~ubuntu-branches/ubuntu/saucy/ncbi-tools6/saucy-proposed

Int4 SeqEntryToAsn3Ex (SeqEntryPtr sep, Boolean strip_old, Boolean source_correct, Boolean taxserver, SeqEntryFunc taxfun, SeqEntryFunc taxmerge, Boolean gpipeMode)

1041

1042

{

1042

1043

ToAsn3 ta;

1043

1044

OrgFixPtr ofp = NULL;

1061

1062

if (sep == NULL) {

1062

1063

return ERR_INPUT;

1063

1064

}

1065

1066

RemoveAllNcbiCleanupUserObjects (sep);

1067

1064

1068

update_date_pos = GetUpdateDatePos (sep);

1065

1069

RemoveEmptyTitleAndPubGenAsOnlyPub (sep);

1066

1070

if (source_correct) {

1068

1072

}

1069

1073

toporg(sep);

1070

1074

SeqEntryExplore(sep, (Pointer)(&ta), FindOrg);

1071

1075

1072

1076

if (ta.had_biosource) {

1073

1077

/* entry is in asn.1 spec 3.0 already do the checks only */

1074

1078

retval |= INFO_ASNNEW;

1102

1106

EntryChangeGBSource(sep);

1103

1107

SeqEntryExplore (sep, NULL, FixProtMolInfo);

1104

1108

SeqEntryExplore (sep, NULL, FuseMolInfos);

1105

SeqEntryExplore(sep, NULL, StripProtXref);

1109

if (! gpipeMode) {

1110

SeqEntryExplore(sep, NULL, StripProtXref);

1111

}

1106

1112

SeqEntryExplore(sep, (Pointer)(&qm), CheckMaps);

1107

1113

1108

1114

if (qm.same == TRUE) {

1172

1178

EntryChangeGBSource(sep);

1173

1179

SeqEntryExplore (sep, NULL, FixProtMolInfo);

1174

1180

SeqEntryExplore (sep, NULL, FuseMolInfos);

1175

SeqEntryExplore(sep, NULL, StripProtXref);

1181

if (! gpipeMode) {

1182

SeqEntryExplore(sep, NULL, StripProtXref);

1183

}

1176

1184

SeqEntryExplore(sep, (Pointer)(&qm), CheckMaps);

1177

1185

1178

1186

if (qm.same == TRUE) {

3327

3335

3328

3336

static void ChangeGBDiv (SeqEntryPtr sep, Pointer data, Int4 index, Int2 indent)

3329

3337

{

3330

BioseqPtr bsp=NULL;

3331

BioseqSetPtr bssp;

3332

ValNodePtr descr=NULL, vnp;

3333

CharPtr div;

3334

GBBlockPtr gbp;

3335

MolInfoPtr mfp=NULL;

3336

Int2 i;

3338

BioseqPtr bsp = NULL;

3339

BioseqSetPtr bssp;

3340

ValNodePtr descr = NULL, vnp;

3341

CharPtr div;

3342

GBBlockPtr gbp;

3343

Int2 i;

3344

Boolean is_patent = FALSE;

3345

MolInfoPtr mfp = NULL;

3346

SeqIdPtr sip;

3337

3347

3338

3348

div = (CharPtr) data;

3339

3349

if (IS_Bioseq(sep)) {

3340

3350

bsp = (BioseqPtr)(sep->data.ptrvalue);

3341

3351

descr = bsp->descr;

3352

for (sip = bsp->id; sip != NULL; sip = sip->next) {

3353

if (sip->choice == SEQID_PATENT) {

3354

is_patent = TRUE;

3355

}

3356

}

3342

3357

} else {

3343

3358

bssp = (BioseqSetPtr)(sep->data.ptrvalue);

3344

3359

descr = bssp->descr;

3352

3367

for (vnp = descr; vnp != NULL; vnp = vnp->next) {

3353

3368

if (vnp->choice == Seq_descr_genbank) {

3354

3369

gbp = (GBBlockPtr) vnp->data.ptrvalue;

3355

if (gbp == NULL) {

3356

return;

3357

}

3370

if (gbp == NULL) continue;

3358

3371

if (mfp) {

3359

3372

if (mfp->tech == MI_TECH_htgs_0 ||

3360

3373

mfp->tech == MI_TECH_htgs_1 ||

3365

3378

CheckKeywords(gbp, mfp->tech);

3366

3379

}

3367

3380

}

3368

if (gbp->div == NULL) {

3369

return;

3370

}

3381

if (gbp->div == NULL) continue;

3371

3382

for (i=0; i < TOTAL_TECH; i++) {

3372

3383

if (StringCmp(gbp->div, check_tech[i].name) == 0) {

3373

3384

break;

3401

3412

gbp->div = MemFree(gbp->div);

