddr colAddr="${colIdx}" rowAddr="${rowIdx}"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="${cellW}" height="${cellH}"/><hp:cellMargin left="141" right="141" top="141" bottom="141"/></hp:tc>`;
     }).join("");
     return `<hp:tr>${tdElements}</hp:tr>`;
   }).join("");
@@ -18752,7 +18775,7 @@ function generateTable(rows) {
   const tbl = `<hp:tbl id="${tblId}" zOrder="0" numberingType="TABLE" pageBreak="CELL" repeatHeader="0" rowCnt="${rowCnt}" colCnt="${colCnt}" cellSpacing="0" borderFillIDRef="1" noShading="0">${tblInner}</hp:tbl>`;
   return `<hp:p paraPrIDRef="0" styleIDRef="0"><hp:run charPrIDRef="0">${tbl}</hp:run></hp:p>`;
 }
-function blocksToSectionXml(blocks) {
+function blocksToSectionXml(blocks, theme) {
   const paraXmls = [];
   let isFirst = true;
   const orderedCounters = {};
@@ -18781,7 +18804,11 @@ function blocksToSectionXml(blocks) {
         break;
       }
       case "blockquote":
-        xml = generateParagraph(block.text || "", PARA_QUOTE);
+        xml = generateParagraph(
+          block.text || "",
+          PARA_QUOTE,
+          theme.hasQuoteOption ? CHAR_QUOTE : CHAR_NORMAL
+        );
         break;
       case "list_item": {
         const indent = block.indent || 0;
@@ -18814,7 +18841,7 @@ function blocksToSectionXml(blocks) {
             paraXmls.push(`<hp:p paraPrIDRef="0" styleIDRef="0">${secRun}</hp:p>`);
             isFirst = false;
           }
-          xml = generateTable(block.rows);
+          xml = generateTable(block.rows, theme);
         }
         break;
     }
@@ -19254,7 +19281,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-DJCMY3OO.js");
+    const mod = await import("./parser-OMPBVEFU.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -19265,8 +19292,8 @@ async function parsePdf(buffer, options) {
     };
   }
   try {
-    const { markdown, blocks, metadata, outline, warnings, isImageBased } = await parsePdfDocument(buffer, options);
-    return { success: true, fileType: "pdf", markdown, blocks, metadata, outline, warnings, isImageBased };
+    const { markdown, blocks, metadata, outline, warnings, isImageBased, pageQuality, qualitySummary } = await parsePdfDocument(buffer, options);
+    return { success: true, fileType: "pdf", markdown, blocks, metadata, outline, warnings, isImageBased, pageQuality, qualitySummary };
   } catch (err) {
     const isImageBased = err instanceof Error && "isImageBased" in err ? true : void 0;
     return { success: false, fileType: "pdf", error: err instanceof Error ? err.message : "PDF \uD30C\uC2F1 \uC2E4\uD328", code: classifyError(err), isImageBased };