1 | /* File: readseq.c |
---|
2 | * main() program for ureadseq.c, ureadseq.h |
---|
3 | * |
---|
4 | * Reads and writes nucleic/protein sequence in various |
---|
5 | * formats. Data files may have multiple sequences. |
---|
6 | * |
---|
7 | * Copyright 1990 by d.g.gilbert |
---|
8 | * biology dept., indiana university, bloomington, in 47405 |
---|
9 | * e-mail: gilbertd@bio.indiana.edu |
---|
10 | * |
---|
11 | * This program may be freely copied and used by anyone. |
---|
12 | * Developers are encourged to incorporate parts in their |
---|
13 | * programs, rather than devise their own private sequence |
---|
14 | * format. |
---|
15 | * |
---|
16 | * This should compile and run with any ANSI C compiler. |
---|
17 | * Please advise me of any bugs, additions or corrections. |
---|
18 | * |
---|
19 | */ |
---|
20 | |
---|
21 | const char *title |
---|
22 | = "readSeq (1Feb93), multi-format molbio sequence reader.\n"; |
---|
23 | |
---|
24 | /* History |
---|
25 | 27 Feb 90. 1st release to public. |
---|
26 | 4 Mar 90. + Gary Olsen format |
---|
27 | + case change |
---|
28 | * minor corrections to NBRF,EMBL,others |
---|
29 | * output 1 file per sequence for gcg, unknown |
---|
30 | * define -DNOSTR for c-libraries w/o strstr |
---|
31 | - readseq.p, pascal version, becomes out-of-date |
---|
32 | 24 May 90. + Phylip 3.2 output format (no input) |
---|
33 | 20 Jul 90. + Phylip 3.3 output (no input yet) |
---|
34 | + interactive output re-direction |
---|
35 | + verbose progress info |
---|
36 | * interactive help output |
---|
37 | * dropped line no.s on NBRF output |
---|
38 | * patched in HyperGCG XCMD corrections, |
---|
39 | - except for seq. documentation handling |
---|
40 | * dropped the IG special nuc codes, as IG has |
---|
41 | adopted the standard IUB codes (now if only |
---|
42 | everyone would adopt a standard format !) |
---|
43 | 11 Oct 90. * corrected bug in reading/writing of EMBL format |
---|
44 | |
---|
45 | 17 Oct 91. * corrected bug in reading Olsen format |
---|
46 | (serious-deletion) |
---|
47 | 10 Nov 91. * corrected bug in reading some GCG format files |
---|
48 | (serious-last line duplicated) |
---|
49 | + add format name parsing (-fgb, -ffasta, ...) |
---|
50 | + Phylip v3.4 output format (== v3.2, sequential) |
---|
51 | + add checksum output to all forms that have document |
---|
52 | + skip mail headers in seq file |
---|
53 | + add pipe for standard input == seq file (with -p) |
---|
54 | * fold in parts of MacApp Seq object |
---|
55 | * strengthen format detection |
---|
56 | * clarify program structure |
---|
57 | * remove fixed sequence size limit (now dynamic, sizeof memory) |
---|
58 | * check and fold in accumulated bug reports: |
---|
59 | * Now ANSI-C fopen(..,"w") & check open failure |
---|
60 | * Define -DFIXTOUPPER for nonANSI C libraries that mess |
---|
61 | up toupper/tolower |
---|
62 | = No command-line changes; callers of readseq main() should be okay |
---|
63 | - ureadseq.h functions have changed; client programs need to note. |
---|
64 | + added Unix and VMS Make scripts, including validation tests |
---|
65 | |
---|
66 | 4 May 92. + added 32 bit CRC checksum as alternative to GCG 6.5bit checksum |
---|
67 | (-DBIGCHECKSUM) |
---|
68 | Aug 92 = fixed Olsen format input to handle files w/ more sequences, |
---|
69 | not to mess up when more than one seq has same identifier, |
---|
70 | and to convert number masks to symbols. |
---|
71 | = IG format fix to understand ^L |
---|
72 | |
---|
73 | 25-30 Dec 92 |
---|
74 | * revised command-line & interactive interface. Suggested form is now |
---|
75 | readseq infile -format=genbank -output=outfile -item=1,3,4 ... |
---|
76 | but remains compatible with prior commandlines: |
---|
77 | readseq infile -f2 -ooutfile -i3 ... |
---|
78 | + added GCG MSF multi sequence file format |
---|
79 | + added PIR/CODATA format |
---|
80 | + added NCBI ASN.1 sequence file format |
---|
81 | + added Pretty, multi sequence pretty output (only) |
---|
82 | + added PAUP multi seq format |
---|
83 | + added degap option |
---|
84 | + added Gary Williams (GWW, G.Williams@CRC.AC.UK) reverse-complement option. |
---|
85 | + added support for reading Phylip formats (interleave & sequential) |
---|
86 | * string fixes, dropped need for compiler flags NOSTR, FIXTOUPPER, NEEDSTRCASECMP |
---|
87 | * changed 32bit checksum to default, -DSMALLCHECKSUM for GCG version |
---|
88 | |
---|
89 | 1Feb93 |
---|
90 | = revert GenBank output to a fixed left number width which |
---|
91 | other software depends on. |
---|
92 | = fix for MSF input to handle symbols in names |
---|
93 | = fix bug for possible memory overrun when truncating seqs for |
---|
94 | Phylip or Paup formats (thanks Anthony Persechini) |
---|
95 | |
---|
96 | */ |
---|
97 | |
---|
98 | |
---|
99 | |
---|
100 | /* |
---|
101 | Readseq has been tested with: |
---|
102 | Macintosh MPW C |
---|
103 | GNU gcc |
---|
104 | SGI cc |
---|
105 | VAX-VMS cc |
---|
106 | Any ANSI C compiler should be able to handle this. |
---|
107 | Old-style C compilers barf all over the source. |
---|
108 | |
---|
109 | |
---|
110 | How do I build the readseq program if I have an Ansi C compiler? |
---|
111 | #-------------------- |
---|
112 | # Unix ANSI C |
---|
113 | # Use the supplied Makefile this way: |
---|
114 | % make CC=name-of-c-compiler |
---|
115 | # OR do this... |
---|
116 | % gcc readseq.c ureadseq.c -o readseq |
---|
117 | |
---|
118 | #-------------------- |
---|
119 | $!VAX-VMS cc |
---|
120 | $! Use the supplied Make.Com this way: |
