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

Pdfium swallows 'fi' or 'ff' in some tested files because it doesn't load the embedded font file correctly. The root cause is that there is incorrect keyword like 'ngendstream' in the stream of the embedded font file. Pdfium tries to find another correct keyword but uses wrong offset rather than accumulated offset.

BUG=524043
R=thestig@chromium.org, tsepez@chromium.org

Review URL: https://codereview.chromium.org/1307353005 .

4 files changed, 25 insertions, 69 deletions

diff --git a/core/src/fpdfapi/fpdf_page/fpdf_page_parser.cpp b/core/src/fpdfapi/fpdf_page/fpdf_page_parser.cpp
index 059dd4c2a6..4e5ef1c898 100644
--- a/core/src/fpdfapi/fpdf_page/fpdf_page_parser.cpp
+++ b/core/src/fpdfapi/fpdf_page/fpdf_page_parser.cpp
@@ -11,7 +11,6 @@
 
 #define REQUIRE_PARAMS(count)  \
   if (m_ParamCount != count) { \
-    m_bAbort = TRUE;           \
     return;                    \
   }
 
@@ -34,7 +33,6 @@ CPDF_StreamContentParser::CPDF_StreamContentParser(
       m_Level(level),
       m_ParamStartPos(0),
       m_ParamCount(0),
-      m_bAbort(FALSE),
       m_pCurStates(new CPDF_AllStates),
       m_pLastTextObject(nullptr),
       m_DefFontSize(0),
diff --git a/core/src/fpdfapi/fpdf_page/fpdf_page_parser_old.cpp b/core/src/fpdfapi/fpdf_page/fpdf_page_parser_old.cpp
index 48e9b98d3a..c9bcff6db6 100644
--- a/core/src/fpdfapi/fpdf_page/fpdf_page_parser_old.cpp
+++ b/core/src/fpdfapi/fpdf_page/fpdf_page_parser_old.cpp
@@ -9,27 +9,7 @@
 #include "../../../include/fxcodec/fx_codec.h"
 #include "pageint.h"
 #include <limits.h>
-const FX_CHAR* const _PDF_OpCharType =
-    "IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII"
-    "IIVIIIIVIIVIIIIIVVIIIIIIIIIIIIII"
-    "IIVVVVVVIVVVVVVIVVVVVIIVVIIIIIII"
-    "IIVVVVVVVVVVVVVVIVVVIIVVIVVIIIII"
-    "IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII"
-    "IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII"
-    "IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII"
-    "IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII";
-FX_BOOL _PDF_HasInvalidOpChar(const FX_CHAR* op) {
-  if (!op) {
-    return FALSE;
-  }
-  uint8_t ch;
-  while ((ch = *op++)) {
-    if (_PDF_OpCharType[ch] == 'I') {
-      return TRUE;
-    }
-  }
-  return FALSE;
-}
+
 class CPDF_StreamParserAutoClearer {
  public:
   CPDF_StreamParserAutoClearer(CPDF_StreamParser** scoped_variable,
@@ -61,13 +41,7 @@ FX_DWORD CPDF_StreamContentParser::Parse(const uint8_t* pData,
       case CPDF_StreamParser::EndOfData:
         return m_pSyntax->GetPos();
       case CPDF_StreamParser::Keyword:
-        if (!OnOperator((char*)syntax.GetWordBuf()) &&
-            _PDF_HasInvalidOpChar((char*)syntax.GetWordBuf())) {
-          m_bAbort = TRUE;
-        }
-        if (m_bAbort) {
-          return m_pSyntax->GetPos();
-        }
+        OnOperator((char*)syntax.GetWordBuf());
         ClearAllParams();
         break;
       case CPDF_StreamParser::Number:
@@ -1126,10 +1100,6 @@ void CPDF_ContentParser::Continue(IFX_Pause* pPause) {
         m_CurrentOffset +=
             m_pParser->Parse(m_pData + m_CurrentOffset,
                              m_Size - m_CurrentOffset, PARSE_STEP_LIMIT);
-        if (m_pParser->ShouldAbort()) {
-          m_InternalStage = PAGEPARSE_STAGE_CHECKCLIP;
-          continue;
-        }
       }
     }
     if (m_InternalStage == PAGEPARSE_STAGE_CHECKCLIP) {
diff --git a/core/src/fpdfapi/fpdf_page/pageint.h b/core/src/fpdfapi/fpdf_page/pageint.h
index 6bec07268c..c85523b833 100644
--- a/core/src/fpdfapi/fpdf_page/pageint.h
+++ b/core/src/fpdfapi/fpdf_page/pageint.h
@@ -162,7 +162,6 @@ class CPDF_StreamContentParser {
                            int level);
   ~CPDF_StreamContentParser();
 
