int main(int argc, char *argv[])
{
embInitPV("genret", argc, argv, "GEMBASSY", "1.0.3");
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPStr inseq = NULL;
AjPStr gene = NULL;
AjPStr access = NULL;
AjBool accid = ajTrue;
AjPStr argument = NULL;
AjPFile outfile = NULL;
AjPStr seqid = NULL;
AjPStr restid = NULL;
AjBool valid = ajFalse;
AjBool isseq = ajFalse;
AjBool isgbk = ajFalse;
AjPFilebuff buff = NULL;
AjPFile tmpfile = NULL;
AjPStr tmpname = NULL;
AjPStr regexstr = NULL;
AjPStrTok token = NULL;
AjPRegexp regex = NULL;
AjPStr url = NULL;
AjPStr base = NULL;
AjPStr head = NULL;
AjPStr line = NULL;
seqall = ajAcdGetSeqall("sequence");
access = ajAcdGetString("access");
gene = ajAcdGetString("gene");
argument = ajAcdGetString("argument");
accid = ajAcdGetBoolean("accid");
outfile = ajAcdGetOutfile("outfile");
if(
ajStrMatchC(access, "translation") ||
ajStrMatchC(access, "get_exon") ||
ajStrMatchC(access, "get_exons") ||
ajStrMatchC(access, "get_cdsseq") ||
ajStrMatchC(access, "get_gbkseq") ||
ajStrMatchC(access, "get_geneseq") ||
ajStrMatchC(access, "get_intron") ||
ajStrMatchC(access, "getseq") ||
ajStrMatchC(access, "seq") ||
ajStrMatchC(access, "around_startcodon") ||
ajStrMatchC(access, "around_stopcodon") ||
ajStrMatchC(access, "before_startcodon") ||
ajStrMatchC(access, "before_stopcodon") ||
ajStrMatchC(access, "after_startcodon") ||
ajStrMatchC(access, "after_stopcodon")
)
{
isseq = ajTrue;
}
else if(ajStrMatchC(access, "annotate") ||
ajStrMatchC(access, "output"))
{
isgbk = ajTrue;
}
else
{
ajFmtPrintF(outfile, "gene,%S\n", access);
}
base = ajStrNewC("rest.g-language.org");
ajStrExchangeCC(&argument, " ", "/");
ajStrExchangeCC(&argument, ",", "/");
ajStrExchangeCC(&argument, "\t", "/");
ajStrExchangeCC(&argument, "\r", "/");
ajStrExchangeCC(&argument, "\n", "/");
if(ajStrMatchC(gene, "*"))
{
ajStrInsertK(&gene, 0, '.');
}
if(ajStrPrefixC(gene, "@") || ajStrPrefixC(gene, "list::"))
{
ajStrExchangeCC(&gene, "@", "");
ajStrExchangeCC(&gene, "list::", "");
ajStrAssignS(&tmpname, gene);
tmpfile = ajFileNewInNameS(tmpname);
if(!tmpfile)
{
ajDie("List file (%S) open error\n", tmpname);
}
gene = ajStrNew();
while(ajReadline(tmpfile, &line))
{
ajStrAppendS(&gene, line);
}
ajFileClose(&tmpfile);
ajStrDel(&tmpname);
ajStrDel(&line);
}
tmpname = ajStrNew();
gAssignUniqueName(&tmpname);
while(ajSeqallNext(seqall, &seq))
{
inseq = ajStrNew();
if(!accid)
{
if(gFormatGenbank(seq, &inseq))
{
tmpfile = ajFileNewOutNameS(tmpname);
if(!tmpfile)
{
ajDie("Output file (%S) open error\n", tmpname);
}
ajFmtPrintF(tmpfile, "%S", inseq);
ajFileClose(&tmpfile);
ajFmtPrintS(&url, "http://%S/upload/upl.pl", base);
gFilePostSS(url, tmpname, &restid);
ajStrDel(&url);
ajSysFileUnlinkS(tmpname);
}
else
{
ajWarn("Sequence does not have features\n"
"Proceeding with sequence accession ID\n");
accid = ajTrue;
}
}
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
if(ajStrGetLen(seqid) == 0)
{
ajStrAssignS(&seqid, ajSeqGetNameS(seq));
}
if(ajStrGetLen(seqid) == 0)
{
ajWarn("No valid header information\n");
}
if(accid)
{
ajStrAssignS(&restid, seqid);
if(ajStrGetLen(seqid) == 0)
{
ajDie("Cannot proceed without header with -accid\n");
}
if(!gValID(seqid))
{
ajDie("Invalid accession ID:%S, exiting\n", seqid);
}
}
url = ajStrNew();
if(isgbk)
{
ajFmtPrintS(&url, "http://%S/%S/%S", base, restid, access);
}
else
{
ajFmtPrintS(&url, "http://%S/%S/*/%S/%S", base, restid, access, argument);
}
if(!gFilebuffURLS(url, &buff))
{
ajDie("GET error from %S\n", url);
}
while(ajBuffreadLine(buff, &line))
{
if(isgbk){
ajFmtPrintF(outfile, "%S", line);
continue;
}
ajStrRemoveLastNewline(&line);
regex = ajRegCompC("^>");
if(ajRegExec(regex, line))
{
head = ajStrNew();
ajStrAssignS(&head, line);
ajStrTrimStartC(&head, ">");
valid = ajFalse;
token = ajStrTokenNewC(ajStrNewS(gene), " ,\t\r\n");
while(ajStrTokenNextParse(token, ®exstr))
{
if(ajStrGetLen(regexstr))
{
