kordoc 4.2.0 → 4.2.1

package/dist/index.d.cts

@@ -203,7 +203,7 @@ interface OutlineItem {
 }
 /** 구조화된 에러 코드 — 프로그래밍적 에러 핸들링용 */
 type ErrorCode = "EMPTY_INPUT" | "UNSUPPORTED_FORMAT" | "ENCRYPTED" | "DRM_PROTECTED" | "CORRUPTED" | "DECOMPRESSION_BOMB" | "ZIP_BOMB" | "IMAGE_BASED_PDF" | "NO_SECTIONS" | "PARSE_ERROR" | "MISSING_DEPENDENCY";
-type FileType = "hwpx" | "hwp" | "hwp3" | "hwpml" | "pdf" | "xlsx" | "xls" | "docx" | "unknown";
+type FileType = "hwpx" | "hwp" | "hwp3" | "hwpml" | "pdf" | "xlsx" | "xls" | "docx" | "image" | "unknown";
 interface ParseResultBase {
     fileType: FileType;
     /** 페이지/섹션 수 — PDF: 실제 페이지 수, HWP/HWPX: 섹션 수, XLSX: 시트 수 */
@@ -1609,6 +1609,8 @@ declare const VERSION: string;
  * ```
  */
 declare function parse(input: string | ArrayBuffer | Buffer, options?: ParseOptions): Promise<ParseResult>;
+/** 이미지(PNG/JPEG/WebP)를 OCR 로 Markdown 변환 — 텍스트층이 없으므로 OCR 상시 적용 */
+declare function parseImage(buffer: ArrayBuffer, options?: ParseOptions): Promise<ParseResult>;
 /** HWP 3.x (구버전 한컴 워드프로세서) 파일을 Markdown 으로 변환. */
 declare function parseHwp3(buffer: ArrayBuffer, options?: ParseOptions): Promise<ParseResult>;
 /** HWPX 파일을 Markdown으로 변환 */
@@ -1664,4 +1666,4 @@ interface FillFormOutput {
  */
 declare function fillForm(input: string | ArrayBuffer | Buffer, values: Record<string, FillInput>, outputFormat?: FillOutputFormat): Promise<FillFormOutput>;
 
-export { type BlockCapability, type BlockCapabilityInfo, type BlockDiff, type BlockEdit, type BlockSourceRef, type BorderDef, type BorderFillDef, type BoundingBox, type CellCapability, type CellContext, type CellDiff, type CellProfile, type CharPrDef, type ChunkOptions, DEFAULT_REDACT_RULES, type DiffChangeType, type DiffResult, type DocChunk, type DocumentMetadata, type DocumentQualitySummary, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillInput, type FillOutputFormat, type FillResult, type FillValue, type FormField, type FormFieldSchema, type FormFieldType, type FormResult, type FormSchemaResult, type FormatProfile, type GongmunFont, type GongmunLintFinding, type GongmunNumbering, type GongmunOptions, type GongmunPreset, type GongmunPresetInput, type HwpxFillResult, HwpxSession, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type MeasureOptions, type OcrProvider, type OutlineItem, PRESET_ALIAS, type PageMargin, type PageQuality, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PatchOptions, type PatchResult, type PatchSkip, type PlaceSealResult, type PrintOptions, type PrintPreset, type RedactHit, type RedactOptions, type RedactRule, type RedactTextResult, type RenderSvgOptions, type RenderSvgResult, SPACE_EM_FIXED, SPACE_EM_FONT, type ScanCell, type ScanParaKind, type ScanParagraph, type ScanTable, type SealOp, type SealPlacement, type SectionScan, type SpliceEdit, type TRange, type TableProfile, VERSION, type ValidateIssue, type ValidateResult, ValueCursor, type WarningCode, type WatchOptions, type WrapMode, type WrapResult, applySplices, blocksToChunks, blocksToMarkdown, blocksToPdf, buildParagraphSplices, buildRangeSplices, charWidthEm1000, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, extractFormSchema, fillForm, fillFormFields, fillHwpx, fillWithUniqueGuard, fitRatioForFewerLines, formatFillValue, gongmunLintWarnings, hwpxToProfile, incompatibleGongmunWarnings, inferFieldType, isHwpxFile, isKnownFont, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, lintGongmunText, markdownToHwpx, markdownToPdf, measureTextWidth, normalizeGongmunPreset, openHwpxDocument, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, patchHwp, patchHwpx, patchHwpxBlocks, placeSealHwpx, redactMarkdown, redactText, renderHtml, renderHwpxToSvg, scanSectionXml, simulateWrap, simulateWrapKeepWord, unknownFontWarnings, validateHwpx };
