Improve copy/paste by inserting spaces into textChunks if we deem it appropriate.

 Add test re same. PR mozilla#5783.

Author: speedplane

src/core/evaluator.js

@@ -1037,8 +1037,66 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         } else {
           textChunk.height += Math.abs(height * scaleCtmX * scaleLineX);
         }
+        addSpaceIfNecessary(textChunk, font);
         return textChunk;
       }
+      function getChunkPos(chunk, font) {
+        var tx = chunk.transform;
+        var angle = Math.atan2(tx[1], tx[0]);
+        if (font.vertical) {
+          angle += Math.PI / 2;
+        }
+        var fontHeight = Math.sqrt((tx[2] * tx[2]) + (tx[3] * tx[3]));
+        var fontAscent = fontHeight;
+        if (font.ascent) {
+          fontAscent = font.ascent * fontAscent;
+        } else if (font.descent) {
+          fontAscent = (1 + font.descent) * fontAscent;
+        }
+        return {
+          x : (angle === 0) ? tx[4] :(tx[4] + (fontAscent * Math.sin(angle))),
+          y: tx[5]
+        };
+      }
+      function addSpaceIfNecessary(newChunk, font) {
+        // If the new chunk starts with a space, it does not need one.
+        if (newChunk.str[0] === ' ' || newChunk.str[0] === '-') {
+          return;
+        }
+        if (bidiTexts.length === 0) {
+          return;
+        }
+        // If the last chunk ends with a space it does not need one.
+        var lastChunk = bidiTexts[bidiTexts.length - 1];
+        if (lastChunk.str.length === 0) {
+          return;
+        }
+        var lastChar = lastChunk.str[lastChunk.str.length - 1];
+        if (lastChar === ' ' || lastChar === '-') {
+          return;
+        }
+        var lastPos = getChunkPos(lastChunk, font);
+        var newPos = getChunkPos(newChunk, font);
+        var yDiff = Math.abs(lastPos.y  - newPos.y);
+        if (yDiff >= lastChunk.height || yDiff >= newChunk.height) {
+          // On different lines, add a space.
+          lastChunk.str += ' ';
+        } else {
+          var wordSpacing = textState.wordSpacing > 0 ?
+            // Standard wordSpacing
+            textState.wordSpacing:
+            // Hueristic for wordSpacing
+            newChunk.width / newChunk.str.length * 0.6;
+          var addSpace = newPos.x >= lastPos.x ?
+            // Left to right
+            newPos.x >= lastPos.x + lastChunk.width + wordSpacing:
+            // Right to left
+            lastPos.x >= newPos.x + newChunk.width + wordSpacing;
+          if (addSpace) {
+            lastChunk.str += ' ';
+          }
+        }
+      }
 
       var timeSlotManager = new TimeSlotManager();
 

test/pdfs/.gitignore

@@ -118,3 +118,4 @@
 !issue5481.pdf
 !issue5567.pdf
 !issue5701.pdf
+!US6205527_page1.pdf

test/pdfs/US6205527_page1.pdf

@@ -0,0 +1,52 @@
+/* -*- Mode: Java; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
+/* vim: set shiftwidth=2 tabstop=2 autoindent cindent expandtab: */
+/* globals PDFJS, expect, it, describe, Promise, combineUrl, waitsFor,
+           isArray, MissingPDFException */
+
+'use strict';
+
+function waitsForPromiseResolved(promise, successCallback) {
+  var data;
+  promise.then(function(val) {
+    data = val;
+    successCallback(data);
+  },
+  function(error) {
+    // Shouldn't get here.
+    expect(false).toEqual(true);
+  });
+  waitsFor(function() {
+    return data !== undefined;
+  }, 20000);
+}
+
+describe('text-extract', function() {
+  var pdfURL = combineUrl(window.location.href, '../pdfs/US6205527_page1.pdf');
+  var resolvePromise;
+  var pagePromise = new Promise(function (resolve) {
+    resolvePromise = resolve;
+  });
+  PDFJS.getDocument(pdfURL).then(function(doc) {
+    doc.getPage(1).then(function(data) {
+      resolvePromise(data);
+    });
+  });
+  var page;
+  waitsForPromiseResolved(pagePromise, function(data) {
+    page = data;
+  });
+  it('gets text content', function () {
+      waitsForPromiseResolved(pagePromise, function (data) {
+        var textPromise = page.getTextContent();
+        waitsForPromiseResolved(textPromise, function (data) {
+          expect(!!data.items).toEqual(true);
+          var text = data.items.map(function (d) { return d.str; }).join('');
+          // Make sure the text is ordered properly.
+          expect(text.indexOf('Disclosed is an apparatus, a system, a') > 0)
+            .toEqual(true);
+          expect(text.indexOf('device to the computer system; (b) preparing ' +
+            'a storage. media of the peripheral storage') > 0).toEqual(true);
+        });
+      });
+    });
+});

test/unit/text_extract.js

@@ -51,6 +51,7 @@
   <script src="stream_spec.js"></script>
   <script src="parser_spec.js"></script>
   <script src="api_spec.js"></script>
+  <script src="text_extract.js"></script>
   <script src="metadata_spec.js"></script>
   <script src="util_spec.js"></script>
   <script src="cmap_spec.js"></script>