---|
121 | $ @make |
---|
122 | $! OR, do this: |
---|
123 | $ cc readseq, ureadseq |
---|
124 | $ link readseq, ureadseq, sys$library:vaxcrtl/lib |
---|
125 | $ readseq :== $ MyDisk:[myacct]readseq |
---|
126 | |
---|
127 | #-------------------- |
---|
128 | # Macintosh Simple Input/Output Window application |
---|
129 | # requires MPW-C and SIOW library (from APDA) |
---|
130 | # also uses files macinit.c, macinit.r, readseqSIOW.make |
---|
131 | # |
---|
132 | Buildprogram readseqSIOW |
---|
133 | |
---|
134 | #-------------------- |
---|
135 | #MPW-C v3 tool |
---|
136 | C ureadseq.c |
---|
137 | C readseq.c |
---|
138 | link -w -o readseq -t MPST -c 'MPS ' ¶ |
---|
139 | readseq.c.o Ureadseq.c.o ¶ |
---|
140 | "{Libraries}"Interface.o ¶ |
---|
141 | "{Libraries}"ToolLibs.o ¶ |
---|
142 | "{Libraries}"Runtime.o ¶ |
---|
143 | "{CLibraries}"StdClib.o |
---|
144 | readseq -i1 ig.seq |
---|
145 | |
---|
146 | # MPW-C with NCBI tools |
---|
147 | |
---|
148 | set NCBI "{Boot}@molbio:ncbi:"; EXPORT NCBI |
---|
149 | set NCBILIB1 "{NCBI}"lib:libncbi.o; export NCBILIB1 |
---|
150 | set NCBILIB2 "{NCBI}"lib:libncbiobj.o; export NCBILIB2 |
---|
151 | set NCBILIB3 "{NCBI}"lib:libncbicdr.o; export NCBILIB3 |
---|
152 | set NCBILIB4 "{NCBI}"lib:libvibrant.o; export NCBILIB4 |
---|
153 | |
---|
154 | C ureadseq.c |
---|
155 | C -d NCBI -i "{NCBI}"include: ureadasn.c |
---|
156 | C -d NCBI -i "{NCBI}"include: readseq.c |
---|
157 | link -w -o readseq -t MPST -c 'MPS ' ¶ |
---|
158 | ureadseq.c.o ureadasn.c.o readseq.c.o ¶ |
---|
159 | {NCBILIB4} {NCBILIB2} {NCBILIB1} ¶ |
---|
160 | "{Libraries}"Interface.o ¶ |
---|
161 | "{Libraries}"ToolLibs.o ¶ |
---|
162 | "{Libraries}"Runtime.o ¶ |
---|
163 | "{CLibraries}"CSANELib.o ¶ |
---|
164 | "{CLibraries}"Math.o ¶ |
---|
165 | "{CLibraries}"StdClib.o |
---|
166 | |
---|
167 | ===========================================================*/ |
---|
168 | |
---|
169 | |
---|
170 | |
---|
171 | #include <stdio.h> |
---|
172 | #include <string.h> |
---|
173 | #define __NO_CTYPE |
---|
174 | #include <ctype.h> |
---|
175 | |
---|
176 | #include "ureadseq.h" |
---|
177 | |
---|
178 | #pragma segment readseq |
---|
179 | |
---|
180 | |
---|
181 | |
---|
182 | static char inputfilestore[256], *inputfile = inputfilestore; |
---|
183 | |
---|
184 | const char *formats[kMaxFormat+1] = { |
---|
185 | " 1. IG/Stanford", |
---|
186 | " 2. GenBank/GB", |
---|
187 | " 3. NBRF", |
---|
188 | " 4. EMBL", |
---|
189 | " 5. GCG", |
---|
190 | " 6. DNAStrider", |
---|
191 | " 7. Fitch", |
---|
192 | " 8. Pearson/Fasta", |
---|
193 | " 9. Zuker (in-only)", |
---|
194 | "10. Olsen (in-only)", |
---|
195 | "11. Phylip3.2", |
---|
196 | "12. Phylip", |
---|
197 | "13. Plain/Raw", |
---|
198 | "14. PIR/CODATA", |
---|
199 | "15. MSF", |
---|
200 | "16. ASN.1", |
---|
201 | "17. PAUP/NEXUS", |
---|
202 | "18. Pretty (out-only)", |
---|
203 | "" }; |
---|
204 | |
---|
205 | #define kFormCount 30 |
---|
206 | #define kMaxFormName 15 |
---|
207 | |
---|
208 | const struct formatTable { |
---|
209 | char *name; |
---|
210 | short num; |
---|
211 | } formname[] = { |
---|
212 | {"ig", kIG}, |
---|
213 | {"stanford", kIG}, |
---|
214 | {"genbank", kGenBank}, |
---|
215 | {"gb", kGenBank}, |
---|
216 | {"nbrf", kNBRF}, |
---|
217 | {"embl", kEMBL}, |
---|
218 | {"gcg", kGCG}, |
---|
219 | {"uwgcg", kGCG}, |
---|
220 | {"dnastrider", kStrider}, |
---|
221 | {"strider", kStrider}, |
---|
222 | {"fitch", kFitch}, |
---|
223 | {"pearson", kPearson}, |
---|
224 | {"fasta", kPearson}, |
---|
225 | {"zuker", kZuker}, |
---|
226 | {"olsen", kOlsen}, |
---|
227 | {"phylip", kPhylip}, |
---|
228 | {"phylip3.2", kPhylip2}, |
---|
229 | {"phylip3.3", kPhylip3}, |
---|
230 | {"phylip3.4", kPhylip4}, |
---|
231 | {"phylip-interleaved", kPhylip4}, |
---|
232 | {"phylip-sequential", kPhylip2}, |
---|
233 | {"plain", kPlain}, |
---|
234 | {"raw", kPlain}, |
---|
235 | {"pir", kPIR}, |
---|
236 | {"codata", kPIR}, |
---|
237 | {"asn.1", kASN1}, |
---|
238 | {"msf", kMSF}, |
---|
239 | {"paup", kPAUP}, |
---|
240 | {"nexus", kPAUP}, |
---|
241 | {"pretty", kPretty}, |
---|
242 | }; |
---|
243 | |
---|
244 | const char *kASN1headline = "Bioseq-set ::= {\nseq-set {\n"; |
---|
245 | |
---|
246 | /* GWW table for getting the complement of a nucleotide (IUB codes) */ |
---|
247 | /* ! "#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[ \]^_`abcdefghijklmnopqrstuvwxyz{|}~ */ |
---|
248 | const char compl[] = " !\"#$%&'()*+,-./0123456789:;<=>?@TVGHNNCDNNMNKNNYRYSAABWNRN[\\]^_`tvghnncdnnmnknnyrysaabwnrn{|}~"; |
---|
249 | |
---|
250 | |
---|
251 | |
---|
252 | char *formatstr( short format) |
---|
253 | { |
---|
254 | if (format < 1 || format > kMaxFormat) { |
---|
255 | switch (format) { |
---|
256 | case kASNseqentry : |
---|
257 | case kASNseqset : return formats[kASN1-1]; |
---|
258 | case kPhylipInterleave: |
---|
259 | case kPhylipSequential: return formats[kPhylip-1]; |
---|
260 | default: return "(unknown)"; |
---|
261 | } |
---|
262 | } |
---|
263 | else return formats[format-1]; |
---|
264 | } |
---|
265 | |
---|
266 | int rs_isdigit(int c){ |
---|
267 | return isdigit(c); |
---|
268 | } |
---|
269 | |
---|
270 | int parseformat( char *name2) |
---|
271 | { |
---|
272 | #define kDupmatch -2 |
---|
273 | int namelen, maxlen, i, match, matchat; |
---|
274 | char lname[kMaxFormName+1]; |
---|
275 | |
---|
276 | skipwhitespace(name2); |
---|
277 | namelen = strlen(name2); |
---|
278 | if (namelen == 0) |
---|
279 | return kNoformat; |
---|
