@wsqc2026/markitdown-typescript 0.1.0

package/dist/converter-utils/docx/latex-dict.d.ts

@@ -0,0 +1,34 @@
+/**
+ * LaTeX dictionary for OMML to LaTeX conversion.
+ * Ported from Python markitdown latex_dict.py
+ */
+export declare const CHARS: Set<string>;
+export declare const BLANK = "";
+export declare const BACKSLASH = "\\";
+export declare const ALN = "&";
+export declare const BRK = "\\\\";
+export declare const FUNC_PLACE = "{fe}";
+/** Unicode → LaTeX accent mapping */
+export declare const CHR: Record<string, string>;
+/** Big operators */
+export declare const CHR_BO: Record<string, string>;
+/** Text symbol mapping (Unicode → LaTeX) */
+export declare const T: Record<string, string>;
+/** Function name templates */
+export declare const FUNC: Record<string, string>;
+export declare const CHR_DEFAULT: Record<string, string>;
+export declare const POS: Record<string, string>;
+export declare const POS_DEFAULT: Record<string, string>;
+export declare const SUB = "_{{{0}}}";
+export declare const SUP = "^{{{0}}}";
+export declare const F: Record<string, string>;
+export declare const F_DEFAULT = "\\frac{{{num}}}{{{den}}}";
+export declare const D = "\\left{left}{text}\\right{right}";
+export declare const D_DEFAULT: Record<string, string>;
+export declare const RAD = "\\sqrt[{deg}]{{{text}}}";
+export declare const RAD_DEFAULT = "\\sqrt{{{text}}}";
+export declare const ARR = "\\begin{{array}}{{c}}{text}\\end{{array}}";
+export declare const LIM_FUNC: Record<string, string>;
+export declare const LIM_TO: [string, string];
+export declare const LIM_UPP = "\\overset{{{lim}}}{{{text}}}";
+export declare const M = "\\begin{{matrix}}{text}\\end{{matrix}}";

package/dist/converter-utils/docx/omml.d.ts

@@ -0,0 +1,25 @@
+/**
+ * Office Math Markup Language (OMML) to LaTeX converter.
+ * Ported from Python markitdown omml.py (adapted from dwml)
+ */
+export declare const OMML_NS = "http://schemas.openxmlformats.org/officeDocument/2006/math";
+interface XmlNode {
+    tag: string;
+    attrib: Record<string, string>;
+    text: string | null;
+    children: XmlNode[];
+}
+/**
+ * Parse XML string into XmlNode tree with ElementTree-style {namespace}tag format.
+ * This is a minimal parser specifically for OMML math XML.
+ */
+declare function parseOmmlXml(xmlStr: string): XmlNode;
+/** Find child elements matching a tag */
+declare function findAll(node: XmlNode, tag: string): XmlNode[];
+/** Find first child element matching tag, returns null if not found */
+declare function find(node: XmlNode, tag: string): XmlNode | null;
+/** Convert an oMath XmlNode element to LaTeX string */
+export declare function oMathToLatex(element: XmlNode): string;
+/** Parse XML string, find all oMath elements, yield LaTeX for each */
+export declare function loadString(xmlString: string): Generator<string>;
+export { find, findAll, parseOmmlXml, type XmlNode };

package/dist/converter-utils/docx/pre-process.d.ts

@@ -0,0 +1,9 @@
+/**
+ * DOCX pre-processing: converts OMML math equations to LaTeX before mammoth.
+ * Ported from Python markitdown pre_process.py
+ */
+/**
+ * Pre-process a DOCX buffer to convert OMML math equations to LaTeX.
+ * Returns a new buffer with equations replaced.
+ */
+export declare function preProcessDocx(inputBuffer: Buffer): Promise<Buffer>;

