ms-word doc plugin: now can read doc files with erorrs in formatting structures

jni/NativeFormats/fbreader/src/formats/doc/OleMainStream.cpp

@@ -83,13 +83,21 @@ bool OleMainStream::open() {
 		return true;
 	}
 
-	return readPieceTable(headerBuffer, tableEntry) &&
+	result = readPieceTable(headerBuffer, tableEntry);
-			readBookmarks(headerBuffer, tableEntry) &&
-			readStylesheet(headerBuffer, tableEntry) &&
-			//readSectionsInfoTable(headerBuffer, tableEntry) && //it doesn't uses now
-			readParagraphStyleTable(headerBuffer, tableEntry) &&
-			readCharInfoTable(headerBuffer, tableEntry);
 
+	if (!result) {
+		ZLLogger::Instance().println("OleMainStream", "error during reading piece table");
+		return false;
+	}
+
+	//result of reading following structures doesn't check, because all these
+	//problems can be ignored, and document can be showed anyway, maybe with wrong formatting
+	readBookmarks(headerBuffer, tableEntry);
+	readStylesheet(headerBuffer, tableEntry);
+	//readSectionsInfoTable(headerBuffer, tableEntry); //it isn't used now
+	readParagraphStyleTable(headerBuffer, tableEntry);
+	readCharInfoTable(headerBuffer, tableEntry);
+	return true;
 }
 
 const OleMainStream::Pieces &OleMainStream::getPieces() const {
@@ -491,7 +499,7 @@ bool OleMainStream::readCharInfoTable(const char *headerBuffer, const OleEntry &
 	}
 
 	char *formatPageBuffer = new char[OleStorage::BBD_BLOCK_SIZE];
-	for (size_t index = 0; index < size; ++index) {
+	for (size_t index = 0; index < charBlocks.size(); ++index) {
 		seek(charBlocks.at(index) * OleStorage::BBD_BLOCK_SIZE, true);
 		if (read(formatPageBuffer, OleStorage::BBD_BLOCK_SIZE) != OleStorage::BBD_BLOCK_SIZE) {
 			return false;
@@ -540,15 +548,15 @@ bool OleMainStream::readParagraphStyleTable(const char *headerBuffer, const OleE
 	}
 
 	char *formatPageBuffer = new char[OleStorage::BBD_BLOCK_SIZE];
-	for (size_t index = 0; index < size; ++index) {
+	for (size_t index = 0; index < paragraphBlocks.size(); ++index) {
 		seek(paragraphBlocks.at(index) * OleStorage::BBD_BLOCK_SIZE, true);
 		if (read(formatPageBuffer, OleStorage::BBD_BLOCK_SIZE) != OleStorage::BBD_BLOCK_SIZE) {
 			return false;
 		}
-		unsigned int cpara = OleUtil::getU1Byte(formatPageBuffer, 0x1ff); //offset with cpara (count of paragraphs)
+		unsigned int paragraphsCount = OleUtil::getU1Byte(formatPageBuffer, 0x1ff); //offset with 'cpara' value (count of paragraphs)
-		for (unsigned int index2 = 0; index2 < cpara; ++index2) {
+		for (unsigned int index2 = 0; index2 < paragraphsCount; ++index2) {
 			unsigned int offset = OleUtil::getU4Bytes(formatPageBuffer, index2 * 4);
-			unsigned int papxOffset = OleUtil::getU1Byte(formatPageBuffer, (cpara + 1) * 4 + index2 * 13) * 2;
+			unsigned int papxOffset = OleUtil::getU1Byte(formatPageBuffer, (paragraphsCount + 1) * 4 + index2 * 13) * 2;
 			if (papxOffset <= 0) {
 				continue;
 			}
@@ -609,7 +617,7 @@ bool OleMainStream::readSectionsInfoTable(const char *headerBuffer, const OleEnt
 
 	//reading the section properties
 	char tmpBuffer[2];
-	for (size_t index = 0; index < decriptorsCount; ++index) {
+	for (size_t index = 0; index < sectPage.size(); ++index) {
 		if (sectPage.at(index) == 0xffffffffUL) { //check for invalid record, to make default section info
 			SectionInfo sectionInfo;
 			sectionInfo.charPos = charPos.at(index);
@@ -617,14 +625,18 @@ bool OleMainStream::readSectionsInfoTable(const char *headerBuffer, const OleEnt
 			continue;
 		}
 		//getting number of bytes to read
-		seek(sectPage.at(index), true);
+		if (!seek(sectPage.at(index), true)) {
+			continue;
+		}
 		if (read(tmpBuffer, 2) != 2) {
-			return false;
+			continue;
 		}
 		size_t bytes = 2 + (size_t)OleUtil::getU2Bytes(tmpBuffer, 0);
 
+		if (!seek(sectPage.at(index), true)) {
+			continue;
+		}
 		char *formatPageBuffer = new char[bytes];
-		seek(sectPage.at(index), true);
 		if (read(formatPageBuffer, bytes) != bytes) {
 			delete formatPageBuffer;
 			continue;

jni/NativeFormats/fbreader/src/formats/doc/OleStorage.cpp

@@ -110,6 +110,12 @@ bool OleStorage::readBBD(char *oleBuf) {
 	char buffer[mySectorSize];
 	unsigned int bbdNumberBlocks = OleUtil::getU4Bytes(oleBuf, 0x2c); //number of big blocks
 
