lib/dnaLoad.c File Reference

#include "common.h"
#include "dnaseq.h"
#include "fa.h"
#include "twoBit.h"
#include "nib.h"
#include "dnaLoad.h"

Include dependency graph for dnaLoad.c:

Go to the source code of this file.

Data Structures
struct	dnaLoadStack
struct	dnaLoad
Functions
dnaLoadStack *	dnaLoadStackNew (char *fileName)
void	dnaLoadStackFree (struct dnaLoadStack **pDls)
void	dnaLoadStackFreeList (struct dnaLoadStack **pList)
void	dnaLoadClose (struct dnaLoad **pDl)
dnaLoad *	dnaLoadOpen (char *fileName)
dnaSeq *	dnaLoadSingle (char *fileName, int *retStart, int *retEnd, int *retParentSize)
static struct dnaSeq *	dnaLoadNextFromStack (struct dnaLoad *dl)
static struct dnaSeq *	dnaLoadStackOrSingle (struct dnaLoad *dl)
dnaSeq *	dnaLoadNext (struct dnaLoad *dl)
dnaSeq *	dnaLoadAll (char *fileName)
int	dnaLoadCurStart (struct dnaLoad *dl)
int	dnaLoadCurEnd (struct dnaLoad *dl)
int	dnaLoadCurSize (struct dnaLoad *dl)
Variables
static char const	rcsid [] = "$Id: dnaLoad.c,v 1.8 2005/03/28 17:21:28 hiram Exp $"

---

Function Documentation

struct dnaSeq* dnaLoadAll

(

char *

fileName

)

[read]

Definition at line 260 of file dnaLoad.c.

References dnaLoadClose(), dnaLoadNext(), dnaLoadOpen(), slAddHead, and slReverse().

{
struct dnaLoad *dl = dnaLoadOpen(fileName);
struct dnaSeq *seqList = NULL, *seq;
while ((seq = dnaLoadNext(dl)) != NULL)
    {
    slAddHead(&seqList, seq);
    }
dnaLoadClose(&dl);
slReverse(seqList);
return seqList;
}

Here is the call graph for this function:

void dnaLoadClose

(

struct dnaLoad **

pDl

)

Definition at line 85 of file dnaLoad.c.

References dnaLoadStackFreeList(), freeMem(), freez(), dnaLoad::stack, and dnaLoad::topFileName.

Referenced by dnaLoadAll().

{
struct dnaLoad *dl = *pDl;
if (dl != NULL)
    {
    dnaLoadStackFreeList(&dl->stack);
    freeMem(dl->topFileName);
    freez(pDl);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

int dnaLoadCurEnd

(

struct dnaLoad *

dl

)

Definition at line 286 of file dnaLoad.c.

References dnaLoad::curEnd.

{
return dl->curEnd;
}

int dnaLoadCurSize

(

struct dnaLoad *

dl

)

Definition at line 295 of file dnaLoad.c.

References dnaLoad::curSize.

{
return dl->curSize;
}

int dnaLoadCurStart

(

struct dnaLoad *

dl

)

Definition at line 277 of file dnaLoad.c.

References dnaLoad::curStart.

{
return dl->curStart;
}

struct dnaSeq* dnaLoadNext

(

struct dnaLoad *

dl

)

[read]

Definition at line 251 of file dnaLoad.c.

References dnaLoad::curEnd, dnaLoad::curSize, dnaLoad::curStart, and dnaLoadStackOrSingle().

Referenced by dnaLoadAll().

{
struct dnaSeq *seq;
dl->curSize = dl->curStart = dl->curEnd = 0;
seq = dnaLoadStackOrSingle(dl);
return seq;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static struct dnaSeq* dnaLoadNextFromStack

(

struct dnaLoad *

dl

)

[static, read]

Definition at line 161 of file dnaLoad.c.

