Exemple #1
0
static bits64 bptDataStart(struct bptFile *bpt)
/* Return offset of first bit of data (as opposed to index) in file.  In hind sight I wish
 * this were stored in the header, but fortunately it's not that hard to compute. */
{
bits64 offset = bpt->rootOffset;
for (;;)
    {
    /* Seek to block start */
    udcSeek(bpt->udc, offset);

    /* Read block header,  break if we are leaf. */
    UBYTE isLeaf;
    UBYTE reserved;
    bits16 childCount;
    udcMustReadOne(bpt->udc, isLeaf);
    if (isLeaf)
         break;
    udcMustReadOne(bpt->udc, reserved);
    boolean isSwapped = bpt->isSwapped;
    childCount = udcReadBits16(bpt->udc, isSwapped);

    /* Read and discard first key. */
    char keyBuf[bpt->keySize];
    udcMustRead(bpt->udc, keyBuf, bpt->keySize);

    /* Get file offset of sub-block. */
    offset = udcReadBits64(bpt->udc, isSwapped);
    }
return offset;
}
static void bbiSummaryOnDiskRead(struct bbiFile *bbi, struct bbiSummaryOnDisk *sum)
/* Read in summary from file. */
{
struct udcFile *udc = bbi->udc;
boolean isSwapped = bbi->isSwapped;
sum->chromId = udcReadBits32(udc, isSwapped);
sum->start = udcReadBits32(udc, isSwapped);
sum->end = udcReadBits32(udc, isSwapped);
sum->validCount = udcReadBits32(udc, isSwapped);
udcMustReadOne(udc, sum->minVal);
udcMustReadOne(udc, sum->maxVal);
udcMustReadOne(udc, sum->sumData);
udcMustReadOne(udc, sum->sumSquares);
}
Exemple #3
0
static boolean rFind(struct bptFile *bpt, bits64 blockStart, void *key, void *val)
/* Find value corresponding to key.  If found copy value to memory pointed to by val and return 
 * true. Otherwise return false. */
{
/* Seek to start of block. */
udcSeek(bpt->udc, blockStart);

/* Read block header. */
UBYTE isLeaf;
UBYTE reserved;
bits16 i, childCount;
udcMustReadOne(bpt->udc, isLeaf);
udcMustReadOne(bpt->udc, reserved);
boolean isSwapped = bpt->isSwapped;
childCount = udcReadBits16(bpt->udc, isSwapped);

UBYTE keyBuf[bpt->keySize];   /* Place to put a key, buffered on stack. */

if (isLeaf)
    {
    for (i=0; i<childCount; ++i)
        {
	udcMustRead(bpt->udc, keyBuf, bpt->keySize);
	udcMustRead(bpt->udc, val, bpt->valSize);
	if (memcmp(key, keyBuf, bpt->keySize) == 0)
	    return TRUE;
	}
    return FALSE;
    }
else
    {
    /* Read and discard first key. */
    udcMustRead(bpt->udc, keyBuf, bpt->keySize);

    /* Scan info for first file offset. */
    bits64 fileOffset = udcReadBits64(bpt->udc, isSwapped);

    /* Loop through remainder. */
    for (i=1; i<childCount; ++i)
	{
	udcMustRead(bpt->udc, keyBuf, bpt->keySize);
	if (memcmp(key, keyBuf, bpt->keySize) < 0)
	    break;
	fileOffset = udcReadBits64(bpt->udc, isSwapped);
	}
    return rFind(bpt, fileOffset, key, val);
    }
}
struct cirTreeFile *cirTreeFileAttach(char *fileName, struct udcFile *udc)
/* Open up r-tree index file on previously open file, with cirTree
 * header at current file position. */
{
/* Open file and allocate structure to hold info from header etc. */
struct cirTreeFile *crt = needMem(sizeof(*crt));
crt->fileName = fileName;
crt->udc = udc;

/* Read magic number at head of file and use it to see if we are proper file type, and
 * see if we are byte-swapped. */
bits32 magic;
boolean isSwapped = FALSE;
udcMustReadOne(udc, magic);
if (magic != cirTreeSig)
    {
    magic = byteSwap32(magic);
    isSwapped = crt->isSwapped = TRUE;
    if (magic != cirTreeSig)
       errAbort("%s is not a chromosome id r-tree index file", fileName);
    }

