feat: extract more info from the HTML content

Report the following elements in the `data` section of the report:
- `audio`
- `video`
- `object`
- `script`
- `iframe`
- `embed`
- `object`
- `canvas`
- `map`

Add a new `properties` section to the report to list boolean statements
on the document:
- `hasMathML`
- `hasPageBreaks`

Add units and integration tests.

Closes #26.

.eslintrc.js

@@ -5,6 +5,7 @@ module.exports = {
   },
   "rules": {
     "no-console": 0,
+    "no-param-reassign": ["error", { "props": false }],
     "no-underscore-dangle": ["error", { "allowAfterThis": true }],
   },
   "plugins": [

src/checker/checker-nightmare.js

@@ -61,13 +61,22 @@ function checkSingle(spineItem, epub, nightmare) {
         (results.assertions == null)
           ? 'No'
           : results.assertions.assertions.length} issues found`);
-      if (results.data != null && results.data.images != null) {
-        results.data.images.forEach((img) => {
-          const imageFullPath = path.resolve(path.dirname(spineItem.filepath), img.path);
-          const imageRelPath = path.relative(epub.basedir, imageFullPath);
-          img.filepath = imageFullPath;
-          img.path = imageRelPath;
-          img.location = `${spineItem.relpath}#epubcfi(${img.cfi})`;
+      // Resolve path and locators for extracted data
+      if (results.data != null) {
+        Object.getOwnPropertyNames(results.data).forEach((key) => {
+          if (!Array.isArray(results.data[key])) return;
+          results.data[key].forEach((item) => {
+            if (item.src !== undefined) {
+              const fullpath = path.resolve(path.dirname(spineItem.filepath), item.src);
+              const relpath = path.relative(epub.basedir, fullpath);
+              item.path = fullpath;
+              item.src = relpath;
+              if (item.cfi !== undefined) {
+                item.location = `${spineItem.relpath}#epubcfi(${item.cfi})`;
+                delete item.cfi;
+              }
+            }
+          });
         });
       }
       return results;

src/checker/checker.js

@@ -3,11 +3,13 @@
 const checker = require('./checker-nightmare.js');
 const winston = require('winston');
 
-function finalize(results, report) {
-  // Copy assertions
-  results
-    .filter(res => res.assertions != null)
-    .forEach(res => report.addContentDocAssertion(res.assertions));
+function consolidate(results, report) {
+  // Integrate checker results to the report
+  results.forEach((res) => {
+    report.addContentDocAssertion(res.assertions);
+    report.addProperties(res.properties);
+    report.addData(res.data);
+  });
   // Get a flat array of all the headings in the documents
   const headings = []
     .concat(...results.map(docResult => docResult.outlines.headings))
@@ -16,16 +18,11 @@ function finalize(results, report) {
   // Aggregated array of the HTML outlines
   const htmlOutlines = results.map(docResult => docResult.outlines.html);
   report.addHTMLOutlines(htmlOutlines);
-  // Get a flat array of the extracted images
-  const images = []
-    .concat(...results.map(docResult => docResult.data.images))
-    .filter(e => e !== undefined);
-  report.addImages(images);
   return report;
 }
 
 module.exports.check = function check(epub, report) {
   winston.info('Checking documents...');
   return checker.check(epub)
-    .then(results => finalize(results, report));
+    .then(results => consolidate(results, report));
 };

src/report/report-builders.js

@@ -108,6 +108,13 @@ class ReportBuilder {
     withAssertion(this._json, assertions);
     return this;
   }
+  withData(data) {
+    Object.getOwnPropertyNames(data).forEach((key) => {
+      this._json.data[key] = (Array.isArray(this._json.data[key]))
+        ? this._json.data[key].push(data[key])
+        : data[key];
+    });
+  }
   withEPUBOutline(outline) {
     this._json.outlines.toc = outline;
     return this;
@@ -120,10 +127,6 @@ class ReportBuilder {
     this._json.outlines.html = outline;
     return this;
   }
-  withImages(images) {
-    this._json.data.images = images;
-    return this;
-  }
   withProperties(properties) {
     Object.getOwnPropertyNames(properties).forEach((key) => {
       this._json.properties[key] = (key in properties)

