Merge to XFA: Early object closing in PDF via premature 'endobj' keyword

Adobe acrobat defines keywords "\nendstream\n" and "\nendobj\n" in object's stream.
However, Pdfium searches "endstream" and "endobj" as the keywords in the streams of objects.
Some words like "(endstream)" are wrongly treated as keywords in pdfium.
This fix follows Adobe's solution.

BUG=493126
R=tsepez@chromium.org

Review URL: https://codereview.chromium.org/1219133005 .

1 files changed, 116 insertions, 45 deletions

diff --git a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
index 714db002de..55fba8df98 100644
--- a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
+++ b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
@@ -2439,65 +2439,134 @@ CPDF_Object* CPDF_SyntaxParser::GetObjectByStrict(
   }
   return NULL;
 }
+unsigned int CPDF_SyntaxParser::ReadEOLMarkers(FX_FILESIZE pos) {
+  unsigned char byte1 = 0;
+  unsigned char byte2 = 0;
+  GetCharAt(pos, byte1);
+  GetCharAt(pos+1, byte2);
+  unsigned int markers = 0;
+  if (byte1 == '\r' && byte2 == '\n') {
+    markers = 2;
+  } else if (byte1 == '\r' || byte1 == '\n') {
+    markers = 1;
+  }
+  return markers;
+}
 CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
                                            PARSE_CONTEXT* pContext,
                                            FX_DWORD objnum,
                                            FX_DWORD gennum) {
   CPDF_Object* pLenObj = pDict->GetElement(FX_BSTRC("Length"));
-  FX_FILESIZE len = 0;
+  FX_FILESIZE len = -1;
   if (pLenObj && ((pLenObj->GetType() != PDFOBJ_REFERENCE) ||
-                  ((((CPDF_Reference*)pLenObj)->GetObjList() != NULL) &&
+                  ((((CPDF_Reference*)pLenObj)->GetObjList()) &&
                    ((CPDF_Reference*)pLenObj)->GetRefObjNum() != objnum))) {
     len = pLenObj->GetInteger();
   }
-
-  ToNextLine();
-  FX_FILESIZE StreamStartPos = m_Pos;
+  //Check whether end of line markers follow the keyword 'stream'.
+  unsigned int numMarkers = ReadEOLMarkers(m_Pos);
+  m_Pos += numMarkers;
+  FX_FILESIZE streamStartPos = m_Pos;
   if (pContext) {
-    pContext->m_DataStart = m_Pos;
-  }
-
-  CPDF_CryptoHandler* pCryptoHandler =
-      objnum == (FX_DWORD)m_MetadataObjnum ? NULL : m_pCryptoHandler;
-  if (pCryptoHandler == NULL) {
-    pdfium::base::CheckedNumeric<FX_FILESIZE> pos = m_Pos;
-    pos += len;
-    if (pos.IsValid() && pos.ValueOrDie() < m_FileLen) {
-      m_Pos = pos.ValueOrDie();
-    }
-    GetNextWord();
-    if (m_WordSize < 9 || FXSYS_memcmp(m_WordBuffer, "endstream", 9)) {
-      m_Pos = StreamStartPos;
-      FX_FILESIZE offset = FindTag(FX_BSTRC("endstream"), 0);
-      if (offset >= 0) {
-        FX_FILESIZE curPos = m_Pos;
-        m_Pos = StreamStartPos;
-        FX_FILESIZE endobjOffset = FindTag(FX_BSTRC("endobj"), 0);
-        if (endobjOffset < offset && endobjOffset >= 0) {
-          offset = endobjOffset;
+    pContext->m_DataStart = streamStartPos;
+  }
+  const unsigned int ENDSTREAM_LEN = sizeof("endstream") - 1;
+  const unsigned int ENDOBJ_LEN = sizeof("endobj") - 1;
+  CPDF_CryptoHandler* pCryptoHandler = 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);
+      GetNextWord();
+      nextMarkers = ReadEOLMarkers(m_Pos);
+      if (m_WordSize == ENDSTREAM_LEN && prevMarkers != 0 && nextMarkers != 0 &&
+          FXSYS_memcmp(m_WordBuffer, "endstream", ENDSTREAM_LEN) == 0) {
+        bSearchForKeyword = FALSE;
+      }
+    }
+    if (bSearchForKeyword) {
+      //If len is not available, len needs to be calculated
+      //by searching the keywords "endstream" or "endobj".
+      m_Pos = streamStartPos;
+      FX_FILESIZE endStreamOffset = 0;
