Merge pull request #9986 from Snuffleupagus/issue-9984

Attempt to combine separate beginText/endText sequences in `getTextContent` (issue 9984)

src/core/evaluator.js

@@ -1510,6 +1510,17 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         textContentItem.str.length = 0;
       }
 
+      function isIdenticalSetFont(name, size) {
+        return (textState.font &&
+                name === textState.fontName && size === textState.fontSize);
+      }
+
+      function handleBeginText() {
+        flushTextContentItem();
+        textState.textMatrix = IDENTITY_MATRIX.slice();
+        textState.textLineMatrix = IDENTITY_MATRIX.slice();
+      }
+
       function enqueueChunk() {
         let length = textContent.items.length;
         if (length > 0) {
@@ -1535,6 +1546,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         task.ensureNotTerminated();
         timeSlotManager.reset();
         var stop, operation = {}, args = [];
+        let pendingBeginText = false;
         while (!(stop = timeSlotManager.check())) {
           // The arguments parsed by read() are not used beyond this loop, so
           // we can reuse the same array on every iteration, thus avoiding
@@ -1545,16 +1557,30 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
             break;
           }
           textState = stateManager.state;
-          var fn = operation.fn;
+          var fn = operation.fn | 0;
           args = operation.args;
           var advance, diff;
 
-          switch (fn | 0) {
+          if (pendingBeginText) {
+            if (fn === OPS.setFont) {
+              const fontNameArg = args[0].name, fontSizeArg = args[1];
+              // For multiple identical Tf (setFont) commands, first check if
+              // the following command is Tm (setTextMatrix) before continuing.
+              if (isIdenticalSetFont(fontNameArg, fontSizeArg)) {
+                continue;
+              }
+            }
+            if (fn !== OPS.setTextMatrix) {
+              handleBeginText();
+            }
+            pendingBeginText = false;
+          }
+
+          switch (fn) {
             case OPS.setFont:
               // Optimization to ignore multiple identical Tf commands.
               var fontNameArg = args[0].name, fontSizeArg = args[1];
-              if (textState.font && fontNameArg === textState.fontName &&
+              if (isIdenticalSetFont(fontNameArg, fontSizeArg)) {
-                  fontSizeArg === textState.fontSize) {
                 break;
               }
 
@@ -1642,9 +1668,15 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               textState.wordSpacing = args[0];
               break;
             case OPS.beginText:
-              flushTextContentItem();
+              // Optimization to attempt to combine separate BT/ET sequences,
-              textState.textMatrix = IDENTITY_MATRIX.slice();
+              // by checking the next operator(s) before flushing text content
-              textState.textLineMatrix = IDENTITY_MATRIX.slice();
+              // and resetting the text/textLine matrices (see above).
+              if (combineTextItems) {
+                pendingBeginText = true;
+                break;
+              }
+
+              handleBeginText();
               break;
             case OPS.showSpacedText:
               var items = args[0];

test/pdfs/.gitignore

@@ -72,6 +72,7 @@
 !issue9458.pdf
 !issue9915_reduced.pdf
 !issue9940.pdf
+!issue9984.pdf
 !bad-PageLabels.pdf
 !decodeACSuccessive.pdf
 !filled-background.pdf

test/test_manifest.json

@@ -1359,6 +1359,13 @@
        "link": false,
        "type": "eq"
     },
+    {  "id": "issue9984-text",
+       "file": "pdfs/issue9984.pdf",
+       "md5": "41be5f1b43f61892978cfc57c74ccf4c",
+       "rounds": 1,
+       "link": false,
+       "type": "text"
+    },
     {  "id": "issue8570",
        "file": "pdfs/issue8570.pdf",
        "md5": "0355731adb72df233eaa10464dcc8c51",