openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 1 | // Copyright 2007 Google Inc. |
| 2 | // Author: Lincoln Smith |
| 3 | // |
| 4 | // Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | // you may not use this file except in compliance with the License. |
| 6 | // You may obtain a copy of the License at |
| 7 | // |
| 8 | // http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | // |
| 10 | // Unless required by applicable law or agreed to in writing, software |
| 11 | // distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | // See the License for the specific language governing permissions and |
| 14 | // limitations under the License. |
| 15 | // |
| 16 | // Classes to implement an Encoder for the format described in |
| 17 | // RFC 3284 - The VCDIFF Generic Differencing and Compression Data Format. |
| 18 | // The RFC text can be found at http://www.faqs.org/rfcs/rfc3284.html |
| 19 | // |
| 20 | // The RFC describes the possibility of using a secondary compressor |
| 21 | // to further reduce the size of each section of the VCDIFF output. |
| 22 | // That feature is not supported in this implementation of the encoder |
| 23 | // and decoder. |
| 24 | // No secondary compressor types have been publicly registered with |
| 25 | // the IANA at http://www.iana.org/assignments/vcdiff-comp-ids |
| 26 | // in the more than five years since the registry was created, so there |
| 27 | // is no standard set of compressor IDs which would be generated by other |
| 28 | // encoders or accepted by other decoders. |
| 29 | |
| 30 | #include <config.h> |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 31 | #include <memory> // auto_ptr |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 32 | #include "checksum.h" |
| 33 | #include "encodetable.h" |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 34 | #include "google/output_string.h" |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 35 | #include "google/vcencoder.h" |
| 36 | #include "jsonwriter.h" |
| 37 | #include "logging.h" |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 38 | #include "vcdiffengine.h" |
| 39 | |
| 40 | namespace open_vcdiff { |
| 41 | |
| 42 | HashedDictionary::HashedDictionary(const char* dictionary_contents, |
| 43 | size_t dictionary_size) |
| 44 | : engine_(new VCDiffEngine(dictionary_contents, dictionary_size)) { } |
| 45 | |
| 46 | HashedDictionary::~HashedDictionary() { delete engine_; } |
| 47 | |
| 48 | bool HashedDictionary::Init() { |
| 49 | return const_cast<VCDiffEngine*>(engine_)->Init(); |
| 50 | } |
| 51 | |
| 52 | class VCDiffStreamingEncoderImpl { |
| 53 | public: |
| 54 | VCDiffStreamingEncoderImpl(const HashedDictionary* dictionary, |
| 55 | VCDiffFormatExtensionFlags format_extensions, |
| 56 | bool look_for_target_matches); |
| 57 | |
| 58 | // These functions are identical to their counterparts |
| 59 | // in VCDiffStreamingEncoder. |
| 60 | bool StartEncoding(OutputStringInterface* out); |
| 61 | |
| 62 | bool EncodeChunk(const char* data, size_t len, OutputStringInterface* out); |
| 63 | |
| 64 | bool FinishEncoding(OutputStringInterface* out); |
| 65 | |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 66 | private: |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 67 | const VCDiffEngine* engine_; |
| 68 | |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 69 | std::auto_ptr<CodeTableWriterInterface> coder_; |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 70 | |
| 71 | const VCDiffFormatExtensionFlags format_extensions_; |
| 72 | |
| 73 | // Determines whether to look for matches within the previously encoded |
| 74 | // target data, or just within the source (dictionary) data. Please see |
| 75 | // vcencoder.h for a full explanation of this parameter. |
| 76 | const bool look_for_target_matches_; |
| 77 | |
| 78 | // This state variable is used to ensure that StartEncoding(), EncodeChunk(), |
| 79 | // and FinishEncoding() are called in the correct order. It will be true |
| 80 | // if StartEncoding() has been called, followed by zero or more calls to |
| 81 | // EncodeChunk(), but FinishEncoding() has not yet been called. It will |
| 82 | // be false initially, and also after FinishEncoding() has been called. |
| 83 | bool encode_chunk_allowed_; |
| 84 | |
| 85 | // Making these private avoids implicit copy constructor & assignment operator |
| 86 | VCDiffStreamingEncoderImpl(const VCDiffStreamingEncoderImpl&); // NOLINT |
| 87 | void operator=(const VCDiffStreamingEncoderImpl&); |
| 88 | }; |
| 89 | |
| 90 | inline VCDiffStreamingEncoderImpl::VCDiffStreamingEncoderImpl( |
| 91 | const HashedDictionary* dictionary, |
| 92 | VCDiffFormatExtensionFlags format_extensions, |
| 93 | bool look_for_target_matches) |
| 94 | : engine_(dictionary->engine()), |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 95 | format_extensions_(format_extensions), |
