Return to gzread.c CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / sudo / zlib

1.8.7

    1: /* gzread.c -- zlib functions for reading gzip files
    2:  * Copyright (C) 2004, 2005, 2010, 2011 Mark Adler
    3:  * For conditions of distribution and use, see copyright notice in zlib.h
    4:  */
    5: 
    6: #include "gzguts.h"
    7: 
    8: /* Local functions */
    9: local int gz_load OF((gz_statep, unsigned char *, unsigned, unsigned *));
   10: local int gz_avail OF((gz_statep));
   11: local int gz_look OF((gz_statep));
   12: local int gz_decomp OF((gz_statep));
   13: local int gz_fetch OF((gz_statep));
   14: local int gz_skip OF((gz_statep, z_off64_t));
   15: 
   16: /* Use read() to load a buffer -- return -1 on error, otherwise 0.  Read from
   17:    state->fd, and update state->eof, state->err, and state->msg as appropriate.
   18:    This function needs to loop on read(), since read() is not guaranteed to
   19:    read the number of bytes requested, depending on the type of descriptor. */
   20: local int gz_load(state, buf, len, have)
   21:     gz_statep state;
   22:     unsigned char *buf;
   23:     unsigned len;
   24:     unsigned *have;
   25: {
   26:     int ret;
   27: 
   28:     *have = 0;
   29:     do {
   30:         ret = read(state->fd, buf + *have, len - *have);
   31:         if (ret <= 0)
   32:             break;
   33:         *have += ret;
   34:     } while (*have < len);
   35:     if (ret < 0) {
   36:         gz_error(state, Z_ERRNO, zstrerror());
   37:         return -1;
   38:     }
   39:     if (ret == 0)
   40:         state->eof = 1;
   41:     return 0;
   42: }
   43: 
   44: /* Load up input buffer and set eof flag if last data loaded -- return -1 on
   45:    error, 0 otherwise.  Note that the eof flag is set when the end of the input
   46:    file is reached, even though there may be unused data in the buffer.  Once
   47:    that data has been used, no more attempts will be made to read the file.
   48:    If strm->avail_in != 0, then the current data is moved to the beginning of
   49:    the input buffer, and then the remainder of the buffer is loaded with the
   50:    available data from the input file. */
   51: local int gz_avail(state)
   52:     gz_statep state;
   53: {
   54:     unsigned got;
   55:     z_streamp strm = &(state->strm);
   56: 
   57:     if (state->err != Z_OK && state->err != Z_BUF_ERROR)
   58:         return -1;
   59:     if (state->eof == 0) {
   60:         if (strm->avail_in)
   61:             memmove(state->in, strm->next_in, strm->avail_in);
   62:         if (gz_load(state, state->in + strm->avail_in,
   63:                     state->size - strm->avail_in, &got) == -1)
   64:             return -1;
   65:         strm->avail_in += got;
   66:         strm->next_in = state->in;
   67:     }
   68:     return 0;
   69: }
   70: 
   71: /* Look for gzip header, set up for inflate or copy.  state->x.have must be 0.
   72:    If this is the first time in, allocate required memory.  state->how will be
   73:    left unchanged if there is no more input data available, will be set to COPY
   74:    if there is no gzip header and direct copying will be performed, or it will
   75:    be set to GZIP for decompression.  If direct copying, then leftover input
   76:    data from the input buffer will be copied to the output buffer.  In that
   77:    case, all further file reads will be directly to either the output buffer or
   78:    a user buffer.  If decompressing, the inflate state will be initialized.
