Revert "Revert "Fix the issue that pdfium swallows 'fi' or 'ff' in some pdf files""

This reverts commit fa9756f77ad6145940d3dc697814b84f5755ae17.

TBR=thestig@chromium.org
Review URL: https://codereview.chromium.org/1307353005/

14 files changed, 428 insertions, 0 deletions

diff --git a/testing/resources/pixel/bug_524043_1.in b/testing/resources/pixel/bug_524043_1.in
new file mode 100644
index 0000000000..61ac6f4bb2
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_1.in
@@ -0,0 +1,61 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 1:
+%   /Length identifies intended end of stream, despite embedded keywords.
+%   Both should render as text.
+{{object 6 0}} <<
+  /Length 107
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+endstream
+endobj
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_1_expected.pdf.0.png b/testing/resources/pixel/bug_524043_1_expected.pdf.0.png
new file mode 100644
index 0000000000..a044bb0c37
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_1_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_2.in b/testing/resources/pixel/bug_524043_2.in
new file mode 100644
index 0000000000..0ee43da106
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_2.in
@@ -0,0 +1,62 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 2:
+%   /Length incorrectly identifies middle of stream.
+%   The stream is blocked by the keyword "endstream"
+%   Only the text before the keyword can be rendered.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream
+0 50 Td
+/F2 13 Tf
+(It's wrong when you see the text!) Tj
+ET
+endstream
+endobj
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_2_expected.pdf.0.png b/testing/resources/pixel/bug_524043_2_expected.pdf.0.png
new file mode 100644
index 0000000000..19de52a360
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_2_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_3.in b/testing/resources/pixel/bug_524043_3.in
new file mode 100644
index 0000000000..716345ede1
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_3.in
@@ -0,0 +1,62 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 3:
+%   /Length incorrectly identifies middle of stream.
+%   "eendstream" shouldn't block the stream.
+%   Both should render as text.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+eendstream
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+endstream
+endobj
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_3_expected.pdf.0.png b/testing/resources/pixel/bug_524043_3_expected.pdf.0.png
new file mode 100644
index 0000000000..a044bb0c37
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_3_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_4.in b/testing/resources/pixel/bug_524043_4.in
new file mode 100644
index 0000000000..6cdb3d18a7
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_4.in
@@ -0,0 +1,61 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 4:
+%   /Length incorrectly identifies middle of stream.
+%   "endstream." shouldn't block the stream. 
+%   Both should render as text even "endstream" is missing.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream.
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+endobj
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_4_expected.pdf.0.png b/testing/resources/pixel/bug_524043_4_expected.pdf.0.png
new file mode 100644
index 0000000000..a044bb0c37
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_4_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_5.in b/testing/resources/pixel/bug_524043_5.in
new file mode 100644
index 0000000000..799674628d
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_5.in
@@ -0,0 +1,61 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 5:
+%   /Length incorrectly identifies middle of stream.
+%   "endstream%" shouldn't block the stream. 
+%   Both should render as text even "endobj" is missing.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream%
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+endstream
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_5_expected.pdf.0.png b/testing/resources/pixel/bug_524043_5_expected.pdf.0.png
new file mode 100644
index 0000000000..a044bb0c37
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_5_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_6.in b/testing/resources/pixel/bug_524043_6.in
new file mode 100644
index 0000000000..318b20750d
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_6.in
@@ -0,0 +1,60 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 6:
+%   /Length incorrectly identifies middle of stream.
+%   "endstream+" shouldn't block the stream. 
+%   Nothing will be rendered when both "endstream" and "endobj" are missing.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream+
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_6_expected.pdf.0.png b/testing/resources/pixel/bug_524043_6_expected.pdf.0.png
new file mode 100644
index 0000000000..3edcc2db89
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_6_expected.pdf.0.png
diff --git a/testing/resources/pixel/bug_524043_7.in b/testing/resources/pixel/bug_524043_7.in
new file mode 100644
index 0000000000..c95f2d94e1
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_7.in
@@ -0,0 +1,61 @@
+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+{{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
+      /F2 5 0 R
+    >>
+  >>
+  /Contents 6 0 R
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Helvetica
+>>
+endobj
+% Case 7:
+%   /Length incorrectly identifies middle of stream.
+%   "endstream" will block the stream. 
+%   The text before the keyword "endstream" can be rendered although
+%   both "endstream" and "endobj" are missing at the end of stream.
+{{object 6 0}} <<
+  /Length 87
+>>
+stream
+BT
+20 50 Td
+/F1 12 Tf
+(endobj is text) Tj
+endstream
+0 50 Td
+/F2 13 Tf
+(endstream is text per /Length) Tj
+ET
+{{xref}}
+trailer <<
+  /Size 6
+  /Root 1 0 R
+>>
+{{startxref}}
+%%EOF
diff --git a/testing/resources/pixel/bug_524043_7_expected.pdf.0.png b/testing/resources/pixel/bug_524043_7_expected.pdf.0.png
new file mode 100644
index 0000000000..19de52a360
--- /dev/null
+++ b/testing/resources/pixel/bug_524043_7_expected.pdf.0.png