+	if (myDIFAT.size() < bbdNumberBlocks) {
+		//TODO maybe add check on myDIFAT == bbdNumberBlocks
+		ZLLogger::Instance().println("OleStorage", "Wrong number of FAT blocks value");
+		return false;
+	}
+
 	for (unsigned int i = 0; i < bbdNumberBlocks; ++i) {
 		int bbdSector = myDIFAT.at(i);
 		if (bbdSector >= (int)(myStreamSize / mySectorSize) || bbdSector < 0) {
@@ -118,7 +124,7 @@ bool OleStorage::readBBD(char *oleBuf) {
 		}
 		myInputStream->seek(BBD_BLOCK_SIZE + bbdSector * mySectorSize, true);
 		if (myInputStream->read(buffer, mySectorSize) != mySectorSize) {
-			ZLLogger::Instance().println("OleStorage", "Can't read BBD!");
+			ZLLogger::Instance().println("OleStorage", "Error during reading BBD!");
 			return false;
 		}
 		for (unsigned int j = 0; j < mySectorSize; j += 4) {
@@ -140,13 +146,20 @@ bool OleStorage::readSBD(char *oleBuf) {
 	char buffer[mySectorSize];
 	for (int i = 0; i < sbdCount; ++i) {
 		if (i != 0) {
+			if (sbdCur < 0 || (unsigned int)sbdCur >= myBBD.size()) {
+				ZLLogger::Instance().println("OleStorage", "error during parsing SBD");
+				return false;
+			}
 			sbdCur = myBBD.at(sbdCur);
 		}
 		if (sbdCur <= 0) {
 			break;
 		}
 		myInputStream->seek(BBD_BLOCK_SIZE + sbdCur * mySectorSize, true);
-		myInputStream->read(buffer, mySectorSize);
+		if (myInputStream->read(buffer, mySectorSize) != mySectorSize) {
+			ZLLogger::Instance().println("OleStorage", "reading error during parsing SBD");
+			return false;
+		}
 		for (unsigned int j = 0; j < mySectorSize; j += 4) {
 			mySBD.push_back(OleUtil::get4Bytes(buffer, j));
 		}
@@ -165,7 +178,10 @@ bool OleStorage::readProperties(char *oleBuf) {
 	char buffer[mySectorSize];
 	do {
 		myInputStream->seek(BBD_BLOCK_SIZE + propCur * mySectorSize, true);
-		myInputStream->read(buffer, mySectorSize);
+		if (myInputStream->read(buffer, mySectorSize) != mySectorSize) {
+			ZLLogger::Instance().println("OleStorage", "Error during reading properties");
+			return false;
+		}
 		for (unsigned int j = 0; j < mySectorSize; j += 128) {
 			myProperties.push_back(std::string(buffer + j, 128));
 		}
@@ -212,6 +228,10 @@ bool OleStorage::readOleEntry(int propNumber, OleEntry &e) {
 	int nameLength = OleUtil::getU2Bytes(property.c_str(), 0x40); //offset for value entry's name length
 	e.name.clear();
 	e.name.reserve(33); //max size of entry name
+
+	if ((unsigned int)nameLength >= property.size()) {
+		return false;
+	}
 	for (int i = 0; i < nameLength; i+=2) {
 		char c = property.at(i);
 		if (c != 0) {

jni/NativeFormats/fbreader/src/formats/doc/OleStreamReader.cpp

@@ -198,14 +198,19 @@ bool OleStreamReader::fillBuffer(OleMainStream &stream) {
 		return false;
 	}
 
+	if (!stream.seek(piece.offset, true)) {
+		//TODO maybe in that case we should take next piece?
+		return false;
+	}
 	char *textBuffer = new char[piece.length];
-
+	size_t readedBytes = stream.read(textBuffer, piece.length);
-	stream.seek(piece.offset, true);
+	if (readedBytes != (unsigned int)piece.length) {
-	stream.read(textBuffer, piece.length);
+		ZLLogger::Instance().println("OleStreamReader", "not all bytes has been readed from piece");
+	}
 
 	myBuffer.clear();
 	if (!piece.isANSI) {
-		for (int i = 0; i < piece.length; i += 2) {
+		for (unsigned int i = 0; i < readedBytes; i += 2) {
 			ZLUnicodeUtil::Ucs2Char ch = OleUtil::getU2Bytes(textBuffer, i);
 			myBuffer.push_back(ch);
 		}
@@ -219,7 +224,7 @@ bool OleStreamReader::fillBuffer(OleMainStream &stream) {
 			}
 		}
 		std::string utf8String;
-		myConverter->convert(utf8String, std::string(textBuffer, piece.length));
+		myConverter->convert(utf8String, std::string(textBuffer, readedBytes));
 		ZLUnicodeUtil::utf8ToUcs2(myBuffer, utf8String);
 	}
 	myCurBufferPosition = 0;