| 1 | /* ------------- File format converting subroutine ------------- */ |
|---|
| 2 | |
|---|
| 3 | #include <stdio.h> |
|---|
| 4 | #include <stdlib.h> |
|---|
| 5 | #include "convert.h" |
|---|
| 6 | #include "global.h" |
|---|
| 7 | |
|---|
| 8 | /* ------------------------------------------------------------- |
|---|
| 9 | * Function convert(). |
|---|
| 10 | * For given input file and type, convert the file |
|---|
| 11 | * to desired out type and save the result in |
|---|
| 12 | * the out file. |
|---|
| 13 | */ |
|---|
| 14 | void convert(inf, outf, intype, outype) |
|---|
| 15 | char *inf, *outf; |
|---|
| 16 | int intype, outype; |
|---|
| 17 | { |
|---|
| 18 | int dd; /* copy stdin to outfile after first line */ |
|---|
| 19 | dd = 0; |
|---|
| 20 | if (outype == PHYLIP2){ |
|---|
| 21 | outype = PHYLIP; |
|---|
| 22 | dd = 1; |
|---|
| 23 | } |
|---|
| 24 | |
|---|
| 25 | if(Cmpstr(inf, outf)==EQ) |
|---|
| 26 | error(45, "Input file and output file must be different file.\n"); |
|---|
| 27 | if(intype==GENBANK&&outype==MACKE) { |
|---|
| 28 | genbank_to_macke(inf, outf); |
|---|
| 29 | } else if(intype==GENBANK&&outype==GENBANK) { |
|---|
| 30 | genbank_to_genbank(inf, outf); |
|---|
| 31 | } else if(intype==GENBANK&&outype==PAUP) { |
|---|
| 32 | to_paup(inf, outf, GENBANK); |
|---|
| 33 | } else if(intype==GENBANK&&outype==PHYLIP) { |
|---|
| 34 | to_phylip(inf, outf, GENBANK,dd); |
|---|
| 35 | } else if(intype==GENBANK&&outype==PROTEIN) { |
|---|
| 36 | error(69, "Sorry, cannot convert from GENBANK to SWISSPROT, Exit"); |
|---|
| 37 | } else if(intype==GENBANK&&outype==EMBL) { |
|---|
| 38 | genbank_to_embl(inf, outf); |
|---|
| 39 | } else if(intype==GENBANK&&outype==PRINTABLE) { |
|---|
| 40 | to_printable(inf, outf, GENBANK); |
|---|
| 41 | } else if(intype==GENBANK&&outype==ALMA) { |
|---|
| 42 | genbank_to_alma(inf, outf); |
|---|
| 43 | } else if(intype==MACKE&&outype==GENBANK) { |
|---|
| 44 | macke_to_genbank(inf, outf); |
|---|
| 45 | } else if(intype==MACKE&&outype==PAUP) { |
|---|
| 46 | to_paup(inf, outf, MACKE); |
|---|
| 47 | } else if(intype==MACKE&&outype==PHYLIP) { |
|---|
| 48 | to_phylip(inf, outf, MACKE,dd); |
|---|
| 49 | } else if(intype==MACKE&&outype==PROTEIN) { |
|---|
| 50 | macke_to_embl(inf, outf); |
|---|
| 51 | } else if(intype==MACKE&&outype==EMBL) { |
|---|
| 52 | macke_to_embl(inf, outf); |
|---|
| 53 | } else if(intype==MACKE&&outype==PRINTABLE) { |
|---|
| 54 | to_printable(inf, outf, MACKE); |
|---|
| 55 | } else if(intype==MACKE&&outype==ALMA) { |
|---|
| 56 | macke_to_alma(inf, outf); |
|---|
| 57 | } else if(intype==PAUP&&outype==GENBANK) { |
|---|
| 58 | error(6, "Sorry, cannot convert from Paup to GENBANK, Exit.\n"); |
|---|
| 59 | } else if(intype==PAUP&&outype==MACKE) { |
|---|