References AllocVar, cloneString(), dnaLoad::curEnd, dnaLoad::curSize, dnaLoad::curStart, dnaSeq::dna, dnaLoadSingle(), dnaLoadStackFree(), dnaLoadStackNew(), faMixedSpeedReadNext(), dnaLoad::finished, lineFileNextReal(), name, twoBitIndex::name, needLargeMem(), dnaLoadStack::next, twoBitIndex::next, dnaSeq::size, slAddHead, dnaLoad::stack, dnaLoadStack::tbi, dnaLoadStack::textFile, dnaLoadStack::textIsFa, trimSpaces(), TRUE, dnaLoadStack::twoBit, and twoBitReadSeqFrag().

Referenced by dnaLoadStackOrSingle().

{
struct dnaLoadStack *dls;
struct dnaSeq *seq = NULL;
while ((dls = dl->stack) != NULL)
    {
    if (dls->twoBit)
        {
        if (dls->tbi != NULL)
            {
            seq = twoBitReadSeqFrag(dls->twoBit, dls->tbi->name, 0, 0);
            dls->tbi = dls->tbi->next;
            return seq;
            }
        else
            {
            dl->stack = dls->next;
            dnaLoadStackFree(&dls);
            }
        }
    else if (dls->textIsFa)
        {
        DNA *dna;
        char *name;
        int size;
        if (faMixedSpeedReadNext(dls->textFile, &dna, &size, &name))
            {
            AllocVar(seq);
            seq->dna = needLargeMem(size+1);
            memcpy((void *)seq->dna, (void *)dna, size);
            seq->dna[size] = 0;
            seq->size = size;
            seq->name = cloneString(name);
            dl->curStart = 0;
            dl->curEnd = size;
            dl->curSize = size;
            return seq;
            }
        else
            {
            dl->stack = dls->next;
            dnaLoadStackFree(&dls);
            }
        }
    else        /* It's a file full of file names. */
        {
        char *line;
        if (lineFileNextReal(dls->textFile, &line))
            {
            line  = trimSpaces(line);
            if ((seq = dnaLoadSingle(line, &dl->curStart, &dl->curEnd, &dl->curSize)) != NULL)
                 return seq;
            else
                 {
                 struct dnaLoadStack *newDls;
                 newDls = dnaLoadStackNew(line);
                 slAddHead(&dl->stack, newDls);
                 }
            }
        else
            {
            dl->stack = dls->next;
            dnaLoadStackFree(&dls);
            }
        }
    }
dl->finished = TRUE;
return NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct dnaLoad* dnaLoadOpen

(

char *

fileName

)

[read]

Definition at line 97 of file dnaLoad.c.

References AllocVar, cloneString(), and dnaLoad::topFileName.

Referenced by dnaLoadAll().

{
struct dnaLoad *dl;
AllocVar(dl);
dl->topFileName = cloneString(fileName);
return dl;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct dnaSeq* dnaLoadSingle	(	char *	fileName,
		int *	retStart,
		int *	retEnd,
		int *	retParentSize
	)			[read]

Definition at line 107 of file dnaLoad.c.

References cloneString(), freez(), dnaSeq::name, name, NIB_MASK_MIXED, nibIsFile(), nibLoadAllMasked(), nibOpenVerify(), nibParseName(), PATH_LEN, dnaSeq::size, twoBitClose(), twoBitIsRange(), twoBitOpen(), twoBitParseRange(), twoBitReadSeqFrag(), and twoBitSeqSize().

Referenced by dnaLoadNextFromStack(), and dnaLoadStackOrSingle().

{
struct dnaSeq *seq = NULL;
unsigned start = 0, end = 0;
int parentSize = 0;
if (nibIsFile(fileName))
    {
    /* Save offset out of fileName for auto-lifting */
    char filePath[PATH_LEN];
    char name[PATH_LEN];
    nibParseName(0, fileName, filePath, name, &start, &end);

    if (end != 0)       /* It's just a range. */
        {
        FILE *f;
        int size;
        nibOpenVerify(filePath, &f, &size);
        parentSize = size;
        }
    seq =  nibLoadAllMasked(NIB_MASK_MIXED, fileName);
    if (end == 0)
         parentSize = end = seq->size;
    freez(&seq->name);
    seq->name = cloneString(name);
    }
else if (twoBitIsRange(fileName))
    {
    /* Save offset out of fileName for auto-lifting */
    char *rangeSpec = cloneString(fileName);
    int start, end;
    char *file, *seqName;
    twoBitParseRange(rangeSpec, &file, &seqName, &start, &end);

