void save_hash(HashFile *hf, options_t *opt) {
HashTableStats(hf->h, stderr);
#ifdef _WIN32
_setmode(_fileno(stdout), _O_BINARY);
#endif
HashFileSave(hf, stdout, opt->prepend_mode ? HASHFILE_PREPEND : 0);
HashFileDestroy(hf);
}
int main(int argc, char *argv[])
{
char *fname; // name of dictionary file
FILE *wordf; // handle for dictionary file
Item word; // current word from file
int size = 7919; // default size of hash table
HashTable htab; // the hash table
int nwords; // # words read and stored
int nfound; // # words found during search tests
// set up parameters
switch (argc) {
case 2: fname = argv[1]; break;
case 3: fname = argv[1]; size = atoi(argv[2]); break;
default: fname = NULL; usage(argv[0]); break;
}
// access the word file
if (eq(fname,"-")) {
wordf = stdin;
printf("Reading words from stdin\n");
}
else {
wordf = fopen(fname,"r");
if (wordf == NULL) {
printf("Can't open %s\n",fname);
exit(1);
}
printf("Reading words from %s\n",fname);
}
// build hash table, containing all words from file
nwords = 0; nfound = 0;
htab = newHashTable(size);
while ((word = ItemGet(wordf)) != NULL) {
if (eq(word,"")) { dropItem(word); continue; }
HashTableInsert(htab,word);
nwords++;
if (HashTableSearch(htab,word) != NULL)
nfound++;
dropItem(word);
}
// examine hash table
HashTableStats(htab);
// tests
// warning: we are assuming that "!aaaaaa!" etc.
// do not occur in the input; this is not guaranteed
assert(nfound == nwords);
assert(HashTableSearch(htab,"!aaaaaa!") == NULL) ;
assert(HashTableSearch(htab,"!xxxxxx!") == NULL) ;
assert(HashTableSearch(htab,"!yyyyyy!") == NULL) ;
assert(HashTableSearch(htab,"!zzzzzz!") == NULL) ;
printf("Testing completed OK\n");
// clean up
fclose(wordf);
dropHashTable(htab);
return 0;
}
int main(int argc, char **argv) {
HashFile *hf;
sff_common_header *ch;
sff_read_header *rh;
int i, dot, arg;
char *sff;
char hdr[31];
uint64_t index_offset = 0;
uint32_t index_size, index_skipped;
FILE *fp, *fpout = NULL;
int copy_archive = 1;
/* process command line arguments of the form -arg */
for (argc--, argv++; argc > 0; argc--, argv++) {
if (**argv != '-' || strcmp(*argv, "--") == 0)
break;
if (strcmp(*argv, "-o") == 0 && argc > 1) {
if (NULL == (fpout = fopen(argv[1], "wb+"))) {
perror(argv[1]);
return 1;
}
argv++;
argc--;
} else if (strcmp(*argv, "-t") == 0) {
copy_archive = 0;
} else if (**argv == '-') {
usage();
}
}
if (argc < 1)
usage();
if (copy_archive == 0 && argc != 1) {
fprintf(stderr, "-t option only supported with a single sff argument\n");
return 1;
}
/* Create the hash table */
hf = HashFileCreate(0, HASH_DYNAMIC_SIZE);
hf->nheaders = 0;
hf->headers = NULL;
for (arg = 0; arg < argc; arg++) {
/* open (and read) the entire sff file */
sff = argv[arg];
printf("Indexing %s:\n", sff);
if (fpout) {
if (NULL == (fp = fopen(sff, "rb"))) {
perror(sff);
return 1;
}
} else {
if (NULL == (fp = fopen(sff, "rb+"))) {
perror(sff);
return 1;
}
}
/* Read the common header */
ch = fread_sff_common_header(fp);
if (ch->index_len && !fpout) {
fprintf(stderr, "Archive already contains index.\nReplacing the"
" index requires the \"-o outfile\" option.\n");
return 1;
}
/* Add the SFF common header as a hash file-header */
hf->nheaders++;
hf->headers = (HashFileSection *)realloc(hf->headers, hf->nheaders *
sizeof(*hf->headers));
hf->headers[hf->nheaders-1].pos = 0;
hf->headers[hf->nheaders-1].size = ch->header_len;
hf->headers[hf->nheaders-1].cached_data = NULL;
/* Read the index items, adding to the hash */
index_skipped = 0;
dot = 0;
printf(" |\r|");
for (i = 0; i < ch->nreads; i++) {
int dlen;
uint32_t offset;
HashData hd;
HashFileItem *hfi;
if (i >= dot * (ch->nreads/69)) {
putchar('.');
fflush(stdout);
dot++;
}
/* Skip old index if present */
offset = ftell(fp);
if (offset == ch->index_offset) {
fseek(fp, ch->index_len, SEEK_CUR);
index_skipped = ch->index_len;
continue;
}
hfi = (HashFileItem *)calloc(1, sizeof(*hfi));
rh = fread_sff_read_header(fp);
dlen = (2*ch->flow_len + 3*rh->nbases + 7) & ~7;
fseek(fp, dlen, SEEK_CUR);
hfi->header = hf->nheaders;
hfi->footer = 0;
hfi->pos = offset - index_skipped;
hfi->size = (ftell(fp) - index_skipped) - hfi->pos;
hd.p = hfi;
HashTableAdd(hf->h, rh->name, rh->name_len, hd, NULL);
}
printf("\n");
HashTableStats(hf->h, stdout);
index_offset = ftell(fp) - index_skipped;
/* Copy the archive if needed, minus the old index */
if (fpout && copy_archive) {
char block[8192];
size_t len;
uint64_t pos = 0;
printf("\nCopying archive\n");
fseek(fp, 0, SEEK_SET);
while (len = fread(block, 1, 8192, fp)) {
/* Skip previous index */
if (pos < ch->index_offset && pos+len > ch->index_offset) {
len = ch->index_offset - pos;
fseek(fp, ch->index_offset + ch->index_len, SEEK_SET);
}
if (len && len != fwrite(block, 1, len, fpout)) {
fprintf(stderr, "Failed to output new archive\n");
return 1;
}
pos += len;
}
}
if (!fpout) {
/* Save the hash */
printf("Saving index\n");
fseek(fp, 0, SEEK_END);
index_size = HashFileSave(hf, fp, 0);
HashFileDestroy(hf);
/* Update the common header */
fseek(fp, 0, SEEK_SET);
fread(hdr, 1, 31, fp);
*(uint64_t *)(hdr+8) = be_int8(index_offset);
*(uint32_t *)(hdr+16) = be_int4(index_size);
fseek(fp, 0, SEEK_SET);
fwrite(hdr, 1, 31, fp);
}
fclose(fp);
}
if (fpout) {
/* Save the hash */
printf("Saving index\n");
if (!copy_archive) {
hf->archive = strdup(argv[0]);
index_offset = 0;
}
fseek(fpout, 0, SEEK_END);
index_size = HashFileSave(hf, fpout, 0);
HashFileDestroy(hf);
/* Update the common header to indicate index location */
if (copy_archive) {
fseek(fpout, 0, SEEK_SET);
fread(hdr, 1, 31, fpout);
*(uint64_t *)(hdr+8) = be_int8(index_offset);
*(uint32_t *)(hdr+16) = be_int4(index_size);
fseek(fpout, 0, SEEK_SET);
fwrite(hdr, 1, 31, fpout);
}
fclose(fpout);
}
return 0;
}