| 60 | error(81, "Sorry, cannot convert from Paup to AE2, Exit.\n"); |
|---|
| 61 | } else if(intype==PAUP&&outype==PHYLIP) { |
|---|
| 62 | error(8, "Sorry, cannot convert from Paup to Phylip, Exit.\n"); |
|---|
| 63 | } else if(intype==PAUP&&outype==PROTEIN) { |
|---|
| 64 | error(71, "Sorry, cannot convert from Paup to SWISSPROT, Exit."); |
|---|
| 65 | } else if(intype==PAUP&&outype==EMBL) { |
|---|
| 66 | error(78, "Sorry, cannot convert from Paup to SWISSPROT, Exit."); |
|---|
| 67 | } else if(intype==PHYLIP&&outype==GENBANK) { |
|---|
| 68 | error(85, "Sorry, cannot convert from Phylip to GenBank, Exit.\n"); |
|---|
| 69 | } else if(intype==PHYLIP&&outype==MACKE) { |
|---|
| 70 | error(88, "Sorry, cannot convert from Phylip to AE2, Exit.\n"); |
|---|
| 71 | } else if(intype==PHYLIP&&outype==PAUP) { |
|---|
| 72 | error(89, "Sorry, cannot convert from Phylip to Paup, Exit.\n"); |
|---|
| 73 | } else if(intype==PHYLIP&&outype==PROTEIN) { |
|---|
| 74 | error(72, "Sorry, cannot convert from Phylip to SWISSPROT, Exit.\n"); |
|---|
| 75 | } else if(intype==PHYLIP&&outype==EMBL) { |
|---|
| 76 | error(79, "Sorry, cannot convert from Phylip to SWISSPROT, Exit.\n"); |
|---|
| 77 | } else if(intype==PROTEIN&&outype==MACKE) { |
|---|
| 78 | embl_to_macke(inf, outf, PROTEIN); |
|---|
| 79 | } else if(intype==PROTEIN&&outype==GENBANK) { |
|---|
| 80 | error(78, "GenBank doesn't maintain protein data."); |
|---|
| 81 | /* embl_to_genbank(inf, outf, PROTEIN); */ |
|---|
| 82 | /* protein_to_genbank(inf, outf); */ |
|---|
| 83 | } else if(intype==PROTEIN&&outype==PAUP) { |
|---|
| 84 | to_paup(inf, outf, PROTEIN); |
|---|
| 85 | } else if(intype==PROTEIN&&outype==PHYLIP) { |
|---|
| 86 | to_phylip(inf, outf, PROTEIN,dd); |
|---|
| 87 | } else if(intype==PROTEIN&&outype==EMBL) { |
|---|
| 88 | error(58, "EMBL doesn't maintain protein data."); |
|---|
| 89 | /* embl_to_embl(inf, outf, PROTEIN); */ |
|---|
| 90 | } else if(intype==PROTEIN&&outype==PRINTABLE) { |
|---|
| 91 | to_printable(inf, outf, PROTEIN); |
|---|
| 92 | } else if(intype==EMBL&&outype==EMBL) { |
|---|
| 93 | embl_to_embl(inf, outf); |
|---|
| 94 | } else if(intype==EMBL&&outype==GENBANK) { |
|---|
| 95 | embl_to_genbank(inf, outf); |
|---|
| 96 | } else if(intype==EMBL&&outype==MACKE) { |
|---|
| 97 | embl_to_macke(inf, outf, EMBL); |
|---|
| 98 | } else if(intype==EMBL&&outype==PROTEIN) { |
|---|
| 99 | embl_to_embl(inf, outf); |
|---|
| 100 | } else if(intype==EMBL&&outype==PAUP) { |
|---|
| 101 | to_paup(inf, outf, EMBL); |
|---|
| 102 | } else if(intype==EMBL&&outype==PHYLIP) { |
|---|
| 103 | to_phylip(inf, outf, EMBL,dd); |
|---|
| 104 | } else if(intype==EMBL&&outype==PRINTABLE) { |
|---|
| 105 | to_printable(inf, outf, EMBL); |
|---|
| 106 | } else if(intype==EMBL&&outype==ALMA) { |
|---|