280 | else if (rs_isdigit(*name2)) { |
---|
281 | i = atol( name2); |
---|
282 | if (i < kMinFormat | i > kMaxFormat) return kNoformat; |
---|
283 | else return i; |
---|
284 | } |
---|
285 | |
---|
286 | /* else match character name */ |
---|
287 | maxlen = min( kMaxFormName, namelen); |
---|
288 | for (i=0; i<maxlen; i++) lname[i] = to_lower(name2[i]); |
---|
289 | lname[maxlen]=0; |
---|
290 | matchat = kNoformat; |
---|
291 | |
---|
292 | for (i=0; i<kFormCount; i++) { |
---|
293 | match = strncmp( lname, formname[i].name, maxlen); |
---|
294 | if (match == 0) { |
---|
295 | if (strlen(formname[i].name) == namelen) return (formname[i].num); |
---|
296 | else if (matchat == kNoformat) matchat = i; |
---|
297 | else matchat = kDupmatch; /* 2 or more partial matches */ |
---|
298 | } |
---|
299 | } |
---|
300 | if (matchat == kNoformat || matchat == kDupmatch) |
---|
301 | return kNoformat; |
---|
302 | else |
---|
303 | return formname[matchat].num; |
---|
304 | } |
---|
305 | |
---|
306 | |
---|
307 | |
---|
308 | static void dumpSeqList(char *list, short format) |
---|
309 | { |
---|
310 | long i, l, listlen; |
---|
311 | char s[256]; |
---|
312 | |
---|
313 | listlen = strlen(list); |
---|
314 | printf("Sequences in %s (format is %s)\n", inputfile, formatstr(format)); |
---|
315 | for (i=0, l=0; i < listlen; i++) { |
---|
316 | if (list[i] == (char)NEWLINE) { |
---|
317 | s[l] = '\0'; l = 0; |
---|
318 | puts(s); |
---|
319 | } |
---|
320 | else if (l < 255) |
---|
321 | s[l++] = list[i]; |
---|
322 | } |
---|
323 | putchar('\n'); |
---|
324 | } |
---|
325 | |
---|
326 | |
---|
327 | |
---|
328 | void usage() |
---|
329 | { |
---|
330 | short i, midi; |
---|
331 | |
---|
332 | fprintf(stderr,title); |
---|
333 | fprintf(stderr, |
---|
334 | "usage: readseq [-options] in.seq > out.seq\n"); |
---|
335 | fprintf(stderr," options\n"); |
---|
336 | /* ? add -d[igits] to allow digits in sequence data, &/or option to specify seq charset !? */ |
---|
337 | fprintf(stderr, " -a[ll] select All sequences\n"); |
---|
338 | fprintf(stderr, " -c[aselower] change to lower case\n"); |
---|
339 | fprintf(stderr, " -C[ASEUPPER] change to UPPER CASE\n"); |
---|
340 | fprintf(stderr, " -degap[=-] remove gap symbols\n"); |
---|
341 | fprintf(stderr, " -i[tem=2,3,4] select Item number(s) from several\n"); |
---|
342 | fprintf(stderr, " -l[ist] List sequences only\n"); |
---|
343 | fprintf(stderr, " -o[utput=]out.seq redirect Output\n"); |
---|
344 | fprintf(stderr, " -p[ipe] Pipe (command line, <stdin, >stdout)\n"); |
---|
345 | fprintf(stderr, " -r[everse] change to Reverse-complement\n"); |
---|
346 | fprintf(stderr, " -v[erbose] Verbose progress\n"); |
---|
347 | fprintf(stderr, " -f[ormat=]# Format number for output, or\n"); |
---|
348 | fprintf(stderr, " -f[ormat=]Name Format name for output:\n"); |
---|
349 | midi = (kMaxFormat+1) / 2; |
---|
350 | for (i = kMinFormat-1; i < midi; i++) |
---|
351 | fprintf( stderr, " %-20s %-20s\n", |
---|
352 | formats[i], formats[midi+i]); |
---|
353 | |
---|
354 | /* new output format options, esp. for pretty format: */ |
---|
355 | fprintf(stderr, " \n"); |
---|
356 | fprintf(stderr, " Pretty format options: \n"); |
---|
357 | fprintf(stderr, " -wid[th]=# sequence line width\n"); |
---|
358 | fprintf(stderr, " -tab=# left indent\n"); |
---|
359 | fprintf(stderr, " -col[space]=# column space within sequence line on output\n"); |
---|
360 | fprintf(stderr, " -gap[count] count gap chars in sequence numbers\n"); |
---|
361 | fprintf(stderr, " -nameleft, -nameright[=#] name on left/right side [=max width]\n"); |
---|
362 | fprintf(stderr, " -nametop name at top/bottom\n"); |
---|
363 | fprintf(stderr, " -numleft, -numright seq index on left/right side\n"); |
---|
364 | fprintf(stderr, " -numtop, -numbot index on top/bottom\n"); |
---|
365 | fprintf(stderr, " -match[=.] use match base for 2..n species\n"); |
---|
366 | fprintf(stderr, " -inter[line=#] blank line(s) between sequence blocks\n"); |
---|
367 | |
---|
368 | /****** not ready yet |
---|
369 | fprintf(stderr, " -code=none,rtf,postscript,ps code syntax\n"); |
---|
370 | fprintf(stderr, " -namefont=, -numfont=, -seqfont=font font choice\n"); |
---|
371 | fprintf(stderr, " font suggestions include times,courier,helvetica\n"); |
---|
372 | fprintf(stderr, " -namefontsize=, -numfontsize=, -seqfontsize=#\n"); |
---|
373 | fprintf(stderr, " fontsize suggestions include 9,10,12,14\n"); |
---|
374 | fprintf(stderr, " -namefontstyle=, -numfontstyle=, -seqfontstyle= style fontstyle for names\n"); |
---|
375 | fprintf(stderr, " fontstyle options are plain,italic,bold,bold-italic\n"); |
---|
376 | ******/ |
---|
377 | } |
---|
378 | |
---|
379 | void erralert(short err) |
---|
380 | { |
---|
381 | switch (err) { |
---|
382 | case 0 : |
---|
383 | break; |
---|
384 | case eFileNotFound: fprintf(stderr, "File not found: %s\n", inputfile); |
---|
385 | break; |
---|
386 | case eFileCreate: fprintf(stderr, "Can't open output file.\n"); |
---|
387 | break; |
---|
388 | case eASNerr: fprintf(stderr, "Error in ASN.1 sequence routines.\n"); |
---|
389 | break; |
---|
390 | case eNoData: fprintf(stderr, "No data in file.\n"); |
---|
391 | break; |
---|
392 | case eItemNotFound: fprintf(stderr, "Specified item not in file.\n"); |
---|
393 | break; |
---|
394 | case eUnequalSize: fprintf(stderr, |
---|
395 | "This format requires equal length sequences.\nSequence truncated or padded to fit.\n"); |
---|
396 | break; |
---|
397 | case eUnknownFormat: fprintf(stderr, "Error: this format is unknown to me.\n"); |