package/dist/converter.d.ts

@@ -0,0 +1,23 @@
+import type { ConvertOptions, ConvertResult, StreamInfo } from "./types.js";
+export type ConverterContext = {
+    buffer: Buffer;
+    info: StreamInfo;
+    opts: ConvertOptions;
+};
+export type Matcher = (ctx: ConverterContext) => boolean;
+export type TransformStep = (ctx: ConverterContext) => Promise<ConvertResult>;
+export type Converter = {
+    name: string;
+    match: Matcher;
+    convert: TransformStep;
+};
+export declare function byMime(...mimes: string[]): Matcher;
+export declare function byExt(...exts: string[]): Matcher;
+export declare function byUrl(pattern: RegExp): Matcher;
+export declare function anyOf(...matchers: Matcher[]): Matcher;
+export declare function allOf(...matchers: Matcher[]): Matcher;
+export declare function hasCharset(): Matcher;
+/**
+ * Create a converter from a name, matcher, and a conversion function.
+ */
+export declare function converter(name: string, match: Matcher, convert: TransformStep): Converter;

package/dist/converters/audio.d.ts

@@ -0,0 +1 @@
+export declare const audioConverter: import("../converter.js").Converter;

package/dist/converters/bing-serp.d.ts

@@ -0,0 +1 @@
+export declare const bingSerpConverter: import("../converter.js").Converter;

package/dist/converters/csv.d.ts

@@ -0,0 +1 @@
+export declare const csvConverter: import("../converter.js").Converter;

package/dist/converters/docx.d.ts

@@ -0,0 +1 @@
+export declare const docxConverter: import("../converter.js").Converter;

package/dist/converters/epub.d.ts

@@ -0,0 +1 @@
+export declare const epubConverter: import("../converter.js").Converter;

package/dist/converters/exiftool.d.ts

@@ -0,0 +1 @@
+export declare function exiftoolMetadata(buffer: Buffer, exiftoolPath?: string): Promise<Record<string, string>>;

package/dist/converters/html.d.ts

@@ -0,0 +1 @@
+export declare const htmlConverter: import("../converter.js").Converter;

package/dist/converters/image.d.ts

@@ -0,0 +1 @@
+export declare const imageConverter: import("../converter.js").Converter;

package/dist/converters/index.d.ts

@@ -0,0 +1,17 @@
+export { audioConverter } from "./audio.js";
+export { bingSerpConverter } from "./bing-serp.js";
+export { csvConverter } from "./csv.js";
+export { docxConverter } from "./docx.js";
+export { epubConverter } from "./epub.js";
+export { htmlConverter } from "./html.js";
+export { imageConverter } from "./image.js";
+export { ipynbConverter } from "./ipynb.js";
+export { outlookMsgConverter } from "./outlook-msg.js";
+export { pdfConverter } from "./pdf.js";
+export { plainTextConverter } from "./plain-text.js";
+export { pptxConverter } from "./pptx.js";
+export { rssConverter } from "./rss.js";
+export { wikipediaConverter } from "./wikipedia.js";
+export { xlsConverter, xlsxConverter } from "./xlsx.js";
+export { youtubeConverter } from "./youtube.js";
+export { createZipConverter } from "./zip.js";

package/dist/converters/ipynb.d.ts

@@ -0,0 +1 @@
+export declare const ipynbConverter: import("../converter.js").Converter;

package/dist/converters/outlook-msg.d.ts

@@ -0,0 +1 @@
+export declare const outlookMsgConverter: import("../converter.js").Converter;

package/dist/converters/pdf.d.ts

@@ -0,0 +1 @@
+export declare const pdfConverter: import("../converter.js").Converter;

package/dist/converters/plain-text.d.ts

@@ -0,0 +1 @@
+export declare const plainTextConverter: import("../converter.js").Converter;

package/dist/converters/pptx.d.ts

@@ -0,0 +1 @@
+export declare const pptxConverter: import("../converter.js").Converter;

package/dist/converters/rss.d.ts

@@ -0,0 +1 @@
+export declare const rssConverter: import("../converter.js").Converter;

package/dist/converters/wikipedia.d.ts

@@ -0,0 +1 @@
+export declare const wikipediaConverter: import("../converter.js").Converter;