/* Read rest of defined bits of header, byte swapping as needed. */
crt->blockSize = udcReadBits32(udc, isSwapped);
crt->itemCount = udcReadBits64(udc, isSwapped);
crt->startChromIx = udcReadBits32(udc, isSwapped);
crt->startBase = udcReadBits32(udc, isSwapped);
crt->endChromIx = udcReadBits32(udc, isSwapped);
crt->endBase = udcReadBits32(udc, isSwapped);
crt->fileSize = udcReadBits64(udc, isSwapped);
crt->itemsPerSlot = udcReadBits32(udc, isSwapped);

/* Skip over reserved bits of header. */
bits32 reserved32;
udcMustReadOne(udc, reserved32);

/* Save position of root block of r tree. */
crt->rootOffset = udcTell(udc);

return crt;
}
Exemple #5
0
static void rTraverse(struct bptFile *bpt, bits64 blockStart, void *context, 
    void (*callback)(void *context, void *key, int keySize, void *val, int valSize) )
/* Recursively go across tree, calling callback at leaves. */
{
/* Seek to start of block. */
udcSeek(bpt->udc, blockStart);

/* Read block header. */
UBYTE isLeaf;
UBYTE reserved;
bits16 i, childCount;
udcMustReadOne(bpt->udc, isLeaf);
udcMustReadOne(bpt->udc, reserved);
boolean isSwapped = bpt->isSwapped;
childCount = udcReadBits16(bpt->udc, isSwapped);

char keyBuf[bpt->keySize], valBuf[bpt->valSize];
if (isLeaf)
    {
    for (i=0; i<childCount; ++i)
        {
	udcMustRead(bpt->udc, keyBuf, bpt->keySize);
	udcMustRead(bpt->udc, valBuf, bpt->valSize);
	callback(context, keyBuf, bpt->keySize, valBuf, bpt->valSize);
	}
    }
else
    {
    bits64 fileOffsets[childCount];
    /* Loop through to get file offsets of children. */
    for (i=0; i<childCount; ++i)
	{
	udcMustRead(bpt->udc, keyBuf, bpt->keySize);
	fileOffsets[i] = udcReadBits64(bpt->udc, isSwapped);
	}
    /* Loop through recursing on child offsets. */
    for (i=0; i<childCount; ++i)
	rTraverse(bpt, fileOffsets[i], context, callback);
    }
}
Exemple #6
0
struct crTreeFile *crTreeFileOpen(char *fileName)
/* Open up r-tree index file - reading headers and verifying things. */
{
/* Open file and allocate structure to hold info from header etc. */
struct udcFile *udc = udcFileOpen(fileName, udcDefaultDir());
struct crTreeFile *crt = needMem(sizeof(*crt));
fileName = crt->fileName = cloneString(fileName);
crt->udc = udc;

/* Read magic number at head of file and use it to see if we are proper file type, and
 * see if we are byte-swapped. */
bits32 magic;
boolean isSwapped = FALSE;
udcMustReadOne(udc, magic);
if (magic != crTreeSig)
    {
    magic = byteSwap32(magic);
    isSwapped = crt->isSwapped = TRUE;
    if (magic != crTreeSig)
       errAbort("%s is not a chromosome r-tree index file", fileName);
    }

/* Read rest of high level header including notably the offsets to the
 * chromosome and range indexes. */
bits32 reserved32;
udcMustReadOne(udc, reserved32);
crt->chromOffset = udcReadBits64(udc, isSwapped);
crt->cirOffset = udcReadBits64(udc, isSwapped);

/* Read in the chromosome index header. */
udcSeek(udc, crt->chromOffset);
crt->chromBpt = bptFileAttach(fileName, udc);

/* Read in range index header. */
udcSeek(udc, crt->cirOffset);
crt->cir = cirTreeFileAttach(fileName, udc);

return crt;
}
Exemple #7
0
struct bptFile *bptFileAttach(char *fileName, struct udcFile *udc)
/* Open up index file on previously open file, with header at current file position. */
{
/* Open file and allocate structure to hold info from header etc. */
struct bptFile *bpt = needMem(sizeof(*bpt));
bpt->fileName = fileName;
bpt->udc = udc;

