Add odt2md

package.json

@@ -17,7 +17,8 @@
   "bin": {
     "wgd": "src/wikigdrive.sh",
     "wikigdrive": "src/wikigdrive.sh",
-    "wikigdrivectl": "src/wikigdrivectl.sh"
+    "wikigdrivectl": "src/wikigdrivectl.sh",
+    "odt2md": "src/odt2md.sh"
   },
   "main": "src/cli/wikigdrive.ts",
   "scripts": {

src/cli/odt2md.ts

@@ -0,0 +1,89 @@
+'use strict';
+
+import path from 'path';
+import minimist from 'minimist';
+import {fileURLToPath} from 'url';
+import {Buffer} from 'buffer';
+import fs from 'fs';
+
+import {OdtProcessor} from '../odt/OdtProcessor.js';
+import {UnMarshaller} from '../odt/UnMarshaller.js';
+import {DocumentContent, DocumentStyles, LIBREOFFICE_CLASSES} from '../odt/LibreOffice.js';
+import {OdtToMarkdown} from '../odt/OdtToMarkdown.js';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+
+process.env.GIT_SHA = process.env.GIT_SHA || 'dev';
+
+async function usage() {
+  const pkg = JSON.parse(new TextDecoder().decode(fs.readFileSync(path.resolve(__dirname, '..', '..', 'package.json'))));
+
+  const commandUsage = 'echo "test" | odt2md\n\nor\n\nodt2md filename.odt';
+
+  console.log(
+    `${pkg.name} version: ${pkg.version}, ${process.env.GIT_SHA}\n\nUsage:\n${commandUsage.trim()}\n`);
+}
+
+async function main() {
+  const inputArr = [];
+
+  process.stdin.on( 'data', function(data) { inputArr.push(data); } );
+
+  await new Promise(resolve => {
+    setTimeout(() => {
+      process.stdin.destroy();
+      resolve(null);
+    }, 50);
+    process.stdin.on( 'end', resolve);
+  });
+
+  const argv = minimist(process.argv.slice(2));
+
+  if (inputArr.length === 0) {
+    if (argv._.length < 1 || argv.h || argv.help) {
+      await usage();
+      process.exit(1);
+    }
+
+    inputArr.push(fs.readFileSync(path.resolve(process.cwd(), argv._[0])));
+  }
+
+  if (inputArr.length === 0) {
+    console.error('No input');
+    process.exit(1);
+  }
+
+  const processor = new OdtProcessor();
+  await processor.loadFromBuffer(Buffer.concat(inputArr));
+  if (!processor.getContentXml()) {
+    throw Error('No odt processed');
+  }
+
+  const parser = new UnMarshaller(LIBREOFFICE_CLASSES, 'DocumentContent');
+  const document: DocumentContent = parser.unmarshal(processor.getContentXml());
+  if (!document) {
+    throw Error('No document unmarshalled');
+  }
+  const parserStyles = new UnMarshaller(LIBREOFFICE_CLASSES, 'DocumentStyles');
+  const styles: DocumentStyles = parserStyles.unmarshal(processor.getStylesXml());
+  if (!styles) {
+    throw Error('No styles unmarshalled');
+  }
+  const converter = new OdtToMarkdown(document, styles, processor.getFileNameMap(), processor.getXmlMap());
+  const markdown = await converter.convert();
+  console.log(markdown);
+}
+
+try {
+  await main();
+  process.exit(0);
+} catch (err) {
+  if (err.isUsageError) {
+    console.error(err.message);
+    await usage();
+  } else {
+    console.error(err);
+  }
+  process.exit(1);
+}

src/containers/transform/TaskLocalFileTransform.ts

@@ -139,8 +139,8 @@ export class TaskLocalFileTransform extends QueueTask {
     const picturesDirAbsolute = destinationPath + '/' + this.realFileName.replace(/.md$/, '.assets/');
 
     if (SINGLE_THREADED_TRANSFORM) {
-      const processor = new OdtProcessor(odtPath, true);
-      await processor.load();
+      const processor = new OdtProcessor(true);
+      await processor.load(odtPath);
       await processor.unzipAssets(destinationPath, this.realFileName);
       const content = processor.getContentXml();
       const stylesXml = processor.getStylesXml();

src/odt/OdtProcessor.ts

@@ -18,17 +18,33 @@ export class OdtProcessor {
   private fileNameMap: { [name: string]: string };
   private xmlMap: { [name: string]: string };
 
-  constructor(private odtPath: string, private contentAddressable = false) {
+  constructor(private contentAddressable = false) {
     this.fileNameMap = {};
     this.xmlMap = {};
   }
 