| 96 | look_for_target_matches_(look_for_target_matches), |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 97 | encode_chunk_allowed_(false) { |
| 98 | if (format_extensions & VCD_FORMAT_JSON) { |
| 99 | coder_.reset(new JSONCodeTableWriter()); |
| 100 | } else { |
| 101 | // This implementation of the encoder uses the default |
| 102 | // code table. A VCDiffCodeTableWriter could also be constructed |
| 103 | // using a custom code table. |
| 104 | coder_.reset(new VCDiffCodeTableWriter( |
| 105 | (format_extensions & VCD_FORMAT_INTERLEAVED) != 0)); |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 106 | } |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 107 | } |
| 108 | |
| 109 | inline bool VCDiffStreamingEncoderImpl::StartEncoding( |
| 110 | OutputStringInterface* out) { |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 111 | if (!coder_->Init(engine_->dictionary_size())) { |
| 112 | VCD_DFATAL << "Internal error: " |
| 113 | "Initialization of code table writer failed" << VCD_ENDL; |
openvcdiff | d184578 | 2009-03-20 21:56:15 +0000 | [diff] [blame] | 114 | return false; |
| 115 | } |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 116 | coder_->WriteHeader(out, format_extensions_); |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 117 | encode_chunk_allowed_ = true; |
| 118 | return true; |
| 119 | } |
| 120 | |
| 121 | inline bool VCDiffStreamingEncoderImpl::EncodeChunk( |
| 122 | const char* data, |
| 123 | size_t len, |
| 124 | OutputStringInterface* out) { |
| 125 | if (!encode_chunk_allowed_) { |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 126 | VCD_ERROR << "EncodeChunk called before StartEncoding" << VCD_ENDL; |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 127 | return false; |
| 128 | } |
| 129 | if ((format_extensions_ & VCD_FORMAT_CHECKSUM) != 0) { |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 130 | coder_->AddChecksum(ComputeAdler32(data, len)); |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 131 | } |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 132 | engine_->Encode(data, len, look_for_target_matches_, out, coder_.get()); |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 133 | return true; |
| 134 | } |
| 135 | |
| 136 | inline bool VCDiffStreamingEncoderImpl::FinishEncoding( |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 137 | OutputStringInterface* out) { |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 138 | if (!encode_chunk_allowed_) { |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 139 | VCD_ERROR << "FinishEncoding called before StartEncoding" << VCD_ENDL; |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 140 | return false; |
| 141 | } |
| 142 | encode_chunk_allowed_ = false; |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 143 | coder_->FinishEncoding(out); |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 144 | return true; |
| 145 | } |
| 146 | |
| 147 | VCDiffStreamingEncoder::VCDiffStreamingEncoder( |
| 148 | const HashedDictionary* dictionary, |
| 149 | VCDiffFormatExtensionFlags format_extensions, |
| 150 | bool look_for_target_matches) |
| 151 | : impl_(new VCDiffStreamingEncoderImpl(dictionary, |
| 152 | format_extensions, |
| 153 | look_for_target_matches)) { } |
| 154 | |
| 155 | VCDiffStreamingEncoder::~VCDiffStreamingEncoder() { delete impl_; } |
| 156 | |
| 157 | bool VCDiffStreamingEncoder::StartEncodingToInterface( |
| 158 | OutputStringInterface* out) { |
| 159 | return impl_->StartEncoding(out); |
| 160 | } |
| 161 | |
| 162 | bool VCDiffStreamingEncoder::EncodeChunkToInterface( |
| 163 | const char* data, |
| 164 | size_t len, |
| 165 | OutputStringInterface* out) { |
| 166 | return impl_->EncodeChunk(data, len, out); |
| 167 | } |
| 168 | |
| 169 | bool VCDiffStreamingEncoder::FinishEncodingToInterface( |
| 170 | OutputStringInterface* out) { |
| 171 | return impl_->FinishEncoding(out); |
| 172 | } |
| 173 | |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 174 | bool VCDiffEncoder::EncodeToInterface(const char* target_data, |
| 175 | size_t target_len, |
| 176 | OutputStringInterface* out) { |
| 177 | out->clear(); |
| 178 | if (!encoder_) { |
| 179 | if (!dictionary_.Init()) { |
openvcdiff@gmail.com | 732fff2 | 2010-08-04 18:00:00 +0000 | [diff] [blame] | 180 | VCD_ERROR << "Error initializing HashedDictionary" << VCD_ENDL; |
openvcdiff | 311c714 | 2008-08-26 19:29:25 +0000 | [diff] [blame] | 181 | return false; |
| 182 | } |
| 183 | encoder_ = new VCDiffStreamingEncoder(&dictionary_, |
| 184 | flags_, |
| 185 | look_for_target_matches_); |
| 186 | } |
| 187 | if (!encoder_->StartEncodingToInterface(out)) { |
| 188 | return false; |
| 189 | } |
| 190 | if (!encoder_->EncodeChunkToInterface(target_data, target_len, out)) { |
| 191 | return false; |
| 192 | } |
| 193 | return encoder_->FinishEncodingToInterface(out); |
| 194 | } |
| 195 | |
| 196 | } // namespace open_vcdiff |