3402

3413

} else if (StringCmp(gbp->div, "UNC") == 0) {

3403

3414

gbp->div = MemFree(gbp->div);

3415

} else if (StringCmp(gbp->div, "PAT") == 0 && is_patent) {

3416

gbp->div = MemFree(gbp->div);

3404

3417

}

3405

3418

}

3406

3419

}

3415

3428

OrgNamePtr onp;

3416

3429

} GBSourceData, PNTR GBSourcePtr;

3417

3430

3418

static Boolean AbbrevStrICmp (CharPtr str, CharPtr gbpsrc)

3431

static Boolean AbbrevStrIEql (CharPtr str, CharPtr gbpsrc)

3419

3432

3420

3433

{

3421

3434

Char buf [200];

3442

3455

*ptr = '\0';

3443

3456

StringCat (ptr, str);

3444

3457

3445

return StringICmp (buf, gbpsrc);

3458

return (Boolean) (StringICmp (buf, gbpsrc) == 0);

3446

3459

}

3447

3460

3448

3461

static Boolean CanDeleteGBSource (GBSourcePtr gsp, CharPtr gbpsrc)

3477

3490

ptr++;

3478

3491

ch = *ptr;

3479

3492

}

3493

} else {

3494

str = StringStr (gbpsrc, "strain)");

3495

if (str != NULL) return FALSE; /* do not handle this case for now */

3480

3496

}

3481

3497

if (foundStrain) {

3482

3498

*str = '\0';

3505

3521

if (StringDoesHaveText (gsp->common) && StringICmp (gsp->common, gbpsrc) == 0) return TRUE;

3506

3522

if (StringDoesHaveText (gsp->oldname) && StringICmp (gsp->oldname, gbpsrc) == 0) return TRUE;

3507

3523

3508

if (StringDoesHaveText (gsp->taxname) && AbbrevStrICmp (gsp->taxname, gbpsrc) == 0) return TRUE;

3509

if (StringDoesHaveText (gsp->oldname) && AbbrevStrICmp (gsp->oldname, gbpsrc) == 0) return TRUE;

3524

if (StringDoesHaveText (gsp->taxname) && AbbrevStrIEql (gsp->taxname, gbpsrc)) return TRUE;

3525

if (StringDoesHaveText (gsp->oldname) && AbbrevStrIEql (gsp->oldname, gbpsrc)) return TRUE;

3510

3526

3511

3527

3512

3528

return FALSE;

3513

3529

}

3514

3530

3531

static void TrimPeriodFromEnd (CharPtr str)

3532

3533

{

3534

size_t len;

3535

3536

len = StringLen (str);

3537

if (len < 2) return;

3538

if (str [len - 1] == '.') {

3539

str [len - 1] = '\0';

3540

}

3541

}

3542

3515

3543

static void ChangeGBSource (SeqEntryPtr sep, Pointer data, Int4 index, Int2 indent)

3516

3544

{

3517

3545

BioseqPtr bsp;

3544

3572

} else if (len > 6 && StringCmp (s + len - 5, " rRNA.") == 0) {

3545

3573

s [len - 6] = '\0';

3546

3574

}

3575

TrimPeriodFromEnd (s);

3576

3547

3577

if (*(s+StringLen(s)-1) =='.') {

3548

3578

*(s+StringLen(s)-1) = '\0';

3549

3579

}

3580

3550

3581

if (CanDeleteGBSource (gsp, s)) {

3551

3582

gbp->source = MemFree (gbp->source);

3552

3583

}

3608

3639

if (StringDoesHaveText (orp->taxname)) {

3609

3640

gsd.taxname = StringSave (orp->taxname);

3610

3641

TrimSpacesAndJunkFromEnds (gsd.taxname, FALSE);

3642

TrimPeriodFromEnd (gsd.taxname);

3611

3643

}

3612

3644

if (StringDoesHaveText (orp->common)) {

3613

3645

gsd.common = StringSave (orp->common);

3614

3646

TrimSpacesAndJunkFromEnds (gsd.common, FALSE);

3647

TrimPeriodFromEnd (gsd.common);

3615

3648

}

3616

3649

onp = orp->orgname;

3617

3650

if (onp != NULL) {

3621

3654

if (omp->subtype == ORGMOD_strain) {

3622

3655

gsd.strain = StringSave (omp->subname);

3623

3656

TrimSpacesAndJunkFromEnds (gsd.strain, FALSE);

3657

TrimPeriodFromEnd (gsd.strain);

3624

3658

} else if (omp->subtype == ORGMOD_old_name) {

3625

3659

gsd.oldname = StringSave (omp->subname);

3626

3660

TrimSpacesAndJunkFromEnds (gsd.oldname, FALSE);

3661

TrimPeriodFromEnd (gsd.oldname);

3627

3662

}

3628

3663

}

3629

3664

if (StringDoesHaveText (onp->div)) {

Older »