   79:    gz_look() will return 0 on success or -1 on failure. */
   80: local int gz_look(state)
   81:     gz_statep state;
   82: {
   83:     z_streamp strm = &(state->strm);
   84: 
   85:     /* allocate read buffers and inflate memory */
   86:     if (state->size == 0) {
   87:         /* allocate buffers */
   88:         state->in = malloc(state->want);
   89:         state->out = malloc(state->want << 1);
   90:         if (state->in == NULL || state->out == NULL) {
   91:             if (state->out != NULL)
   92:                 free(state->out);
   93:             if (state->in != NULL)
   94:                 free(state->in);
   95:             gz_error(state, Z_MEM_ERROR, "out of memory");
   96:             return -1;
   97:         }
   98:         state->size = state->want;
   99: 
  100:         /* allocate inflate memory */
  101:         state->strm.zalloc = Z_NULL;
  102:         state->strm.zfree = Z_NULL;
  103:         state->strm.opaque = Z_NULL;
  104:         state->strm.avail_in = 0;
  105:         state->strm.next_in = Z_NULL;
  106:         if (inflateInit2(&(state->strm), 15 + 16) != Z_OK) {    /* gunzip */
  107:             free(state->out);
  108:             free(state->in);
  109:             state->size = 0;
  110:             gz_error(state, Z_MEM_ERROR, "out of memory");
  111:             return -1;
  112:         }
  113:     }
  114: 
  115:     /* get at least the magic bytes in the input buffer */
  116:     if (strm->avail_in < 2) {
  117:         if (gz_avail(state) == -1)
  118:             return -1;
  119:         if (strm->avail_in == 0)
  120:             return 0;
  121:     }
  122: 
  123:     /* look for gzip magic bytes -- if there, do gzip decoding (note: there is
  124:        a logical dilemma here when considering the case of a partially written
  125:        gzip file, to wit, if a single 31 byte is written, then we cannot tell
  126:        whether this is a single-byte file, or just a partially written gzip
  127:        file -- for here we assume that if a gzip file is being written, then
  128:        the header will be written in a single operation, so that reading a
  129:        single byte is sufficient indication that it is not a gzip file) */
  130:     if (strm->avail_in > 1 &&
  131:             strm->next_in[0] == 31 && strm->next_in[1] == 139) {
  132:         inflateReset(strm);
  133:         state->how = GZIP;
  134:         state->direct = 0;
  135:         return 0;
  136:     }
  137: 
  138:     /* no gzip header -- if we were decoding gzip before, then this is trailing
  139:        garbage.  Ignore the trailing garbage and finish. */
  140:     if (state->direct == 0) {
  141:         strm->avail_in = 0;
  142:         state->eof = 1;
  143:         state->x.have = 0;
  144:         return 0;
  145:     }
  146: 
  147:     /* doing raw i/o, copy any leftover input to output -- this assumes that
  148:        the output buffer is larger than the input buffer, which also assures
  149:        space for gzungetc() */
  150:     state->x.next = state->out;
  151:     if (strm->avail_in) {
  152:         memcpy(state->x.next, strm->next_in, strm->avail_in);
  153:         state->x.have = strm->avail_in;
  154:         strm->avail_in = 0;
  155:     }
  156:     state->how = COPY;
  157:     state->direct = 1;
  158:     return 0;
  159: }
  160: 
  161: /* Decompress from input to the provided next_out and avail_out in the state.
  162:    On return, state->x.have and state->x.next point to the just decompressed
  163:    data.  If the gzip stream completes, state->how is reset to LOOK to look for
  164:    the next gzip stream or raw data, once state->x.have is depleted.  Returns 0
  165:    on success, -1 on failure. */
  166: local int gz_decomp(state)
  167:     gz_statep state;
  168: {
  169:     int ret = Z_OK;
  170:     unsigned had;
  171:     z_streamp strm = &(state->strm);
  172: 
  173:     /* fill output buffer up to end of deflate stream */
  174:     had = strm->avail_out;
  175:     do {
  176:         /* get more input for inflate() */
  177:         if (strm->avail_in == 0 && gz_avail(state) == -1)
  178:             return -1;
  179:         if (strm->avail_in == 0) {
  180:             gz_error(state, Z_BUF_ERROR, "unexpected end of file");
  181:             break;
  182:         }
  183: 
  184:         /* decompress and handle errors */
  185:         ret = inflate(strm, Z_NO_FLUSH);
  186:         if (ret == Z_STREAM_ERROR || ret == Z_NEED_DICT) {
  187:             gz_error(state, Z_STREAM_ERROR,
  188:                      "internal error: inflate stream corrupt");
  189:             return -1;
  190:         }
  191:         if (ret == Z_MEM_ERROR) {
  192:             gz_error(state, Z_MEM_ERROR, "out of memory");
  193:             return -1;
  194:         }
  195:         if (ret == Z_DATA_ERROR) {              /* deflate stream invalid */
  196:             gz_error(state, Z_DATA_ERROR,
  197:                      strm->msg == NULL ? "compressed data error" : strm->msg);
  198:             return -1;
  199:         }
  200:     } while (strm->avail_out && ret != Z_STREAM_END);
  201: 
  202:     /* update available output */
  203:     state->x.have = had - strm->avail_out;
  204:     state->x.next = strm->next_out - state->x.have;
  205: 
  206:     /* if the gzip stream completed successfully, look for another */
  207:     if (ret == Z_STREAM_END)
  208:         state->how = LOOK;
  209: 
  210:     /* good decompression */
  211:     return 0;
  212: }
  213: 
  214: /* Fetch data and put it in the output buffer.  Assumes state->x.have is 0.
