| 1 | /*
 | 
| 2 |  * Souffle - A Datalog Compiler
 | 
| 3 |  * Copyright (c) 2021, The Souffle Developers. All rights reserved
 | 
| 4 |  * Licensed under the Universal Permissive License v 1.0 as shown at:
 | 
| 5 |  * - https://opensource.org/licenses/UPL
 | 
| 6 |  * - <souffle root>/licenses/SOUFFLE-UPL.txt
 | 
| 7 |  */
 | 
| 8 | 
 | 
| 9 | /************************************************************************
 | 
| 10 |  *
 | 
| 11 |  * @file gzfstream.h
 | 
| 12 |  * A simple zlib wrapper to provide gzip file streams.
 | 
| 13 |  *
 | 
| 14 |  ***********************************************************************/
 | 
| 15 | 
 | 
| 16 | #pragma once
 | 
| 17 | 
 | 
| 18 | #include <cstdio>
 | 
| 19 | #include <cstring>
 | 
| 20 | #include <iostream>
 | 
| 21 | #include <string>
 | 
| 22 | #include <zlib.h>
 | 
| 23 | 
 | 
| 24 | namespace souffle {
 | 
| 25 | 
 | 
| 26 | namespace gzfstream {
 | 
| 27 | 
 | 
| 28 | namespace internal {
 | 
| 29 | 
 | 
| 30 | class gzfstreambuf : public std::streambuf {
 | 
| 31 | public:
 | 
| 32 |     gzfstreambuf() {
 | 
| 33 |         setp(buffer, buffer + (bufferSize - 1));
 | 
| 34 |         setg(buffer + reserveSize, buffer + reserveSize, buffer + reserveSize);
 | 
| 35 |     }
 | 
| 36 | 
 | 
| 37 |     gzfstreambuf(const gzfstreambuf&) = delete;
 | 
| 38 | 
 | 
| 39 |     gzfstreambuf(gzfstreambuf&& old) = default;
 | 
| 40 | 
 | 
| 41 |     gzfstreambuf* open(const std::string& filename, std::ios_base::openmode mode) {
 | 
| 42 |         if (is_open()) {
 | 
| 43 |             return nullptr;
 | 
| 44 |         }
 | 
| 45 |         if ((mode ^ std::ios::in ^ std::ios::out) == 0) {
 | 
| 46 |             return nullptr;
 | 
| 47 |         }
 | 
| 48 | 
 | 
| 49 |         this->mode = mode;
 | 
| 50 |         std::string gzmode((mode & std::ios::in) != 0 ? "rb" : "wb");
 | 
| 51 |         fileHandle = gzopen(filename.c_str(), gzmode.c_str());
 | 
| 52 | 
 | 
| 53 |         if (fileHandle == nullptr) {
 | 
| 54 |             return nullptr;
 | 
| 55 |         }
 | 
| 56 |         isOpen = true;
 | 
| 57 | 
 | 
| 58 |         return this;
 | 
| 59 |     }
 | 
| 60 | 
 | 
| 61 |     gzfstreambuf* close() {
 | 
| 62 |         if (is_open()) {
 | 
| 63 |             sync();
 | 
| 64 |             isOpen = false;
 | 
| 65 |             if (gzclose(fileHandle) == Z_OK) {
 | 
| 66 |                 return this;
 | 
| 67 |             }
 | 
| 68 |         }
 | 
| 69 |         return nullptr;
 | 
| 70 |     }
 | 
| 71 | 
 | 
| 72 |     bool is_open() const {
 | 
| 73 |         return isOpen;
 | 
| 74 |     }
 | 
| 75 | 
 | 
| 76 |     ~gzfstreambuf() override {
 | 
| 77 |         try {
 | 
| 78 |             close();
 | 
| 79 |         } catch (...) {
 | 
| 80 |             // Don't throw exceptions.
