CollectionReader.cpp

   1 /**
   2  * © 2008 by David Roden <droden@gmail.com>
   3  */
   4
   5 #include <stdlib.h>
   6 #include <string.h>
   7 #include "CollectionReader.h"
   8 #include "GlobalSettings.h"
   9
  10 CollectionReader::CollectionReader(ReaderInput* readerInput) {
  11         this->readerInput = readerInput;
  12         firstLink = true;
  13 }
  14
  15 CollectionReader::~CollectionReader() {
  16 }
  17
  18 bool CollectionReader::isLineBreakPresent() {
  19         size_t indexOfLineBreak = growingBuffer.indexOf('\n');
  20         GlobalSettings::isVerbose() && (indexOfLineBreak != (size_t) -1) && fprintf(stderr, "[%s:%d] found line break at %d.\n", __FILE__, __LINE__, indexOfLineBreak);
  21         GlobalSettings::isVerbose() && (indexOfLineBreak == (size_t) -1) && fprintf(stderr, "[%s:%d] could not find line break.\n", __FILE__, __LINE__);
  22         return indexOfLineBreak != (size_t) -1;
  23 }
  24
  25 StringTag* CollectionReader::readStringTag(bool header) {
  26         if (header) {
  27                 uint16_t unknown = 0;
  28                 if (!ensureBufferCapacity(2)) {
  29                         return NULL;
  30                 }
  31                 growingBuffer.read(&unknown, 2);
  32         }
  33         uint8_t tagName = 0;
  34         if (!ensureBufferCapacity(1)) {
  35                 return NULL;
  36         }
  37         growingBuffer.read(&tagName, 1);
  38         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag name %d.\n", __FILE__, __LINE__, tagName);
  39         uint16_t tagLength = 0;
  40         if (!ensureBufferCapacity(2)) {
  41                 return NULL;
  42         }
  43         growingBuffer.read(&tagLength, 2);
  44         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag length %d.\n", __FILE__, __LINE__, tagLength);
  45         char* tagValue = (char*) malloc(tagLength + 1);
  46         if (!ensureBufferCapacity(tagLength)) {
  47                 return NULL;
  48         }
  49         growingBuffer.read(tagValue, tagLength);
  50         tagValue[tagLength] = '\0';
  51         StringTag* stringTag = new StringTag(tagName, tagValue);
  52         free(tagValue);
  53         return stringTag;
  54 }
  55
  56 BlobTag* CollectionReader::readBlobTag(bool header) {
  57         if (header) {
  58                 uint16_t unknown = 0;
  59                 if (!ensureBufferCapacity(2)) {
  60                         return NULL;
  61                 }
  62                 growingBuffer.read(&unknown, 2);
  63         }
  64         uint8_t tagName = 0;
  65         if (!ensureBufferCapacity(1)) {
  66                 return NULL;
  67         }
  68         growingBuffer.read(&tagName, 1);
  69         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag name %d.\n", __FILE__, __LINE__, tagName);
  70         uint32_t tagLength = 0;
  71         if (!ensureBufferCapacity(4)) {
  72                 return NULL;
  73         }
  74         growingBuffer.read(&tagLength, 4);
  75         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag length %d.\n", __FILE__, __LINE__, tagLength);
  76         void* tagValue = (char*) malloc(tagLength);
  77         if (!ensureBufferCapacity(tagLength)) {
  78                 return NULL;
  79         }
  80         growingBuffer.read(tagValue, tagLength);
  81         BlobTag* blobTag = new BlobTag(tagName, tagValue, tagLength);
  82         free(tagValue);
  83         return blobTag;
  84 }
  85
  86 HashTag* CollectionReader::readHashTag() {
  87         if (!ensureBufferCapacity(1)) {
  88                 return NULL;
  89         }
  90         uint8_t tagId = 0;
  91         growingBuffer.read(&tagId, 1);
  92         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag id %d.\n", __FILE__, __LINE__, tagId);
  93         if (!ensureBufferCapacity(16)) {
  94                 return NULL;
  95         }
  96         char hash[16];
  97         growingBuffer.read(hash, 16);
  98         return new HashTag(tagId, hash);
  99 }
 100
 101 ED2KLink* CollectionReader::getNextLink() {
 102         if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
 103                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
 104                 return NULL;
 105         }
 106         if (firstLink) {
 107                 identifyCollectionType();
 108                 if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
 109                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
 110                         return NULL;
 111                 }
 112         }
 113         if (isTextCollection) {
 114                 while (!readerInput->isEOF() && !isLineBreakPresent()) {
 115                         readMoreBytes();
 116                 }
 117                 if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
 118                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
 119                         return NULL;
 120                 }
 121                 size_t indexOfLineBreak = growingBuffer.indexOf('\n');
 122                 char* line;
 123                 if (indexOfLineBreak == (size_t) -1) {
 124                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not find line break, using remainder of file.\n", __FILE__, __LINE__);
 125                         indexOfLineBreak = growingBuffer.getRemaining();
 126                         line = (char*) malloc(indexOfLineBreak + 1);
 127                         growingBuffer.read(line, indexOfLineBreak);
 128                         line[indexOfLineBreak] = '\0';
 129                 } else {
 130                         line = (char*) malloc(indexOfLineBreak + 1);
 131                         growingBuffer.read(line, indexOfLineBreak + 1);
 132                         if (line[indexOfLineBreak] == '\n') {
 133                                 line[indexOfLineBreak] = '\0';
 134                         }
