aurochs 0.6.2 → 0.6.4

package/dist/@aurochs/pdf/src/domain/document/index.d.ts

@@ -3,5 +3,5 @@
  *
  * Exports types for PDF document structure.
  */
-export type { PdfElement, PdfPage, PdfDocument, PdfEmbeddedFont } from './types';
+export type { PdfElement, PdfPage, PdfDocument, PdfEmbeddedFont, PdfFontToUnicode, PdfFontMetrics } from './types';
 export { PDF_UNITS, isPdfPath, isPdfText, isPdfImage } from './types';

package/dist/@aurochs/pdf/src/domain/document/types.d.ts

@@ -1,6 +1,7 @@
 import { PdfImage } from '../image';
 import { PdfPath } from '../path';
 import { PdfText } from '../text';
+import { CIDOrdering } from '../font';
 export type PdfElement = PdfPath | PdfText | PdfImage;
 /**
  * Represents a parsed PDF page.
@@ -45,6 +46,29 @@ export type PdfPage = {
      */
     readonly elements: readonly PdfElement[];
 };
+/**
+ * ToUnicode mapping data for round-trip preservation.
+ * Used to reconstruct ToUnicode CMap when writing PDF.
+ */
+export type PdfFontToUnicode = {
+    /** Source bytes (hex) → Unicode string mapping. Key is uppercase hex (e.g., "8140" → "ア"). */
+    readonly byteMapping: ReadonlyMap<string, string>;
+    /** Source code byte lengths from codespace ranges (descending order). */
+    readonly sourceCodeByteLengths: readonly number[];
+};
+/**
+ * Font metrics for PDF writing.
+ */
+export type PdfFontMetrics = {
+    /** Ascender height in 1/1000 em units */
+    readonly ascender: number;
+    /** Descender depth in 1/1000 em units (negative) */
+    readonly descender: number;
+    /** Glyph widths: character code → width in 1/1000 em units */
+    readonly widths: ReadonlyMap<number, number>;
+    /** Default glyph width when not found in widths */
+    readonly defaultWidth: number;
+};
 /**
  * Embedded font data extracted from PDF.
  *
@@ -59,6 +83,16 @@ export type PdfEmbeddedFont = {
     readonly data: Uint8Array;
     /** MIME type */
     readonly mimeType: string;
+    /** Original BaseFont name from PDF (e.g., "/ZRDQJE+Hiragino-Sans"). Includes subset prefix. */
+    readonly baseFontName?: string;
+    /** ToUnicode CMap information for round-trip preservation. */
+    readonly toUnicode?: PdfFontToUnicode;
+    /** Font metrics for accurate text layout. */
+    readonly metrics?: PdfFontMetrics;
+    /** CID ordering (Japan1, GB1, CNS1, Korea1, Identity). */
+    readonly ordering?: CIDOrdering;
+    /** Number of bytes per character code (1 for single-byte, 2 for CID fonts). */
+    readonly codeByteWidth?: 1 | 2;
 };
 export type PdfDocument = {
     readonly pages: readonly PdfPage[];

package/dist/@aurochs/pdf/src/writer/content-stream/text-operators.d.ts

@@ -1,4 +1,5 @@
 import { PdfText } from '../../domain/text';
+import { PdfEmbeddedFont } from '../../domain/document';
 /**
  * Context for text serialization.
  */
@@ -8,6 +9,11 @@ export type TextSerializationContext = {
      * e.g., "Helvetica" -> "F1"
      */
     readonly fontNameToResource: ReadonlyMap<string, string>;
+    /**
+     * Embedded fonts for CID font detection.
+     * When present, CID fonts will use hex string output.
+     */
+    readonly embeddedFonts?: readonly PdfEmbeddedFont[];
 };
 /**
  * Serialize a PdfText element to PDF content stream operators.

package/dist/@aurochs/pdf/src/writer/document/font-builder.d.ts

@@ -20,6 +20,22 @@ export declare function buildType1Font(fontName: string, tracker: PdfObjectTrack
  * @returns The font dictionary object number
  */
 export declare function buildEmbeddedFont(font: PdfEmbeddedFont, tracker: PdfObjectTracker): number;
+/**
+ * Build a Type0 CID font with Identity-H encoding.
+ *
+ * Structure:
+ * - Type0 font dictionary (top-level)
+ * - CIDFont dictionary (descendant)
+ * - FontDescriptor
+ * - CIDToGIDMap (Identity)
+ * - ToUnicode CMap stream
+ * - Embedded font file
+ *
+ * @param font - The embedded font data with CID information
+ * @param tracker - Object tracker for allocation
+ * @returns The Type0 font dictionary object number
+ */
+export declare function buildType0Font(font: PdfEmbeddedFont, tracker: PdfObjectTracker): number;
 /**
  * Build fonts for a document.
  * Returns a map of font name to object number.

package/dist/@aurochs/pdf/src/writer/document/index.d.ts

@@ -3,6 +3,7 @@
  */
 export { PdfObjectTracker, type PdfObjectEntry, } from './object-tracker';
 export { buildResourceDict, buildEmptyResourceDict, type ResourceRefs, } from './resource-builder';
-export { buildType1Font, buildEmbeddedFont, buildFonts, } from './font-builder';
+export { buildType1Font, buildEmbeddedFont, buildType0Font, buildFonts, } from './font-builder';
+export { generateToUnicodeStream, } from './tounicode-writer';
 export { buildImageXObject, buildImages, } from './image-builder';
 export { buildPage, type PageBuildResult, type BuildPageOptions, } from './page-builder';

package/dist/@aurochs/pdf/src/writer/document/page-builder.d.ts

@@ -1,4 +1,4 @@
-import { PdfPage } from '../../domain/document';
+import { PdfPage, PdfEmbeddedFont } from '../../domain/document';
 import { PdfImage } from '../../domain/image';
 import { ResourceRefs } from './resource-builder';
 import { PdfObjectTracker } from './object-tracker';
@@ -23,6 +23,8 @@ export type BuildPageOptions = {
     readonly fontObjMap: ReadonlyMap<string, number>;
     readonly imageObjMap: ReadonlyMap<number, number>;
     readonly tracker: PdfObjectTracker;
+    /** Embedded fonts for CID font text serialization. */
+    readonly embeddedFonts?: readonly PdfEmbeddedFont[];
 };
 /**
  * Build a page object.

package/dist/@aurochs/pdf/src/writer/document/tounicode-writer.d.ts

@@ -0,0 +1,8 @@
+import { PdfFontToUnicode } from '../../domain/document';
+/**
+ * Generate a ToUnicode CMap stream.
+ *
+ * @param toUnicode - The ToUnicode mapping data
+ * @returns CMap stream content as UTF-8 encoded bytes
+ */
+export declare function generateToUnicodeStream(toUnicode: PdfFontToUnicode): Uint8Array;