Mercurial > repos > dawe > srf2fastq
annotate srf2fastq/io_lib-1.12.2/progs/hash_sff.c @ 0:d901c9f41a6a default tip
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
author | dawe |
---|---|
date | Tue, 07 Jun 2011 17:48:05 -0400 |
parents | |
children |
rev | line source |
---|---|
0
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
1 /* |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
2 * This adds a hash table index (".hsh" v1.00 format) to an SFF archive. |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
3 * It does this either inline on the file itself (provided it doesn't already |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
4 * have an index) or by producing a new indexed SFF archive. |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
5 * |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
6 * It has been coded to require only the memory needed to store the index |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
7 * and so does quite a lot of I/O but with minimised memory. For a 460,000 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
8 * SFF archive it took about 22 seconds real time on a 1.7GHz P4 when copying |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
9 * or 10 seconds when updating inline. |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
10 */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
11 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
12 /* ---------------------------------------------------------------------- */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
13 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
14 #include <stdio.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
15 #include <string.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
16 #include <errno.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
17 #include <stdlib.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
18 #include <io_lib/hash_table.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
19 #include <io_lib/sff.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
20 #include <io_lib/os.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
21 #include <io_lib/mFILE.h> |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
22 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
23 /* |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
24 * Override the sff.c functions to use FILE pointers instead. This means |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
25 * we don't have to load the entire archive into memory, which is optimal when |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
26 * dealing with a single file (ie in sff/sff.c), but not when indexing it. |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
27 * |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
28 * Done with minimal error checking I'll admit... |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
29 */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
30 static sff_read_header *fread_sff_read_header(FILE *fp) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
31 sff_read_header *h; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
32 unsigned char rhdr[16]; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
33 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
34 if (16 != fread(rhdr, 1, 16, fp)) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
35 return NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
36 h = decode_sff_read_header(rhdr); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
37 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
38 if (h->name_len != fread(h->name, 1, h->name_len, fp)) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
39 return free_sff_read_header(h), NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
40 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
41 /* Pad to 8 chars */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
42 fseek(fp, (ftell(fp) + 7)& ~7, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
43 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
44 return h; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
45 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
46 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
47 static sff_common_header *fread_sff_common_header(FILE *fp) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
48 sff_common_header *h; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
49 unsigned char chdr[31]; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
50 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
51 if (31 != fread(chdr, 1, 31, fp)) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
52 return NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
53 h = decode_sff_common_header(chdr); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
54 if (h->flow_len != fread(h->flow, 1, h->flow_len, fp)) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
55 return free_sff_common_header(h), NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
56 if (h->key_len != fread(h->key , 1, h->key_len, fp)) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
57 return free_sff_common_header(h), NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
58 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
59 /* Pad to 8 chars */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
60 fseek(fp, (ftell(fp) + 7)& ~7, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
61 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
62 return h; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
63 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
64 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
65 void usage(void) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
66 fprintf(stderr, "Usage: hash_sff [-o outfile] [-t] sff_file ...\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
67 exit(1); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
68 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
69 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
70 int main(int argc, char **argv) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
71 HashFile *hf; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
72 sff_common_header *ch; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
73 sff_read_header *rh; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
74 int i, dot, arg; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
75 char *sff; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
76 char hdr[31]; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
77 uint64_t index_offset = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
78 uint32_t index_size, index_skipped; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
79 FILE *fp, *fpout = NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
80 int copy_archive = 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
81 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
82 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
83 /* process command line arguments of the form -arg */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
84 for (argc--, argv++; argc > 0; argc--, argv++) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
85 if (**argv != '-' || strcmp(*argv, "--") == 0) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
86 break; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
87 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
88 if (strcmp(*argv, "-o") == 0 && argc > 1) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
89 if (NULL == (fpout = fopen(argv[1], "wb+"))) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
90 perror(argv[1]); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
91 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
92 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
93 argv++; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
94 argc--; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
95 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
96 } else if (strcmp(*argv, "-t") == 0) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
97 copy_archive = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
98 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
99 } else if (**argv == '-') { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
100 usage(); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
101 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
102 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
103 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
104 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
105 if (argc < 1) |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
106 usage(); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
107 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
108 if (copy_archive == 0 && argc != 1) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
109 fprintf(stderr, "-t option only supported with a single sff argument\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
110 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
111 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
112 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
113 /* Create the hash table */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
114 hf = HashFileCreate(0, HASH_DYNAMIC_SIZE); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
115 hf->nheaders = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
116 hf->headers = NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
117 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
118 for (arg = 0; arg < argc; arg++) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
119 /* open (and read) the entire sff file */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
120 sff = argv[arg]; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
121 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
122 printf("Indexing %s:\n", sff); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
123 if (fpout) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
124 if (NULL == (fp = fopen(sff, "rb"))) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
125 perror(sff); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
126 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
127 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
128 } else { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
