Fixes #478 (/Index problem) (#479)

* Add files via upload Fixing problem of incomplete analysis of the /Index entry. * Delete RawDataParser.php Wrong subdirectory. * Add files via upload Fix problem of uncomplete analysis of /Index entry. * Update RawDataParser.php optical changes * Update RawDataParser.php optical changes * Update RawDataParser.php optical changes * Add files via upload After adding a description to the file, the valid /Index entry now contains two entries (consisting of 2 values: first object number, number of objects): /Index[2 1 21 2] * Update RawDataParserTest.php Adding test for issue 479 * Update RawDataParserTest.php Forgot a { * Update RawDataParser.php Code style update * Update RawDataParserTest.php Added more description and more checks. * Update PageTest.php Issue #331 is fixed by issue #479: test updated * Update RawDataParserTest.php optical fix * Update PageTest.php optical changes * Update RawDataParser.php change to remove the native_function_invocation message * Update tests/Integration/PageTest.php Co-authored-by: Konrad Abicht <[email protected]> * Update RawDataParser.php Added comments... * Update RawDataParser.php Changes for CS fixer * Update PageTest.php Comment update * Update tests/Integration/PageTest.php Co-authored-by: Konrad Abicht <[email protected]> Co-authored-by: Konrad Abicht <[email protected]>
smalot · Nov 22, 2021 · 768d1d6 · 768d1d6
1 parent a216ccd
commit 768d1d6
Show file tree

Hide file tree

Showing 4 changed files with 65 additions and 10 deletions.
diff --git a/samples/bugs/Issue479.pdf b/samples/bugs/Issue479.pdf
diff --git a/src/Smalot/PdfParser/RawData/RawDataParser.php b/src/Smalot/PdfParser/RawData/RawDataParser.php
@@ -269,8 +269,11 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
             ) {
                 $valid_crs = true;
             } elseif (('/' == $v[0]) && ('Index' == $v[1]) && (isset($sarr[($k + 1)]))) {
-                // first object number in the subsection
-                $index_first = (int) ($sarr[($k + 1)][1][0][1]);
+                // initialize list for: first object number in the subsection / number of objects
+                $index_blocks = [];
+                for ($m = 0; $m < \count($sarr[($k + 1)][1]); $m += 2) {
+                    $index_blocks[] = [$sarr[($k + 1)][1][$m][1], $sarr[($k + 1)][1][$m + 1][1]];
+                }
             } elseif (('/' == $v[0]) && ('Prev' == $v[1]) && (isset($sarr[($k + 1)]) && ('numeric' == $sarr[($k + 1)][0]))) {
                 // get previous xref offset
                 $prevxref = (int) ($sarr[($k + 1)][1]);
@@ -432,8 +435,9 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
             }
 
             // fill xref
-            if (isset($index_first)) {
-                $obj_num = $index_first;
+            if (isset($index_blocks)) {
+                // load the first object number of the first /Index entry
+                $obj_num = $index_blocks[0][0];
             } else {
                 $obj_num = 0;
             }
@@ -463,6 +467,21 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
                             break;
                 }
                 ++$obj_num;
+                if (isset($index_blocks)) {
+                    // reduce the number of remaining objects
+                    --$index_blocks[0][1];
+                    if (0 == $index_blocks[0][1]) {
+                        // remove the actual used /Index entry
+                        array_shift($index_blocks);
+                        if (0 < \count($index_blocks)) {
+                            // load the first object number of the following /Index entry
+                            $obj_num = $index_blocks[0][0];
+                        } else {
+                            // if there are no more entries, remove $index_blocks to avoid actions on an empty array
+                            unset($index_blocks);
+                        }
+                    }
+                }
             }
         } // end decoding data
         if (isset($prevxref)) {

diff --git a/tests/Integration/PageTest.php b/tests/Integration/PageTest.php
@@ -486,12 +486,17 @@ public function testGetPages()
         $document = $this->getParserInstance()->parseFile($filename);
         $pages = $document->getPages();
 
-        // This should actually be 3 pages, but as long as the cause for issue #331
-        // has not been found and the issue is not fixed, we'll settle for 2 here.
-        // We still test for the count, so in case the bug should be fixed
-        // unknowingly, we don't forget to resolve the issue as well and make sure
-        // this assertion is present.
-        $this->assertCount(2, $pages);
+        /*
+         * The problem of issue #331 is fixed by the pull request of the issue #479.
+         * The original Issue331.pdf was modified so for the updated version (actual
+         * version) a new xref was added and now the valid /Index has the following value:
+         *    [1 1 3 1 7 1 175 1 178 1 219 2]
+         * This means, that there a 6 pairs containing the values for 'first object id'
+         * and 'number of objects'. Till now only the first entry was used and so the
+         * objects of all following entries gots a wrong id.
+         * By the fix of issue #479 now the expected number of pages is counted.
+         */
+        $this->assertCount(3, $pages);
 
         foreach ($pages as $page) {
             $this->assertTrue($page instanceof Page);

diff --git a/tests/Integration/RawData/RawDataParserTest.php b/tests/Integration/RawData/RawDataParserTest.php
@@ -119,4 +119,35 @@ public function testDecodeObjectHeaderIssue405()
 
         $this->assertStringContainsString('Bug fix: PR #405', $pages[0]->getText());
     }
+
+    /**
+     * Tests buggy behavior of decodeXrefStream.
+     *
+     * When PDF has more than one entry in the /Index area (for example by changing
+     * the document description), only the first entry is used.
+     * If the fix is not used the array returned by getDetails() contains only the entry
+     * with the key 'Pages'. All other entries like 'Author', 'Creator', 'Title',
+     * 'Subject' (which come from the 'Info' object) are not listed, because the
+     * 'Info' object gets a wrong object id during parsing the data into the xref structure.
+     * So the object id listed at the /Info entry is not valid and the data of the info object
+     * cannot be loaded during executing Document::buildDetails().
+     *
+     * @see https://github.com/smalot/pdfparser/pull/479
+     */
+    public function testDecodeXrefStreamIssue479()
+    {
+        $filename = $this->rootDir.'/samples/bugs/Issue479.pdf';
+
+        $parser = $this->getParserInstance();
+        $document = $parser->parseFile($filename);
+        $details = $document->getDetails();
+
+        $this->assertArrayHasKey('Author', $details);
+        $this->assertArrayHasKey('CreationDate', $details);
+        $this->assertArrayHasKey('Creator', $details);
+        $this->assertArrayHasKey('ModDate', $details);
+        $this->assertArrayHasKey('Producer', $details);
+        $this->assertArrayHasKey('Subject', $details);
+        $this->assertArrayHasKey('Title', $details);
+    }
 }