| 1 | /* ------------------------------------------------------------ */ |
|---|
| 2 | /* */ |
|---|
| 3 | /* Format Conversion Program. */ |
|---|
| 4 | /* */ |
|---|
| 5 | /* Woese Lab., Dept. of Microbiology, UIUC */ |
|---|
| 6 | /* */ |
|---|
| 7 | /* ------------------------------------------------------------ */ |
|---|
| 8 | /* if using Dec. Unix, <sys/time.h> doesn't not exist */ |
|---|
| 9 | |
|---|
| 10 | #include <stdio.h> |
|---|
| 11 | #include <stdlib.h> |
|---|
| 12 | #include <ctype.h> |
|---|
| 13 | #include <string.h> |
|---|
| 14 | #include "convert.h" |
|---|
| 15 | #include "global.h" |
|---|
| 16 | |
|---|
| 17 | /* --------------------------------------------------------------- |
|---|
| 18 | * Main program: |
|---|
| 19 | * File conversion; convert from one file format to |
|---|
| 20 | * the other. eg. Genbank<-->Macke format. |
|---|
| 21 | */ |
|---|
| 22 | int main(int argc, char *argv[]) { |
|---|
| 23 | int intype, outtype; |
|---|
| 24 | /* int Lenstr(), file_exist(); */ |
|---|
| 25 | /* void error(), genbank_to_macke(), macke_to_genbank(); */ |
|---|
| 26 | /* void Getstr(), change_file_suffix(); */ |
|---|
| 27 | /* char *Dupstr(), *Reallocspace(); */ |
|---|
| 28 | /* char error_msg[LINENUM]; */ |
|---|
| 29 | char temp[LINENUM]; |
|---|
| 30 | char choice[LINENUM]; |
|---|
| 31 | |
|---|
| 32 | intype = outtype = UNKNOWN; |
|---|
| 33 | if(argc < 3 || argv[1][0] != '-') { |
|---|
| 34 | fprintf(stderr, |
|---|
| 35 | "---------------------------------------------------------------\n"); |
|---|
| 36 | fprintf(stderr, |
|---|
| 37 | "\n convert_aln - an alignment and file converter written by\n"); |
|---|
| 38 | fprintf(stderr, |
|---|
| 39 | " WenMin Kuan for the RDP database project. Please\n"); |
|---|
| 40 | fprintf(stderr, |
|---|
| 41 | " report errors or deficiencies to kuan@phylo.life.uiuc.edu\n"); |
|---|
| 42 | fprintf(stderr, |
|---|
| 43 | "\n Command line usage:\n\n"); |
|---|
| 44 | fprintf(stderr, |
|---|
| 45 | " $ convert_aln -infmt input_file -outfmt output_file, where \n"); |
|---|
| 46 | fprintf(stderr, |
|---|
| 47 | " infmt can be either GenBank, EMBL, AE2, SwissProt or ALMA\n"); |
|---|
| 48 | fprintf(stderr, |
|---|
| 49 | " outfmt GenBank, EMBL, AE2, PAUP, PHYLIP, GCG, Printable or ALMA\n\n"); |
|---|
| 50 | fprintf(stderr, |
|---|
| 51 | "---------------------------------------------------------------\n\n"); |
|---|
| 52 | fprintf(stderr, "Select input format (<CR> means default)\n\n"); |
|---|
| 53 | fprintf(stderr, " (1) GenBank [default]\n"); |
|---|
| 54 | fprintf(stderr, " (2) EMBL\n"); |
|---|
| 55 | fprintf(stderr, " (3) AE2\n"); |
|---|
| 56 | fprintf(stderr, " (4) SwissProt\n"); |
|---|
| 57 | fprintf(stderr, " (5) ALMA\n"); |
|---|
| 58 | fprintf(stderr, " (6) Quit\n"); |
|---|
| 59 | fprintf(stderr, " ? "); |
|---|
| 60 | Getstr(choice, LINENUM); |
|---|
| 61 | switch(choice[0]) { |
|---|
| 62 | case '1': intype = GENBANK; |
|---|
| 63 | break; |
|---|
| 64 | case '2': intype = EMBL; |
|---|
| 65 | break; |
|---|
| 66 | case '3': intype = MACKE; |
|---|
| 67 | break; |
|---|
| 68 | case '4': intype = PROTEIN; |
|---|
| 69 | break; |
|---|
| 70 | case '5': intype = ALMA; |
|---|
| 71 | break; |
|---|
