Daniel Veillard | e2d034d | 1999-07-27 19:52:06 +0000 | [diff] [blame] | 1 | /* |
| 2 | * xmlIO.c : implementation of the I/O interfaces used by the parser |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * Daniel.Veillard@w3.org |
| 7 | */ |
| 8 | |
Daniel Veillard | b05deb7 | 1999-08-10 19:04:08 +0000 | [diff] [blame] | 9 | /* |
| 10 | * TODO: plug-in a generic transfer library, like libwww if available |
| 11 | */ |
| 12 | |
Daniel Veillard | e2d034d | 1999-07-27 19:52:06 +0000 | [diff] [blame] | 13 | #include "config.h" |
| 14 | |
| 15 | #include <sys/types.h> |
| 16 | #include <sys/stat.h> |
| 17 | #include <fcntl.h> |
| 18 | #include <malloc.h> |
| 19 | #ifdef HAVE_UNISTD_H |
| 20 | #include <unistd.h> |
| 21 | #endif |
| 22 | #ifdef HAVE_ZLIB_H |
| 23 | #include <zlib.h> |
| 24 | #endif |
Daniel Veillard | b05deb7 | 1999-08-10 19:04:08 +0000 | [diff] [blame] | 25 | #include <string.h> |
Daniel Veillard | e2d034d | 1999-07-27 19:52:06 +0000 | [diff] [blame] | 26 | |
| 27 | #include "xmlIO.h" |
| 28 | |
| 29 | /* #define DEBUG_INPUT */ |
| 30 | /* #define VERBOSE_FAILURE */ |
| 31 | |
| 32 | #ifdef DEBUG_INPUT |
| 33 | #define MINLEN 40 |
| 34 | #else |
| 35 | #define MINLEN 4000 |
| 36 | #endif |
| 37 | |
| 38 | /** |
| 39 | * xmlAllocParserInputBuffer: |
| 40 | * @enc: the charset encoding if known |
| 41 | * |
| 42 | * Create a buffered parser input for progressive parsing |
| 43 | * |
| 44 | * Returns the new parser input or NULL |
| 45 | */ |
| 46 | xmlParserInputBufferPtr |
| 47 | xmlAllocParserInputBuffer(xmlCharEncoding enc) { |
| 48 | xmlParserInputBufferPtr ret; |
| 49 | |
| 50 | ret = (xmlParserInputBufferPtr) malloc(sizeof(xmlParserInputBuffer)); |
| 51 | if (ret == NULL) { |
| 52 | fprintf(stderr, "xmlAllocParserInputBuffer : out of memory!\n"); |
| 53 | return(NULL); |
| 54 | } |
| 55 | memset(ret, 0, (size_t) sizeof(xmlParserInputBuffer)); |
| 56 | ret->buffer = xmlBufferCreate(); |
| 57 | ret->encoder = xmlGetCharEncodingHandler(enc); |
| 58 | ret->fd = -1; |
| 59 | |
| 60 | return(ret); |
| 61 | } |
| 62 | |
| 63 | /** |
| 64 | * xmlFreeParserInputBuffer: |
| 65 | * @in: a buffered parser input |
| 66 | * |
| 67 | * Free up the memory used by a buffered parser input |
| 68 | */ |
| 69 | void |
| 70 | xmlFreeParserInputBuffer(xmlParserInputBufferPtr in) { |
| 71 | if (in->buffer != NULL) { |
| 72 | xmlBufferFree(in->buffer); |
| 73 | in->buffer = NULL; |
| 74 | } |
| 75 | #ifdef HAVE_ZLIB_H |
| 76 | if (in->gzfile != NULL) |
| 77 | gzclose(in->gzfile); |
| 78 | #endif |
| 79 | if (in->fd >= 0) |
| 80 | close(in->fd); |
| 81 | memset(in, 0xbe, (size_t) sizeof(xmlParserInputBuffer)); |
| 82 | free(in); |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * xmlParserInputBufferCreateFilename: |
| 87 | * @filename: a C string containing the filename |
| 88 | * @enc: the charset encoding if known |
| 89 | * |