  215:    Data is either copied from the input file or decompressed from the input
  216:    file depending on state->how.  If state->how is LOOK, then a gzip header is
  217:    looked for to determine whether to copy or decompress.  Returns -1 on error,
  218:    otherwise 0.  gz_fetch() will leave state->how as COPY or GZIP unless the
  219:    end of the input file has been reached and all data has been processed.  */
  220: local int gz_fetch(state)
  221:     gz_statep state;
  222: {
  223:     z_streamp strm = &(state->strm);
  224: 
  225:     do {
  226:         switch(state->how) {
  227:         case LOOK:      /* -> LOOK, COPY (only if never GZIP), or GZIP */
  228:             if (gz_look(state) == -1)
  229:                 return -1;
  230:             if (state->how == LOOK)
  231:                 return 0;
  232:             break;
  233:         case COPY:      /* -> COPY */
  234:             if (gz_load(state, state->out, state->size << 1, &(state->x.have))
  235:                     == -1)
  236:                 return -1;
  237:             state->x.next = state->out;
  238:             return 0;
  239:         case GZIP:      /* -> GZIP or LOOK (if end of gzip stream) */
  240:             strm->avail_out = state->size << 1;
  241:             strm->next_out = state->out;
  242:             if (gz_decomp(state) == -1)
  243:                 return -1;
  244:         }
  245:     } while (state->x.have == 0 && (!state->eof || strm->avail_in));
  246:     return 0;
  247: }
  248: 
  249: /* Skip len uncompressed bytes of output.  Return -1 on error, 0 on success. */
  250: local int gz_skip(state, len)
  251:     gz_statep state;
  252:     z_off64_t len;
  253: {
  254:     unsigned n;
  255: 
  256:     /* skip over len bytes or reach end-of-file, whichever comes first */
  257:     while (len)
  258:         /* skip over whatever is in output buffer */
  259:         if (state->x.have) {
  260:             n = GT_OFF(state->x.have) || (z_off64_t)state->x.have > len ?
  261:                 (unsigned)len : state->x.have;
  262:             state->x.have -= n;
  263:             state->x.next += n;
  264:             state->x.pos += n;
  265:             len -= n;
  266:         }
  267: 
  268:         /* output buffer empty -- return if we're at the end of the input */
  269:         else if (state->eof && state->strm.avail_in == 0)
  270:             break;
  271: 
  272:         /* need more data to skip -- load up output buffer */
  273:         else {
  274:             /* get more output, looking for header if required */
  275:             if (gz_fetch(state) == -1)
  276:                 return -1;
  277:         }
  278:     return 0;
  279: }
  280: 
  281: /* -- see zlib.h -- */
  282: int ZEXPORT gzread(file, buf, len)
  283:     gzFile file;
  284:     voidp buf;
  285:     unsigned len;
  286: {
  287:     unsigned got, n;
  288:     gz_statep state;
  289:     z_streamp strm;
  290: 
  291:     /* get internal structure */
  292:     if (file == NULL)
  293:         return -1;
  294:     state = (gz_statep)file;
  295:     strm = &(state->strm);
  296: 
  297:     /* check that we're reading and that there's no (serious) error */
  298:     if (state->mode != GZ_READ ||
  299:             (state->err != Z_OK && state->err != Z_BUF_ERROR))
  300:         return -1;
  301: 
  302:     /* since an int is returned, make sure len fits in one, otherwise return
  303:        with an error (this avoids the flaw in the interface) */
  304:     if ((int)len < 0) {
  305:         gz_error(state, Z_DATA_ERROR, "requested length does not fit in int");
  306:         return -1;
  307:     }
  308: 
  309:     /* if len is zero, avoid unnecessary operations */
  310:     if (len == 0)
  311:         return 0;
  312: 
  313:     /* process a skip request */
  314:     if (state->seek) {
  315:         state->seek = 0;
  316:         if (gz_skip(state, state->skip) == -1)
  317:             return -1;
  318:     }
  319: 
  320:     /* get len bytes to buf, or less than len if at the end */
  321:     got = 0;
  322:     do {
  323:         /* first just try copying data from the output buffer */
  324:         if (state->x.have) {
  325:             n = state->x.have > len ? len : state->x.have;
  326:             memcpy(buf, state->x.next, n);
  327:             state->x.next += n;
  328:             state->x.have -= n;