---|
398 | break; |
---|
399 | case eOneFormat: fprintf(stderr, |
---|
400 | "Warning: This format permits only 1 sequence per file.\n"); |
---|
401 | break; |
---|
402 | case eMemFull: fprintf(stderr, "Out of storage memory. Sequence truncated.\n"); |
---|
403 | break; |
---|
404 | default: fprintf(stderr, "readSeq error = %d\n", err); |
---|
405 | break; |
---|
406 | } |
---|
407 | } /* erralert */ |
---|
408 | |
---|
409 | |
---|
410 | int chooseFormat( boolean quietly) |
---|
411 | { |
---|
412 | char sform[128]; |
---|
413 | int midi, i, outform; |
---|
414 | |
---|
415 | if (quietly) |
---|
416 | return kPearson; /* default */ |
---|
417 | else { |
---|
418 | midi = (kMaxFormat+1) / 2; |
---|
419 | for (i = kMinFormat-1; i < midi; i++) |
---|
420 | fprintf( stderr, " %-20s %-20s\n", |
---|
421 | formats[i], formats[midi+i]); |
---|
422 | fprintf(stderr,"\nChoose an output format (name or #): \n"); |
---|
423 | gets(sform); |
---|
424 | outform = parseformat(sform); |
---|
425 | if (outform == kNoformat) outform = kPearson; |
---|
426 | return outform; |
---|
427 | } |
---|
428 | } |
---|
429 | |
---|
430 | |
---|
431 | |
---|
432 | /* read paramater(s) */ |
---|
433 | |
---|
434 | boolean checkopt( boolean casesense, char *sopt, const char *smatch, short minword) |
---|
435 | { |
---|
436 | long lenopt, lenmatch; |
---|
437 | boolean result; |
---|
438 | short minmaxw; |
---|
439 | |
---|
440 | lenopt = strlen(sopt); |
---|
441 | lenmatch= strlen(smatch); |
---|
442 | minmaxw= max(minword, min(lenopt, lenmatch)); |
---|
443 | |
---|
444 | if (casesense) |
---|
445 | result= (!strncmp( sopt, smatch, minmaxw)); |
---|
446 | else |
---|
447 | result= (!Strncasecmp( sopt, smatch, minmaxw )); |
---|
448 | /* if (result) { */ |
---|
449 | /* fprintf(stderr,"true checkopt(opt=%s,match=%s,param=%s)\n", sopt, smatch, *sparam); */ |
---|
450 | /* } */ |
---|
451 | return result; |
---|
452 | } |
---|
453 | |
---|
454 | |
---|
455 | #define kMaxwhichlist 50 |
---|
456 | |
---|
457 | /* global for readopt(), main() */ |
---|
458 | boolean chooseall = false, quietly = false, gotinputfile = false, |
---|
459 | listonly = false, closeout = false, verbose = false, |
---|
460 | manyout = false, dolower = false, doupper = false, doreverse= false, |
---|
461 | askout = true, dopipe= false, interleaved = false; |
---|
462 | short nfile = 0, iwhichlist=0, nwhichlist = 0; |
---|
463 | short whichlist[kMaxwhichlist+1]; |
---|
464 | long whichSeq = 0, outform = kNoformat; |
---|
465 | char onamestore[128], *oname = onamestore; |
---|
466 | FILE *foo = NULL; |
---|
467 | |
---|
468 | void resetGlobals() |
---|
469 | /* need this when used from SIOW, as these globals are not reinited automatically |
---|
470 | between calls to local main() */ |
---|
471 | { |
---|
472 | chooseall = false; quietly = false; gotinputfile = false; |
---|
473 | listonly = false; closeout = false; verbose = false; |
---|
474 | manyout = false; dolower = false; doupper = false; doreverse= false; |
---|
475 | askout = true; dopipe= false; interleaved = false; |
---|
476 | nfile = 0; iwhichlist=0; nwhichlist = 0; |
---|
477 | whichSeq = 0; outform = kNoformat; |
---|
478 | oname = onamestore; |
---|
479 | foo = NULL; |
---|
480 | |
---|
481 | gPrettyInit(gPretty); |
---|
482 | } |
---|
483 | |
---|
484 | |
---|
485 | #define kOptOkay 1 |
---|
486 | #define kOptNone 0 |
---|
487 | |
---|
488 | int readopt( char *sopt) |
---|
489 | { |
---|
490 | char sparamstore[256], *sparam= sparamstore; |
---|
491 | short n, slen= strlen(sopt); |
---|
492 | |
---|
493 | /* fprintf(stderr,"readopt( %s) == ", sopt); */ |
---|
494 | |
---|
495 | if (*sopt == '?') { |
---|
496 | usage(); |
---|
497 | return kOptNone; /*? eOptionBad or kOptNone */ |
---|
498 | } |
---|
499 | |
---|
500 | else if (*sopt == '-') { |
---|
501 | |
---|
502 | char *cp= strchr(sopt,'='); |
---|
503 | *sparam= '\0'; |
---|
504 | if (cp) { |
---|
505 | strcpy(sparam, cp+1); |
---|
506 | *cp= 0; |
---|
507 | } |
---|
508 | |
---|
509 | if (checkopt( false, sopt, "-help", 2)) { |
---|
510 | usage(); |
---|
511 | return kOptNone; |
---|
512 | } |
---|
513 | |
---|
514 | if (checkopt( false, sopt, "-all", 2)) { |
---|
515 | whichSeq= 1; chooseall= true; |
---|
516 | return kOptOkay; |
---|
517 | } |
---|
518 | |
---|
519 | if (checkopt( false, sopt, "-colspace", 4)) { /* test before -c[ase] */ |
---|
520 | n= atoi( sparam); |
---|
521 | gPretty.spacer = n; |
---|
522 | return kOptOkay; |
---|
523 | } |
---|
524 | |
---|
525 | if (checkopt( true, sopt, "-caselower", 2)) { |
---|
526 | dolower= true; |
---|
527 | return kOptOkay; |
---|
528 | } |
---|
529 | if (checkopt( true, sopt, "-CASEUPPER", 2)) { |
---|
530 | doupper= true; |
---|
531 | return kOptOkay; |
---|
532 | } |
---|
533 | |
---|
534 | if (checkopt( false, sopt, "-pipe", 2)) { |
---|
535 | dopipe= true; askout= false; |
---|
536 | return kOptOkay; |
---|
537 | } |
---|
538 | |
---|
539 | if (checkopt( false, sopt, "-list", 2)) { |
---|
540 | listonly = true; askout = false; |
---|
541 | return kOptOkay; |
---|
542 | } |
---|
543 | |
---|
544 | if (checkopt( false, sopt, "-reverse", 2)) { |
---|
545 | doreverse = true; |
---|
546 | return kOptOkay; |
---|
547 | } |
---|
548 | |
---|
549 | if (checkopt( false, sopt, "-verbose", 2)) { |
---|
550 | verbose = true; |
---|
551 | return kOptOkay; |
---|
552 | } |
---|
553 | |
---|
554 | if (checkopt( false, sopt, "-match", 5)) { |
---|
555 | gPretty.domatch= true; |
---|
556 | if (*sparam >= ' ') gPretty.matchchar= *sparam; |