    /* Load sequence. */
        {
        struct twoBitFile *tbf = twoBitOpen(file);
        parentSize = twoBitSeqSize(tbf, seqName);
        seq = twoBitReadSeqFrag(tbf, seqName, start, end);
        twoBitClose(&tbf);
        }
    if (end == 0)
        end = seq->size;
    freez(&rangeSpec);
    }
if (retStart != NULL)
    *retStart = start;
if (retEnd != NULL)
    *retEnd = end;
if (retParentSize != NULL)
    *retParentSize = parentSize;
return seq;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaLoadStackFree

(

struct dnaLoadStack **

pDls

)

Definition at line 60 of file dnaLoad.c.

References freez(), lineFileClose(), dnaLoadStack::textFile, dnaLoadStack::twoBit, and twoBitClose().

Referenced by dnaLoadNextFromStack(), and dnaLoadStackFreeList().

{
struct dnaLoadStack *dls = *pDls;
if (dls != NULL)
    {
    lineFileClose(&dls->textFile);
    twoBitClose(&dls->twoBit);
    freez(pDls);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaLoadStackFreeList

(

struct dnaLoadStack **

pList

)

Definition at line 72 of file dnaLoad.c.

References dnaLoadStackFree(), and dnaLoadStack::next.

Referenced by dnaLoadClose().

{
struct dnaLoadStack *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    dnaLoadStackFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct dnaLoadStack* dnaLoadStackNew

(

char *

fileName

)

[read]

Definition at line 35 of file dnaLoad.c.

References AllocVar, twoBitFile::indexList, lineFileNextReal(), lineFileOpen(), lineFileReuse(), dnaLoadStack::tbi, dnaLoadStack::textFile, dnaLoadStack::textIsFa, trimSpaces(), TRUE, dnaLoadStack::twoBit, twoBitIsFile(), and twoBitOpen().

Referenced by dnaLoadNextFromStack(), and dnaLoadStackOrSingle().

{
struct dnaLoadStack *dls;
AllocVar(dls);
if (twoBitIsFile(fileName))
    {
    dls->twoBit = twoBitOpen(fileName);
    dls->tbi = dls->twoBit->indexList;
    }
else
    {
    char *line;
    dls->textFile = lineFileOpen(fileName, TRUE);
    if (lineFileNextReal(dls->textFile, &line))
        {
        line = trimSpaces(line);
        if (line[0] == '>')
            dls->textIsFa = TRUE;
        lineFileReuse(dls->textFile);
        }
    }
return dls;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static struct dnaSeq* dnaLoadStackOrSingle

(

struct dnaLoad *

dl

)

[static, read]

Definition at line 233 of file dnaLoad.c.

References dnaLoad::curEnd, dnaLoad::curSize, dnaLoad::curStart, dnaLoadNextFromStack(), dnaLoadSingle(), dnaLoadStackNew(), dnaLoad::finished, dnaLoad::stack, dnaLoad::topFileName, and TRUE.

Referenced by dnaLoadNext().

{
struct dnaSeq *seq = NULL;
if (dl->finished)
    return NULL;
if (dl->stack == NULL)
    {
    if ((seq = dnaLoadSingle(dl->topFileName, &dl->curStart, &dl->curEnd, &dl->curSize)) != NULL)
        {
        dl->finished = TRUE;
        return seq;
        }
    dl->stack = dnaLoadStackNew(dl->topFileName);
    }
return dnaLoadNextFromStack(dl);
}

Here is the call graph for this function:

Here is the caller graph for this function:

---

Variable Documentation

char const rcsid[] = "$Id: dnaLoad.c,v 1.8 2005/03/28 17:21:28 hiram Exp $" [static]

Definition at line 10 of file dnaLoad.c.

---