/*****************************************************************************
*
* PrintIdDefLine
* SeqEntryExplore callback routine that prints the seqids and definition
* lines.
*
*****************************************************************************/
static void PrintIdDefLine (SeqEntryPtr sep, Pointer data,
Int4 index, Int2 indent)
{
BioseqPtr bsp;
FILE * fp;
Char buf[40];
CharPtr title = NULL;
if (IS_Bioseq (sep)) {
*buf = '\0';
bsp = (BioseqPtr) sep->data.ptrvalue;
fp = (FILE *) data;
title = BioseqGetTitle(bsp); /* this does not deal with all cases */
SeqIdPrint(bsp->id, buf, PRINTID_FASTA_LONG); /* print SeqId */
if (pmon != NULL)
MonitorStrValue(pmon, buf);
fprintf(fp, ">%s", buf);
if (title != NULL)
fprintf(fp, " %s", title);
fprintf(fp, "\n");
}
return;
}
static void PrintSequence (BioseqPtr bsp, SeqFeatPtr sfp,
FILE *fp, Boolean is_na)
{
Char buffer [255];
Uint1 code;
Int2 count;
Uint1 repr;
Uint1 residue;
SeqPortPtr spp;
CharPtr title;
CharPtr tmp;
if (bsp != NULL && fp != NULL) {
if ((Boolean) ISA_na (bsp->mol) == is_na) {
repr = Bioseq_repr (bsp);
if (repr == Seq_repr_raw || repr == Seq_repr_const) {
title = BioseqGetTitle (bsp);
tmp = StringMove (buffer, ">");
tmp = SeqIdPrint (bsp->id, tmp, PRINTID_FASTA_LONG);
tmp = StringMove (tmp, " ");
StringNCpy (tmp, title, 200);
fprintf (fp, "%s\n", buffer);
if (pmon != NULL)
MonitorStrValue(pmon, buffer);
if (is_na) {
code = Seq_code_iupacna;
} else {
code = Seq_code_iupacaa;
}
if (sfp != NULL) {
spp = SeqPortNewByLoc (sfp->location, code);
} else {
spp = SeqPortNew (bsp, 0, -1, 0, code);
}
if (spp != NULL) {
count = 0;
while ((residue = SeqPortGetResidue (spp)) != SEQPORT_EOF) {
if (! IS_residue (residue)) {
buffer [count] = '\0';
fprintf (fp, "%s\n", buffer);
count = 0;
switch (residue) {
case SEQPORT_VIRT :
fprintf (fp, "[Gap]\n");
break;
case SEQPORT_EOS :
fprintf (fp, "[EOS]\n");
break;
default :
fprintf (fp, "[Invalid Residue]\n");
break;
}
} else {
buffer [count] = residue;
count++;
if (count >= CHARSPERLINE) {
buffer [count] = '\0';
fprintf (fp, "%s\n", buffer);
count = 0;
}
}
}
if (count != 0) {
buffer [count] = '\0';
fprintf (fp, "%s\n", buffer);
}
SeqPortFree (spp);
}
}
}
}
}
static void GetPubFromGenBank (SeqEntryPtr sep, Pointer data,
Int4 index, Int2 indent)
{
BioseqPtr bsp;
BioseqContextPtr bcp;
SeqFeatPtr sfp;
ValNodePtr vnp;
FILE * fp;
CharPtr accession = NULL;
ValNode tpub;
TextSeqIdPtr tsip;
PubdescPtr pdp;
Char buf[40];
if (! IS_Bioseq(sep))
return;
fp = (FILE *) data; /* get the output file pointer */
bsp = (BioseqPtr) sep->data.ptrvalue;
if (pmon != NULL)
{
*buf = '\0';
SeqIdPrint(bsp->id, buf, PRINTID_FASTA_LONG);
MonitorStrValue(pmon, buf);
}
for (vnp = bsp->id; ((vnp != NULL) && (accession == NULL)); vnp = vnp->next)
{
switch (vnp->choice)
{
case SEQID_GENBANK:
case SEQID_EMBL:
case SEQID_DDBJ:
tsip = (TextSeqIdPtr)(vnp->data.ptrvalue);
if (tsip->accession != NULL)
accession = tsip->accession;
break;
default:
break;
}
}
if (accession == NULL) return;
bcp = BioseqContextNew(bsp);
if (bcp == NULL)
return;
vnp = NULL;
tpub.choice = PUB_Equiv;
tpub.next = NULL;
/* get any pub descriptors */
while ((vnp = BioseqContextGetSeqDescr(bcp, (Int2)Seq_descr_pub, vnp, NULL)) != NULL)
{
pdp = (PubdescPtr)(vnp->data.ptrvalue); /* it's a Pubdesc */
tpub.data.ptrvalue = pdp->pub; /* make Pub-equiv into a Pub */
PubWrite(accession, &tpub, fp);
}
sfp = NULL;
while ((sfp = BioseqContextGetSeqFeat(bcp, 0, sfp, NULL, 0)) != NULL)
{
if (sfp->data.choice == 6) /* a pub feature */
{
pdp = (PubdescPtr)(sfp->data.value.ptrvalue);
tpub.data.ptrvalue = pdp->pub;
PubWrite(accession, &tpub, fp);
}
/* get any feature citations */
if (sfp->cit != NULL)
{
for (vnp = sfp->cit->data.ptrvalue; vnp != NULL; vnp = vnp->next)
PubWrite(accession, vnp, fp);
}
}
BioseqContextFree(bcp);
return;
}
void BioseqRawToRaw(BioseqPtr bsp, Boolean idonly,
short whichSeq, short *seqnum,
char **seq, char **seqid, long *seqlen)
{
SeqPortPtr spp;
SeqIdPtr bestid;
Uint1 repr, code, residue;
CharPtr tmp, title;
long outlen, outmax;
char localid[256], *sp;
/* !!! this may be called several times for a single sequence
because SeqEntryExplore looks for parts and joins them...
assume seq, seqid, seqlen may contain data (or NULL)
*/
if (bsp == NULL) return;
repr = Bioseq_repr(bsp);
if (!(repr == Seq_repr_raw || repr == Seq_repr_const)) return;
(*seqnum)++;
if (!(whichSeq == *seqnum || whichSeq == 0)) return;
bestid = SeqIdFindBest(bsp->id, (Uint1) 0);
title = BioseqGetTitle(bsp);
if (idonly) {
sprintf(localid, " %d) ", *seqnum);
tmp= localid + strlen(localid)-1;
}
else {
strcpy(localid," ");
tmp= localid;
}
tmp = SeqIdPrint(bestid, tmp, PRINTID_FASTA_SHORT);
tmp = StringMove(tmp, " ");
StringNCpy(tmp, title, 200);
/* fprintf(stderr,"BioseqRawToRaw: localid='%s'\n",localid); */
/* < seqid is fixed storage */
/* strcpy( *seqid, localid); */
/* < seqid is variable sized */
outmax= strlen(localid) + 3;
if (*seqid==NULL) {
*seqid= (char*) malloc(outmax);
if (*seqid==NULL) return;
strcpy(*seqid, localid);
}
else {
outmax += strlen(*seqid) + 2;
*seqid= (char*) realloc( *seqid, outmax);
if (*seqid==NULL) return;
if (!idonly) strcat(*seqid, "; ");
strcat(*seqid, localid);
}
if (idonly) {
strcat(*seqid,"\n");
return;
}
if (ISA_na(bsp->mol)) code = Seq_code_iupacna;
else code = Seq_code_iupacaa;
spp = SeqPortNew(bsp, 0, -1, 0, code);
SeqPortSeek(spp, 0, SEEK_SET);
sp= *seq;
if (sp==NULL) { /* this is always true now !? */
outlen= 0;
outmax= 500;
sp= (char*) malloc(outmax);
}
else {
outlen= strlen(sp);
outmax= outlen + 500;
sp= (char*) realloc( sp, outmax);
}
if (sp==NULL) return;
while ((residue = SeqPortGetResidue(spp)) != SEQPORT_EOF) {
if (outlen>=outmax) {
outmax= outlen + 500;
sp= (char*) realloc(sp, outmax);
if (sp==NULL) return;
}
sp[outlen++] = residue;
}
sp= (char*) realloc(sp, outlen+1);
if (sp!=NULL) sp[outlen]= '\0';
*seq= sp;
*seqlen= outlen;
SeqPortFree(spp);
return;
}