regex = ajRegComp(regexstr);
if(ajRegExec(regex, line))
{
valid = ajTrue;
if(ajStrIsAlnum(regexstr))
{
ajStrExchangeSC(&gene, regexstr, "");
}
}
ajRegFree(®ex);
}
}
}
else
{
if(valid)
{
if(isseq)
{
ajStrFmtWrap(&line, 60);
ajFmtPrintF(outfile, ">%S\n%S\n", head, line);
}
else
{
ajFmtPrintF(outfile, "%S,%S\n", head, line);
}
valid = ajFalse;
}
}
}
ajFileClose(&outfile);
ajStrDel(&restid);
ajStrDel(&seqid);
ajStrDel(&inseq);
}
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&access);
ajStrDel(&gene);
embExit();
}
int main(int argc, char** argv)
{
AjPFile outf = NULL;
AjPFile cachef = NULL;
AjIList iterator = NULL;
AjPList aliases = NULL;
AjPList dbas = NULL;
AjPList species = NULL;
AjPStr alias = NULL;
AjPStr dbname = NULL;
AjPStr spname = NULL;
AjPStr svrname = NULL;
AjPStr svrurl = NULL;
AjPStr dbcurl = NULL;
AjPTime svrtime = NULL;
EnsEDatabaseadaptorGroup dbag = ensEDatabaseadaptorGroupNULL;
EnsPDatabaseadaptor dba = NULL;
EnsPDatabaseconnection dbc = NULL;
embInit("cacheensembl", argc, argv);
ensInit();
svrname = ajAcdGetString("servername");
outf = ajAcdGetOutfile("outfile");
cachef = ajAcdGetOutfile("cachefile");
dbcurl = ajStrNew();
svrurl = ajStrNew();
dbname = ajStrNew();
ajNamSvrGetUrl(svrname, &svrurl);
if(!svrurl)
ajFatal("Could not resolve server name '%S'.", svrname);
dbc = ensDatabaseconnectionNewUrl(svrurl);
ensRegistryLoadDatabaseconnection(dbc);
ensDatabaseconnectionDel(&dbc);
/* Write the server file header. */
svrtime = ajTimeNewTodayFmt("cachefile");
ajFmtPrintF(cachef, "# %S %D\n", ajFileGetNameS(cachef), svrtime);
ajTimeDel(&svrtime);
ajFmtPrintF(cachef,
"# Automatically generated by cacheensembl "
"for server '%S'.\n\n",
svrname);
/*
** Get all Ensembl Database Adaptor objects and write them as
** EMBOSS Database definitions.
*/
aliases = ajListstrNew();
dbas = ajListNew();
species = ajListstrNew();
ensRegistryRetrieveAllSpecies(species);
while(ajListstrPop(species, &spname))
{
ensRegistryGetAllDatabaseadaptors(ensEDatabaseadaptorGroupNULL,
spname,
dbas);
while(ajListPop(dbas, (void**) &dba))
{
dbag = ensDatabaseadaptorGetGroup(dba);
if(dbag == ensEDatabaseadaptorGroupNULL)
{
ajDebug("cacheensembl main got unexpected "
"Ensembl Database Adaptor Group %d.\n",
dbag);
continue;
}
ajStrAssignS(&dbname, ensDatabaseadaptorGetSpecies(dba));
if(dbag != ensEDatabaseadaptorGroupCore)
{
ajStrAppendC(&dbname, "_");
ajStrAppendC(&dbname, ensDatabaseadaptorGroupToChar(dbag));
}
dbc = ensDatabaseadaptorGetDatabaseconnection(dba);
ensDatabaseconnectionFetchUrl(dbc, &dbcurl);
if(outf)
ajFmtPrintF(outf, "%S\n", dbname);
ajFmtPrintF(cachef, "DBNAME %S [\n", dbname);
ajFmtPrintF(cachef, " release: \"%s\"\n", ensSoftwareGetVersion());
ajFmtPrintF(cachef, " server: \"%S\"\n", svrname);
ajFmtPrintF(cachef, " url: \"%S\"\n", dbcurl);
ajFmtPrintF(cachef, "]\n");
ajFmtPrintF(cachef, "\n");
if(dbag != ensEDatabaseadaptorGroupCore)
continue;
ensRegistryAliasFetchAllbySpecies(
ensDatabaseadaptorGetSpecies(dba),
aliases);
/*
** Format all aliases to lower case,
** sort them alphabetically and remove duplicates.
*/
iterator = ajListIterNew(aliases);
while(!ajListIterDone(iterator))
{
alias = ajListstrIterGet(iterator);
ajStrFmtLower(&alias);
}
ajListIterDel(&iterator);
ajListSortUnique(aliases,
cacheensembl_stringcompare,
cacheensembl_stringdelete);
alias = NULL;
if(ajListGetLength(aliases) > 0)
{
while(ajListstrPop(aliases, &alias))
{
/*
** Reject any aliases with other than alpha-numeric
** characters like white space.
*/
if(ajStrIsAlnum(alias))
ajFmtPrintF(cachef,
"ALIAS %S %S\n",
alias,
ensDatabaseadaptorGetSpecies(dba));
ajStrDel(&alias);
}
ajFmtPrintF(cachef, "\n");
}
/* Ensembl Database Adaptor objects *must not* be deleted. */
}
ajStrDel(&spname);
}
ajListstrFree(&aliases);
ajListFree(&dbas);
ajStrDel(&dbcurl);
ajStrDel(&svrurl);
ajStrDel(&dbname);
ajStrDel(&svrname);
ajFileClose(&outf);
ajFileClose(&cachef);
embExit();
return EXIT_SUCCESS;
}