| 72 | case '6': exit(0); |
|---|
| 73 | case '\0': intype = GENBANK; |
|---|
| 74 | break; /* default selection */ |
|---|
| 75 | default : error(16, "Unknown input file format, EXIT."); |
|---|
| 76 | } |
|---|
| 77 | argv=(char**)calloc(1,sizeof(char*)*5); |
|---|
| 78 | |
|---|
| 79 | fprintf(stderr, "\nInput file name? "); |
|---|
| 80 | Getstr(temp, LINENUM); |
|---|
| 81 | argv[2] = Dupstr(temp); |
|---|
| 82 | if(!file_exist(temp)) |
|---|
| 83 | error(77, "Input file not found, EXIT."); |
|---|
| 84 | |
|---|
| 85 | /* output file information */ |
|---|
| 86 | fprintf(stderr, "\nSelect output format (<CR> means default)\n\n"); |
|---|
| 87 | fprintf(stderr, " (1) GenBank\n"); |
|---|
| 88 | fprintf(stderr, " (2) EMBL\n"); |
|---|
| 89 | fprintf(stderr, " (3) AE2 [default]\n"); |
|---|
| 90 | fprintf(stderr, " (4) PAUP\n"); |
|---|
| 91 | fprintf(stderr, " (5) PHYLIP\n"); |
|---|
| 92 | fprintf(stderr, " (A) PHYLIP2 (insert stdin in first line)\n"); |
|---|
| 93 | fprintf(stderr, " (6) GCG\n"); |
|---|
| 94 | fprintf(stderr, " (7) Printable\n"); |
|---|
| 95 | fprintf(stderr, " (8) ALMA\n"); |
|---|
| 96 | fprintf(stderr, " (9) Quit\n"); |
|---|
| 97 | fprintf(stderr, " ? "); |
|---|
| 98 | Getstr(choice, LINENUM); |
|---|
| 99 | switch(choice[0]) { |
|---|
| 100 | case '1': outtype = GENBANK; |
|---|
| 101 | break; |
|---|
| 102 | case '2': outtype = EMBL; |
|---|
| 103 | break; |
|---|
| 104 | case '3': outtype = MACKE; |
|---|
| 105 | break; |
|---|
| 106 | case '4': outtype = PAUP; |
|---|
| 107 | break; |
|---|
| 108 | case '5': outtype = PHYLIP; |
|---|
| 109 | break; |
|---|
| 110 | case 'A': outtype = PHYLIP2; |
|---|
| 111 | break; |
|---|
| 112 | case '6': outtype = GCG; |
|---|
| 113 | break; |
|---|
| 114 | case '7': outtype = PRINTABLE; |
|---|
| 115 | break; |
|---|
| 116 | case '8': outtype = ALMA; |
|---|
| 117 | break; |
|---|
| 118 | case '9': exit(0); |
|---|
| 119 | case '\0': outtype = MACKE; |
|---|
| 120 | break; |
|---|
| 121 | default : error(66, "Unknown output file format, EXIT."); |
|---|
| 122 | } |
|---|
| 123 | change_file_suffix(argv[2], temp, outtype); |
|---|
| 124 | if(outtype!=GCG) { |
|---|
| 125 | fprintf(stderr, "\nOutput file name [%s]? ", temp); |
|---|
| 126 | Getstr(temp, LINENUM); |
|---|
| 127 | if(Lenstr(temp)==0) |
|---|
| 128 | change_file_suffix(argv[2], temp, outtype); |
|---|
| 129 | } |
|---|
| 130 | argv[4] = Dupstr(temp); |
|---|
| 131 | argc = 5; |
|---|
| 132 | } else { /* processing command line */ |
|---|
| 133 | /* input file */ |
|---|
| 134 | if(argv[1][1]=='G'||argv[1][1]=='g') |
|---|
| 135 | intype=GENBANK; |
|---|
| 136 | else if(argv[1][1]=='E'||argv[1][1]=='e') |
|---|
| 137 | intype=EMBL; |
|---|
| 138 | else if((argv[1][1]=='A'||argv[1][1]=='a') |
|---|
| 139 | &&(argv[1][2]=='E'|| argv[1][2]=='e')) |
|---|
| 140 | intype=MACKE; |
|---|
| 141 | else if(argv[1][1]=='S'||argv[1][1]=='s') |
|---|
| 142 | intype = PROTEIN; |
|---|
| 143 | else if((argv[1][1]=='A'||argv[1][1]=='a') |
|---|
| 144 | &&(argv[1][2]=='L'||argv[1][2]=='l')) |
|---|
| 145 | intype = ALMA; |
|---|
| 146 | else error(67, "UNKNOWN input file type, EXIT."); |
|---|
| 147 | |
|---|
| 148 | /* output file */ |
|---|
| 149 | if((argv[3][1]=='G'||argv[3][1]=='g') |