+export { type BlockCapability, type BlockCapabilityInfo, type BlockDiff, type BlockEdit, type BlockSourceRef, type BorderDef, type BorderFillDef, type BoundingBox, type CellCapability, type CellContext, type CellDiff, type CellProfile, type CharPrDef, type ChunkOptions, DEFAULT_REDACT_RULES, type DiffChangeType, type DiffResult, type DocChunk, type DocumentMetadata, type DocumentQualitySummary, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillInput, type FillOutputFormat, type FillResult, type FillValue, type FormField, type FormFieldSchema, type FormFieldType, type FormResult, type FormSchemaResult, type FormatProfile, type GongmunFont, type GongmunLintFinding, type GongmunNumbering, type GongmunOptions, type GongmunPreset, type GongmunPresetInput, type HwpxFillResult, HwpxSession, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type MeasureOptions, type OcrProvider, type OutlineItem, PRESET_ALIAS, type PageMargin, type PageQuality, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PatchOptions, type PatchResult, type PatchSkip, type PlaceSealResult, type PrintOptions, type PrintPreset, type RedactHit, type RedactOptions, type RedactRule, type RedactTextResult, type RenderSvgOptions, type RenderSvgResult, SPACE_EM_FIXED, SPACE_EM_FONT, type ScanCell, type ScanParaKind, type ScanParagraph, type ScanTable, type SealOp, type SealPlacement, type SectionScan, type SpliceEdit, type TRange, type TableProfile, VERSION, type ValidateIssue, type ValidateResult, ValueCursor, type WarningCode, type WatchOptions, type WrapMode, type WrapResult, applySplices, blocksToChunks, blocksToMarkdown, blocksToPdf, buildParagraphSplices, buildRangeSplices, charWidthEm1000, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, extractFormSchema, fillForm, fillFormFields, fillHwpx, fillWithUniqueGuard, fitRatioForFewerLines, formatFillValue, gongmunLintWarnings, hwpxToProfile, incompatibleGongmunWarnings, inferFieldType, isHwpxFile, isKnownFont, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, lintGongmunText, markdownToHwpx, markdownToPdf, measureTextWidth, normalizeGongmunPreset, openHwpxDocument, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parseImage, parsePdf, parseXls, parseXlsx, patchHwp, patchHwpx, patchHwpxBlocks, placeSealHwpx, redactMarkdown, redactText, renderHtml, renderHwpxToSvg, scanSectionXml, simulateWrap, simulateWrapKeepWord, unknownFontWarnings, validateHwpx };

package/dist/index.d.ts

@@ -203,7 +203,7 @@ interface OutlineItem {
 }
 /** 구조화된 에러 코드 — 프로그래밍적 에러 핸들링용 */
 type ErrorCode = "EMPTY_INPUT" | "UNSUPPORTED_FORMAT" | "ENCRYPTED" | "DRM_PROTECTED" | "CORRUPTED" | "DECOMPRESSION_BOMB" | "ZIP_BOMB" | "IMAGE_BASED_PDF" | "NO_SECTIONS" | "PARSE_ERROR" | "MISSING_DEPENDENCY";
-type FileType = "hwpx" | "hwp" | "hwp3" | "hwpml" | "pdf" | "xlsx" | "xls" | "docx" | "unknown";
+type FileType = "hwpx" | "hwp" | "hwp3" | "hwpml" | "pdf" | "xlsx" | "xls" | "docx" | "image" | "unknown";
 interface ParseResultBase {
     fileType: FileType;
     /** 페이지/섹션 수 — PDF: 실제 페이지 수, HWP/HWPX: 섹션 수, XLSX: 시트 수 */
@@ -1609,6 +1609,8 @@ declare const VERSION: string;
  * ```
  */
 declare function parse(input: string | ArrayBuffer | Buffer, options?: ParseOptions): Promise<ParseResult>;
+/** 이미지(PNG/JPEG/WebP)를 OCR 로 Markdown 변환 — 텍스트층이 없으므로 OCR 상시 적용 */
+declare function parseImage(buffer: ArrayBuffer, options?: ParseOptions): Promise<ParseResult>;
 /** HWP 3.x (구버전 한컴 워드프로세서) 파일을 Markdown 으로 변환. */
 declare function parseHwp3(buffer: ArrayBuffer, options?: ParseOptions): Promise<ParseResult>;
 /** HWPX 파일을 Markdown으로 변환 */
@@ -1664,4 +1666,4 @@ interface FillFormOutput {
  */
 declare function fillForm(input: string | ArrayBuffer | Buffer, values: Record<string, FillInput>, outputFormat?: FillOutputFormat): Promise<FillFormOutput>;
 