 | 
| 81 |         }
 | 
| 82 |     }
 | 
| 83 | 
 | 
| 84 | protected:
 | 
| 85 |     int_type overflow(int c = EOF) override {
 | 
| 86 |         if (((mode & std::ios::out) == 0) || !isOpen) {
 | 
| 87 |             return EOF;
 | 
| 88 |         }
 | 
| 89 | 
 | 
| 90 |         if (c != EOF) {
 | 
| 91 |             *pptr() = c;
 | 
| 92 |             pbump(1);
 | 
| 93 |         }
 | 
| 94 |         const int toWrite = static_cast<int>(pptr() - pbase());
 | 
| 95 |         if (gzwrite(fileHandle, pbase(), static_cast<unsigned int>(toWrite)) != toWrite) {
 | 
| 96 |             return EOF;
 | 
| 97 |         }
 | 
| 98 |         pbump(-toWrite);
 | 
| 99 | 
 | 
| 100 |         return c;
 | 
| 101 |     }
 | 
| 102 | 
 | 
| 103 |     int_type underflow() override {
 | 
| 104 |         if (((mode & std::ios::in) == 0) || !isOpen) {
 | 
| 105 |             return EOF;
 | 
| 106 |         }
 | 
| 107 |         if ((gptr() != nullptr) && (gptr() < egptr())) {
 | 
| 108 |             return traits_type::to_int_type(*gptr());
 | 
| 109 |         }
 | 
| 110 | 
 | 
| 111 |         std::size_t charsPutBack = gptr() - eback();
 | 
| 112 |         if (charsPutBack > reserveSize) {
 | 
| 113 |             charsPutBack = reserveSize;
 | 
| 114 |         }
 | 
| 115 |         memcpy(buffer + reserveSize - charsPutBack, gptr() - charsPutBack, charsPutBack);
 | 
| 116 | 
 | 
| 117 |         int charsRead =
 | 
| 118 |                 gzread(fileHandle, buffer + reserveSize, static_cast<unsigned int>(bufferSize - reserveSize));
 | 
| 119 |         if (charsRead <= 0) {
 | 
| 120 |             return EOF;
 | 
| 121 |         }
 | 
| 122 | 
 | 
| 123 |         setg(buffer + reserveSize - charsPutBack, buffer + reserveSize, buffer + reserveSize + charsRead);
 | 
| 124 | 
 | 
| 125 |         return traits_type::to_int_type(*gptr());
 | 
| 126 |     }
 | 
| 127 | 
 | 
| 128 |     int sync() override {
 | 
| 129 |         if ((pptr() != nullptr) && pptr() > pbase()) {
 | 
| 130 |             const int toWrite = static_cast<int>(pptr() - pbase());
 | 
| 131 |             if (gzwrite(fileHandle, pbase(), static_cast<unsigned int>(toWrite)) != toWrite) {
 | 
| 132 |                 return -1;
 | 
| 133 |             }
 | 
| 134 |             pbump(-toWrite);
 | 
| 135 |         }
 | 
| 136 |         return 0;
 | 
| 137 |     }
 | 
| 138 | 
 | 
| 139 | private:
 | 
| 140 |     static constexpr std::size_t bufferSize = 65536;
 | 
| 141 |     static constexpr std::size_t reserveSize = 16;
 | 
| 142 | 
 | 
| 143 |     char buffer[bufferSize] = {};
 | 
| 144 |     gzFile fileHandle = {};
 | 
| 145 |     bool isOpen = false;
 | 
| 146 |     std::ios_base::openmode mode = std::ios_base::in;
 | 
| 147 | };
 | 
| 148 | 
 | 
| 149 | class gzfstream : virtual public std::ios {
 | 
| 150 | public:
 | 
| 151 |     gzfstream() {
 | 
| 152 |         init(&buf);
 | 
| 153 |     }
 | 
| 154 | 
 | 
| 155 |     gzfstream(const std::string& filename, std::ios_base::openmode mode) {
 | 
| 156 |         init(&buf);
