static void ParseArgv(int argc, char **argv,
char **input_path,
char **output_path,
char **dictionary_path,
int *force,
int *quality,
int *decompress,
int *repeat,
int *verbose,
int *lgwin) {
int k;
*force = 0;
*input_path = 0;
*output_path = 0;
*repeat = 1;
*verbose = 0;
*lgwin = 22;
{
size_t argv0_len = strlen(argv[0]);
*decompress =
argv0_len >= 5 && strcmp(&argv[0][argv0_len - 5], "unbro") == 0;
}
for (k = 1; k < argc; ++k) {
if (!strcmp("--force", argv[k]) ||
!strcmp("-f", argv[k])) {
if (*force != 0) {
goto error;
}
*force = 1;
continue;
} else if (!strcmp("--decompress", argv[k]) ||
!strcmp("--uncompress", argv[k]) ||
!strcmp("-d", argv[k])) {
*decompress = 1;
continue;
} else if (!strcmp("--verbose", argv[k]) ||
!strcmp("-v", argv[k])) {
if (*verbose != 0) {
goto error;
}
*verbose = 1;
continue;
}
if (k < argc - 1) {
if (!strcmp("--input", argv[k]) ||
!strcmp("--in", argv[k]) ||
!strcmp("-i", argv[k])) {
if (*input_path != 0) {
goto error;
}
*input_path = argv[k + 1];
++k;
continue;
} else if (!strcmp("--output", argv[k]) ||
!strcmp("--out", argv[k]) ||
!strcmp("-o", argv[k])) {
if (*output_path != 0) {
goto error;
}
*output_path = argv[k + 1];
++k;
continue;
} else if (!strcmp("--custom-dictionary", argv[k])) {
if (*dictionary_path != 0) {
goto error;
}
*dictionary_path = argv[k + 1];
++k;
continue;
} else if (!strcmp("--quality", argv[k]) ||
!strcmp("-q", argv[k])) {
if (!ParseQuality(argv[k + 1], quality)) {
goto error;
}
++k;
continue;
} else if (!strcmp("--repeat", argv[k]) ||
!strcmp("-r", argv[k])) {
if (!ParseQuality(argv[k + 1], repeat)) {
goto error;
}
++k;
continue;
} else if (!strcmp("--window", argv[k]) ||
!strcmp("-w", argv[k])) {
if (!ParseQuality(argv[k + 1], lgwin)) {
goto error;
}
if (*lgwin < 10 || *lgwin >= 25) {
goto error;
}
++k;
continue;
}
}
goto error;
}
return;
error:
fprintf(stderr,
"Usage: %s [--force] [--quality n] [--decompress]"
" [--input filename] [--output filename] [--repeat iters]"
" [--verbose] [--window n] [--custom-dictionary filename]\n",
argv[0]);
exit(1);
}
/* load a description from a file */
PMSDESCRIPTION LoadSiteDescription(const char *Filename)
{
int error; /* non-zero if error occurred */
int foundPattern; /* non-zero if pattern found */
int foundScoring; /* non-zero if scoring parameters found*/
int foundQuality; /* non-zero if quality filter found */
PMSDESCRIPTION pDescription; /* site description to be loaded */
xmlDocPtr pXMLDocument; /* the XML document in the file */
xmlNode *pNode; /* current node in XML document */
/* parse XML file */
xmlInitParser();
pXMLDocument=xmlParseFile(Filename);
if (!pXMLDocument) {
fprintf(stderr,"Could not parse site description file!\n");
return(NULL);
}
/* get memory for site description */
pDescription=(PMSDESCRIPTION)malloc(sizeof(MSDESCRIPTION));
if (!pDescription) {
fprintf(stderr,"Could not allocate memory for site description!\n");
xmlFreeDoc(pXMLDocument);
xmlCleanupParser();
return(NULL);
}
pDescription->Length=0;
pDescription->Sequence=NULL;
pDescription->RepeatBasePosition=-1;
pDescription->Minimal5Flank=5;
pDescription->Minimal3Flank=5;
pDescription->MaximalMismatch=10;
pDescription->MaxMismatch5Req=1;
pDescription->MaxMismatch3Req=1;
pDescription->CutoffForNs=20;
pDescription->MaximumNs=30;
error=0;
do { /* region of controlled exit */
for(pNode=xmlDocGetRootElement(pXMLDocument);pNode;pNode=pNode->next)
if (pNode->type==XML_ELEMENT_NODE &&
!strcmp((const char*)pNode->name,"SiteDescription"))
break;
if (!pNode) {
fprintf(stderr,"Could not find <SiteDescription> in description file!\n");
error=1;
break;
}
foundPattern=0;
foundScoring=0;
foundQuality=0;
for(pNode=pNode->children;pNode;pNode=pNode->next) {
if (pNode->type != XML_ELEMENT_NODE) continue;
if (!strcmp((const char*)pNode->name,"Pattern")) {
if (foundPattern) {
fprintf(stderr,"Site description contains two patterns!\n");
error=1;
break;
}
foundPattern=1;
error=ParsePatterns(pDescription,pNode);
}
if (!strcmp((const char*)pNode->name,"ScoringRules")) {
if (foundScoring) {
fprintf(stderr,"Site description contains two scoring parameter "
"sets!\n");
error=1;
break;
}
foundScoring=1;
error=ParseScoring(pDescription,pNode);
}
if (!strcmp((const char*)pNode->name,"QualityFilter")) {
if (foundQuality) {
fprintf(stderr,"Site description contains two quality filter "
"parameter sets!\n");
error=1;
break;
}
foundQuality=1;
error=ParseQuality(pDescription,pNode);
}
if (error) break;
}
if (!foundPattern) {
fprintf(stderr,"Site description does not contain a pattern!\n");
error=1;
break;
}
if (pDescription->RepeatBasePosition<pDescription->Minimal5Flank) {
fprintf(stderr,"5' flank is shorter than requested minimal 5' flank "
"length!\n");
error=1;
break;
}
if (pDescription->Length-pDescription->RepeatBasePosition-1
<pDescription->Minimal3Flank) {
fprintf(stderr,"3' flank is shorter than requested minimal 3' flank "
"length!\n");
error=1;
break;
}
} while(0); /* end of region of controlled exit */
xmlFreeDoc(pXMLDocument);
xmlCleanupParser();
if (error) DestroySiteDescription(&pDescription);
return(pDescription);
}