|---|
| 150 | &&(argv[3][2]=='E'||argv[3][2]=='e')) |
|---|
| 151 | outtype = GENBANK; |
|---|
| 152 | else if((argv[3][1]=='E'||argv[3][1]=='e')) |
|---|
| 153 | outtype = EMBL; |
|---|
| 154 | else if((argv[3][1]=='A'||argv[3][1]=='a') |
|---|
| 155 | &&(argv[3][2]=='E'||argv[3][2]=='e')) |
|---|
| 156 | outtype = MACKE; |
|---|
| 157 | else if((argv[3][1]=='P'||argv[3][1]=='p') |
|---|
| 158 | &&(argv[3][2]=='a'||argv[3][2]=='A')) |
|---|
| 159 | outtype = PAUP; |
|---|
| 160 | else if(!strncasecmp(argv[3]+1,"ph",2)){ |
|---|
| 161 | if (!strcasecmp(argv[3]+1,"phylip2")){ |
|---|
| 162 | outtype = PHYLIP2; |
|---|
| 163 | }else{ |
|---|
| 164 | outtype = PHYLIP; |
|---|
| 165 | } |
|---|
| 166 | }else if((argv[3][1]=='G'||argv[3][1]=='g') |
|---|
| 167 | &&(argv[3][2]=='C'||argv[3][2]=='c')) |
|---|
| 168 | outtype = GCG; |
|---|
| 169 | else if((argv[3][1]=='P'||argv[3][1]=='p') |
|---|
| 170 | &&(argv[3][2]=='R'||argv[3][2]=='r')) |
|---|
| 171 | outtype = PRINTABLE; |
|---|
| 172 | else if((argv[3][1]=='A'||argv[3][1]=='a') |
|---|
| 173 | &&(argv[3][2]=='L'||argv[3][2]=='l')) |
|---|
| 174 | outtype = ALMA; |
|---|
| 175 | else error(68, "UNKNOWN output file file, EXIT."); |
|---|
| 176 | } /* command line */ |
|---|
| 177 | |
|---|
| 178 | if(argc==4) { /* default output file */ |
|---|
| 179 | /* argv = (char**)Reallocspace(argv, sizeof(char*)*5); */ |
|---|
| 180 | const char **argv_new = calloc(sizeof(char*), 5); |
|---|
| 181 | memcpy(argv_new, argv, sizeof(char*)*4); |
|---|
| 182 | argv_new[4] = ""; |
|---|
| 183 | argv = (char**)argv_new; |
|---|
| 184 | } |
|---|
| 185 | |
|---|
| 186 | #ifdef log |
|---|
| 187 | if(outtype!=GCG) |
|---|
| 188 | fprintf(stderr, "\n\nConvert file %s to file %s.\n", argv[2], |
|---|
| 189 | argv[4]); |
|---|
| 190 | else fprintf(stderr, "\n\nConvert file %s to GCG files\n", argv[2]); |
|---|
| 191 | #endif |
|---|
| 192 | |
|---|
| 193 | /* check if output file exists and filename's validation */ |
|---|
| 194 | if(file_exist(argv[4])) { |
|---|
| 195 | sprintf(temp, "Output file %s exists, will be overwritten.", argv[4]); |
|---|
| 196 | warning(151, temp); |
|---|
| 197 | } |
|---|
| 198 | |
|---|
| 199 | /* file format transfer... */ |
|---|
| 200 | convert(argv[2], argv[4], intype, outtype); |
|---|
| 201 | return 0; |
|---|
| 202 | |
|---|
| 203 | } |
|---|
| 204 | /* --------------------------------------------------------------- |
|---|
| 205 | * Function file_type |
|---|
| 206 | * According to the first line in the file to decide |
|---|
| 207 | * the file type. File type could be Genbank, Macke,... |
|---|
| 208 | */ |
|---|
| 209 | int |
|---|
| 210 | file_type(filename) |
|---|
| 211 | char *filename; |
|---|
| 212 | { |
|---|
| 213 | /* char line[LINENUM]; */ |
|---|
| 214 | char token[LINENUM], temp[LINENUM]; |
|---|
| 215 | /* int Cmpstr(); */ |
|---|
| 216 | FILE *fp; |
|---|
| 217 | /* void error(); */ |
|---|
| 218 | |
|---|
| 219 | if((fp=fopen(filename, "r"))==NULL) { |
|---|
| 220 | sprintf(temp, "Cannot open file: %s. Exit.\n", filename); |
|---|
| 221 | error(5, temp); |
|---|
| 222 | } |
|---|
| 223 | /* rewind(fp); */ |
|---|
| 224 | fscanf(fp, "%s", token); |
|---|
| 225 | if(Cmpstr(token, "LOCUS")==EQ) |