package/dist/converters/xlsx.d.ts

@@ -0,0 +1,2 @@
+export declare const xlsxConverter: import("../converter.js").Converter;
+export declare const xlsConverter: import("../converter.js").Converter;

package/dist/converters/youtube.d.ts

@@ -0,0 +1 @@
+export declare const youtubeConverter: import("../converter.js").Converter;

package/dist/converters/zip.d.ts

@@ -0,0 +1,8 @@
+import type { Converter } from "../converter.js";
+import type { ConvertResult, StreamInfo } from "../types.js";
+type ConvertBufferFn = (buffer: Buffer, input?: {
+    streamInfo?: StreamInfo;
+    _zipDepth?: number;
+}) => Promise<ConvertResult>;
+export declare function createZipConverter(convertFn: ConvertBufferFn): Converter;
+export {};

package/dist/exceptions.d.ts

@@ -0,0 +1,18 @@
+export declare class MarkItDownError extends Error {
+    constructor(message?: string);
+}
+export declare class MissingDependencyError extends MarkItDownError {
+    constructor(message?: string);
+}
+export declare class UnsupportedFormatError extends MarkItDownError {
+    constructor(message?: string);
+}
+export interface FailedConversionAttempt {
+    converterName: string;
+    error?: Error;
+}
+export declare function getSuggestion(error: unknown): string | undefined;
+export declare class FileConversionError extends MarkItDownError {
+    attempts?: FailedConversionAttempt[];
+    constructor(message?: string, attempts?: FailedConversionAttempt[]);
+}

package/dist/exit-codes.d.ts

@@ -0,0 +1,8 @@
+export declare const EXIT_SUCCESS = 0;
+export declare const EXIT_CONVERSION_FAILURE = 1;
+export declare const EXIT_BAD_ARGUMENTS = 2;
+export declare const EXIT_FILE_NOT_FOUND = 3;
+export declare const EXIT_PERMISSION_DENIED = 4;
+export declare const EXIT_UNSUPPORTED_FORMAT = 5;
+export declare const EXIT_CODE_DESCRIPTIONS: Record<number, string>;
+export declare function getExitCode(error: unknown): number;

package/dist/index.d.ts

@@ -0,0 +1,27 @@
+export type { Converter, ConverterContext, Matcher, TransformStep } from "./converter.js";
+export { allOf, anyOf, byExt, byMime, byUrl, converter, hasCharset } from "./converter.js";
+export { audioConverter } from "./converters/audio.js";
+export { bingSerpConverter } from "./converters/bing-serp.js";
+export { csvConverter } from "./converters/csv.js";
+export { docxConverter } from "./converters/docx.js";
+export { epubConverter } from "./converters/epub.js";
+export { htmlConverter } from "./converters/html.js";
+export { imageConverter } from "./converters/image.js";
+export { ipynbConverter } from "./converters/ipynb.js";
+export { outlookMsgConverter } from "./converters/outlook-msg.js";
+export { pdfConverter } from "./converters/pdf.js";
+export { plainTextConverter } from "./converters/plain-text.js";
+export { pptxConverter } from "./converters/pptx.js";
+export { rssConverter } from "./converters/rss.js";
+export { wikipediaConverter } from "./converters/wikipedia.js";
+export { xlsConverter, xlsxConverter } from "./converters/xlsx.js";
+export { youtubeConverter } from "./converters/youtube.js";
+export { createZipConverter } from "./converters/zip.js";
+export { FileConversionError, MarkItDownError, MissingDependencyError, UnsupportedFormatError, } from "./exceptions.js";
+export type { ConvertInput, MarkItDownOptions } from "./markitdown.js";
+export { createMarkItDown } from "./markitdown.js";
+export { mergeStreamInfo } from "./stream-info.js";
+export { decodeBuffer } from "./transforms/decode-text.js";
+export { htmlToMarkdown } from "./transforms/html-to-markdown.js";
+export type { ConvertOptions, ConvertResult, StreamInfo } from "./types.js";
+export { ConvertOptionsSchema, ConvertResultSchema, StreamInfoSchema } from "./types.js";