PDF a11y: Properly handle pages with only whitespaces.

https://crrev.com/690041 added handling for out of bound indices in
PDFiumPage::GetTextRunInfo(), but failed to consider a valid situation
where all the text in a text run are whitespaces. In this case,
GetTextRunInfo() needs to report the number of characters processed.

Add a test case with just a whitespace to test this condition.

Also update hello-world.in to have a {{streamlen}} macro and regenerate
hello-world.pdf

Bug: 997747
Change-Id: Id9e074080b19e4c642cf75fa6b72c5eb969197a0
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1769842Reviewed-by: Ian Prest <iapres@microsoft.com>
Commit-Queue: Lei Zhang <thestig@chromium.org>
Cr-Commit-Position: refs/heads/master@{#690811}

chrome/browser/pdf/pdf_extension_test.cc

@@ -2373,6 +2373,13 @@ IN_PROC_BROWSER_TEST_F(PDFExtensionAccessibilityTextExtractionTest,
   RunTextExtractionTest(FILE_PATH_LITERAL("font-change.pdf"));
 }
 
+// Test one property of pdf_private/accessibility_crash_2.pdf, where a page has
+// only whitespace characters.
+IN_PROC_BROWSER_TEST_F(PDFExtensionAccessibilityTextExtractionTest,
+                       OnlyWhitespaceText) {
+  RunTextExtractionTest(FILE_PATH_LITERAL("whitespace.pdf"));
+}
+
 class PDFExtensionAccessibilityTreeDumpTest
     : public PDFExtensionTest,
       public ::testing::WithParamInterface<size_t> {
@@ -2566,4 +2573,4 @@ IN_PROC_BROWSER_TEST_P(PDFExtensionAccessibilityTreeDumpTest,
 
 IN_PROC_BROWSER_TEST_P(PDFExtensionAccessibilityTreeDumpTest, TextDirection) {
   RunPDFTest(FILE_PATH_LITERAL("text-direction.pdf"));
-}
+}
\ No newline at end of file

chrome/test/data/pdf/accessibility/hello-world.in

@@ -29,6 +29,7 @@ endobj
 >>
 endobj
 {{object 5 0}} <<
+  {{streamlen}}
 >>
 stream
 BT

chrome/test/data/pdf/accessibility/whitespace-expected.txt

+ 

chrome/test/data/pdf/accessibility/whitespace.in

+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+endobj
+{{object 2 0}} <<
+  /Type /Pages
+  /MediaBox [ 0 0 200 200 ]
+  /Count 1
+  /Kids [ 3 0 R ]
+>>
+endobj
+{{object 3 0}} <<
+  /Type /Page
+  /Parent 2 0 R
+  /Resources <<
+    /Font <<
+      /F1 4 0 R
+    >>
+  >>
+  /Contents 5 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+{{object 5 0}} <<
+  {{streamlen}}
+>>
+stream
+BT
+20 100 Td
+/F1 16 Tf
+( ) Tj
+ET
+endstream
+endobj
+{{xref}}
+{{trailer}}
+{{startxref}}
+%%EOF

pdf/pdfium/pdfium_page.cc

@@ -254,21 +254,27 @@ FPDF_TEXTPAGE PDFiumPage::GetTextPage() {
 void PDFiumPage::GetTextRunInfo(
     int start_char_index,
     PP_PrivateAccessibilityTextRunInfo* text_run_info) {
-  if (start_char_index < 0) {
+  FPDF_PAGE page = GetPage();
+  FPDF_TEXTPAGE text_page = GetTextPage();
+  int chars_count = FPDFText_CountChars(text_page);
+  // Check to make sure |start_char_index| is within bounds.
+  if (start_char_index < 0 || start_char_index >= chars_count) {
     text_run_info->len = 0;
     text_run_info->font_size = 0;
     text_run_info->bounds = pp::FloatRect();
     text_run_info->direction = PP_PRIVATEDIRECTION_NONE;
     return;
   }
-  FPDF_PAGE page = GetPage();
-  FPDF_TEXTPAGE text_page = GetTextPage();
-  int chars_count = FPDFText_CountChars(text_page);
 
   int actual_start_char_index = GetFirstNonUnicodeWhiteSpaceCharIndex(
       text_page, start_char_index, chars_count);
+  // Check to see if GetFirstNonUnicodeWhiteSpaceCharIndex() iterated through
+  // all the characters.
   if (actual_start_char_index >= chars_count) {
-    text_run_info->len = 0;
+    // If so, |text_run_info->len| needs to take the number of characters
+    // iterated into account.
+    DCHECK_GT(actual_start_char_index, start_char_index);
+    text_run_info->len = chars_count - start_char_index;
     text_run_info->font_size = 0;
     text_run_info->bounds = pp::FloatRect();
     text_run_info->direction = PP_PRIVATEDIRECTION_NONE;