  329:         }
  330: 
  331:         /* output buffer empty -- return if we're at the end of the input */
  332:         else if (state->eof && strm->avail_in == 0) {
  333:             state->past = 1;        /* tried to read past end */
  334:             break;
  335:         }
  336: 
  337:         /* need output data -- for small len or new stream load up our output
  338:            buffer */
  339:         else if (state->how == LOOK || len < (state->size << 1)) {
  340:             /* get more output, looking for header if required */
  341:             if (gz_fetch(state) == -1)
  342:                 return -1;
  343:             continue;       /* no progress yet -- go back to memcpy() above */
  344:             /* the copy above assures that we will leave with space in the
  345:                output buffer, allowing at least one gzungetc() to succeed */
  346:         }
  347: 
  348:         /* large len -- read directly into user buffer */
  349:         else if (state->how == COPY) {      /* read directly */
  350:             if (gz_load(state, buf, len, &n) == -1)
  351:                 return -1;
  352:         }
  353: 
  354:         /* large len -- decompress directly into user buffer */
  355:         else {  /* state->how == GZIP */
  356:             strm->avail_out = len;
  357:             strm->next_out = buf;
  358:             if (gz_decomp(state) == -1)
  359:                 return -1;
  360:             n = state->x.have;
  361:             state->x.have = 0;
  362:         }
  363: 
  364:         /* update progress */
  365:         len -= n;
  366:         buf = (char *)buf + n;
  367:         got += n;
  368:         state->x.pos += n;
  369:     } while (len);
  370: 
  371:     /* return number of bytes read into user buffer (will fit in int) */
  372:     return (int)got;
  373: }
  374: 
  375: /* -- see zlib.h -- */
  376: int ZEXPORT gzgetc_(file)
  377:     gzFile file;
  378: {
  379:     int ret;
  380:     unsigned char buf[1];
  381:     gz_statep state;
  382: 
  383:     /* get internal structure */
  384:     if (file == NULL)
  385:         return -1;
  386:     state = (gz_statep)file;
  387: 
  388:     /* check that we're reading and that there's no (serious) error */
  389:     if (state->mode != GZ_READ ||
  390:         (state->err != Z_OK && state->err != Z_BUF_ERROR))
  391:         return -1;
  392: 
  393:     /* try output buffer (no need to check for skip request) */
  394:     if (state->x.have) {
  395:         state->x.have--;
  396:         state->x.pos++;
  397:         return *(state->x.next)++;
  398:     }
  399: 
  400:     /* nothing there -- try gzread() */
  401:     ret = gzread(file, buf, 1);
  402:     return ret < 1 ? -1 : buf[0];
  403: }
  404: 
  405: #undef gzgetc
  406: int ZEXPORT gzgetc(file)
  407: gzFile file;
  408: {
  409:     return gzgetc_(file);
  410: }    
  411: 
  412: /* -- see zlib.h -- */
  413: int ZEXPORT gzungetc(c, file)
  414:     int c;
  415:     gzFile file;
  416: {
  417:     gz_statep state;
  418: 
  419:     /* get internal structure */
  420:     if (file == NULL)
  421:         return -1;
  422:     state = (gz_statep)file;
  423: 
  424:     /* check that we're reading and that there's no (serious) error */
  425:     if (state->mode != GZ_READ ||
  426:         (state->err != Z_OK && state->err != Z_BUF_ERROR))
  427:         return -1;
  428: 
  429:     /* process a skip request */
  430:     if (state->seek) {
  431:         state->seek = 0;
  432:         if (gz_skip(state, state->skip) == -1)
  433:             return -1;
  434:     }
  435: 
  436:     /* can't push EOF */
  437:     if (c < 0)
  438:         return -1;
  439: 
  440:     /* if output buffer empty, put byte at end (allows more pushing) */
  441:     if (state->x.have == 0) {
  442:         state->x.have = 1;
  443:         state->x.next = state->out + (state->size << 1) - 1;
  444:         state->x.next[0] = c;
  445:         state->x.pos--;
  446:         state->past = 0;
  447:         return c;
  448:     }
  449: 
  450:     /* if no room, give up (must have already done a gzungetc()) */
  451:     if (state->x.have == (state->size << 1)) {
  452:         gz_error(state, Z_DATA_ERROR, "out of room to push characters");
  453:         return -1;
  454:     }
  455: 
  456:     /* slide output data if needed and insert byte before existing data */
