From 18151f1aa126b8c930506907685e7fed7b4ea9f0 Mon Sep 17 00:00:00 2001
From: msv <msv@opencascade.com>
Date: Fri, 29 Apr 2016 19:31:13 +0300
Subject: [PATCH] 0027454: Application hangs while opening a non-OCAF XML file

1) Add protection against accessing null pointer in PCDM_ReadWriter::FileFormat.

2) In LDOM_XmlReader::ReadRecord, take into account that the character '>' can have no special meaning (e.g., in a text), and we must read the data behind this character to complete the current tag reading. This treatment concerns the mode of work when myTagPerStep is true.

3) Create a test case of reading of XML file not related to OCAF document.

Test case was added.
---
 src/LDOM/LDOM_XmlReader.cxx  | 15 ++++++++++++++-
 src/PCDM/PCDM_ReadWriter.cxx |  8 ++++++--
 tests/bugs/caf/bug27454      | 19 +++++++++++++++++++
 tests/bugs/caf/bug27454_1    | 16 ++++++++++++++++
 4 files changed, 55 insertions(+), 3 deletions(-)
 create mode 100644 tests/bugs/caf/bug27454
 create mode 100644 tests/bugs/caf/bug27454_1

diff --git a/src/LDOM/LDOM_XmlReader.cxx b/src/LDOM/LDOM_XmlReader.cxx
index 18e17cec7c..0655a9a511 100644
--- a/src/LDOM/LDOM_XmlReader.cxx
+++ b/src/LDOM/LDOM_XmlReader.cxx
@@ -107,6 +107,8 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
       }
       else if (myTagPerStep && aHasRead)
       {
+        // in myTagPerStep mode, we should parse the buffer to the end before
+        // getting more characters from the stream.
       }
       else
       {
@@ -225,6 +227,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         aState = STATE_TEXT;
         aStartData = myPtr;
         myPtr = myEndPtr;
+        aHasRead = Standard_False;
       }   // end of checking in STATE_WAITING
       continue;
 
@@ -244,6 +247,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         return XML_HEADER;
       }
       myPtr = myEndPtr - 1;
+      aHasRead = Standard_False;
       continue;
 
       // Checking the characters in STATE_DOCTYPE, seek for "]>" sequence
@@ -264,6 +268,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         }
       }
       myPtr = myEndPtr - 1;
+      aHasRead = Standard_False;
       continue;
 
     state_doctype_markup:
@@ -282,6 +287,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         return XML_DOCTYPE;
       }
       myPtr = myEndPtr - 1;
+      aHasRead = Standard_False;
       continue;
 
         // Checking the characters in STATE_COMMENT, seek for "-->" sequence
@@ -303,6 +309,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         }
       }
       myPtr = myEndPtr - 2;
+      aHasRead = Standard_False;
       continue;
 
         // Checking the characters in STATE_TEXT, seek for "<"
@@ -316,6 +323,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         return XML_TEXT;
       }
       myPtr = myEndPtr;
+      aHasRead = Standard_False;
       continue;
 
         // Checking the characters in STATE_CDATA, seek for "]]"
@@ -334,6 +342,7 @@ LDOM_XmlReader::RecordType LDOM_XmlReader::ReadRecord (Standard_IStream& theIStr
         return XML_CDATA;
       }
       myPtr = myEndPtr - 1;
+      aHasRead = Standard_False;
       continue;
 
         // Checking the characters in STATE_ELEMENT, seek the end of TagName
@@ -489,8 +498,11 @@ attr_name:
           myPtr = aPtr + 1;
           aStartData = NULL;
           aState = STATE_ATTRIBUTE_NAME;
-        } else
+        }
+        else {
           myPtr = myEndPtr;
+          aHasRead = Standard_False;
+        }
         continue;
       }
         // Checking the characters in STATE_ELEMENT_END, seek for ">"
@@ -504,6 +516,7 @@ attr_name:
         return XML_END_ELEMENT;
       }
       myPtr = myEndPtr;
+      aHasRead = Standard_False;
       continue;
     }
   }
diff --git a/src/PCDM/PCDM_ReadWriter.cxx b/src/PCDM/PCDM_ReadWriter.cxx
index 1af94f1247..8ce4da3273 100644
--- a/src/PCDM/PCDM_ReadWriter.cxx
+++ b/src/PCDM/PCDM_ReadWriter.cxx
@@ -166,12 +166,16 @@ TCollection_ExtendedString PCDM_ReadWriter::FileFormat (Standard_IStream& theISt
 {
   TCollection_ExtendedString aFormat;
 
-  Storage_BaseDriver* aFileDriver;
+  Storage_BaseDriver* aFileDriver = 0L;
   if (PCDM::FileDriverType (theIStream, aFileDriver) == PCDM_TOFD_XmlFile)
   {
     return ::TryXmlDriverType (theIStream);
   }
- 
+  if (!aFileDriver)
+  {
+    // type is not recognized, return empty string
+    return aFormat;
+  }
 
   aFileDriver->ReadCompleteInfo (theIStream, theData);
 
diff --git a/tests/bugs/caf/bug27454 b/tests/bugs/caf/bug27454
new file mode 100644
index 0000000000..e146f3b6bb
--- /dev/null
+++ b/tests/bugs/caf/bug27454
@@ -0,0 +1,19 @@
+puts "REQUIRED All: DDocStd_Open : Error"
+
+puts "========"
+puts "OCC27454"
+puts "========"
+puts ""
+###################################################
+## Application hangs while opening a non-OCAF XML file
+###################################################
+
+cpulimit 10
+
+catch {Open [locate_data_file bug27454_test.xml] D -stream} msg
+
+cpulimit
+
+if ![regexp {DDocStd_Open : Error} $msg] {
+  puts "Error: reader did not report an error reading the file of unknown format"
+}
\ No newline at end of file
diff --git a/tests/bugs/caf/bug27454_1 b/tests/bugs/caf/bug27454_1
new file mode 100644
index 0000000000..f146be3130
--- /dev/null
+++ b/tests/bugs/caf/bug27454_1
@@ -0,0 +1,16 @@
+puts "========"
+puts "OCC27454"
+puts "========"
+puts ""
+###################################################
+## Application hangs while opening a non-OCAF XML file
+###################################################
+
+if [info exists D] {unset D}
+Open [locate_data_file shape_tracking.ocaf.xml] D -stream
+
+if {[dtyp D] != "TDocStd_Document"} {
+  puts "Error: XML OCAF document read failure"
+}
+
+Close D
-- 
2.20.1