---|
557 | return kOptOkay; |
---|
558 | } |
---|
559 | if (checkopt( false, sopt, "-degap", 4)) { |
---|
560 | gPretty.degap= true; |
---|
561 | if (*sparam >= ' ') gPretty.gapchar= *sparam; |
---|
562 | return kOptOkay; |
---|
563 | } |
---|
564 | |
---|
565 | if (checkopt( false, sopt, "-interline", 4)) { |
---|
566 | gPretty.interline= atoi( sparam); |
---|
567 | return kOptOkay; |
---|
568 | } |
---|
569 | |
---|
570 | if (checkopt( false, sopt, "-item", 2)) { |
---|
571 | char *cp = sparam; |
---|
572 | nwhichlist= 0; |
---|
573 | whichlist[0]= 0; |
---|
574 | if (*cp == 0) cp= sopt+2; /* compatible w/ old way */ |
---|
575 | do { |
---|
576 | while (*cp!=0 && !rs_isdigit(*cp)) cp++; |
---|
577 | if (*cp!=0) { |
---|
578 | n = atoi( cp); |
---|
579 | whichlist[nwhichlist++]= n; |
---|
580 | while (*cp!=0 && rs_isdigit(*cp)) cp++; |
---|
581 | } |
---|
582 | } while (*cp!=0 && n>0 && nwhichlist<kMaxwhichlist); |
---|
583 | whichlist[nwhichlist++]= 0; /* 0 == stopsign for loop */ |
---|
584 | whichSeq= max(1,whichlist[0]); iwhichlist= 1; |
---|
585 | return kOptOkay; |
---|
586 | } |
---|
587 | |
---|
588 | if (checkopt( false, sopt, "-format", 5)) {/* -format=phylip, -f2, -form=phylip */ |
---|
589 | if (*sparam==0) { for (sparam= sopt+2; isalpha(*sparam); sparam++) ; } |
---|
590 | outform = parseformat( sparam); |
---|
591 | return kOptOkay; |
---|
592 | } |
---|
593 | if (checkopt( false, sopt, "-f", 2)) { /* compatible w/ -fphylip prior version */ |
---|
594 | if (*sparam==0) sparam= sopt+2; |
---|
595 | outform = parseformat( sparam); |
---|
596 | return kOptOkay; |
---|
597 | } |
---|
598 | |
---|
599 | if (checkopt( false, sopt, "-output", 3)) {/* -output=myseq */ |
---|
600 | if (*sparam==0) { for (sparam= sopt+3; isalpha(*sparam); sparam++) ; } |
---|
601 | strcpy( oname, sparam); |
---|
602 | foo = fopen( oname, "w"); |
---|
603 | if (!foo) { erralert(eFileCreate); return eFileCreate; } |
---|
604 | closeout = true; |
---|
605 | askout = false; |
---|
606 | return kOptOkay; |
---|
607 | } |
---|
608 | if (checkopt( false, sopt, "-o", 2)) { /* compatible w/ -omyseq prior version */ |
---|
609 | if (*sparam==0) sparam= sopt+2; |
---|
610 | strcpy( oname, sparam); |
---|
611 | foo = fopen( oname, "w"); |
---|
612 | if (!foo) { erralert(eFileCreate); return eFileCreate; } |
---|
613 | closeout = true; |
---|
614 | askout = false; |
---|
615 | return kOptOkay; |
---|
616 | } |
---|
617 | |
---|
618 | if (checkopt( false, sopt, "-width", 2)) { |
---|
619 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
620 | n= atoi( sparam); |
---|
621 | if (n>0) gPretty.seqwidth = n; |
---|
622 | return kOptOkay; |
---|
623 | } |
---|
624 | |
---|
625 | if (checkopt( false, sopt, "-tab", 4)) { |
---|
626 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
627 | n= atoi( sparam); |
---|
628 | gPretty.tab = n; |
---|
629 | return kOptOkay; |
---|
630 | } |
---|
631 | |
---|
632 | if (checkopt( false, sopt, "-gapcount", 4)) { |
---|
633 | gPretty.baseonlynum = false; |
---|
634 | /* if (*sparam >= ' ') gPretty.gapchar= *sparam; */ |
---|
635 | return kOptOkay; |
---|
636 | } |
---|
637 | if (checkopt( false, sopt, "-nointerleave", 8)) { |
---|
638 | gPretty.noleaves = true; |
---|
639 | return kOptOkay; |
---|
640 | } |
---|
641 | |
---|
642 | if (checkopt( false, sopt, "-nameleft", 7)) { |
---|
643 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
644 | n= atoi( sparam); |
---|
645 | if (n>0 && n<50) gPretty.namewidth = n; |
---|
646 | gPretty.nameleft= true; |
---|
647 | return kOptOkay; |
---|
648 | } |
---|
649 | if (checkopt( false, sopt, "-nameright", 7)) { |
---|
650 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
651 | n= atoi( sparam); |
---|
652 | if (n>0 && n<50) gPretty.namewidth = n; |
---|
653 | gPretty.nameright= true; |
---|
654 | return kOptOkay; |
---|
655 | } |
---|
656 | if (checkopt( false, sopt, "-nametop", 6)) { |
---|
657 | gPretty.nametop= true; |
---|
658 | return kOptOkay; |
---|
659 | } |
---|
660 | |
---|
661 | if (checkopt( false, sopt, "-numleft", 6)) { |
---|
662 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
663 | n= atoi( sparam); |
---|
664 | if (n>0 && n<50) gPretty.numwidth = n; |
---|
665 | gPretty.numleft= true; |
---|
666 | return kOptOkay; |
---|
667 | } |
---|
668 | if (checkopt( false, sopt, "-numright", 6)) { |
---|
669 | if (*sparam==0) { for (sparam= sopt+2; !rs_isdigit(*sparam) && *sparam!=0; sparam++) ; } |
---|
670 | n= atoi( sparam); |
---|
671 | if (n>0 && n<50) gPretty.numwidth = n; |
---|
672 | gPretty.numright= true; |
---|
673 | return kOptOkay; |
---|
674 | } |
---|
675 | |
---|
676 | if (checkopt( false, sopt, "-numtop", 6)) { |
---|
677 | gPretty.numtop= true; |
---|
678 | return kOptOkay; |
---|
679 | } |
---|
680 | if (checkopt( false, sopt, "-numbottom", 6)) { |
---|
681 | gPretty.numbot= true; |
---|
682 | return kOptOkay; |
---|
683 | } |
---|
684 | |
---|
685 | else { |
---|
686 | usage(); |
---|
687 | return eOptionBad; |
---|
688 | } |
---|
689 | } |
---|
690 | |
---|
691 | else { |
---|
692 | strcpy( inputfile, sopt); |
---|
693 | gotinputfile = (*inputfile != 0); |
---|
694 | nfile++; |
---|
695 | return kOptOkay; |
---|
696 | } |
---|
697 | |
---|
698 | /* return kOptNone; -- never here */ |
---|
699 | } |
---|
700 | |
---|
701 | |
---|
702 | |
---|
703 | |
---|
704 | /* this program suffers some as it tries to be a quiet translator pipe |
---|
705 | _and_ a noisy user interactor |
---|
706 | */ |
---|
707 | |
---|
708 | /* return is best for SIOW, okay for others */ |
