Handle non-printing characters at beginning of extraction region

Currently if a text extraction region begins on a non-printing
character then "" will be returned. This is the incorrect behaviour,
instead the call should scan ahead until a printing character is
found and start extracting from there. Also proactively adds a
similar check and scan for the end of the extraction region.

BUG=pdfium:1139

Change-Id: Ia2001ac89740f3d31d2bb69e8000773f8b01091b
Reviewed-on: https://pdfium-review.googlesource.com/41532
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>

1 files changed, 25 insertions, 4 deletions

diff --git a/core/fpdftext/cpdf_textpage.cpp b/core/fpdftext/cpdf_textpage.cpp
index 5465297263..8c110b6814 100644
--- a/core/fpdftext/cpdf_textpage.cpp
+++ b/core/fpdftext/cpdf_textpage.cpp
@@ -510,15 +510,36 @@ WideString CPDF_TextPage::GetPageText(int start, int count) const {
     return L"";
   }
 
+  const int count_chars = CountChars();
   int text_start = TextIndexFromCharIndex(start);
-  if (text_start < 0)
-    return L"";
 
-  count = std::min(count, CountChars() - start);
+  // If the character at |start| is a non-printing character, then
+  // TextIndexFromCharIndex will return -1, so scan ahead to the first printing
+  // character.
+  while (text_start < 0) {
+    if (start >= count_chars)
+      return L"";
+    start++;
+    text_start = TextIndexFromCharIndex(start);
+  }
+
+  count = std::min(count, count_chars - start);
 
   int last = start + count - 1;
   int text_last = TextIndexFromCharIndex(last);
-  if (text_last < 0 || text_last < text_start)
+
+  // If the character at |last| is a non-printing character, then
+  // TextIndexFromCharIndex will return -1, so scan back to the last printing
+  // character.
+  while (text_last < 0) {
+    if (last < text_start)
+      return L"";
+
+    last--;
+    text_last = TextIndexFromCharIndex(last);
+  }
+
+  if (text_last < text_start)
     return L"";
 
   int text_count = text_last - text_start + 1;