-  FX_BOOL ShouldAbort() const { return m_bAbort; }
   CPDF_PageObjects* GetObjectList() const { return m_pObjectList; }
   CPDF_AllStates* GetCurStates() const { return m_pCurStates.get(); }
   FX_BOOL IsColored() const { return m_bColored; }
@@ -305,7 +304,6 @@ class CPDF_StreamContentParser {
   _ContentParam m_ParamBuf1[PARAM_BUF_SIZE];
   FX_DWORD m_ParamStartPos;
   FX_DWORD m_ParamCount;
-  FX_BOOL m_bAbort;
   CPDF_StreamParser* m_pSyntax;
   nonstd::unique_ptr<CPDF_AllStates> m_pCurStates;
   CPDF_ContentMark m_CurContentMark;
diff --git a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
index 7482f0b8e4..cd923f3e22 100644
--- a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
+++ b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
@@ -2455,8 +2455,8 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
     len = pLenObj->GetInteger();
   }
   // Check whether end of line markers follow the keyword 'stream'.
-  unsigned int numMarkers = ReadEOLMarkers(m_Pos);
-  m_Pos += numMarkers;
+  // The stream starts after end of line markers.
+  m_Pos += ReadEOLMarkers(m_Pos);
   FX_FILESIZE streamStartPos = m_Pos;
   if (pContext) {
     pContext->m_DataStart = streamStartPos;
@@ -2467,19 +2467,18 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
       objnum == (FX_DWORD)m_MetadataObjnum ? nullptr : m_pCryptoHandler;
   if (!pCryptoHandler) {
     FX_BOOL bSearchForKeyword = TRUE;
-    unsigned int prevMarkers = 0;
-    unsigned int nextMarkers = 0;
     if (len >= 0) {
       pdfium::base::CheckedNumeric<FX_FILESIZE> pos = m_Pos;
       pos += len;
       if (pos.IsValid() && pos.ValueOrDie() < m_FileLen) {
         m_Pos = pos.ValueOrDie();
       }
-      prevMarkers = ReadEOLMarkers(m_Pos);
+      m_Pos += ReadEOLMarkers(m_Pos);
+      FXSYS_memset(m_WordBuffer, 0, ENDSTREAM_LEN + 1);
       GetNextWord();
-      nextMarkers = ReadEOLMarkers(m_Pos);
-      if (m_WordSize == ENDSTREAM_LEN && prevMarkers != 0 && nextMarkers != 0 &&
-          FXSYS_memcmp(m_WordBuffer, "endstream", ENDSTREAM_LEN) == 0) {
+      if (FXSYS_memcmp(m_WordBuffer, "endstream", ENDSTREAM_LEN) == 0 &&
+          IsWholeWord(m_Pos - ENDSTREAM_LEN, m_FileLen,
+                      FX_BSTRC("endstream").GetPtr(), ENDSTREAM_LEN, TRUE)) {
         bSearchForKeyword = FALSE;
       }
     }
@@ -2494,22 +2493,12 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
           // Can't find any "endstream".
           break;
         }
-        prevMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 1);
-        nextMarkers =
-            ReadEOLMarkers(streamStartPos + endStreamOffset + ENDSTREAM_LEN);
-        if (prevMarkers != 0 && nextMarkers != 0) {
+        if (IsWholeWord(m_Pos - ENDSTREAM_LEN, m_FileLen,
+                        FX_BSTRC("endstream").GetPtr(), ENDSTREAM_LEN, TRUE)) {
           // Stop searching when the keyword "endstream" is found.
+          endStreamOffset = m_Pos - streamStartPos - ENDSTREAM_LEN;
           break;
-        } else {
-          unsigned char ch = 0x00;
-          GetCharAt(streamStartPos + endStreamOffset + ENDSTREAM_LEN, ch);
-          if (ch == 0x09 || ch == 0x20) {
-            //"endstream" is treated as a keyword
-            // when it is followed by a tab or whitespace
-            break;
-          }
         }
-        m_Pos += ENDSTREAM_LEN;
       }
       m_Pos = streamStartPos;
       FX_FILESIZE endObjOffset = 0;
@@ -2519,14 +2508,12 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
           // Can't find any "endobj".