 | 
| 157 |         open(filename, mode);
 | 
| 158 |     }
 | 
| 159 | 
 | 
| 160 |     gzfstream(const gzfstream&) = delete;
 | 
| 161 | 
 | 
| 162 |     gzfstream(gzfstream&&) = delete;
 | 
| 163 | 
 | 
| 164 |     ~gzfstream() override = default;
 | 
| 165 | 
 | 
| 166 |     void open(const std::string& filename, std::ios_base::openmode mode) {
 | 
| 167 |         if (buf.open(filename, mode) == nullptr) {
 | 
| 168 |             clear(rdstate() | std::ios::badbit);
 | 
| 169 |         }
 | 
| 170 |     }
 | 
| 171 | 
 | 
| 172 |     bool is_open() {
 | 
| 173 |         return buf.is_open();
 | 
| 174 |     }
 | 
| 175 | 
 | 
| 176 |     void close() {
 | 
| 177 |         if (buf.is_open()) {
 | 
| 178 |             if (buf.close() == nullptr) {
 | 
| 179 |                 clear(rdstate() | std::ios::badbit);
 | 
| 180 |             }
 | 
| 181 |         }
 | 
| 182 |     }
 | 
| 183 | 
 | 
| 184 |     gzfstreambuf* rdbuf() const {
 | 
| 185 |         return &buf;
 | 
| 186 |     }
 | 
| 187 | 
 | 
| 188 | protected:
 | 
| 189 |     mutable gzfstreambuf buf;
 | 
| 190 | };
 | 
| 191 | 
 | 
| 192 | }  // namespace internal
 | 
| 193 | 
 | 
| 194 | class igzfstream : public internal::gzfstream, public std::istream {
 | 
| 195 | public:
 | 
| 196 |     igzfstream() : internal::gzfstream(), std::istream(&buf) {}
 | 
| 197 | 
 | 
| 198 |     explicit igzfstream(const std::string& filename, std::ios_base::openmode mode = std::ios::in)
 | 
| 199 |             : internal::gzfstream(filename, mode), std::istream(&buf) {}
 | 
| 200 | 
 | 
| 201 |     igzfstream(const igzfstream&) = delete;
 | 
| 202 | 
 | 
| 203 |     igzfstream(igzfstream&&) = delete;
 | 
| 204 | 
 | 
| 205 |     internal::gzfstreambuf* rdbuf() const {
 | 
| 206 |         return internal::gzfstream::rdbuf();
 | 
| 207 |     }
 | 
| 208 | 
 | 
| 209 |     void open(const std::string& filename, std::ios_base::openmode mode = std::ios::in) {
 | 
| 210 |         internal::gzfstream::open(filename, mode);
 | 
| 211 |     }
 | 
| 212 | };
 | 
| 213 | 
 | 
| 214 | class ogzfstream : public internal::gzfstream, public std::ostream {
 | 
| 215 | public:
 | 
| 216 |     ogzfstream() : std::ostream(&buf) {}
 | 
| 217 | 
 | 
| 218 |     explicit ogzfstream(const std::string& filename, std::ios_base::openmode mode = std::ios::out)
 | 
| 219 |             : internal::gzfstream(filename, mode), std::ostream(&buf) {}
 | 
| 220 | 
 | 
| 221 |     ogzfstream(const ogzfstream&) = delete;
 | 
| 222 | 
 | 
| 223 |     ogzfstream(ogzfstream&&) = delete;
 | 
| 224 | 
 | 
| 225 |     internal::gzfstreambuf* rdbuf() const {
 | 
| 226 |         return internal::gzfstream::rdbuf();
 | 
| 227 |     }
 | 
| 228 | 
 | 
| 229 |     void open(const std::string& filename, std::ios_base::openmode mode = std::ios::out) {
 | 
| 230 |         internal::gzfstream::open(filename, mode);
 | 
| 231 |     }
 | 
| 232 | };
 | 
| 233 | 
 | 
| 234 | } /* namespace gzfstream */
 | 
| 235 | 
 | 
| 236 | } /* namespace souffle */
 |