@adobe/helix-md2docx 1.4.26 → 2.0.0

package/CHANGELOG.md

@@ -1,3 +1,15 @@
+# [2.0.0](https://github.com/adobe/helix-md2docx/compare/v1.4.26...v2.0.0) (2022-09-23)
+
+
+### Features
+
+* provide resource loader and converter plugin capability ([#136](https://github.com/adobe/helix-md2docx/issues/136)) ([0a6b703](https://github.com/adobe/helix-md2docx/commit/0a6b7034ca9118016a24197647d49835cddad83d)), closes [#133](https://github.com/adobe/helix-md2docx/issues/133) [#134](https://github.com/adobe/helix-md2docx/issues/134) [#135](https://github.com/adobe/helix-md2docx/issues/135)
+
+
+### BREAKING CHANGES
+
+* svg2png was renamed to a more generic image2png
+
 ## [1.4.26](https://github.com/adobe/helix-md2docx/compare/v1.4.25...v1.4.26) (2022-09-12)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-md2docx",
-  "version": "1.4.26",
+  "version": "2.0.0",
   "description": "Helix Service that converts markdown to word documents",
   "main": "src/index.js",
   "type": "module",
@@ -44,13 +44,13 @@
   },
   "devDependencies": {
     "@adobe/eslint-config-helix": "1.3.2",
-    "@adobe/helix-mediahandler": "1.2.2",
+    "@adobe/helix-mediahandler": "1.2.4",
     "@semantic-release/changelog": "6.0.1",
     "@semantic-release/exec": "6.0.3",
     "@semantic-release/git": "10.0.1",
     "c8": "7.12.0",
     "dotenv": "16.0.2",
-    "eslint": "8.23.0",
+    "eslint": "8.23.1",
     "eslint-import-resolver-exports": "1.0.0-beta.3",
     "eslint-plugin-header": "3.1.1",
     "eslint-plugin-import": "2.26.0",
@@ -59,6 +59,7 @@
     "lint-staged": "13.0.3",
     "mocha": "10.0.0",
     "mocha-multi-reporters": "1.5.1",
+    "nock": "13.2.9",
     "semantic-release": "19.0.5",
     "unist-util-inspect": "7.0.1",
     "yauzl": "2.10.0"

package/src/mdast2docx/handlers/image.js

@@ -136,10 +136,10 @@ export default async function image(ctx, node) {
     }
   }
 
-  if (data.ext === 'svg' || data.svgBuffer) {
+  if (data.originalType === 'image/svg') {
     // create a fake image run for the svg image
     const ir = new ImageRun({
-      data: data.svgBuffer,
+      data: data.originalBuffer,
       transformation: data.dimensions,
     });
     ir.imageData.fileName = data.svgKey;

package/src/mdast2docx/index.d.ts

@@ -10,14 +10,38 @@
  * governing permissions and limitations under the License.
  */
 
+import {RequestOptions, Response} from "@adobe/helix-fetch";
+
+declare interface ImageToPngOptions {
+  /**
+   * image data
+   */
+  data:Buffer|ArrayBuffer;
+  /**
+   * informational src information
+   */
+  src?:string;
+  /**
+   * image content type, if available.
+   */
+  type?:string;
+}
+
+declare interface ImageToPngResult {
+  data:Buffer|ArrayBuffer;
+  type?:string;
+  width:number;
+  height:number;
+}
+
+declare type ImageToPngConverter = (opts:ImageToPngOptions) => Promise<ImageToPngResult>;
+
 /**
- * Converts an SVG to a PNG.
- *
- * @function
- * @param {string} svg the input svg
- * @return {string} the base64 encoded png
+ * Loader used for loading resources for urls starting with `res:`
  */
-declare type Svg2PngConverter = (svg:string) => Promise<string>;
+declare interface ResourceLoader {
+  fetch(url:string, opts:RequestOptions): Promise<Response>
+}
 
 declare interface Mdast2DocxOptions {
   /**
@@ -30,7 +54,15 @@ declare interface Mdast2DocxOptions {
    */
   stylesXML: string;
 
-  svg2png?: Svg2PngConverter;
+  /**
+   * Optional loader for (image) resources
+   */
+  resourceLoader?:ResourceLoader;
+
+  /**
+   * Optional image2png converter
+   */
+  image2png?: ImageToPngConverter
 }
 