| 107 | embl_to_alma(inf, outf); |
|---|
| 108 | } else if(intype==ALMA&&outype==MACKE) { |
|---|
| 109 | alma_to_macke(inf, outf); |
|---|
| 110 | } else if(intype==ALMA&&outype==GENBANK) { |
|---|
| 111 | alma_to_genbank(inf, outf); |
|---|
| 112 | } else if(intype==ALMA&&outype==PAUP) { |
|---|
| 113 | to_paup(inf, outf, ALMA); |
|---|
| 114 | } else if(intype==ALMA&&outype==PHYLIP) { |
|---|
| 115 | to_phylip(inf, outf, ALMA,dd); |
|---|
| 116 | } else if(intype==ALMA&&outype==PRINTABLE) { |
|---|
| 117 | to_printable(inf, outf, ALMA); |
|---|
| 118 | } else if(outype==GCG) { |
|---|
| 119 | to_gcg(intype, inf); |
|---|
| 120 | } else error(90, "Unidentified input type or output type, Exit\n"); |
|---|
| 121 | } |
|---|
| 122 | |
|---|
| 123 | /* ------------------- COMMON SUBROUTINES ----------------------- */ |
|---|
| 124 | |
|---|
| 125 | /* -------------------------------------------------------------- |
|---|
| 126 | * Function init(). |
|---|
| 127 | * Initialize data structure at the very beginning of running |
|---|
| 128 | * any program. |
|---|
| 129 | */ |
|---|
| 130 | void |
|---|
| 131 | init() { |
|---|
| 132 | |
|---|
| 133 | /* initialize macke format */ |
|---|
| 134 | data.macke.seqabbr = NULL; |
|---|
| 135 | data.macke.name = NULL; |
|---|
| 136 | data.macke.atcc = NULL; |
|---|
| 137 | data.macke.rna = NULL; |
|---|
| 138 | data.macke.date = NULL; |
|---|
| 139 | data.macke.nbk = NULL; |
|---|
| 140 | data.macke.acs = NULL; |
|---|
| 141 | data.macke.who = NULL; |
|---|
| 142 | data.macke.remarks = NULL; |
|---|
| 143 | data.macke.numofrem = 0; |
|---|
| 144 | data.macke.rna_or_dna = 'd'; |
|---|
| 145 | data.macke.journal = NULL; |
|---|
| 146 | data.macke.title = NULL; |
|---|
| 147 | data.macke.author = NULL; |
|---|
| 148 | data.macke.strain = NULL; |
|---|
| 149 | data.macke.subspecies = NULL; |
|---|
| 150 | /* initialize gnebank format */ |
|---|
| 151 | data.gbk.locus = NULL; |
|---|
| 152 | data.gbk.definition = NULL; |
|---|
| 153 | data.gbk.accession = NULL; |
|---|
| 154 | data.gbk.keywords = NULL; |
|---|
| 155 | data.gbk.source = NULL; |
|---|
| 156 | data.gbk.organism = NULL; |
|---|
| 157 | data.gbk.numofref = 0; |
|---|
| 158 | data.gbk.reference = NULL; |
|---|
| 159 | data.gbk.comments.orginf.exist = 0; |
|---|
| 160 | data.gbk.comments.orginf.source = NULL; |
|---|
| 161 | data.gbk.comments.orginf.cc = NULL; |
|---|
| 162 | data.gbk.comments.orginf.formname = NULL; |
|---|
| 163 | data.gbk.comments.orginf.nickname = NULL; |
|---|
| 164 | data.gbk.comments.orginf.commname = NULL; |
|---|
| 165 | data.gbk.comments.orginf.hostorg = NULL; |
|---|
| 166 | data.gbk.comments.seqinf.exist = 0; |
|---|
| 167 | data.gbk.comments.seqinf.RDPid = NULL; |
|---|
| 168 | data.gbk.comments.seqinf.gbkentry = NULL; |
|---|