-export { type BlockCapability, type BlockCapabilityInfo, type BlockDiff, type BlockEdit, type BlockSourceRef, type BorderDef, type BorderFillDef, type BoundingBox, type CellCapability, type CellContext, type CellDiff, type CellProfile, type CharPrDef, type ChunkOptions, DEFAULT_REDACT_RULES, type DiffChangeType, type DiffResult, type DocChunk, type DocumentMetadata, type DocumentQualitySummary, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillInput, type FillOutputFormat, type FillResult, type FillValue, type FormField, type FormFieldSchema, type FormFieldType, type FormResult, type FormSchemaResult, type FormatProfile, type GongmunFont, type GongmunLintFinding, type GongmunNumbering, type GongmunOptions, type GongmunPreset, type GongmunPresetInput, type HwpxFillResult, HwpxSession, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type MeasureOptions, type OcrProvider, type OutlineItem, PRESET_ALIAS, type PageMargin, type PageQuality, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PatchOptions, type PatchResult, type PatchSkip, type PlaceSealResult, type PrintOptions, type PrintPreset, type RedactHit, type RedactOptions, type RedactRule, type RedactTextResult, type RenderSvgOptions, type RenderSvgResult, SPACE_EM_FIXED, SPACE_EM_FONT, type ScanCell, type ScanParaKind, type ScanParagraph, type ScanTable, type SealOp, type SealPlacement, type SectionScan, type SpliceEdit, type TRange, type TableProfile, VERSION, type ValidateIssue, type ValidateResult, ValueCursor, type WarningCode, type WatchOptions, type WrapMode, type WrapResult, applySplices, blocksToChunks, blocksToMarkdown, blocksToPdf, buildParagraphSplices, buildRangeSplices, charWidthEm1000, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, extractFormSchema, fillForm, fillFormFields, fillHwpx, fillWithUniqueGuard, fitRatioForFewerLines, formatFillValue, gongmunLintWarnings, hwpxToProfile, incompatibleGongmunWarnings, inferFieldType, isHwpxFile, isKnownFont, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, lintGongmunText, markdownToHwpx, markdownToPdf, measureTextWidth, normalizeGongmunPreset, openHwpxDocument, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parsePdf, parseXls, parseXlsx, patchHwp, patchHwpx, patchHwpxBlocks, placeSealHwpx, redactMarkdown, redactText, renderHtml, renderHwpxToSvg, scanSectionXml, simulateWrap, simulateWrapKeepWord, unknownFontWarnings, validateHwpx };
+export { type BlockCapability, type BlockCapabilityInfo, type BlockDiff, type BlockEdit, type BlockSourceRef, type BorderDef, type BorderFillDef, type BoundingBox, type CellCapability, type CellContext, type CellDiff, type CellProfile, type CharPrDef, type ChunkOptions, DEFAULT_REDACT_RULES, type DiffChangeType, type DiffResult, type DocChunk, type DocumentMetadata, type DocumentQualitySummary, type ErrorCode, type ExtractedImage, type FileType, type FillFormOutput, type FillInput, type FillOutputFormat, type FillResult, type FillValue, type FormField, type FormFieldSchema, type FormFieldType, type FormResult, type FormSchemaResult, type FormatProfile, type GongmunFont, type GongmunLintFinding, type GongmunNumbering, type GongmunOptions, type GongmunPreset, type GongmunPresetInput, type HwpxFillResult, HwpxSession, type HwpxTheme, type IRBlock, type IRBlockType, type IRCell, type IRTable, type ImageData, type InlineStyle, type MarkdownToHwpxOptions, type MeasureOptions, type OcrProvider, type OutlineItem, PRESET_ALIAS, type PageMargin, type PageQuality, type ParseFailure, type ParseOptions, type ParseResult, type ParseSuccess, type ParseWarning, type PatchOptions, type PatchResult, type PatchSkip, type PlaceSealResult, type PrintOptions, type PrintPreset, type RedactHit, type RedactOptions, type RedactRule, type RedactTextResult, type RenderSvgOptions, type RenderSvgResult, SPACE_EM_FIXED, SPACE_EM_FONT, type ScanCell, type ScanParaKind, type ScanParagraph, type ScanTable, type SealOp, type SealPlacement, type SectionScan, type SpliceEdit, type TRange, type TableProfile, VERSION, type ValidateIssue, type