---|
709 | #ifdef SIOW |
---|
710 | #define Exit(a) return(a) |
---|
711 | siow_main( int argc, char *argv[]) |
---|
712 | |
---|
713 | #else |
---|
714 | #define Exit(a) exit(a) |
---|
715 | |
---|
716 | main( int argc, char *argv[]) |
---|
717 | #endif |
---|
718 | { |
---|
719 | boolean closein = false; |
---|
720 | short ifile, nseq, atseq, format, err = 0, seqtype = kDNA, |
---|
721 | nlines, seqout = 0, phylvers = 2; |
---|
722 | long i, skiplines, seqlen, seqlen0; |
---|
723 | unsigned long checksum= 0, checkall= 0; |
---|
724 | char *seq, *cp, *firstseq = NULL, *seqlist, *progname, tempname[256]; |
---|
725 | char seqid[256], *seqidptr = seqid; |
---|
726 | char stempstore[256], *stemp = stempstore; |
---|
727 | FILE *ftmp, *fin, *fout; |
---|
728 | long outindexmax= 0, noutindex= 0, *outindex = NULL; |
---|
729 | |
---|
730 | #define exit_main(err) { \ |
---|
731 | if (closeout) fclose(fout); \ |
---|
732 | if (closein) fclose(fin); \ |
---|
733 | if (*tempname!=0) remove(tempname);\ |
---|
734 | Exit(err); } |
---|
735 | |
---|
736 | #define indexout() if (interleaved) {\ |
---|
737 | if (noutindex>=outindexmax) {\ |
---|
738 | outindexmax= noutindex + 20;\ |
---|
739 | outindex= (long*) realloc(outindex, sizeof(long)*outindexmax);\ |
---|
740 | if (outindex==NULL) { err= eMemFull; erralert(err); exit_main(err); }\ |
---|
741 | }\ |
---|
742 | outindex[noutindex++]= ftell(fout);\ |
---|
743 | } |
---|
744 | |
---|
745 | |
---|
746 | resetGlobals(); |
---|
747 | foo = stdout; |
---|
748 | progname = argv[0]; |
---|
749 | *oname = 0; |
---|
750 | *tempname = 0; |
---|
751 | /* initialize gPretty ?? -- done in header */ |
---|
752 | |
---|
753 | for (i=1; i < argc; i++) { |
---|
754 | err= readopt( argv[i]); |
---|
755 | if (err <= 0) exit_main(err); |
---|
756 | } |
---|
757 | |
---|
758 | /* pipe input from stdin !? */ |
---|
759 | if (dopipe && !gotinputfile) { |
---|
760 | int c; |
---|
761 | tmpnam(tempname); |
---|
762 | inputfile = tempname; |
---|
763 | ftmp = fopen( inputfile, "w"); |
---|
764 | if (!ftmp) { erralert(eFileCreate); exit_main(eFileCreate); } |
---|
765 | while ((c = getc(stdin)) != EOF) fputc(c, ftmp); |
---|
766 | fclose(ftmp); |
---|
767 | gotinputfile= true; |
---|
768 | } |
---|
769 | |
---|
770 | quietly = (dopipe || (gotinputfile && (listonly || whichSeq != 0))); |
---|
771 | |
---|
772 | if (verbose || (!quietly && !gotinputfile)) fprintf( stderr, title); |
---|
773 | ifile = 1; |
---|
774 | |
---|
775 | /* UI: Choose output */ |
---|
776 | if (askout && !closeout && !quietly) { |
---|
777 | askout = false; |
---|
778 | fprintf(stderr,"\nName of output file (?=help, defaults to display): \n"); |
---|
779 | gets(oname= onamestore); |
---|
780 | skipwhitespace(oname); |
---|
781 | if (*oname == '?') { usage(); exit_main(0); } |
---|
782 | else if (*oname != 0) { |
---|
783 | closeout = true; |
---|
784 | foo = fopen( oname, "w"); |
---|
785 | if (!foo) { erralert(eFileCreate); exit_main(eFileCreate); } |
---|
786 | } |
---|
787 | } |
---|
788 | |
---|
789 | fout = foo; |
---|
790 | if (outform == kNoformat) outform = chooseFormat(quietly); |
---|
791 | |
---|
792 | /* set up formats ... */ |
---|
793 | switch (outform) { |
---|
794 | case kPhylip2: |
---|
795 | interleaved= false; |
---|
796 | phylvers = 2; |
---|
797 | outform = kPhylip; |
---|
798 | break; |
---|
799 | |
---|
800 | case kPhylip4: |
---|
801 | interleaved= true; |
---|
802 | phylvers = 4; |
---|
803 | outform = kPhylip; |
---|
804 | break; |
---|
805 | |
---|
806 | case kMSF: |
---|
807 | case kPAUP: |
---|
808 | interleaved= true; |
---|
809 | break; |
---|
810 | |
---|
811 | case kPretty: |
---|
812 | gPretty.isactive= true; |
---|
813 | interleaved= true; |
---|
814 | break; |
---|
815 | |
---|
816 | } |
---|
817 | |
---|
818 | if (gPretty.isactive && gPretty.noleaves) interleaved= false; |
---|
819 | if (interleaved) { |
---|
820 | fout = ftmp = tmpfile(); |
---|
821 | outindexmax= 30; noutindex= 0; |
---|
822 | outindex = (long*) malloc(outindexmax*sizeof(long)); |
---|
823 | if (outindex==NULL) { err= eMemFull; erralert(err); exit_main(err); } |
---|
824 | } |
---|
825 | |
---|
826 | /* big loop over all input files */ |
---|
827 | do { |
---|
828 | /* select next input file */ |
---|
829 | gotinputfile = (*tempname != 0); |
---|
830 | while ((ifile < argc) && (!gotinputfile)) { |
---|
831 | if (*argv[ifile] != '-') { |
---|
832 | strcpy( inputfile, argv[ifile]); |
---|
833 | gotinputfile = (*inputfile != 0); |
---|
834 | --nfile; |
---|
835 | } |
---|
836 | ifile++; |
---|
837 | } |
---|
838 | |
---|
839 | while (!gotinputfile) { |
---|
840 | fprintf(stderr,"\nName an input sequence or -option: \n"); |
---|
841 | inputfile= inputfilestore; |
---|
842 | |
---|
843 | gets(stemp= stempstore); |
---|
844 | if (*stemp==0) goto fini; /* !! need this to finish work during interactive use */ |
---|
845 | stemp= strtok(stempstore, " \n\r\t"); |
---|
846 | while (stemp) { |
---|
847 | err= readopt( stemp); /* will read inputfile if it exists */ |
---|
848 | if (err<0) exit_main(err); |
---|
849 | stemp= strtok( NULL, " \n\r\t"); |
---|
850 | } |
---|
851 | } |
---|
852 | /* thanks to AJB@UK.AC.DARESBURY.DLVH for this PHYLIP3 fix: */ |
---|
853 | /* head for end (interleave if needed) */ |
---|
854 | if (*inputfile == 0) break; |
---|
855 | |
---|
856 | format = seqFileFormat( inputfile, &skiplines, &err); |
---|
857 | |
---|
858 | if (err == 0) { |
---|
859 | #ifdef NCBI |
---|
860 | if (format == kASNseqentry || format == kASNseqset) |
---|
