(feat) parse docx or pdf

Signed-off-by: Dan Selman <[email protected]>

Author: dselman

packages/markdown-cli/lib/Commands.js

@@ -21,6 +21,8 @@ const CommonMarkTransformer = require('@accordproject/markdown-common').CommonMa
 const CiceroMarkTransformer = require('@accordproject/markdown-cicero').CiceroMarkTransformer;
 const SlateTransformer = require('@accordproject/markdown-slate').SlateTransformer;
 const HtmlTransformer = require('@accordproject/markdown-html').HtmlTransformer;
+const PdfTransformer = require('@accordproject/markdown-pdf').PdfTransformer;
+const DocxTransformer = require('@accordproject/markdown-docx').DocxTransformer;
 
 /**
  * Utility class that implements the commands exposed by the CLI.
@@ -62,7 +64,7 @@ class Commands {
         argv = Commands.setDefaultFileArg(argv, 'sample', 'sample.md', ((argv, argDefaultName) => { return argDefaultName; }));
 
         if(argv.verbose) {
-            Logger.info(`parse sample ${argv.sample} printing intermediate transformations.`);
+            Logger.info(`parse sample ${argv.sample} (or docx, pdf) printing intermediate transformations.`);
         }
 
         return argv;
@@ -80,7 +82,7 @@ class Commands {
      * @param {boolean} [options.verbose] verbose output
      * @returns {object} Promise to the result of parsing
      */
-    static parse(samplePath, outputPath, options) {
+    static async parse(samplePath, outputPath, options) {
         const { cicero, slate, html, verbose } = options;
         const commonOptions = {};
         commonOptions.tagInfo = true;
@@ -89,9 +91,24 @@ class Commands {
         const ciceroMark = new CiceroMarkTransformer();
         const slateMark = new SlateTransformer();
         const htmlMark = new HtmlTransformer();
+        const docx = new DocxTransformer();
+        const pdf = new PdfTransformer();
+
+        let result = null;
+
+        if(samplePath.endsWith('.pdf')) {
+            const pdfBuffer = Fs.readFileSync(samplePath);
+            result = await pdf.toCiceroMark(pdfBuffer, 'json');
+        }
+        else if(samplePath.endsWith('.docx')) {
+            const docxBuffer = Fs.readFileSync(samplePath);
+            result = await docx.toCiceroMark(docxBuffer, 'json');
+        }
+        else {
+            const markdownText = Fs.readFileSync(samplePath, 'utf8');
+            result = commonMark.fromMarkdown(markdownText, 'json');
+        }
 
-        const markdownText = Fs.readFileSync(samplePath, 'utf8');
-        let result = commonMark.fromMarkdown(markdownText, 'json');
         if(verbose) {
             Logger.info('=== CommonMark ===');
             Logger.info(JSON.stringify(result, null, 4));
@@ -150,7 +167,7 @@ class Commands {
     }
 
     /**
-     * Parse a sample markdown
+     * Parse a sample markdown/pdf/docx
      *
      * @param {string} dataPath to the sample file
      * @param {string} outputPath to an output file

packages/markdown-cli/package.json

@@ -57,6 +57,8 @@
     "@accordproject/markdown-cicero": "0.9.4",
     "@accordproject/markdown-slate": "0.9.4",
     "@accordproject/markdown-html": "0.9.4",
+    "@accordproject/markdown-pdf": "0.9.4",
+    "@accordproject/markdown-docx": "0.9.4",
     "immutable": "^4.0.0-rc.12",
     "jsome": "2.5.0",
     "slate": "^0.47.8",

packages/markdown-docx/src/DocxTransformer.js

@@ -34,7 +34,7 @@ class DocxTransformer {
 
     /**
      * Converts an html string to a CiceroMark DOM
-     * @param {string} input - html string
+     * @param {Buffer} input - docx buffer
      * @param {string} [format] result format, defaults to 'concerto'. Pass
      * 'json' to return the JSON data.
      * @returns {promise} promise to the CiceroMark DOM

packages/markdown-docx/src/DocxTransformer.test.js

@@ -48,6 +48,7 @@ describe.only('import docx', () => {
     getDocxFiles().forEach(([file, docx], i) => {
         it(`converts ${file} to ciceromark`, async () => {
             const json = await docxTransformer.toCiceroMark(docx, 'json');
+            console.log(JSON.stringify(json, null, 4));
             expect(json).toMatchSnapshot(); // (1)
         });
     });

packages/markdown-pdf/src/PdfTransformer.js

@@ -34,7 +34,7 @@ class PdfTransformer {
 
     /**
      * Converts an pdf buffer to a CiceroMark DOM
-     * @param {string} input - html string
+     * @param {Buffer} input - pdf buffer
      * @param {string} [format] result format, defaults to 'concerto'. Pass
      * 'json' to return the JSON data.
      * @returns {promise} a Promise to the CiceroMark DOM

packages/markdown-pdf/src/PdfTransformer.test.js

@@ -48,6 +48,7 @@ describe.only('pdf import', () => {
     getPdfFiles().forEach(([file, pdfContent], i) => {
         it(`converts ${file} to cicero mark`, async () => {
             const ciceroMarkDom = await pdfTransformer.toCiceroMark(pdfContent, 'json');
+            console.log(JSON.stringify(ciceroMarkDom, null, 4));
             expect(ciceroMarkDom).toMatchSnapshot(); // (1)
         });
     });