ValidateResult, ValueCursor, type WarningCode, type WatchOptions, type WrapMode, type WrapResult, applySplices, blocksToChunks, blocksToMarkdown, blocksToPdf, buildParagraphSplices, buildRangeSplices, charWidthEm1000, compare, detectFormat, detectOle2Format, detectZipFormat, diffBlocks, extractFormFields, extractFormSchema, fillForm, fillFormFields, fillHwpx, fillWithUniqueGuard, fitRatioForFewerLines, formatFillValue, gongmunLintWarnings, hwpxToProfile, incompatibleGongmunWarnings, inferFieldType, isHwpxFile, isKnownFont, isLabelCell, isOldHwpFile, isPdfFile, isZipFile, lintGongmunText, markdownToHwpx, markdownToPdf, measureTextWidth, normalizeGongmunPreset, openHwpxDocument, parse, parseDocx, parseHwp, parseHwp3, parseHwpml, parseHwpx, parseImage, parsePdf, parseXls, parseXlsx, patchHwp, patchHwpx, patchHwpxBlocks, placeSealHwpx, redactMarkdown, redactText, renderHtml, renderHwpxToSvg, scanSectionXml, simulateWrap, simulateWrapKeepWord, unknownFontWarnings, validateHwpx };

package/dist/index.js

@@ -8,7 +8,7 @@ import {
   flattenLayoutTables,
   inlineImagesIntoMarkdown,
   mapPuaText
-} from "./chunk-WF2E5P2G.js";
+} from "./chunk-7IMNZ2BL.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -27,7 +27,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-JP5YH3Z4.js";
+} from "./chunk-HTCKGSSW.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -247,6 +247,14 @@ function isPdfFile(buffer) {
   const b = magicBytes(buffer);
   return b[0] === 37 && b[1] === 80 && b[2] === 68 && b[3] === 70;
 }
+function isImageFile(buffer) {
+  if (buffer.byteLength < 12) return false;
+  const b = new Uint8Array(buffer, 0, 12);
+  if (b[0] === 137 && b[1] === 80 && b[2] === 78 && b[3] === 71) return true;
+  if (b[0] === 255 && b[1] === 216 && b[2] === 255) return true;
+  if (b[0] === 82 && b[1] === 73 && b[2] === 70 && b[3] === 70 && b[8] === 87 && b[9] === 69 && b[10] === 66 && b[11] === 80) return true;
+  return false;
+}
 function isHwpmlFile(buffer) {
   const bytes = new Uint8Array(buffer, 0, Math.min(512, buffer.byteLength));
   const head = new TextDecoder("utf-8", { fatal: false }).decode(bytes).replace(/^\uFEFF/, "");
@@ -259,6 +267,7 @@ function detectFormat(buffer) {
   if (isOldHwpFile(buffer)) return "hwp";
   if (isPdfFile(buffer)) return "pdf";
   if (isHwpmlFile(buffer)) return "hwpml";
+  if (isImageFile(buffer)) return "image";
   return "unknown";
 }
 function detectOle2Format(buffer) {
@@ -30446,10 +30455,28 @@ async function parse(input, options) {
       return parseHwpml(buffer, opts);
     case "pdf":
       return parsePdf(buffer, opts);
+    case "image":
+      return parseImage(buffer, opts);
     default:
       return { success: false, fileType: "unknown", error: "\uC9C0\uC6D0\uD558\uC9C0 \uC54A\uB294 \uD30C\uC77C \uD615\uC2DD\uC785\uB2C8\uB2E4.", code: "UNSUPPORTED_FORMAT" };
   }
 }
+async function parseImage(buffer, options) {
+  try {
+    const { parseImageDocument } = await import("./image-ocr-DALT7PFQ.js");
+    const { blocks, warnings } = await parseImageDocument(buffer, options);
+    return {
+      success: true,
+      fileType: "image",
+      markdown: blocksToMarkdown(blocks),
+      blocks,
+      warnings: warnings.length ? warnings : void 0,
+      pageCount: 1
+    };
+  } catch (err) {
+    return { success: false, fileType: "image", error: sanitizeError(err), code: classifyError(err) };
+  }
+}
 async function parseHwp3(buffer, options) {
   try {
     const { markdown, blocks, metadata, outline, warnings } = parseHwp3Document(buffer, options);
@@ -30494,7 +30521,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-ND2YGQZR.js");
+    const mod = await import("./parser-33TK5SMH.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -30633,6 +30660,7 @@ export {
   parseHwp3,
   parseHwpml,
   parseHwpx,
+  parseImage,
   parsePdf,
   parseXls,
   parseXlsx,