+      while (endStreamOffset >= 0) {
+        endStreamOffset = FindTag(FX_BSTRC("endstream"), 0);
+        if (endStreamOffset < 0) {
+          //Can't find any "endstream".
+          break;
+        }
+        prevMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 1);
+        nextMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset + ENDSTREAM_LEN);
+        if (prevMarkers != 0 && nextMarkers != 0) {
+          //Stop searching when the keyword "endstream" is found.
+          break;
         } else {
-          m_Pos = curPos;
+          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;
+          }
         }
-        uint8_t byte1, byte2;
-        GetCharAt(StreamStartPos + offset - 1, byte1);
-        GetCharAt(StreamStartPos + offset - 2, byte2);
-        if (byte1 == 0x0a && byte2 == 0x0d) {
-          len -= 2;
-        } else if (byte1 == 0x0a || byte1 == 0x0d) {
-          len--;
+        m_Pos += ENDSTREAM_LEN;
+      }
+      m_Pos = streamStartPos;
+      FX_FILESIZE endObjOffset = 0;
+      while (endObjOffset >= 0) {
+        endObjOffset = FindTag(FX_BSTRC("endobj"), 0);
+        if (endObjOffset < 0) {
+          //Can't find any "endobj".
+          break;
+        }
+        prevMarkers = ReadEOLMarkers(streamStartPos + endObjOffset - 1);
+        nextMarkers = ReadEOLMarkers(streamStartPos + endObjOffset + ENDOBJ_LEN);
+        if (prevMarkers != 0 && nextMarkers != 0) {
+          //Stop searching when the keyword "endobj" is found.
+          break;
         }
-        len = (FX_DWORD)offset;
-        pDict->SetAtInteger(FX_BSTRC("Length"), len);
+        m_Pos += ENDOBJ_LEN;
+      }
+      if (endStreamOffset < 0 && endObjOffset < 0) {
+        //Can't find "endstream" or "endobj".
+        return nullptr;
+      }
+      if (endStreamOffset < 0 && endObjOffset >= 0) {
+        //Correct the position of end stream.
+        endStreamOffset = endObjOffset;
+      } else if (endStreamOffset >= 0 && endObjOffset < 0) {
+        //Correct the position of end obj.
+        endObjOffset = endStreamOffset;
+      } else if (endStreamOffset > endObjOffset) {
+        endStreamOffset = endObjOffset;
+      }
+      len = endStreamOffset;
+      numMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 2);
+      if (numMarkers == 2) {
+        len -= 2;
       } else {
-        m_Pos = StreamStartPos;
-        if (FindTag(FX_BSTRC("endobj"), 0) < 0) {
-          return NULL;
+        numMarkers = ReadEOLMarkers(streamStartPos + endStreamOffset - 1);
+        if (numMarkers == 1) {
+          len -= 1;
         }
       }
+      if (len <= 0) {
+        return nullptr;
+      }
+      pDict->SetAtInteger(FX_BSTRC("Length"), len);
     }
-    m_Pos = StreamStartPos;
+    m_Pos = streamStartPos;
+  }
+  if (len <= 0) {
+    return nullptr;
   }
-  CPDF_Stream* pStream;
   uint8_t* pData = FX_Alloc(uint8_t, len);
   ReadBlock(pData, len);
   if (pCryptoHandler) {
@@ -2511,14 +2580,16 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
     len = dest_buf.GetSize();
     dest_buf.DetachBuffer();
   }
-  pStream = new CPDF_Stream(pData, len, pDict);
+  CPDF_Stream* pStream = new CPDF_Stream(pData, len, pDict);
   if (pContext) {
     pContext->m_DataEnd = pContext->m_DataStart + len;
   }
-  StreamStartPos = m_Pos;
+  streamStartPos = m_Pos;
   GetNextWord();
-  if (m_WordSize == 6 && 0 == FXSYS_memcmp(m_WordBuffer, "endobj", 6)) {
-    m_Pos = StreamStartPos;
+  numMarkers = ReadEOLMarkers(m_Pos);
+  if (m_WordSize == ENDOBJ_LEN && numMarkers != 0 && 
+    FXSYS_memcmp(m_WordBuffer, "endobj", ENDOBJ_LEN) == 0) {
+    m_Pos = streamStartPos;
   }
   return pStream;
 }