/* Read magic number at head of file and use it to see if we are proper file type, and
 * see if we are byte-swapped. */
bits32 magic;
boolean isSwapped = FALSE;
udcMustReadOne(udc, magic);
if (magic != bptSig)
    {
    magic = byteSwap32(magic);
    isSwapped = bpt->isSwapped = TRUE;
    if (magic != bptSig)
       errAbort("%s is not a bpt b-plus tree index file", fileName);
    }

/* Read rest of defined bits of header, byte swapping as needed. */
bpt->blockSize = udcReadBits32(udc, isSwapped);
bpt->keySize = udcReadBits32(udc, isSwapped);
bpt->valSize = udcReadBits32(udc, isSwapped);
bpt->itemCount = udcReadBits64(udc, isSwapped);

/* Skip over reserved bits of header. */
bits32 reserved32;
udcMustReadOne(udc, reserved32);
udcMustReadOne(udc, reserved32);

/* Save position of root block of b+ tree. */
bpt->rootOffset = udcTell(udc);

return bpt;
}
Exemple #8
0
static void rFindMulti(struct bptFile *bpt, bits64 blockStart, void *key, struct slRef **pList)
/* Find values corresponding to key and add them to pList.  You'll need to 
 * Do a slRefFreeListAndVals() on the list when done. */
{
/* Seek to start of block. */
udcSeek(bpt->udc, blockStart);

/* Read block header. */
UBYTE isLeaf;
UBYTE reserved;
bits16 i, childCount;
udcMustReadOne(bpt->udc, isLeaf);
udcMustReadOne(bpt->udc, reserved);
boolean isSwapped = bpt->isSwapped;
childCount = udcReadBits16(bpt->udc, isSwapped);

int keySize = bpt->keySize;
UBYTE keyBuf[keySize];   /* Place to put a key, buffered on stack. */
UBYTE valBuf[bpt->valSize];   /* Place to put a value, buffered on stack. */

if (isLeaf)
    {
    for (i=0; i<childCount; ++i)
        {
	udcMustRead(bpt->udc, keyBuf, keySize);
	udcMustRead(bpt->udc, valBuf, bpt->valSize);
	if (memcmp(key, keyBuf, keySize) == 0)
	    {
	    void *val = cloneMem(valBuf, bpt->valSize);
	    refAdd(pList, val);
	    }
	}
    }
else
    {
    /* Read first key and first file offset. */
    udcMustRead(bpt->udc, keyBuf, keySize);
    bits64 lastFileOffset = udcReadBits64(bpt->udc, isSwapped);
    bits64 fileOffset = lastFileOffset;
    int lastCmp = memcmp(key, keyBuf, keySize);

    /* Loop through remainder. */
    for (i=1; i<childCount; ++i)
	{
	udcMustRead(bpt->udc, keyBuf, keySize);
	fileOffset = udcReadBits64(bpt->udc, isSwapped);
	int cmp = memcmp(key, keyBuf, keySize);
	if (lastCmp >= 0 && cmp <= 0)
	    {
	    bits64 curPos = udcTell(bpt->udc);
	    rFindMulti(bpt, lastFileOffset, key, pList);
	    udcSeek(bpt->udc, curPos);
	    }
	if (cmp < 0)
	    return;
	lastCmp = cmp;
	lastFileOffset = fileOffset;
	}
    /* If made it all the way to end, do last one too. */
    rFindMulti(bpt, fileOffset, key, pList);
    }
}
static void rFindOverlappingBlocks(struct cirTreeFile *crt, int level, bits64 indexFileOffset,
	bits32 chromIx, bits32 start, bits32 end, struct fileOffsetSize **retList)
/* Recursively find blocks with data. */
{
struct udcFile *udc = crt->udc;

/* Seek to start of block. */
udcSeek(udc, indexFileOffset);

/* Read block header. */
UBYTE isLeaf;
UBYTE reserved;
bits16 i, childCount;
udcMustReadOne(udc, isLeaf);
udcMustReadOne(udc, reserved);
boolean isSwapped = crt->isSwapped;
childCount = udcReadBits16(udc, isSwapped);

verbose(3, "rFindOverlappingBlocks %llu %u:%u-%u.  childCount %d. isLeaf %d\n", indexFileOffset, chromIx, start, end, (int)childCount, (int)isLeaf);

