* © 2008 by David Roden <droden@gmail.com>
*/
+#include <stdlib.h>
+#include <string.h>
#include "CollectionReader.h"
+#include "GlobalSettings.h"
CollectionReader::CollectionReader(ReaderInput* readerInput) {
this->readerInput = readerInput;
+ firstLink = true;
}
CollectionReader::~CollectionReader() {
- delete readerInput;
+}
+
+bool CollectionReader::isLineBreakPresent() {
+ size_t indexOfLineBreak = growingBuffer.indexOf('\n');
+ GlobalSettings::isVerbose() && (indexOfLineBreak != (size_t) -1) && fprintf(stderr, "[%s:%d] found line break at %d.\n", __FILE__, __LINE__, indexOfLineBreak);
+ GlobalSettings::isVerbose() && (indexOfLineBreak == (size_t) -1) && fprintf(stderr, "[%s:%d] could not find line break.\n", __FILE__, __LINE__);
+ 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) || !tagValue) {
+ 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;
+}
+
+HashTag* CollectionReader::readHashTag() {
+ if (!ensureBufferCapacity(1)) {
+ return NULL;
+ }
+ uint8_t tagId = 0;
+ growingBuffer.read(&tagId, 1);
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag id %d.\n", __FILE__, __LINE__, tagId);
+ if (!ensureBufferCapacity(16)) {
+ return NULL;
+ }
+ char hash[16];
+ growingBuffer.read(hash, 16);
+ return new HashTag(tagId, hash);
+}
+
+ED2KLink* CollectionReader::getNextLink() {
+ if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
+ return NULL;
+ }
+ if (firstLink) {
+ identifyCollectionType();
+ if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
+ return NULL;
+ }
+ }
+ if (isTextCollection) {
+ while (!readerInput->isEOF() && !isLineBreakPresent()) {
+ readMoreBytes();
+ }
+ if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
+ return NULL;
+ }
+ size_t indexOfLineBreak = growingBuffer.indexOf('\n');
+ char* line;
+ if (indexOfLineBreak == (size_t) -1) {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not find line break, using remainder of file.\n", __FILE__, __LINE__);
+ indexOfLineBreak = growingBuffer.getRemaining();
+ line = (char*) malloc(indexOfLineBreak + 1);
+ growingBuffer.read(line, indexOfLineBreak);
+ line[indexOfLineBreak] = '\0';
+ } else {
+ line = (char*) malloc(indexOfLineBreak + 1);
+ growingBuffer.read(line, indexOfLineBreak + 1);
+ if (line[indexOfLineBreak] == '\n') {
+ line[indexOfLineBreak] = '\0';
+ }
+ if (line[indexOfLineBreak - 1] == '\r') {
+ line[indexOfLineBreak - 1] = '\0';
+ }
+ }
+ growingBuffer.cut();
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] got line: %s\n", __FILE__, __LINE__, line);
+ ED2KLink* ed2kLink = ED2KLink::parseED2KLink(line);
+ free(line);
+ firstLink = false;
+ 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 files.\n", __FILE__, __LINE__, fileCollectionCount);
+ collectionFileIndex = 0;
+ firstLink = false;
+ }
+ if (collectionFileIndex < fileCollectionCount) {
+ uint32_t fileTagCount = 0;
+ if (!ensureBufferCapacity(4)) {
+ return NULL;
+ }
+ growingBuffer.read(&fileTagCount, 4);
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d file tags.\n", __FILE__, __LINE__, fileTagCount);
+ StringTag* filenameTag = NULL;
+ uint64_t size = 0;
+ HashTag* hashTag = NULL;
+ for (uint32_t fileTagIndex = 0; fileTagIndex < fileTagCount; fileTagIndex++) {
+ uint8_t tagType = 0;
+ if (!ensureBufferCapacity(1)) {
+ return NULL;
+ }
+ growingBuffer.read(&tagType, 1);
+ tagType &= 0x7f;
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] reading tag type %d.\n", __FILE__, __LINE__, tagType);
+ if (tagType == 0x01) {
+ hashTag = readHashTag();
+ if (!hashTag) {
+ return NULL;
+ }
+ } else if (tagType == 0x02) {
+ StringTag* stringTag = readStringTag();
+ if (!stringTag) {
+ return NULL;
+ }
+ if (stringTag->getId() == 0x01) {
+ filenameTag = stringTag;
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read file name “%s”.\n", __FILE__, __LINE__, (char*) stringTag->getValue());
+ } else if (stringTag->getId() == 0xf6) {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read file comment “%s”.\n", __FILE__, __LINE__, (char*) stringTag->getValue());
+ delete stringTag;
+ }
+ } else if ((tagType == 0x03) || (tagType == 0x08) || (tagType == 0x09) || (tagType == 0x0b)) {
+ if (!ensureBufferCapacity(1)) {
+ return NULL;
+ }
+ uint8_t id = 0;
+ growingBuffer.read(&id, 1);
+ if (id == 0x02) {
+ int neededBytes = (tagType == 0x03) ? 4 : ((tagType == 0x08) ? 2 : ((tagType == 0x09) ? 1 : 8));
+ if (!ensureBufferCapacity(neededBytes)) {
+ return NULL;
+ }
+ growingBuffer.read(&size, neededBytes);
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] file size is %llu bytes.\n", __FILE__, __LINE__, size);
+ } else if (id == 0xf7) {
+ uint8_t fileRating = 0;
+ if (!ensureBufferCapacity(1)) {
+ return NULL;
+ }
+ growingBuffer.read(&fileRating, 1);
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read file rating %d.\n", __FILE__, __LINE__, fileRating);
+ }
+ }
+ }
+ collectionFileIndex++;
+ growingBuffer.cut();
+ if (filenameTag && hashTag) {
+ ED2KLink* ed2kLink = new ED2KLink((char*) filenameTag->getValue(), size, hashTag->getValue());
+ delete filenameTag;
+ delete hashTag;
+ return ed2kLink;
+ } else {
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] not enough data to decode file.\n", __FILE__, __LINE__);
+ }
+ if (filenameTag) {
+ delete filenameTag;
+ }
+ if (hashTag) {
+ delete hashTag;
+ }
+ } 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;
+
+ readBytes = readerInput->read(buffer, 1024);
+ growingBuffer.write(buffer, readBytes);
+}
+
+void CollectionReader::identifyCollectionType() {
+ size_t readBytes;
+
+ readBytes = readerInput->read(&version, 4);
+ if ((readBytes < 4) || readerInput->isEOF()) {
+ return;
+ }
+ if (version == 0x01) {
+ isTextCollection = false;
+ this->version = 1;
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 1\n", __FILE__, __LINE__);
+ } else if (version == 0x02) {
+ isTextCollection = false;
+ this->version = 2;
+ GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 2\n", __FILE__, __LINE__);
+ } 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__);
+ }
}
projects / ecparse.git / blobdiff