| 90 | * Create a buffered parser input for the progressive parsing of a file |
| 91 | * If filename is "-' then we use stdin as the input. |
| 92 | * Automatic support for ZLIB/Compress compressed document is provided |
| 93 | * by default if found at compile-time. |
| 94 | * |
| 95 | * Returns the new parser input or NULL |
| 96 | */ |
| 97 | xmlParserInputBufferPtr |
| 98 | xmlParserInputBufferCreateFilename(const char *filename, xmlCharEncoding enc) { |
| 99 | xmlParserInputBufferPtr ret; |
| 100 | #ifdef HAVE_ZLIB_H |
| 101 | gzFile input; |
| 102 | #else |
| 103 | int input = -1; |
| 104 | #endif |
| 105 | |
| 106 | if (filename == NULL) return(NULL); |
| 107 | |
| 108 | if (!strcmp(filename, "-")) { |
| 109 | #ifdef HAVE_ZLIB_H |
| 110 | input = gzdopen (fileno(stdin), "r"); |
| 111 | if (input == NULL) { |
| 112 | #ifdef VERBOSE_FAILURE |
| 113 | fprintf (stderr, "Cannot read from stdin\n"); |
| 114 | perror ("gzdopen failed"); |
| 115 | #endif |
| 116 | return(NULL); |
| 117 | } |
| 118 | #else |
| 119 | #ifdef WIN32 |
| 120 | input = -1; |
| 121 | #else |
| 122 | input = fileno(stdin); |
| 123 | #endif |
| 124 | if (input < 0) { |
| 125 | #ifdef VERBOSE_FAILURE |
| 126 | fprintf (stderr, "Cannot read from stdin\n"); |
| 127 | perror ("open failed"); |
| 128 | #endif |
| 129 | return(NULL); |
| 130 | } |
| 131 | #endif |
| 132 | } else { |
| 133 | #ifdef HAVE_ZLIB_H |
| 134 | input = gzopen (filename, "r"); |
| 135 | if (input == NULL) { |
| 136 | #ifdef VERBOSE_FAILURE |
| 137 | fprintf (stderr, "Cannot read file %s :\n", filename); |
| 138 | perror ("gzopen failed"); |
| 139 | #endif |
| 140 | return(NULL); |
| 141 | } |
| 142 | #else |
| 143 | #ifdef WIN32 |
| 144 | input = _open (filename, O_RDONLY | _O_BINARY); |
| 145 | #else |
| 146 | input = open (filename, O_RDONLY); |
| 147 | #endif |
| 148 | if (input < 0) { |
| 149 | #ifdef VERBOSE_FAILURE |
| 150 | fprintf (stderr, "Cannot read file %s :\n", filename); |
| 151 | perror ("open failed"); |
| 152 | #endif |
| 153 | return(NULL); |
| 154 | } |
| 155 | #endif |
| 156 | } |
| 157 | /* |
| 158 | * TODO : get the 4 first bytes and |
| 159 | * if enc == XML_CHAR_ENCODING_NONE |
| 160 | * plug some encoding conversion routines here. !!! |
| 161 | * enc = xmlDetectCharEncoding(buffer); |
| 162 | */ |
| 163 | |
| 164 | ret = xmlAllocParserInputBuffer(enc); |
| 165 | if (ret != NULL) { |
| 166 | #ifdef HAVE_ZLIB_H |
| 167 | ret->gzfile = input; |
| 168 | #else |
| 169 | ret->fd = input; |
| 170 | #endif |
| 171 | } |
| 172 | xmlParserInputBufferRead(ret, 4); |
| 173 | |
| 174 | return(ret); |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * xmlParserInputBufferCreateFile: |
| 179 | * @file: a FILE* |
| 180 | * @enc: the charset encoding if known |
| 181 | * |
| 182 | * Create a buffered parser input for the progressive parsing of a FILE * |
| 183 | * buffered C I/O |
| 184 | * |