  457:     if (state->x.next == state->out) {
  458:         unsigned char *src = state->out + state->x.have;
  459:         unsigned char *dest = state->out + (state->size << 1);
  460:         while (src > state->out)
  461:             *--dest = *--src;
  462:         state->x.next = dest;
  463:     }
  464:     state->x.have++;
  465:     state->x.next--;
  466:     state->x.next[0] = c;
  467:     state->x.pos--;
  468:     state->past = 0;
  469:     return c;
  470: }
  471: 
  472: /* -- see zlib.h -- */
  473: char * ZEXPORT gzgets(file, buf, len)
  474:     gzFile file;
  475:     char *buf;
  476:     int len;
  477: {
  478:     unsigned left, n;
  479:     char *str;
  480:     unsigned char *eol;
  481:     gz_statep state;
  482: 
  483:     /* check parameters and get internal structure */
  484:     if (file == NULL || buf == NULL || len < 1)
  485:         return NULL;
  486:     state = (gz_statep)file;
  487: 
  488:     /* check that we're reading and that there's no (serious) error */
  489:     if (state->mode != GZ_READ ||
  490:         (state->err != Z_OK && state->err != Z_BUF_ERROR))
  491:         return NULL;
  492: 
  493:     /* process a skip request */
  494:     if (state->seek) {
  495:         state->seek = 0;
  496:         if (gz_skip(state, state->skip) == -1)
  497:             return NULL;
  498:     }
  499: 
  500:     /* copy output bytes up to new line or len - 1, whichever comes first --
  501:        append a terminating zero to the string (we don't check for a zero in
  502:        the contents, let the user worry about that) */
  503:     str = buf;
  504:     left = (unsigned)len - 1;
  505:     if (left) do {
  506:         /* assure that something is in the output buffer */
  507:         if (state->x.have == 0 && gz_fetch(state) == -1)
  508:             return NULL;                /* error */
  509:         if (state->x.have == 0) {       /* end of file */
  510:             state->past = 1;            /* read past end */
  511:             break;                      /* return what we have */
  512:         }
  513: 
  514:         /* look for end-of-line in current output buffer */
  515:         n = state->x.have > left ? left : state->x.have;
  516:         eol = memchr(state->x.next, '\n', n);
  517:         if (eol != NULL)
  518:             n = (unsigned)(eol - state->x.next) + 1;
  519: 
  520:         /* copy through end-of-line, or remainder if not found */
  521:         memcpy(buf, state->x.next, n);
  522:         state->x.have -= n;
  523:         state->x.next += n;
  524:         state->x.pos += n;
  525:         left -= n;
  526:         buf += n;
  527:     } while (left && eol == NULL);
  528: 
  529:     /* return terminated string, or if nothing, end of file */
  530:     if (buf == str)
  531:         return NULL;
  532:     buf[0] = 0;
  533:     return str;
  534: }
  535: 
  536: /* -- see zlib.h -- */
  537: int ZEXPORT gzdirect(file)
  538:     gzFile file;
  539: {
  540:     gz_statep state;
  541: 
  542:     /* get internal structure */
  543:     if (file == NULL)
  544:         return 0;
  545:     state = (gz_statep)file;
  546: 
  547:     /* if the state is not known, but we can find out, then do so (this is
  548:        mainly for right after a gzopen() or gzdopen()) */
  549:     if (state->mode == GZ_READ && state->how == LOOK && state->x.have == 0)
  550:         (void)gz_look(state);
  551: 
  552:     /* return 1 if transparent, 0 if processing a gzip stream */
  553:     return state->direct;
  554: }
  555: 
  556: /* -- see zlib.h -- */
  557: int ZEXPORT gzclose_r(file)
  558:     gzFile file;
  559: {
  560:     int ret, err;
  561:     gz_statep state;
  562: 
  563:     /* get internal structure */
  564:     if (file == NULL)
  565:         return Z_STREAM_ERROR;
  566:     state = (gz_statep)file;
  567: 
  568:     /* check that we're reading */
  569:     if (state->mode != GZ_READ)
  570:         return Z_STREAM_ERROR;
  571: 
  572:     /* free memory and close file */
  573:     if (state->size) {
  574:         inflateEnd(&(state->strm));
  575:         free(state->out);
  576:         free(state->in);
  577:     }
  578:     err = state->err == Z_BUF_ERROR ? Z_BUF_ERROR : Z_OK;
  579:     gz_error(state, Z_OK, NULL);
  580:     free(state->path);
  581:     ret = close(state->fd);
  582:     free(state);
  583:     return ret ? Z_ERRNO : err;
  584: }