 135                         if (line[indexOfLineBreak - 1] == '\r') {
 136                                 line[indexOfLineBreak - 1] = '\0';
 137                         }
 138                 }
 139                 growingBuffer.cut();
 140                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] got line: %s\n", __FILE__, __LINE__, line);
 141                 ED2KLink* ed2kLink = ED2KLink::parseED2KLink(line);
 142                 free(line);
 143                 return ed2kLink;
 144         }
 145         if (firstLink) {
 146                 /* read header */
 147                 if (!ensureBufferCapacity(4)) {
 148                         return NULL;
 149                 }
 150                 uint32_t headerTagCount = 0;
 151                 growingBuffer.read(&headerTagCount, 4);
 152                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d header tags.\n", __FILE__, __LINE__, headerTagCount);
 153                 for (uint32_t headerTagIndex = 0; headerTagIndex < headerTagCount; headerTagIndex++) {
 154                         uint8_t tagType = 0;
 155                         if (!ensureBufferCapacity(1)) {
 156                                 return NULL;
 157                         }
 158                         growingBuffer.read(&tagType, 1);
 159                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag type %d.\n", __FILE__, __LINE__, tagType);
 160                         if (tagType == 0x02) {
 161                                 StringTag* stringTag = readStringTag(true);
 162                                 if (stringTag->getId() == 0x01) {
 163                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_FILENAME: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 164                                 } else if (stringTag->getId() == 0x31) {
 165                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHOR: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 166                                 } else {
 167                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown string in header: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 168                                 }
 169                         } else if (tagType == 0x07) {
 170                                 BlobTag* blobTag = readBlobTag(true);
 171                                 if (blobTag->getId() == 0x32) {
 172                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHORKEY: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
 173                                 } else {
 174                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown block in header: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
 175                                 }
 176                         } else {
 177                                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown tag type: %02x.\n", __FILE__, __LINE__, tagType);
 178                         }
 179                 }
 180                 fileCollectionCount = 0;
 181                 if (!ensureBufferCapacity(4)) {
 182                         return NULL;
 183                 }
 184                 growingBuffer.read(&fileCollectionCount, 4);
 185                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d files.\n", __FILE__, __LINE__, fileCollectionCount);
 186                 collectionFileIndex = 0;
 187                 firstLink = false;
 188         }
 189         if (collectionFileIndex < fileCollectionCount) {
 190         } else {
 191                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] reached end of collection.\n", __FILE__, __LINE__);
 192         }
 193         return NULL;
 194 }
 195
 196 bool CollectionReader::ensureBufferCapacity(size_t byteCount) {
 197         while (!readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
 198                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] only %d bytes remaning, need at least %d, reading more bytes.\n", __FILE__, __LINE__, growingBuffer.getRemaining(), byteCount);
 199                 readMoreBytes();
 200         }
 201         if (readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
 202                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] need %d more bytes, but file is EOF.\n", __FILE__, __LINE__, (byteCount - growingBuffer.getRemaining()));
 203                 return false;
 204         }
 205         return true;
 206 }
 207
 208 void CollectionReader::readMoreBytes() {
 209         char buffer[1024];
 210         size_t readBytes;
 211
 212         readBytes = readerInput->read(buffer, 1024);
 213         growingBuffer.write(buffer, readBytes);
 214 }
 215
 216 void CollectionReader::identifyCollectionType() {
 217         size_t readBytes;
 218
 219         readBytes = readerInput->read(&version, 4);
 220         if ((readBytes < 4) || readerInput->isEOF()) {
 221                 return;
 222         }
 223         if (version == 0x01) {
 224                 isTextCollection = false;
 225                 this->version = 1;
 226                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 1\n", __FILE__, __LINE__);
 227         } else if (version == 0x02) {
 228                 isTextCollection = false;
 229                 this->version = 2;
 230                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 2\n", __FILE__, __LINE__);
 231         } else if (!strncmp("ed2k", (char*) &version, 4)) {
 232                 isTextCollection = true;
 233                 growingBuffer.write(&version, 4);
 234                 version = 0;
 235                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified text collection\n", __FILE__, __LINE__);
 236         } else  {
 237                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not identify collection!\n", __FILE__, __LINE__);
 238         }
 239 }
 240