if (isLeaf)
    {
    /* Loop through node adding overlapping leaves to block list. */
    for (i=0; i<childCount; ++i)
        {
	bits32 startChromIx = udcReadBits32(udc, isSwapped);
	bits32 startBase = udcReadBits32(udc, isSwapped);
	bits32 endChromIx = udcReadBits32(udc, isSwapped);
	bits32 endBase = udcReadBits32(udc, isSwapped);
	bits64 offset = udcReadBits64(udc, isSwapped);
	bits64 size = udcReadBits64(udc, isSwapped);
	if (cirTreeOverlaps(chromIx, start, end, startChromIx, startBase, endChromIx, endBase))
	    {
	    struct fileOffsetSize *block;
	    AllocVar(block);
	    block->offset = offset;
	    block->size = size;
	    slAddHead(retList, block);
	    }
	}
    }
else
    {
    /* Read node into arrays. */
    bits32 startChromIx[childCount], startBase[childCount];
    bits32 endChromIx[childCount], endBase[childCount];
    bits64 offset[childCount];
    for (i=0; i<childCount; ++i)
        {
	startChromIx[i] = udcReadBits32(udc, isSwapped);
	startBase[i] = udcReadBits32(udc, isSwapped);
	endChromIx[i] = udcReadBits32(udc, isSwapped);
	endBase[i] = udcReadBits32(udc, isSwapped);
	offset[i] = udcReadBits64(udc, isSwapped);
	}

    /* Recurse into child nodes that we overlap. */
    for (i=0; i<childCount; ++i)
	{
	if (cirTreeOverlaps(chromIx, start, end, startChromIx[i], startBase[i], 
		endChromIx[i], endBase[i]))
	    {
	    rFindOverlappingBlocks(crt, level+1, offset[i], chromIx, start, end, retList);
	    }
	}
    }
}
Exemple #10
0
static int bigWigBlockDumpIntersectingRange(struct bbiFile *bwf, char *chrom, 
	bits32 rangeStart, bits32 rangeEnd, int maxCount, FILE *out)
/* Print out info on parts of block that intersect start-end, block starting at current position. */
{
boolean isSwapped = bwf->isSwapped;
struct udcFile *udc = bwf->udc;
struct bwgSectionHead head;
bwgSectionHeadRead(bwf, &head);
bits16 i;
float val;
int outCount = 0;

switch (head.type)
    {
    case bwgTypeBedGraph:
	{
	fprintf(out, "#bedGraph section %s:%u-%u\n",  chrom, head.start, head.end);
	for (i=0; i<head.itemCount; ++i)
	    {
	    bits32 start = udcReadBits32(udc, isSwapped);
	    bits32 end = udcReadBits32(udc, isSwapped);
	    udcMustReadOne(udc, val);
	    if (rangeIntersection(rangeStart, rangeEnd, start, end) > 0)
		{
		fprintf(out, "%s\t%u\t%u\t%g\n", chrom, start, end, val);
		++outCount;
		if (maxCount != 0 && outCount >= maxCount)
		    break;
		}
	    }
	break;
	}
    case bwgTypeVariableStep:
	{
	fprintf(out, "variableStep chrom=%s span=%u\n", chrom, head.itemSpan);
	for (i=0; i<head.itemCount; ++i)
	    {
	    bits32 start = udcReadBits32(udc, isSwapped);
	    udcMustReadOne(udc, val);
	    if (rangeIntersection(rangeStart, rangeEnd, start, start+head.itemSpan) > 0)
		{
		fprintf(out, "%u\t%g\n", start+1, val);
		++outCount;
		if (maxCount != 0 && outCount >= maxCount)
		    break;
		}
	    }
	break;
	}
    case bwgTypeFixedStep:
	{
	boolean gotStart = FALSE;
	bits32 start = head.start;
	for (i=0; i<head.itemCount; ++i)
	    {
	    udcMustReadOne(udc, val);
	    if (rangeIntersection(rangeStart, rangeEnd, start, start+head.itemSpan) > 0)
	        {
		if (!gotStart)
		    {
		    fprintf(out, "fixedStep chrom=%s start=%u step=%u span=%u\n", 
			    chrom, start, head.itemStep, head.itemSpan);
		    gotStart = TRUE;
		    }
		fprintf(out, "%g\n", val);
		++outCount;
		if (maxCount != 0 && outCount >= maxCount)
		    break;
		}
	    start += head.itemStep;
	    }
	break;
	}
    default:
        internalErr();
	break;
    }
return outCount;
}