129 if (NULL == (fp = fopen(sff, "rb+"))) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
130 perror(sff); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
131 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
132 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
133 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
134 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
135 /* Read the common header */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
136 ch = fread_sff_common_header(fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
137 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
138 if (ch->index_len && !fpout) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
139 fprintf(stderr, "Archive already contains index.\nReplacing the" |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
140 " index requires the \"-o outfile\" option.\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
141 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
142 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
143 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
144 /* Add the SFF common header as a hash file-header */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
145 hf->nheaders++; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
146 hf->headers = (HashFileSection *)realloc(hf->headers, hf->nheaders * |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
147 sizeof(*hf->headers)); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
148 hf->headers[hf->nheaders-1].pos = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
149 hf->headers[hf->nheaders-1].size = ch->header_len; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
150 hf->headers[hf->nheaders-1].cached_data = NULL; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
151 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
152 /* Read the index items, adding to the hash */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
153 index_skipped = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
154 dot = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
155 printf(" |\r|"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
156 for (i = 0; i < ch->nreads; i++) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
157 int dlen; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
158 uint32_t offset; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
159 HashData hd; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
160 HashFileItem *hfi; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
161 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
162 if (i >= dot * (ch->nreads/69)) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
163 putchar('.'); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
164 fflush(stdout); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
165 dot++; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
166 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
167 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
168 /* Skip old index if present */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
169 offset = ftell(fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
170 if (offset == ch->index_offset) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
171 fseek(fp, ch->index_len, SEEK_CUR); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
172 index_skipped = ch->index_len; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
173 continue; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
174 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
175 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
176 hfi = (HashFileItem *)calloc(1, sizeof(*hfi)); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
177 rh = fread_sff_read_header(fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
178 dlen = (2*ch->flow_len + 3*rh->nbases + 7) & ~7; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
179 fseek(fp, dlen, SEEK_CUR); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
180 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
181 hfi->header = hf->nheaders; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
182 hfi->footer = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
183 hfi->pos = offset - index_skipped; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
184 hfi->size = (ftell(fp) - index_skipped) - hfi->pos; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
185 hd.p = hfi; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
186 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
187 HashTableAdd(hf->h, rh->name, rh->name_len, hd, NULL); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
188 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
189 printf("\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
190 HashTableStats(hf->h, stdout); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
191 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
192 index_offset = ftell(fp) - index_skipped; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
193 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
194 /* Copy the archive if needed, minus the old index */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
195 if (fpout && copy_archive) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
196 char block[8192]; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
197 size_t len; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
198 uint64_t pos = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
199 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
200 printf("\nCopying archive\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
201 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
202 fseek(fp, 0, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
203 while (len = fread(block, 1, 8192, fp)) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
204 /* Skip previous index */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
205 if (pos < ch->index_offset && pos+len > ch->index_offset) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
206 len = ch->index_offset - pos; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
207 fseek(fp, ch->index_offset + ch->index_len, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
208 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
209 if (len && len != fwrite(block, 1, len, fpout)) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
210 fprintf(stderr, "Failed to output new archive\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
211 return 1; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
212 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
213 pos += len; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
214 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
215 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
216 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
217 if (!fpout) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
218 /* Save the hash */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
219 printf("Saving index\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
220 fseek(fp, 0, SEEK_END); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
221 index_size = HashFileSave(hf, fp, 0); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
222 HashFileDestroy(hf); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
223 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
224 /* Update the common header */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
225 fseek(fp, 0, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
226 fread(hdr, 1, 31, fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
227 *(uint64_t *)(hdr+8) = be_int8(index_offset); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
228 *(uint32_t *)(hdr+16) = be_int4(index_size); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
229 fseek(fp, 0, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
230 fwrite(hdr, 1, 31, fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
231 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
232 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
233 fclose(fp); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
234 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
235 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
236 if (fpout) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
237 /* Save the hash */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
238 printf("Saving index\n"); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
239 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
240 if (!copy_archive) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
241 hf->archive = strdup(argv[0]); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
242 index_offset = 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
243 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
244 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
245 fseek(fpout, 0, SEEK_END); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
246 index_size = HashFileSave(hf, fpout, 0); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
247 HashFileDestroy(hf); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
248 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
249 /* Update the common header to indicate index location */ |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
250 if (copy_archive) { |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
251 fseek(fpout, 0, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
252 fread(hdr, 1, 31, fpout); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
253 *(uint64_t *)(hdr+8) = be_int8(index_offset); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
254 *(uint32_t *)(hdr+16) = be_int4(index_size); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
255 fseek(fpout, 0, SEEK_SET); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
256 fwrite(hdr, 1, 31, fpout); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
257 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
258 fclose(fpout); |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
259 } |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
260 |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
261 return 0; |
d901c9f41a6a
Migrated tool version 1.0.1 from old tool shed archive to new tool shed repository
dawe
parents:
diff
changeset
|
262 } |