| 169 | data.gbk.comments.seqinf.methods = NULL; |
|---|
| 170 | data.gbk.comments.seqinf.comp5 = ' '; |
|---|
| 171 | data.gbk.comments.seqinf.comp3 = ' '; |
|---|
| 172 | data.gbk.comments.others = NULL; |
|---|
| 173 | /* initialize paup format */ |
|---|
| 174 | data.paup.ntax = 0; |
|---|
| 175 | data.paup.equate = "~=.|><"; |
|---|
| 176 | data.paup.gap = '-'; |
|---|
| 177 | /* initial phylip data */ |
|---|
| 178 | /* initial embl data */ |
|---|
| 179 | data.embl.id = NULL; |
|---|
| 180 | data.embl.dateu = NULL; |
|---|
| 181 | data.embl.datec = NULL; |
|---|
| 182 | data.embl.description = NULL; |
|---|
| 183 | data.embl.os = NULL; |
|---|
| 184 | data.embl.accession = NULL; |
|---|
| 185 | data.embl.keywords = NULL; |
|---|
| 186 | data.embl.dr = NULL; |
|---|
| 187 | data.embl.numofref = 0; |
|---|
| 188 | data.embl.reference = NULL; |
|---|
| 189 | data.embl.comments.orginf.exist = 0; |
|---|
| 190 | data.embl.comments.orginf.source = NULL; |
|---|
| 191 | data.embl.comments.orginf.cc = NULL; |
|---|
| 192 | data.embl.comments.orginf.formname = NULL; |
|---|
| 193 | data.embl.comments.orginf.nickname = NULL; |
|---|
| 194 | data.embl.comments.orginf.commname = NULL; |
|---|
| 195 | data.embl.comments.orginf.hostorg = NULL; |
|---|
| 196 | data.embl.comments.seqinf.exist = 0; |
|---|
| 197 | data.embl.comments.seqinf.RDPid = NULL; |
|---|
| 198 | data.embl.comments.seqinf.gbkentry = NULL; |
|---|
| 199 | data.embl.comments.seqinf.methods = NULL; |
|---|
| 200 | data.embl.comments.seqinf.comp5 = ' '; |
|---|
| 201 | data.embl.comments.seqinf.comp3 = ' '; |
|---|
| 202 | data.embl.comments.others = NULL; |
|---|
| 203 | /* initial alma data */ |
|---|
| 204 | data.alma.id = NULL; |
|---|
| 205 | data.alma.filename = NULL; |
|---|
| 206 | data.alma.format = UNKNOWN; |
|---|
| 207 | data.alma.defgap = '-'; |
|---|
| 208 | data.alma.num_of_sequence = 0; |
|---|
| 209 | data.alma.sequence = NULL; |
|---|
| 210 | /* inital NBRF data format */ |
|---|
| 211 | data.nbrf.id = NULL; |
|---|
| 212 | data.nbrf.description = NULL; |
|---|
| 213 | /* initial sequence data */ |
|---|
| 214 | data.numofseq = 0; |
|---|
| 215 | data.seq_length = 0; |
|---|
| 216 | data.max = INITSEQ; |
|---|
| 217 | data.sequence = (char*)calloc(1,(unsigned)(sizeof(char)*INITSEQ+1)); |
|---|
| 218 | |
|---|
| 219 | data.ids = NULL; |
|---|
| 220 | data.seqs = NULL; |
|---|
| 221 | data.lengths = NULL; |
|---|
| 222 | data.allocated = 0; |
|---|
| 223 | } |
|---|
| 224 | /* -------------------------------------------------------------- |
|---|
| 225 | * Function init_seq_data(). |
|---|
| 226 | * Init. seq. data. |
|---|
| 227 | */ |
|---|
| 228 | void |
|---|
| 229 | init_seq_data() { |
|---|
| 230 | data.numofseq = 0; |
|---|
| 231 | data.seq_length = 0; |
|---|
| 232 | } |
|---|