| 185 | * Returns the new parser input or NULL |
| 186 | */ |
| 187 | xmlParserInputBufferPtr |
| 188 | xmlParserInputBufferCreateFile(FILE *file, xmlCharEncoding enc) { |
| 189 | xmlParserInputBufferPtr ret; |
| 190 | |
| 191 | if (file == NULL) return(NULL); |
| 192 | |
| 193 | ret = xmlAllocParserInputBuffer(enc); |
| 194 | if (ret != NULL) |
| 195 | ret->file = file; |
| 196 | |
| 197 | return(ret); |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * xmlParserInputBufferCreateFd: |
| 202 | * @fd: a file descriptor number |
| 203 | * @enc: the charset encoding if known |
| 204 | * |
| 205 | * Create a buffered parser input for the progressive parsing for the input |
| 206 | * from a file descriptor |
| 207 | * |
| 208 | * Returns the new parser input or NULL |
| 209 | */ |
| 210 | xmlParserInputBufferPtr |
| 211 | xmlParserInputBufferCreateFd(int fd, xmlCharEncoding enc) { |
| 212 | xmlParserInputBufferPtr ret; |
| 213 | |
| 214 | if (fd < 0) return(NULL); |
| 215 | |
| 216 | ret = xmlAllocParserInputBuffer(enc); |
| 217 | if (ret != NULL) |
| 218 | ret->fd = fd; |
| 219 | |
| 220 | return(ret); |
| 221 | } |
| 222 | |
| 223 | /** |
| 224 | * xmlParserInputBufferGrow: |
| 225 | * @in: a buffered parser input |
| 226 | * @len: indicative value of the amount of chars to read |
| 227 | * |
| 228 | * Grow up the content of the input buffer, the old data are preserved |
| 229 | * This routine handle the I18N transcoding to internal UTF-8 |
| 230 | * TODO: one should be able to remove one copy |
| 231 | * |
| 232 | * Returns the number of chars read and stored in the buffer, or -1 |
| 233 | * in case of error. |
| 234 | */ |
| 235 | int |
| 236 | xmlParserInputBufferGrow(xmlParserInputBufferPtr in, int len) { |
| 237 | char *buffer = NULL; |
| 238 | #ifdef HAVE_ZLIB_H |
| 239 | gzFile input = (gzFile) in->gzfile; |
| 240 | #endif |
| 241 | int res = 0; |
| 242 | int nbchars = 0; |
| 243 | int buffree; |
| 244 | |
| 245 | if ((len <= MINLEN) && (len != 4)) |
| 246 | len = MINLEN; |
| 247 | buffree = in->buffer->size - in->buffer->use; |
| 248 | if (buffree <= 0) { |
| 249 | fprintf(stderr, "xmlParserInputBufferGrow : buffer full !\n"); |
| 250 | return(0); |
| 251 | } |
| 252 | if (len > buffree) |
| 253 | len = buffree; |
| 254 | |
| 255 | buffer = malloc((len + 1) * sizeof(char)); |
| 256 | if (buffer == NULL) { |
| 257 | fprintf(stderr, "xmlParserInputBufferGrow : out of memory !\n"); |
| 258 | return(-1); |
| 259 | } |
| 260 | if (in->file != NULL) { |
| 261 | res = fread(&buffer[0], 1, len, in->file); |
| 262 | #ifdef HAVE_ZLIB_H |
| 263 | } else if (in->gzfile != NULL) { |
| 264 | res = gzread(input, &buffer[0], len); |
| 265 | #endif |
| 266 | } else if (in->fd >= 0) { |
| 267 | res = read(in->fd, &buffer[0], len); |
| 268 | } else { |
| 269 | fprintf(stderr, "xmlParserInputBufferGrow : no input !\n"); |
| 270 | free(buffer); |
| 271 | return(-1); |
| 272 | } |
