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 ED2KLink* CollectionReader::getNextLink() {
  87         if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
  88                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
  89                 return NULL;
  90         }
  91         if (firstLink) {
  92                 identifyCollectionType();
  93                 if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
  94                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
  95                         return NULL;
  96                 }
  97         }
  98         if (isTextCollection) {
  99                 while (!readerInput->isEOF() && !isLineBreakPresent()) {
 100                         readMoreBytes();
 101                 }
 102                 if (readerInput->isEOF() && !growingBuffer.getRemaining()) {
 103                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] readInput EOF reached.\n", __FILE__, __LINE__);
 104                         return NULL;
 105                 }
 106                 size_t indexOfLineBreak = growingBuffer.indexOf('\n');
 107                 char* line;
 108                 if (indexOfLineBreak == (size_t) -1) {
 109                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not find line break, using remainder of file.\n", __FILE__, __LINE__);
 110                         indexOfLineBreak = growingBuffer.getRemaining();
 111                         line = (char*) malloc(indexOfLineBreak + 1);
 112                         growingBuffer.read(line, indexOfLineBreak);
 113                         line[indexOfLineBreak] = '\0';
 114                 } else {
 115                         line = (char*) malloc(indexOfLineBreak + 1);
 116                         growingBuffer.read(line, indexOfLineBreak + 1);
 117                         if (line[indexOfLineBreak] == '\n') {
 118                                 line[indexOfLineBreak] = '\0';
 119                         }
 120                         if (line[indexOfLineBreak - 1] == '\r') {
 121                                 line[indexOfLineBreak - 1] = '\0';
 122                         }
 123                 }
 124                 growingBuffer.cut();
 125                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] got line: %s\n", __FILE__, __LINE__, line);
 126                 ED2KLink* ed2kLink = ED2KLink::parseED2KLink(line);
 127                 free(line);
 128                 return ed2kLink;
 129         }
 130         if (firstLink) {
 131                 /* read header */
 132                 if (!ensureBufferCapacity(4)) {
 133                         return NULL;
 134                 }
 135                 uint32_t headerTagCount = 0;
 136                 growingBuffer.read(&headerTagCount, 4);
 137                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d header tags.\n", __FILE__, __LINE__, headerTagCount);
 138                 for (uint32_t headerTagIndex = 0; headerTagIndex < headerTagCount; headerTagIndex++) {
 139                         uint8_t tagType = 0;
 140                         if (!ensureBufferCapacity(1)) {
 141                                 return NULL;
 142                         }
 143                         growingBuffer.read(&tagType, 1);
 144                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] read tag type %d.\n", __FILE__, __LINE__, tagType);
 145                         if (tagType == 0x02) {
 146                                 StringTag* stringTag = readStringTag(true);
 147                                 if (stringTag->getId() == 0x01) {
 148                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_FILENAME: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 149                                 } else if (stringTag->getId() == 0x31) {
 150                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHOR: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 151                                 } else {
 152                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown string in header: “%s”\n", __FILE__, __LINE__, (char*) stringTag->getValue());
 153                                 }
 154                         } else if (tagType == 0x07) {
 155                                 BlobTag* blobTag = readBlobTag(true);
 156                                 if (blobTag->getId() == 0x32) {
 157                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] FT_COLLECTIONAUTHORKEY: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
 158                                 } else {
 159                                         GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown block in header: %d bytes.\n", __FILE__, __LINE__, blobTag->getSize());
 160                                 }
 161                         } else {
 162                                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] unknown tag type: %02x.\n", __FILE__, __LINE__, tagType);
 163                         }
 164                 }
 165                 fileCollectionCount = 0;
 166                 if (!ensureBufferCapacity(4)) {
 167                         return NULL;
 168                 }
 169                 growingBuffer.read(&fileCollectionCount, 4);
 170                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] will read %d file tags.\n", __FILE__, __LINE__, fileCollectionCount);
 171                 collectionFileIndex = 0;
 172                 firstLink = false;
 173         }
 174         if (collectionFileIndex < fileCollectionCount) {
 175         } else {
 176                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] reached end of collection.\n", __FILE__, __LINE__);
 177         }
 178         return NULL;
 179 }
 180
 181 bool CollectionReader::ensureBufferCapacity(size_t byteCount) {
 182         while (!readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
 183                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] only %d bytes remaning, need at least %d, reading more bytes.\n", __FILE__, __LINE__, growingBuffer.getRemaining(), byteCount);
 184                 readMoreBytes();
 185         }
 186         if (readerInput->isEOF() && (growingBuffer.getRemaining() < byteCount)) {
 187                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] need %d more bytes, but file is EOF.\n", __FILE__, __LINE__, (byteCount - growingBuffer.getRemaining()));
 188                 return false;
 189         }
 190         return true;
 191 }
 192
 193 void CollectionReader::readMoreBytes() {
 194         char buffer[1024];
 195         size_t readBytes;
 196
 197         readBytes = readerInput->read(buffer, 1024);
 198         growingBuffer.write(buffer, readBytes);
 199 }
 200
 201 void CollectionReader::identifyCollectionType() {
 202         size_t readBytes;
 203
 204         readBytes = readerInput->read(&version, 4);
 205         if ((readBytes < 4) || readerInput->isEOF()) {
 206                 return;
 207         }
 208         if (version == 0x01) {
 209                 isTextCollection = false;
 210                 this->version = 1;
 211                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 1\n", __FILE__, __LINE__);
 212         } else if (version == 0x02) {
 213                 isTextCollection = false;
 214                 this->version = 2;
 215                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified binary collection, version 2\n", __FILE__, __LINE__);
 216         } else if (!strncmp("ed2k", (char*) &version, 4)) {
 217                 isTextCollection = true;
 218                 growingBuffer.write(&version, 4);
 219                 version = 0;
 220                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] identified text collection\n", __FILE__, __LINE__);
 221         } else  {
 222                 GlobalSettings::isVerbose() && fprintf(stderr, "[%s:%d] could not identify collection!\n", __FILE__, __LINE__);
 223         }
 224 }
 225