861 | seqlist = listASNSeqs( inputfile, skiplines, format, &nseq, &err); |
---|
862 | else |
---|
863 | #endif |
---|
864 | seqlist = listSeqs( inputfile, skiplines, format, &nseq, &err); |
---|
865 | } |
---|
866 | |
---|
867 | if (err != 0) |
---|
868 | erralert(err); |
---|
869 | |
---|
870 | else if (listonly) { |
---|
871 | dumpSeqList(seqlist,format); |
---|
872 | free( seqlist); |
---|
873 | } |
---|
874 | |
---|
875 | else { |
---|
876 | /* choose whichSeq if needed */ |
---|
877 | if (nseq == 1 || chooseall || (quietly && whichSeq == 0)) { |
---|
878 | chooseall= true; |
---|
879 | whichSeq = 1; |
---|
880 | quietly = true; /* no loop */ |
---|
881 | } |
---|
882 | else if (whichSeq > nseq && quietly) { |
---|
883 | erralert(eItemNotFound); |
---|
884 | err= eItemNotFound; |
---|
885 | } |
---|
886 | else if (whichSeq > nseq || !quietly) { |
---|
887 | dumpSeqList(seqlist, format); |
---|
888 | fprintf(stderr,"\nChoose a sequence (# or All): \n"); |
---|
889 | gets(stemp= stempstore); |
---|
890 | skipwhitespace(stemp); |
---|
891 | if (to_lower(*stemp) == 'a') { |
---|
892 | chooseall= true; |
---|
893 | whichSeq = 1; |
---|
894 | quietly = true; /* !? this means we don't ask for another file |
---|
895 | as well as no more whichSeqs... */ |
---|
896 | } |
---|
897 | else if (rs_isdigit(*stemp)) whichSeq= atol(stemp); |
---|
898 | else whichSeq= 1; /* default */ |
---|
899 | } |
---|
900 | free( seqlist); |
---|
901 | |
---|
902 | if (false /*chooseall*/) { /* this isn't debugged yet...*/ |
---|
903 | fin = fopen(inputfile, "r"); |
---|
904 | closein= true; |
---|
905 | } |
---|
906 | |
---|
907 | while (whichSeq > 0 && whichSeq <= nseq) { |
---|
908 | /* need to open multiple output files ? */ |
---|
909 | manyout = ((chooseall || nwhichlist>1) && nseq > 1 |
---|
910 | && (outform == kPlain || outform == kGCG)); |
---|
911 | if (manyout) { |
---|
912 | if ( whichSeq == 1 ) erralert(eOneFormat); |
---|
913 | else if (closeout) { |
---|
914 | sprintf( stemp,"%s_%ld", oname, whichSeq); |
---|
915 | freopen( stemp, "w", fout); |
---|
916 | fprintf( stderr,"Writing sequence %ld to file %s\n", whichSeq, stemp); |
---|
917 | } |
---|
918 | } |
---|
919 | |
---|
920 | if (closein) { |
---|
921 | /* !! this fails... skips most seqs... */ |
---|
922 | /* !! in sequential read, must count seqs already read from whichSeq ... */ |
---|
923 | /* need major revision of ureadseq before we can do this */ |
---|
924 | atseq= whichSeq-1; |
---|
925 | seqidptr= seqid; |
---|
926 | seq = readSeqFp( whichSeq, fin, skiplines, format, |
---|
927 | &seqlen, &atseq, &err, seqidptr); |
---|
928 | skiplines= 0; |
---|
929 | } |
---|
930 | else { |
---|
931 | atseq= 0; |
---|
932 | seqidptr= seqid; |
---|
933 | #ifdef NCBI |
---|
934 | if (format == kASNseqentry || format == kASNseqset) { |
---|
935 | seqidptr= NULL; |
---|
936 | seq = readASNSeq( whichSeq, inputfile, skiplines, format, |
---|
937 | &seqlen, &atseq, &err, &seqidptr); |
---|
938 | } |
---|
939 | else |
---|
940 | #endif |
---|
941 | seq = readSeq( whichSeq, inputfile, skiplines, format, |
---|
942 | &seqlen, &atseq, &err, seqidptr); |
---|
943 | } |
---|
944 | |
---|
945 | |
---|
946 | if (gPretty.degap) { |
---|
947 | char *newseq; |
---|
948 | long newlen; |
---|
949 | newseq= compressSeq( gPretty.gapchar, seq, seqlen, &newlen); |
---|
950 | if (newseq) { |
---|
951 | free(seq); seq= newseq; seqlen= newlen; |
---|
952 | } |
---|
953 | } |
---|
954 | |
---|
955 | if (outform == kMSF) checksum= GCGchecksum(seq, seqlen, &checkall); |
---|
956 | else if (verbose) checksum= seqchecksum(seq, seqlen, &checkall); |
---|
957 | if (verbose) |
---|
958 | fprintf( stderr, "Sequence %ld, length= %ld, checksum= %lX, format= %s, id= %s\n", |
---|
959 | whichSeq, seqlen, checksum, formatstr(format), seqidptr); |
---|
960 | |
---|
961 | if (err != 0) erralert(err); |
---|
962 | else { |
---|
963 | /* format fixes that writeseq doesn't do */ |
---|
964 | switch (outform) { |
---|
965 | case kPIR: |
---|
966 | if (seqout == 0) fprintf( foo,"\\\\\\\n"); |
---|
967 | break; |
---|
968 | case kASN1: |
---|
969 | if (seqout == 0) fprintf( foo, kASN1headline); |
---|
970 | break; |
---|
971 | |
---|
972 | case kPhylip: |
---|
973 | if (seqout == 0) { |
---|
974 | if (!interleaved) { /* bug, nseq is for 1st infile only */ |
---|
975 | if (chooseall) i= nseq; else i=1; |
---|
976 | if (phylvers >= 4) fprintf(foo," %ld %ld\n", i, seqlen); |
---|
977 | else fprintf(foo," %ld %ld YF\n", i, seqlen); |
---|
978 | } |
---|
979 | seqlen0 = seqlen; |
---|
980 | } |
---|
981 | else if (seqlen != seqlen0) { |
---|
982 | erralert(eUnequalSize); |
---|
983 | if (seqlen < seqlen0) seq = (char *)realloc(seq, seqlen0); |
---|
984 | for (i=seqlen; i<seqlen0; i++) seq[i]= gPretty.gapchar; |
---|
985 | seqlen = seqlen0; |
---|
986 | seq[seqlen] = 0; |
---|
987 | } |
---|
988 | break; |
---|
989 | |
---|
990 | case kPAUP: |
---|
991 | if (seqout == 0) { |
---|
992 | seqtype= getseqtype(seq, seqlen); |
---|
993 | seqlen0 = seqlen; |
---|
994 | } |
---|
995 | else if (seqlen != seqlen0) { |
---|
996 | erralert(eUnequalSize); |
---|
997 | if (seqlen < seqlen0) seq = (char *)realloc(seq, seqlen0); |
---|
998 | for (i=seqlen; i<seqlen0; i++) seq[i]= gPretty.gapchar; |
---|
999 | seqlen = seqlen0; |
---|
1000 | seq[seqlen] = 0; |
---|
1001 | } |
---|
1002 | break; |
---|
1003 | |
---|
1004 | } |
---|
1005 | |
---|
1006 | if (doupper) |
---|
1007 | for (i = 0; i<seqlen; i++) seq[i] = to_upper(seq[i]); |
---|
1008 | else if (dolower) |
---|