 /**

package/src/mdast2docx/index.js

@@ -27,7 +27,8 @@ const { Document, Packer } = docx;
 export default async function mdast2docx(mdast, opts = {}) {
   const {
     log = console,
-    svg2png,
+    resourceLoader,
+    image2png,
   } = opts;
 
   let {
@@ -42,7 +43,8 @@ export default async function mdast2docx(mdast, opts = {}) {
     listLevel: -1,
     lists: [],
     log,
-    svg2png,
+    image2png,
+    resourceLoader,
   };
 
   // eslint-disable-next-line no-param-reassign

package/src/mdast2docx/mdast-download-images.js

@@ -40,6 +40,7 @@ function hsize(bytes, decimals = 2) {
  * @returns {Promise<void>}
  */
 export default async function downloadImages(ctx, tree) {
+  const { log, resourceLoader, image2png } = ctx;
   const context = createFetchContext();
   const { fetch } = context;
 
@@ -63,33 +64,77 @@ export default async function downloadImages(ctx, tree) {
       if (node.data) {
         return;
       }
-
+      const idx = String(count).padStart(2, ' ');
+      count += 1;
       let buffer;
       let type;
-      if (node.url.startsWith('data:image/png;base64,')) {
-        buffer = Buffer.from(node.url.split(',').pop(), 'base64');
-        type = 'image/png';
+      let maybeConvert;
+      let dimensions = {
+        width: 100,
+        height: 100,
+      };
+      if (node.url.startsWith('data:')) {
+        const [prefix, data] = node.url.substring(5).split(',');
+        const [typ, enc] = prefix.split(';');
+        if (enc !== 'base64') {
+          log.warn(`[${idx}] Error decoding data url. unknown encoding: ${enc}`);
+          return;
+        }
+        buffer = Buffer.from(data, 'base64');
+        type = typ;
       } else {
-        const idx = String(count).padStart(2, ' ');
-        count += 1;
-        ctx.log.info(`[${idx}] GET ${node.url}`);
-        const ret = await fetch(node.url);
+        log.info(`[${idx}] GET ${node.url}`);
+
+        let doFetch = fetch;
+        if (node.url.startsWith('res:')) {
+          if (!resourceLoader) {
+            log.warn(`[${idx}] Error loading image ${node.url}. resource loader missing.`);
+            return;
+          }
+          doFetch = resourceLoader.fetch.bind(resourceLoader);
+        }
+
+        const ret = await doFetch(node.url);
         if (!ret.ok) {
           const text = await ret.text();
-          ctx.log.error(`[${idx}] ${ret.status} ${text}`);
+          log.error(`[${idx}] ${ret.status} ${text}`);
           return;
         }
         buffer = await ret.buffer();
         type = ret.headers.get('content-type');
-        ctx.log.info(`[${idx}] ${ret.status} ${hsize(buffer.length).padStart(10)} ${type}`);
+        log.info(`[${idx}] ${ret.status} ${hsize(buffer.length).padStart(10)} ${type}`);
       }
-      const dimensions = getDimensions(buffer);
 
-      let svgBuffer;
-      if (type === 'image/svg+xml' && ctx.svg2png) {
-        svgBuffer = buffer;
-        buffer = Buffer.from(await ctx.svg2png(svgBuffer.toString('utf-8')), 'base64');
-        type = 'image/png';
+      maybeConvert = type !== 'image/png' && type !== 'image/jpg' && type !== 'image/jpeg' && type !== 'image/gif';
+      try {
+        dimensions = getDimensions(buffer);
+      } catch (e) {
+        maybeConvert = true;
+        log.warn(`[${idx}] Error detecting dimensions: ${e} ${type}`);
+      }
+
+      let originalBuffer;
+      let originalType;
+      if (maybeConvert && image2png) {
+        try {
+          const result = await image2png({
+            src: node.url,
+            data: buffer,
+            type,
+          });
+          if (result) {
+            originalBuffer = buffer;
+            originalType = type;
+            buffer = result.data;
+            type = result.type;
+            dimensions = {
+              width: result.width,
+              height: result.height,
+            };
+          }
+        } catch (e) {
+          log.warn(`[${idx}] Error to convert to png`, e);
+        }
       }
 
       const ext = mime.getExtension(type);
@@ -98,13 +143,14 @@ export default async function downloadImages(ctx, tree) {
         ext,
         key: `${ref}.${ext}`,
         buffer,
-        svgBuffer,
-        svgKey: `${ref}.svg`,
+        originalBuffer,
+        type,
+        originalType,
         dimensions,
       };
       ctx.images[ref] = node.data;
     } catch (error) {
-      ctx.log.error(`Cannot download image ${node.url}: ${error.message}`);
+      log.error(`Cannot download image ${node.url}: ${error.message}`);
     }
   }, 8);