| 273 | if (res == 0) { |
| 274 | free(buffer); |
| 275 | return(0); |
| 276 | } |
| 277 | if (res < 0) { |
| 278 | perror ("read error"); |
| 279 | free(buffer); |
| 280 | return(-1); |
| 281 | } |
| 282 | if (in->encoder != NULL) { |
| 283 | CHAR *buf; |
| 284 | |
| 285 | buf = (CHAR *) malloc((res + 1) * 2 * sizeof(CHAR)); |
| 286 | if (buf == NULL) { |
| 287 | fprintf(stderr, "xmlParserInputBufferGrow : out of memory !\n"); |
| 288 | free(buffer); |
| 289 | return(-1); |
| 290 | } |
| 291 | nbchars = in->encoder->input(buf, (res + 1) * 2 * sizeof(CHAR), |
| 292 | buffer, res); |
| 293 | buf[nbchars] = 0; |
| 294 | xmlBufferAdd(in->buffer, (CHAR *) buf, nbchars); |
| 295 | free(buf); |
| 296 | } else { |
| 297 | nbchars = res; |
| 298 | buffer[nbchars] = 0; |
| 299 | xmlBufferAdd(in->buffer, (CHAR *) buffer, nbchars); |
| 300 | } |
| 301 | #ifdef DEBUG_INPUT |
| 302 | fprintf(stderr, "I/O: read %d chars, buffer %d/%d\n", |
| 303 | nbchars, in->buffer->use, in->buffer->size); |
| 304 | #endif |
| 305 | free(buffer); |
| 306 | return(nbchars); |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * xmlParserInputBufferRead: |
| 311 | * @in: a buffered parser input |
| 312 | * @len: indicative value of the amount of chars to read |
| 313 | * |
| 314 | * Refresh the content of the input buffer, the old data are considered |
| 315 | * consumed |
| 316 | * This routine handle the I18N transcoding to internal UTF-8 |
| 317 | * |
| 318 | * Returns the number of chars read and stored in the buffer, or -1 |
| 319 | * in case of error. |
| 320 | */ |
| 321 | int |
| 322 | xmlParserInputBufferRead(xmlParserInputBufferPtr in, int len) { |
| 323 | /* xmlBufferEmpty(in->buffer); */ |
| 324 | return(xmlParserInputBufferGrow(in, len)); |
| 325 | } |
| 326 | |
Daniel Veillard | b05deb7 | 1999-08-10 19:04:08 +0000 | [diff] [blame] | 327 | /* |
| 328 | * xmlParserGetDirectory: |
| 329 | * @filename: the path to a file |
| 330 | * |
| 331 | * lookup the directory for that file |
| 332 | * |
| 333 | * Returns a new allocated string containing the directory, or NULL. |
| 334 | */ |
| 335 | char * |
| 336 | xmlParserGetDirectory(const char *filename) { |
| 337 | char *ret = NULL; |
| 338 | char dir[1024]; |
| 339 | char *cur; |
| 340 | char sep = '/'; |
| 341 | |
| 342 | if (filename == NULL) return(NULL); |
| 343 | #ifdef WIN32 |
| 344 | sep = '\\'; |
| 345 | #endif |
| 346 | |
| 347 | strncpy(dir, filename, 1023); |
| 348 | dir[1023] = 0; |
| 349 | cur = &dir[strlen(dir)]; |
| 350 | while (cur > dir) { |
| 351 | if (*cur == sep) break; |
| 352 | cur --; |
| 353 | } |
| 354 | if (*cur == sep) { |
| 355 | if (cur == dir) dir[1] = 0; |
| 356 | else *cur = 0; |
| 357 | ret = strdup(dir); |
| 358 | } else { |
| 359 | if (getcwd(dir, 1024) != NULL) { |
| 360 | dir[1023] = 0; |
| 361 | ret = strdup(dir); |
| 362 | } |
| 363 | } |
| 364 | return(ret); |
| 365 | } |
| 366 | |