obitools/ecopcr
2
0
Fork 0
Code Issues 4 Pull Requests Actions Packages Projects Releases 3 Wiki Activity

git-svn-id: https://www.grenoble.prabi.fr/svn/LECASofts/ecoPCR/branches/refactoring@27 60f365c0-8329-0410-b2a4-ec073aeeaa1d

Browse Source
This commit is contained in:
Eric Coissac
2007-06-04 15:34:00 +00:00
parent 5094a9a9ce
commit 7d9ab96dc1
1 changed files with 65 additions and 91 deletions
Download Patch File Download Diff File Expand all files Collapse all files

156
src/ecofind.c
View File

@ -6,34 +6,19 @@
#include <stdio.h>
#define VERSION "0.1"
/* ----------------------------------------------- */
/* structures */
/* ----------------------------------------------- */
typedef struct {
char *sci_name; // scientific name
int32_t tax_id; // tax id use ecoPCR prefilter
int32_t tax_rank; // taxonomic rank
} tax_info ;
/* ----------------------------------------------- */
/* functions */
/* ----------------------------------------------- */
/**
* display the result in sdtout
**/
static void printresult(tax_info **taxonlist,int *j){
int i;
printf("taxonomy rank \t|\t taxonomy id \t|\t scientific name\n\n");
for(i=0;i<*j;i++){
printf("%d \t|\t %d \t|\t %s\n",
taxonlist[i]->tax_id,
taxonlist[i]->tax_rank,
taxonlist[i]->sci_name);
}
static void printresult(ecotx_t *taxon,ecotaxonomy_t *taxonomy){
char* rankname;
rankname= taxonomy->ranks->label[taxon->rank];
printf("%10d \t| %15s \t|\t %s\n",
taxon->taxid,
rankname,
taxon->name);
}
/**
@ -44,7 +29,7 @@ static void printresult(tax_info **taxonlist,int *j){
static void ExitUsage(stat)
int stat;
{
PP "usage: ecofind [-d datafile] [-t taxon] [-h]\n");
PP "usage: ecofind [-d datafile] [-h] <taxon name pattern> ... \n");
PP "type \"ecofind -h\" for help\n");
if (stat)
exit(stat);
@ -74,7 +59,7 @@ static void PrintHelp()
PP " program located in the tools directory.\n");
PP " Write the datafile radical without any extension.\n\n");
PP "-h : [H]elp - print <this> help\n\n");
PP "-t : [T]axonomic pattern bearing regular expressions\n\n");
PP "<taxon> name pattern bearing regular expressions\n\n");
PP "------------------------------------------\n");
}
@ -87,51 +72,39 @@ static void PrintHelp()
int main(int argc, char **argv)
{
int32_t carg = 0;
int32_t numpattern = 0; /* stores number of pattern */
int32_t nummatch = 0;
int32_t k,j = 0;
int32_t errflag = 0;
int32_t tax_count = 0;
char *prefix = NULL;
char *taxon = NULL;
char pattern[PATTERN_NUMBER][PATTERN_LENGHT]; /* stores the pattern */
tax_info **taxonlist = ECOMALLOC(sizeof(tax_info) * RESULT_LENGTH, /* stores the result */
"error in **taxonlist allocation");
ecotaxonomy_t *taxonomy; /* stores all the taxonomy */
ecotx_t *taxon = NULL;
ecotaxonomy_t *taxonomy;
int re_error;
int re_match;
regex_t re_preg;
while ((carg = getopt(argc, argv, "hd:t:")) != -1) {
switch (carg) {
case 'd': /* path to the database */
prefix = optarg;
break;
case 't': /* pattern list */
strncpy(pattern[numpattern],optarg,PATTERN_LENGHT);
numpattern++;
break;
case 'h': /* display help */
PrintHelp();
exit(0);
break;
case '?': /* bad option */
errflag++;
}
}
/**
* check data integrity
* and exit if needed
**/
if (!prefix || !pattern)
errflag++;
while ((carg = getopt(argc, argv, "hd:")) != -1) {
switch (carg) {
case 'd': /* path to the database */
prefix = optarg;
break;
if (taxonlist == NULL)
errflag++;
case 'h': /* display help */
PrintHelp();
exit(0);
break;
case '?': /* bad option */
errflag++;
}
}
/**
* check the path to the database is given as last argument
**/
if ((argc - optind) < 1)
errflag++;
if (errflag)
ExitUsage(errflag);
@ -139,44 +112,45 @@ int main(int argc, char **argv)
/**
* load taxonomy using libecoPCR functions
**/
printf("-- opening %s database --\n",prefix);
printf("# \n# opening %s database\n",prefix);
taxonomy = read_taxonomy(prefix);
tax_count = taxonomy->taxons->count;
printf("-- %d taxons --\n", taxonomy->taxons->count);
printf("# %d taxons\n", tax_count);
/**
* parse taxonomy
**/
for (k=0;k<numpattern;k++)
for (k=optind;k<argc;k++)
{
j=0;
nummatch=0;
printf("\n-- searching for %s pattern --\n",pattern[k]);
do {
taxon = taxonomy->taxons->taxon[j].name;
re_error = regcomp (&re_preg, pattern[k], REG_NOSUB | REG_EXTENDED);
if (re_error == 0)
{
re_match = regexec (&re_preg, taxon, 0, NULL, 0);
regfree(&re_preg);
if (re_match == 0)
{
taxonlist[nummatch] = ECOMALLOC(sizeof(taxonlist),
"error in taxon j allocation");
taxonlist[nummatch]->sci_name = taxon;
taxonlist[nummatch]->tax_id = taxonomy->taxons->taxon[j].taxid;
taxonlist[nummatch]->tax_rank = taxonomy->taxons->taxon[j].rank;
nummatch++;
}
}
j++;
} while(j < taxonomy->taxons->count);
printf("#\n# searching for '%s' pattern\n",argv[k]);
printf("-- %d records found --\n\n",nummatch);
if (nummatch > 0)
printresult(taxonlist, &nummatch);
re_error = regcomp (&re_preg, argv[k], REG_NOSUB | REG_EXTENDED | REG_ICASE);
if (re_error)
{
fprintf(stderr,"# misformed pattern '%s'\n",argv[k]);
exit(1);
}
nummatch=0;
printf("# taxonomy id \t|\t taxonomy rank \t|\t scientific name\n#\n");
for (j=0,taxon=taxonomy->taxons->taxon;
j < tax_count;
taxon++,j++)
{
re_match = regexec (&re_preg, taxon->name, 0, NULL, 0);
if (re_match == 0)
{
printresult(taxon,taxonomy);
nummatch++;
}
}
printf("# %d records found \n",nummatch);
regfree(&re_preg);
}
return 0;