1009 | for (i = 0; i<seqlen; i++) seq[i] = to_lower(seq[i]); |
---|
1010 | |
---|
1011 | if (outform==kPhylip){ |
---|
1012 | for (i = 0; i<seqlen; i++) if (seq[i] == '.') seq[i] = '?'; |
---|
1013 | } |
---|
1014 | |
---|
1015 | if (doreverse) { |
---|
1016 | long j, k; |
---|
1017 | char ctemp; |
---|
1018 | for (j=0, k=seqlen-1; j <= k; j++, k--) { |
---|
1019 | ctemp = compl[seq[j] - ' ']; |
---|
1020 | seq[j] = compl[seq[k] - ' ']; |
---|
1021 | seq[k] = ctemp; |
---|
1022 | } |
---|
1023 | } |
---|
1024 | |
---|
1025 | if ((gPretty.isactive || outform==kPAUP) && gPretty.domatch && firstseq != NULL) { |
---|
1026 | for (i=0; i<seqlen; i++){ |
---|
1027 | if (seq[i] == gPretty.matchchar) seq[i] = 'o'; |
---|
1028 | if (seq[i]==firstseq[i]) seq[i]= gPretty.matchchar; |
---|
1029 | } |
---|
1030 | } |
---|
1031 | |
---|
1032 | |
---|
1033 | if (gPretty.isactive && gPretty.numtop && seqout == 0) { |
---|
1034 | gPretty.numline = 1; |
---|
1035 | indexout(); |
---|
1036 | (void) writeSeq( fout, seq, seqlen, outform, seqidptr); |
---|
1037 | gPretty.numline = 2; |
---|
1038 | indexout(); |
---|
1039 | (void) writeSeq( fout, seq, seqlen, outform, seqidptr); |
---|
1040 | gPretty.numline = 0; |
---|
1041 | } |
---|
1042 | |
---|
1043 | indexout(); |
---|
1044 | nlines = writeSeq( fout, seq, seqlen, outform, seqidptr); |
---|
1045 | seqout++; |
---|
1046 | } |
---|
1047 | |
---|
1048 | if ((gPretty.isactive || outform==kPAUP) && gPretty.domatch && firstseq == NULL) { |
---|
1049 | firstseq= seq; |
---|
1050 | seq = NULL; |
---|
1051 | } |
---|
1052 | else if (seq!=NULL) { free(seq); seq = NULL; } |
---|
1053 | |
---|
1054 | #ifdef NCBI |
---|
1055 | if ( (format == kASNseqentry || format == kASNseqset) |
---|
1056 | && seqidptr && seqidptr!= seqid) |
---|
1057 | free(seqidptr); |
---|
1058 | #endif |
---|
1059 | if (chooseall) whichSeq++; |
---|
1060 | else if (iwhichlist<nwhichlist) whichSeq= whichlist[iwhichlist++]; |
---|
1061 | else whichSeq= 0; |
---|
1062 | } |
---|
1063 | if (closein) { fclose(fin); closein= false; } |
---|
1064 | } |
---|
1065 | whichSeq = 0; |
---|
1066 | } while (nfile > 0 || !quietly); |
---|
1067 | |
---|
1068 | |
---|
1069 | fini: |
---|
1070 | if (firstseq) { free(firstseq); firstseq= NULL; } |
---|
1071 | if (err || listonly) exit_main(err); |
---|
1072 | |
---|
1073 | if (gPretty.isactive && gPretty.numbot) { |
---|
1074 | gPretty.numline = 2; |
---|
1075 | indexout(); |
---|
1076 | (void) writeSeq( fout, seq, seqlen, outform, seqidptr); |
---|
1077 | gPretty.numline = 1; |
---|
1078 | indexout(); |
---|
1079 | (void) writeSeq( fout, seq, seqlen, outform, seqidptr); |
---|
1080 | gPretty.numline = 0; |
---|
1081 | } |
---|
1082 | |
---|
1083 | if (outform == kMSF) { |
---|
1084 | if (*oname) cp= oname; else cp= inputfile; |
---|
1085 | fprintf(foo,"\n %s MSF: %ld Type: N January 01, 1776 12:00 Check: %lu ..\n\n", |
---|
1086 | cp, seqlen, checkall); |
---|
1087 | } |
---|
1088 | |
---|
1089 | if (outform == kPAUP) { |
---|
1090 | fprintf(foo,"#NEXUS\n"); |
---|
1091 | if (*oname) cp= oname; else cp= inputfile; |
---|
1092 | fprintf(foo,"[%s -- data title]\n\n", cp); |
---|
1093 | /* ! now have header lines for each sequence... put them before "begin data;... */ |
---|
1094 | } |
---|
1095 | |
---|
1096 | if (outform==kPhylip && interleaved) { |
---|
1097 | if (phylvers >= 4) fprintf(foo," %d %ld\n", seqout, seqlen); |
---|
1098 | else fprintf(foo," %d %ld YF\n", seqout, seqlen); |
---|
1099 | } |
---|
1100 | |
---|
1101 | if (interleaved) { |
---|
1102 | /* interleave species lines in true output */ |
---|
1103 | /* nlines is # lines / sequence */ |
---|
1104 | short iline, j, leaf, iseq; |
---|
1105 | char *s = stempstore; |
---|
1106 | |
---|
1107 | indexout(); noutindex--; /* mark eof */ |
---|
1108 | |
---|
1109 | for (leaf=0; leaf<nlines; leaf++) { |
---|
1110 | if (outform == kMSF && leaf == 1) { |
---|
1111 | fputs("//\n\n", foo); |
---|
1112 | } |
---|
1113 | if (outform == kPAUP && leaf==1) { |
---|
1114 | switch (seqtype) { |
---|
1115 | case kDNA : cp= "dna"; break; |
---|
1116 | case kRNA : cp= "rna"; break; |
---|
1117 | case kNucleic : cp= "dna"; break; |
---|
1118 | case kAmino : cp= "protein"; break; |
---|
1119 | case kOtherSeq: cp= "dna"; break; |
---|
1120 | } |
---|
1121 | fprintf(foo,"\nbegin data;\n"); |
---|
1122 | fprintf(foo," dimensions ntax=%d nchar=%ld;\n", seqout, seqlen); |
---|
1123 | /* fix by Ralf Westram (ARB): '-' means 'gap' ('.' means 'missing') */ |
---|
1124 | fprintf(foo," format datatype=%s interleave=yes missing=. gap=%c", cp, gPretty.gapchar); |
---|
1125 | if (gPretty.domatch) fprintf(foo," matchchar=%c", gPretty.matchchar); |
---|
1126 | fprintf(foo,";\n matrix\n"); |
---|
1127 | } |
---|
1128 | |
---|
1129 | for (iseq=0; iseq<noutindex; iseq++) { |
---|
1130 | fseek(ftmp, outindex[iseq], 0); |
---|
1131 | for (iline=0; iline<=leaf; iline++) |
---|
1132 | if (!fgets(s, 256, ftmp)) *s= 0; |
---|
1133 | if (ftell(ftmp) <= outindex[iseq+1]) |
---|
1134 | fputs( s, foo); |
---|
1135 | } |
---|
1136 | |
---|
1137 | for (j=0; j<gPretty.interline; j++) |
---|
1138 | fputs( "\n", foo); /* some want spacer line */ |
---|
1139 | } |
---|
1140 | fclose(ftmp); /* tmp disappears */ |
---|
1141 | fout= foo; |
---|
1142 | } |
---|
1143 | |
---|
1144 | if (outform == kASN1) fprintf( foo, "} }\n"); |
---|
1145 | if (outform == kPAUP) fprintf( foo,";\n end;\n"); |
---|
1146 | |
---|
1147 | if (outindex != NULL) free(outindex); |
---|
1148 | exit_main(0); |
---|
1149 | } |
---|
1150 | |
---|
1151 | |
---|