-  async load() {
-    if (!fs.existsSync(this.odtPath)) {
+  async load(odtPath: string) {
+    if (!fs.existsSync(odtPath)) {
       return;
     }
     const jsZip = new JSZip();
-    const input: Buffer = fs.readFileSync(this.odtPath);
+    const input: Buffer = fs.readFileSync(odtPath);
+    const zip = await jsZip.loadAsync(input);
+
+    this.files = zip.folder('').files;
+
+    if (this.files['content.xml']) {
+      this.contentXml = await this.files['content.xml'].async('string');
+    }
+    if (this.files['styles.xml']) {
+      this.stylesXml = await this.files['styles.xml'].async('string');
+    }
+
+    await this.processMathMl();
+  }
+
+  async loadFromBuffer(input: Buffer): Promise<void> {
+    const jsZip = new JSZip();
     const zip = await jsZip.loadAsync(input);
 
     this.files = zip.folder('').files;

src/odt/executeOdtToMarkdown.ts

@@ -8,8 +8,8 @@ import {generateDocumentFrontMatter} from '../containers/transform/frontmatters/
 import {OdtProcessor} from './OdtProcessor.ts';
 
 export async function executeOdtToMarkdown(workerData) {
-  const processor = new OdtProcessor(workerData.odtPath, true);
-  await processor.load();
+  const processor = new OdtProcessor(true);
+  await processor.load(workerData.odtPath);
   await processor.unzipAssets(workerData.destinationPath, workerData.realFileName);
   const content = processor.getContentXml();
   const stylesXml = processor.getStylesXml();

src/odt2md.sh

@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+
+FULL_PATH="$(readlink -f ${BASH_SOURCE[0]})"
+MAIN_DIR=$(dirname "$FULL_PATH")/..
+NODE_MODULES=$MAIN_DIR/node_modules
+
+POSITIONAL_ARGS=()
+INSPECT=""
+
+ORIG_ARGS=$@
+
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    --inspect)
+      INSPECT="$1"
+      shift # past argument
+      ;;
+    *)
+      if [[ -z "$CMD" ]]; then
+        CMD=$1
+      fi
+      POSITIONAL_ARGS+=("$1") # save positional arg
+      shift # past argument
+      ;;
+  esac
+done
+
+if test "$INSPECT" = "--inspect"; then
+  /usr/bin/env node --inspect --no-warnings --enable-source-maps --experimental-specifier-resolution=node --loader ts-node/esm $MAIN_DIR/src/cli/odt2md.ts $ORIG_ARGS
+else
+  /usr/bin/env node --no-warnings --enable-source-maps --experimental-specifier-resolution=node --loader ts-node/esm $MAIN_DIR/src/cli/odt2md.ts $ORIG_ARGS
+fi

test/odt/OdtLoad.test.ts

@@ -16,8 +16,8 @@ describe('OdtLoad', () => {
   it('test content.xml transform to object', async () => {
     const fileSystem = new FileContentService(__dirname);
     const odtPath = fileSystem.getRealPath() + '/' + 'example_document.odt';
-    const processor = new OdtProcessor(odtPath);
-    await processor.load();
+    const processor = new OdtProcessor();
+    await processor.load(odtPath);
 
     const content = processor.getContentXml();
 

test/odt_md/Issues.test.ts

@@ -64,8 +64,8 @@ describe('MarkDownTransformTest', () => {
 async function transformOdt(id: string) {
   const folder = new FileContentService(__dirname);
   const odtPath = folder.getRealPath() + '/' + id + '.odt';
-  const processor = new OdtProcessor(odtPath);
-  await processor.load();
+  const processor = new OdtProcessor();
+  await processor.load(odtPath);
   if (!processor.getContentXml()) {
     throw Error('No odt processed');
   }

test/odt_md/MarkDownTransform.test.ts

@@ -1,7 +1,7 @@
 import {assert} from 'chai';
 import fs from 'fs';
 
-import {compareTexts, createTmpDir} from '../utils.ts';
+import {compareTexts} from '../utils.ts';
 import {OdtToMarkdown} from '../../src/odt/OdtToMarkdown.ts';
 import {DocumentContent, DocumentStyles, LIBREOFFICE_CLASSES} from '../../src/odt/LibreOffice.ts';
 import {UnMarshaller} from '../../src/odt/UnMarshaller.ts';
@@ -135,8 +135,8 @@ describe('MarkDownTransformTest', () => {
 async function transformOdt(id: string) {
   const folder = new FileContentService(__dirname);
   const odtPath = folder.getRealPath() + '/' + id + '.odt';
-  const processor = new OdtProcessor(odtPath);
-  await processor.load();
+  const processor = new OdtProcessor();
+  await processor.load(odtPath);
   if (!processor.getContentXml()) {
     throw Error('No odt processed');
   }

test/odt_md/RewriteRules.test.ts

@@ -42,8 +42,8 @@ describe('RewriteRulesTest', () => {
 async function transformOdt(id: string) {
   const folder = new FileContentService(__dirname);
   const odtPath = folder.getRealPath() + '/' + id + '.odt';
-  const processor = new OdtProcessor(odtPath);
-  await processor.load();
+  const processor = new OdtProcessor();
+  await processor.load(odtPath);
   if (!processor.getContentXml()) {
     throw Error('No odt processed');
   }