src/report/report.js

@@ -53,6 +53,9 @@ module.exports = class Report {
     this._builder.withAssertion(assertion);
     return this;
   }
+  addData(data) {
+    this._builder.withData(data);
+  }
   addOutline(outline) {
     this._builder.withHOutline(outline);
     return this;
@@ -69,10 +72,6 @@ module.exports = class Report {
     this._builder.withEPUBOutline(navDoc.tocHTML);
     return this;
   }
-  addImages(images) {
-    this._builder.withImages(images);
-    return this;
-  }
   addProperties(properties) {
     this._builder.withProperties(properties);
     return this;
@@ -87,9 +86,9 @@ module.exports = class Report {
         .then(() => {
           if (this.json.data.images != null) {
             this.json.data.images.forEach((img) => {
-              const fromPath = img.filepath;
-              const toPath = path.join(outdir, 'data', img.path);
-              delete img.filepath;
+              const fromPath = img.path;
+              const toPath = path.join(outdir, 'data', img.src);
+              delete img.path;
               return fs.copy(fromPath, toPath, {
                 overwrite: false,
               });

src/scripts/ace-extraction.js

@@ -3,13 +3,89 @@ window.daisy =  window.daisy || {};
 ace = daisy.ace = daisy.ace || {};
 
 ace.createReport = function(report) {
+    let reportData = function(field, array) {
+      if (array.length > 0) report.data[field] = array;
+    }
     report.outlines = report.outlines || {};
     report.outlines.html = ace.getHTMLOutline();
     report.outlines.headings = ace.getHeadings();
     report.data = report.data || {};
-    report.data.images = ace.getImages();
+    reportData('images', ace.getImages());
+    reportData('audios', ace.getAudios());
+    reportData('canvases', ace.getCanvases());
+    reportData('embeds', ace.getEmbeds());
+    reportData('iframes', ace.getIframes());
+    reportData('maps', ace.getMaps());
+    reportData('scripts`', ace.getScripts());
+    reportData('videos', ace.getVideos());
+    report.properties = report.properties || {};
+    report.properties.hasMathML = ace.hasMathML();
+    report.properties.hasPageBreaks = ace.hasPageBreaks();
 };
 
+ace.getAudios = function() {
+  let audioElems = document.querySelectorAll('audio');
+  let audios = [];
+  audioElems.forEach(function(audio) {
+    let audioObj = {
+      controls: audio.hasAttribute('controls'),
+      cfi: window.daisy.epub.createCFI(audio),
+      html: audio.outerHTML,
+      tracks: [],
+    }
+    if (audio.hasAttribute('id')) audioObj.id = audio.getAttribute('id');
+    if (audio.hasAttribute('src')) {
+      audioObj.src = audio.getAttribute('src');
+    } else {
+      audioObj.src = [];
+      audio.querySelectorAll('source').forEach(function(source) {
+        audioObj.src.push({
+          src: source.getAttribute('src'),
+          type: source.getAttribute('type'),
+        });
+      });
+    }
+    audio.querySelectorAll('track').forEach(function(track) {
+      let trackObj = {};
+      if (track.hasAttribute('label')) trackObj.label = track.getAttribute('label');
+      if (track.hasAttribute('kind')) trackObj.kind = track.getAttribute('kind');
+      if (track.hasAttribute('src')) trackObj.src = track.getAttribute('src');
+      if (track.hasAttribute('srclang')) trackObj.srclang = track.getAttribute('srclang');
+      audioObj.tracks.push(trackObj);
+    });
+    audios.push(audioObj);
+  });
+  return audios;
+}
+
+ace.getCanvases = function() {
+  let canvasElems = document.querySelectorAll('canvas');
+  let canvass = [];
+  canvasElems.forEach(function(canvas) {
+    let canvasObj = {
+      cfi: window.daisy.epub.createCFI(canvas),
+    }
+    if (canvas.hasAttribute('id')) canvasObj.id = canvas.getAttribute('id');
+    canvass.push(canvasObj);
+  });
+  return canvass;
+}
+
+ace.getEmbeds = function() {
+  let embedElems = document.querySelectorAll('embed');
+  let embeds = [];
+  embedElems.forEach(function(embed) {
+    let embedObj = {
+      cfi: window.daisy.epub.createCFI(embed),
+    }
+    if (embed.hasAttribute('id')) embedObj.id = embed.getAttribute('id');
+    if (embed.hasAttribute('src')) embedObj.src = embed.getAttribute('src');
+    if (embed.hasAttribute('type')) embedObj.type = embed.getAttribute('type');
+    embeds.push(embedObj);
+  });
+  return embeds;
+}
+
 ace.getHTMLOutline = function() {
   return HTML5Outline(document.body).asHTML();
 }
@@ -34,6 +110,20 @@ ace.getHeadings = function() {
   return headings;
 }
 