|---|
| 226 | return(GENBANK); |
|---|
| 227 | else if(Cmpstr(token, "#-")==EQ) |
|---|
| 228 | return(MACKE); |
|---|
| 229 | else if(Cmpstr(token, "ID")==EQ) |
|---|
| 230 | return(PROTEIN); |
|---|
| 231 | else if(Cmpstr(token, "#NEXUS")==EQ) |
|---|
| 232 | return(PAUP); |
|---|
| 233 | else if(isnum(token)) |
|---|
| 234 | return(PHYLIP); |
|---|
| 235 | else return(UNKNOWN); |
|---|
| 236 | } |
|---|
| 237 | /* ----------------------------------------------------------------- |
|---|
| 238 | * Function isnum(). |
|---|
| 239 | * Return TRUE if the string is an integer number. |
|---|
| 240 | */ |
|---|
| 241 | int |
|---|
| 242 | isnum(string) |
|---|
| 243 | char *string; |
|---|
| 244 | { |
|---|
| 245 | int indi, length, flag; |
|---|
| 246 | |
|---|
| 247 | for(indi=0, length=Lenstr(string), flag=1; flag&&indi<length; indi++) |
|---|
| 248 | if(!isdigit(string[indi])) flag = 0; |
|---|
| 249 | |
|---|
| 250 | return(flag); |
|---|
| 251 | } |
|---|
| 252 | /* ------------------------------------------------------------------ |
|---|
| 253 | * Function file_exist(). |
|---|
| 254 | * Check if file is already existed and also check file |
|---|
| 255 | * name is valid or not. |
|---|
| 256 | */ |
|---|
| 257 | int |
|---|
| 258 | file_exist(file_name) |
|---|
| 259 | char *file_name; |
|---|
| 260 | { |
|---|
| 261 | /* int Lenstr(); */ |
|---|
| 262 | /* void error(); */ |
|---|
| 263 | char temp[TOKENNUM]; |
|---|
| 264 | |
|---|
| 265 | if(Lenstr(file_name)<=0) { |
|---|
| 266 | sprintf(temp, "ilegal file name: %s, EXIT.", file_name); |
|---|
| 267 | error(152, temp); |
|---|
| 268 | } |
|---|
| 269 | return((fopen(file_name, "r")!=NULL)); |
|---|
| 270 | } |
|---|
| 271 | /* ------------------------------------------------------------------- |
|---|
| 272 | * Function change_file_suffix(). |
|---|
| 273 | * Define the default file name by changing suffix. |
|---|
| 274 | */ |
|---|
| 275 | void |
|---|
| 276 | change_file_suffix(old_file, file_name, type) |
|---|
| 277 | char *old_file, *file_name; |
|---|
| 278 | int type; |
|---|
| 279 | { |
|---|
| 280 | int indi, indj; |
|---|
| 281 | /* char *Catstr(); */ |
|---|
| 282 | /* void Cpystr(); */ |
|---|
| 283 | |
|---|
| 284 | for(indi=Lenstr(old_file)-1;indi>=0&&old_file[indi]!='.'; indi--) |
|---|
| 285 | if(indi==0) |
|---|
| 286 | Cpystr(file_name, old_file); |
|---|
| 287 | else { |
|---|
| 288 | for(indj=0; indj<(indi-1); indj++) |
|---|
| 289 | file_name[indj]=old_file[indj]; |
|---|
| 290 | file_name[indj]='\0'; |
|---|
| 291 | } |
|---|
| 292 | switch(type) { |
|---|
| 293 | case GENBANK: Catstr(file_name, ".GB"); |
|---|
| 294 | break; |
|---|
| 295 | case MACKE: Catstr(file_name, ".aln"); |
|---|
| 296 | break; |
|---|
| 297 | case PAUP: Catstr(file_name, ".PAUP"); |
|---|
| 298 | break; |
|---|
| 299 | case PHYLIP: Catstr(file_name, ".PHY"); |
|---|
| 300 | break; |
|---|
| 301 | case EMBL: Catstr(file_name, ".EMBL"); |
|---|
| 302 | break; |
|---|
| 303 | case PRINTABLE: Catstr(file_name, ".prt"); |
|---|
| 304 | break; |
|---|
| 305 | case ALMA: Catstr(file_name, ".ALMA"); |
|---|
| 306 | break; |
|---|
| 307 | default: Catstr(file_name, ".???"); |
|---|
| 308 | } |
|---|
| 309 | } |
|---|