           break;
         }
-        prevMarkers = ReadEOLMarkers(streamStartPos + endObjOffset - 1);
-        nextMarkers =
-            ReadEOLMarkers(streamStartPos + endObjOffset + ENDOBJ_LEN);
-        if (prevMarkers != 0 && nextMarkers != 0) {
+        if (IsWholeWord(m_Pos - ENDOBJ_LEN, m_FileLen,
+                        FX_BSTRC("endobj").GetPtr(), ENDOBJ_LEN, TRUE)) {
           // Stop searching when the keyword "endobj" is found.
+          endObjOffset = m_Pos - streamStartPos - ENDOBJ_LEN;
           break;
         }
-        m_Pos += ENDOBJ_LEN;
       }
       if (endStreamOffset < 0 && endObjOffset < 0) {
         // Can't find "endstream" or "endobj".
@@ -2542,7 +2529,7 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
         endStreamOffset = endObjOffset;
       }
       len = endStreamOffset;
-      numMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 2);
+      int numMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 2);
       if (numMarkers == 2) {
         len -= 2;
       } else {
@@ -2579,8 +2566,9 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
     pContext->m_DataEnd = pContext->m_DataStart + len;
   }
   streamStartPos = m_Pos;
+  FXSYS_memset(m_WordBuffer, 0, ENDOBJ_LEN + 1);
   GetNextWord();
-  numMarkers = ReadEOLMarkers(m_Pos);
+  int numMarkers = ReadEOLMarkers(m_Pos);
   if (m_WordSize == ENDOBJ_LEN && numMarkers != 0 &&
       FXSYS_memcmp(m_WordBuffer, "endobj", ENDOBJ_LEN) == 0) {
     m_Pos = streamStartPos;
@@ -2611,7 +2599,8 @@ int32_t CPDF_SyntaxParser::GetDirectNum() {
 FX_BOOL CPDF_SyntaxParser::IsWholeWord(FX_FILESIZE startpos,
                                        FX_FILESIZE limit,
                                        const uint8_t* tag,
-                                       FX_DWORD taglen) {
+                                       FX_DWORD taglen,
+                                       FX_BOOL checkKeyword) {
   uint8_t type = PDF_CharType[tag[0]];
   FX_BOOL bCheckLeft = type != 'D' && type != 'W';
   type = PDF_CharType[tag[taglen - 1]];
@@ -2620,13 +2609,13 @@ FX_BOOL CPDF_SyntaxParser::IsWholeWord(FX_FILESIZE startpos,
   if (bCheckRight && startpos + (int32_t)taglen <= limit &&
       GetCharAt(startpos + (int32_t)taglen, ch)) {
     uint8_t type = PDF_CharType[ch];
-    if (type == 'N' || type == 'R') {
+    if (type == 'N' || type == 'R' || (checkKeyword && type == 'D')) {
       return FALSE;
     }
   }
   if (bCheckLeft && startpos > 0 && GetCharAt(startpos - 1, ch)) {
     uint8_t type = PDF_CharType[ch];
-    if (type == 'N' || type == 'R') {
+    if (type == 'N' || type == 'R' || (checkKeyword && type == 'D')) {
       return FALSE;
     }
   }
@@ -2682,7 +2671,8 @@ FX_BOOL CPDF_SyntaxParser::SearchWord(const CFX_ByteStringC& tag,
         }
       }
       FX_FILESIZE startpos = bForward ? pos - taglen + 1 : pos;
-      if (!bWholeWord || IsWholeWord(startpos, limit, tag.GetPtr(), taglen)) {
+      if (!bWholeWord ||
+          IsWholeWord(startpos, limit, tag.GetPtr(), taglen, FALSE)) {
         m_Pos = startpos;
         return TRUE;
       }
@@ -2739,7 +2729,7 @@ int32_t CPDF_SyntaxParser::SearchMultiWord(const CFX_ByteStringC& tags,
         if (pPatterns[i].m_Offset == pPatterns[i].m_Len) {
           if (!bWholeWord ||
               IsWholeWord(pos - pPatterns[i].m_Len, limit, pPatterns[i].m_pTag,
-                          pPatterns[i].m_Len)) {
+                          pPatterns[i].m_Len, FALSE)) {
             found = i;
             goto end;
           } else {