+ace.getIframes = function() {
+  let iframeElems = document.querySelectorAll('iframe');
+  let iframes = [];
+  iframeElems.forEach(function(iframe) {
+    let iframeObj = {
+      cfi: window.daisy.epub.createCFI(iframe),
+    }
+    if (iframe.hasAttribute('id')) iframeObj.id = iframe.getAttribute('id');
+    if (iframe.hasAttribute('src')) iframeObj.src = iframe.getAttribute('src');
+    iframes.push(iframeObj);
+  });
+  return iframes;
+}
+
 ace.getImages = function() {
   var findFigure = function(el) {
     while ((el = el.parentElement) && !(el.localName === 'figure'));
@@ -44,7 +134,7 @@ ace.getImages = function() {
   let images = [];
   imgElems.forEach(function(img) {
     let imageObj = {
-      path: img.getAttribute('src'),
+      src: img.getAttribute('src'),
       alt: img.getAttribute('alt'),
       role: img.getAttribute('role'),
       cfi: window.daisy.epub.createCFI(img),
@@ -64,3 +154,90 @@ ace.getImages = function() {
   });
   return images;
 }
+
+ace.getMaps = function() {
+  let mapElems = document.querySelectorAll('map');
+  let maps = [];
+  mapElems.forEach(function(map) {
+    let mapObj = {
+      cfi: window.daisy.epub.createCFI(map),
+      name: map.getAttribute('name'),
+    }
+    if (map.hasAttribute('id')) mapObj.id = map.getAttribute('id');
+    maps.push(mapObj);
+  });
+  return maps;
+}
+
+ace.getObjects = function() {
+  let objectElems = document.querySelectorAll('object');
+  let objects = [];
+  objectElems.forEach(function(object) {
+    let objectObj = {
+      cfi: window.daisy.epub.createCFI(object),
+    }
+    if (object.hasAttribute('id')) objectObj.id = object.getAttribute('id');
+    if (object.hasAttribute('data')) objectObj.data = object.getAttribute('data');
+    if (object.hasAttribute('type')) objectObj.type = object.getAttribute('type');
+    objects.push(objectObj);
+  });
+  return objects;
+}
+
+ace.getScripts = function() {
+  let scriptElems = document.querySelectorAll('script');
+  let scripts = [];
+  scriptElems.forEach(function(script) {
+    let scriptObj = {
+      cfi: window.daisy.epub.createCFI(script),
+    }
+    scriptObj.type = (script.hasAttribute('type'))?script.getAttribute('type'):'text/javascript';
+    if (script.hasAttribute('id')) scriptObj.id = script.getAttribute('id');
+    if (script.hasAttribute('src')) scriptObj.src = script.getAttribute('src');
+    scripts.push(scriptObj);
+  });
+  return scripts;
+}
+
+ace.getVideos = function() {
+  let videoElems = document.querySelectorAll('video');
+  let videos = [];
+  videoElems.forEach(function(video) {
+    let videoObj = {
+      controls: video.hasAttribute('controls'),
+      cfi: window.daisy.epub.createCFI(video),
+      html: video.outerHTML,
+      tracks: [],
+    }
+    if (video.hasAttribute('id')) videoObj.id = video.getAttribute('id');
+    if (video.hasAttribute('src')) {
+      videoObj.src = video.getAttribute('src');
+    } else {
+      videoObj.src = [];
+      video.querySelectorAll('source').forEach(function(source) {
+        videoObj.src.push({
+          src: source.getAttribute('src'),
+          type: source.getAttribute('type'),
+        });
+      });
+    }
+    video.querySelectorAll('track').forEach(function(track) {
+      let trackObj = {};
+      if (track.hasAttribute('label')) trackObj.label = track.getAttribute('label');
+      if (track.hasAttribute('kind')) trackObj.kind = track.getAttribute('kind');
+      if (track.hasAttribute('src')) trackObj.src = track.getAttribute('src');
+      if (track.hasAttribute('srclang')) trackObj.srclang = track.getAttribute('srclang');
+      videoObj.tracks.push(trackObj);
+    });
+    videos.push(videoObj);
+  });
+  return videos;
+}
+
+ace.hasMathML = function() {
+  return document.querySelectorAll('math').length > 0;
+}
+
+ace.hasPageBreaks = function() {
+  return document.querySelectorAll('[epub\\:type~="pagebreak"], [role~="doc-pagebreak"]').length > 0;
+}