parse header only on first link

[ecparse.git] / CollectionReader.cpp

diff --git a/CollectionReader.cpp b/CollectionReader.cpp
index 4aa5173..e33d6b8 100644 (file)
--- a/CollectionReader.cpp
+++ b/CollectionReader.cpp
@@ -22,6 +22,67 @@ bool CollectionReader::isLineBreakPresent() {
        return indexOfLineBreak != (size_t) -1;
 }
 
+StringTag* CollectionReader::readStringTag(bool header) {
+       if (header) {
+               uint16_t unknown = 0;
+               if (!ensureBufferCapacity(2)) {
+                       return NULL;
+               }
+               growingBuffer.read(&unknown, 2);
+       }
+       uint8_t tagName = 0;
+       if (!ensureBufferCapacity(1)) {
+               return NULL;
+       }
+       growingBuffer.read(&tagName, 1);
+       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag name %d.\n", __FILE__, __LINE__, tagName);
+       uint16_t tagLength = 0;
+       if (!ensureBufferCapacity(2)) {
+               return NULL;
+       }
+       growingBuffer.read(&tagLength, 2);
+       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag length %d.\n", __FILE__, __LINE__, tagLength);
+       char* tagValue = (char*) malloc(tagLength + 1);
+       if (!ensureBufferCapacity(tagLength)) {
+               return NULL;
+       }
+       growingBuffer.read(tagValue, tagLength);
+       tagValue[tagLength] = '\0';
+       StringTag* stringTag = new StringTag(tagName, tagValue);
+       free(tagValue);
+       return stringTag;
+}
+
+BlobTag* CollectionReader::readBlobTag(bool header) {
+       if (header) {
+               uint16_t unknown = 0;
+               if (!ensureBufferCapacity(2)) {
+                       return NULL;
+               }
+               growingBuffer.read(&unknown, 2);
+       }
+       uint8_t tagName = 0;
+       if (!ensureBufferCapacity(1)) {
+               return NULL;
+       }
+       growingBuffer.read(&tagName, 1);
+       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag name %d.\n", __FILE__, __LINE__, tagName);
+       uint32_t tagLength = 0;
+       if (!ensureBufferCapacity(4)) {
+               return NULL;
+       }
+       growingBuffer.read(&tagLength, 4);
+       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag length %d.\n", __FILE__, __LINE__, tagLength);
+       void* tagValue = (char*) malloc(tagLength);
+       if (!ensureBufferCapacity(tagLength)) {
+               return NULL;
+       }
+       growingBuffer.read(tagValue, tagLength);
+       BlobTag* blobTag = new BlobTag(tagName, tagValue, tagLength);
+       free(tagValue);
+       return blobTag;
+}
+
 ED2KLink* CollectionReader::getNextLink() {
        if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
                GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
@@ -33,7 +94,6 @@ ED2KLink* CollectionReader::getNextLink() {
                        GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
                        return NULL;
                }
-               firstLink = false;
        }
        if (isTextCollection) {
                while (!readerInput->isEOF() && !isLineBreakPresent()) {
@@ -66,11 +126,70 @@ ED2KLink* CollectionReader::getNextLink() {
                ED2KLink* ed2kLink = ED2KLink::parseED2KLink(line);
                free(line);
                return ed2kLink;
+       }
+       if (firstLink) {
+               /* read header */
+               if (!ensureBufferCapacity(4)) {
+                       return NULL;
+               }
+               uint32_t headerTagCount = 0;
+               growingBuffer.read(&headerTagCount, 4);
+               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d header tags.\n", __FILE__, __LINE__, headerTagCount);
+               for (uint32_t headerTagIndex = 0; headerTagIndex < headerTagCount; headerTagIndex++) {
+                       uint8_t tagType = 0;
+                       if (!ensureBufferCapacity(1)) {
+                               return NULL;
+                       }
+                       growingBuffer.read(&tagType, 1);
+                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag type %d.\n", __FILE__, __LINE__, tagType);
+                       if (tagType == 0x02) {
+                               StringTag* stringTag = readStringTag(true);
+                               if (stringTag->getId() == 0x01) {
+                                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_FILENAME: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
+                               } else if (stringTag->getId() == 0x31) {
+                                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHOR: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
+                               } else {
+                                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown string in header: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
+                               }
+                       } else if (tagType == 0x07) {
+                               BlobTag* blobTag = readBlobTag(true);
+                               if (blobTag->getId() == 0x32) {
+                                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHORKEY: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
+                               } else {
+                                       GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown block in header: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
+                               }
+                       } else {
+                               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown tag type: %02x.\n", __FILE__, __LINE__, tagType);
+                       }
+               }
+               fileCollectionCount = 0;
+               if (!ensureBufferCapacity(4)) {
+                       return NULL;
+               }
+               growingBuffer.read(&fileCollectionCount, 4);
+               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d file tags.\n", __FILE__, __LINE__, fileCollectionCount);
+               collectionFileIndex = 0;
+               firstLink = false;
+       }
+       if (collectionFileIndex < fileCollectionCount) {
        } else {
+               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] reached end of collection.\n", __FILE__, __LINE__);
        }
        return NULL;
 }
 
+bool CollectionReader::ensureBufferCapacity(size_t byteCount) {
+       while (!readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
+               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] only %d bytes remaning, need at least %d, reading more bytes.\n", __FILE__, __LINE__, growingBuffer.getRemaining(), byteCount);
+               readMoreBytes();
+       }
+       if (readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
+               GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] need %d more bytes, but file is EOF.\n", __FILE__, __LINE__, (byteCount - growingBuffer.getRemaining()));
+               return false;
+       }
+       return true;
+}
+
 void CollectionReader::readMoreBytes() {
        char buffer[1024];
        size_t readBytes;
@@ -80,7 +199,6 @@ void CollectionReader::readMoreBytes() {
 }
 
 void CollectionReader::identifyCollectionType() {
-       int version;
        size_t readBytes;
 
        readBytes = readerInput->read(&version, 4);
@@ -98,6 +216,7 @@ void CollectionReader::identifyCollectionType() {
        } else if (!strncmp("ed2k", (char*) &version, 4)) {
                isTextCollection = true;
                growingBuffer.write(&version, 4);
+               version = 0;
                GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified text collection\n", __FILE__, __LINE__);
        } else  {
                GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not identify collection!\n", __FILE__, __LINE__);