GtUword gt_bwtseqfirstmatch(const FMindex *fmindex,
GtUword bound)
{
struct extBitsRetrieval extBits;
GtUword pos;
initExtBitsRetrieval(&extBits);
pos = gt_BWTSeqLocateMatch((const BWTSeq *) fmindex,bound,&extBits);
destructExtBitsRetrieval(&extBits);
return pos;
}
unsigned long gt_bwtseqfirstmatch(const FMindex *fmindex,
unsigned long bound)
{
struct extBitsRetrieval extBits;
unsigned long pos;
initExtBitsRetrieval(&extBits);
pos = gt_BWTSeqLocateMatch((const BWTSeq *) fmindex,bound,&extBits);
destructExtBitsRetrieval(&extBits);
return pos;
}
Bwtseqcontextiterator *gt_Bwtseqcontextiterator_new(const FMindex *fmindex,
GtUword bound)
{
Bwtseqcontextiterator *bsci;
bsci = gt_malloc(sizeof (*bsci));
initExtBitsRetrieval(&bsci->extBits);
bsci->bwtseq = (const BWTSeq *) fmindex;
bsci->bound = bound;
return bsci;
}
BwtSeqpositionextractor *gt_newBwtSeqpositionextractor(
const FMindex *voidbwtseq,
GtUword upperbound)
{
BwtSeqpositionextractor *bspex;
bspex = gt_malloc(sizeof (*bspex));
initExtBitsRetrieval(&bspex->extBits);
bspex->bwtseq = (const BWTSeq *) voidbwtseq;
bspex->upperbound = upperbound;
return bspex;
}
Bwtseqpositioniterator *gt_Bwtseqpositioniterator_new(const FMindex *fmindex,
GtUword lowerbound,
GtUword upperbound)
{
Bwtseqpositioniterator *bspi;
bspi = gt_malloc(sizeof (*bspi));
initExtBitsRetrieval(&bspi->extBits);
bspi->bwtseq = (const BWTSeq *) fmindex;
bspi->currentbound = lowerbound;
bspi->upperbound = upperbound;
return bspi;
}
bool
gt_initEmptyEMIterator(BWTSeqExactMatchesIterator *iter, const BWTSeq *bwtSeq)
{
gt_assert(iter && bwtSeq);
if (!bwtSeq->locateSampleInterval)
{
fputs("Index does not contain locate information.\n"
"Localization of matches impossible!", stderr);
return false;
}
iter->bounds.start = iter->bounds.end = iter->nextMatchBWTPos = 0;
initExtBitsRetrieval(&iter->extBits);
return true;
}
static size_t
BWTSASSReadSufTab(SeqDataSrc src, void *dest, size_t len)
{
const BWTSeq *bwtSeq;
struct extBitsRetrieval extBits;
gt_assert(src);
initExtBitsRetrieval(&extBits);
bwtSeq = ((BWTSeqReaderState *)src)->backLink->bwtSeq;
size_t i;
Seqpos pos = ((BWTSeqReaderState *)src)->nextReadPos;
for (i = 0; i < len; ++i)
((Seqpos *)dest)[i] = BWTSeqLocateMatch(bwtSeq, pos + i, &extBits);
((BWTSeqReaderState *)src)->nextReadPos = pos + len;
destructExtBitsRetrieval(&extBits);
return len;
}
bool
gt_initEMIterator(BWTSeqExactMatchesIterator *iter, const BWTSeq *bwtSeq,
const Symbol *query, size_t queryLen, bool forward)
{
gt_assert(iter && bwtSeq && query);
if (!bwtSeq->locateSampleInterval)
{
fputs("Index does not contain locate information.\n"
"Localization of matches impossible!", stderr);
return false;
}
getMatchBound(bwtSeq, query, queryLen, &iter->bounds, forward);
iter->nextMatchBWTPos = iter->bounds.start;
initExtBitsRetrieval(&iter->extBits);
return true;
}
enum verifyBWTSeqErrCode
gt_BWTSeqVerifyIntegrity(BWTSeq *bwtSeq, const char *projectName,
int checkFlags,
GtUword tickPrint, FILE *fp,
GtLogger *verbosity, GtError *err)
{
Suffixarray suffixArray;
struct extBitsRetrieval extBits;
bool suffixArrayIsInitialized = false, extBitsAreInitialized = false;
enum verifyBWTSeqErrCode retval = VERIFY_BWTSEQ_NO_ERROR;
do
{
GtUword seqLen;
gt_assert(bwtSeq && projectName && err);
gt_error_check(err);
initExtBitsRetrieval(&extBits);
extBitsAreInitialized = true;
if (gt_mapsuffixarray(&suffixArray,
SARR_SUFTAB | SARR_ESQTAB, projectName, verbosity, err))
{
gt_error_set(err, "Cannot load reference suffix array project with"
" demand for suffix table file and encoded sequence"
" for project: %s", projectName);
retval = VERIFY_BWTSEQ_REFLOAD_ERROR;
break;
}
suffixArrayIsInitialized = true;
seqLen = gt_encseq_total_length(suffixArray.encseq) + 1;
if (BWTSeqLength(bwtSeq) != seqLen)
{
gt_error_set(err, "length mismatch for suffix array project %s and "
"bwt sequence index", projectName);
retval = VERIFY_BWTSEQ_LENCOMPARE_ERROR;
break;
}
if (checkFlags & VERIFY_BWTSEQ_SUFVAL
&& BWTSeqHasLocateInformation(bwtSeq))
{
GtUword i;
for (i = 0; i < seqLen && retval == VERIFY_BWTSEQ_NO_ERROR; ++i)
{
if (gt_BWTSeqPosHasLocateInfo(bwtSeq, i, &extBits))
{
GtUword sfxArrayValue = gt_BWTSeqLocateMatch(bwtSeq, i,
&extBits);
if (sfxArrayValue != ESASUFFIXPTRGET(suffixArray.suftab,i))
{
gt_error_set(err, "Failed suffix array value comparison"
" at position "GT_WU": "GT_WU" != "GT_WU"",
i, sfxArrayValue,
ESASUFFIXPTRGET(suffixArray.suftab,i));
retval = VERIFY_BWTSEQ_SUFVAL_ERROR;
break;
}
}
if (tickPrint && !((i + 1) % tickPrint))
putc('.', fp);
}
if (tickPrint)
putc('\n', fp);
if (retval != VERIFY_BWTSEQ_NO_ERROR)
break;
}
else if (checkFlags & VERIFY_BWTSEQ_SUFVAL)
{
gt_error_set(err, "check of suffix array values was requested,"
" but index contains no locate information!");
retval = VERIFY_BWTSEQ_SUFVAL_ERROR;
break;
}
else if (!(checkFlags & VERIFY_BWTSEQ_SUFVAL)
&& BWTSeqHasLocateInformation(bwtSeq))
{
fputs("Not checking suftab values.\n", stderr);
}
if (BWTSeqHasLocateInformation(bwtSeq))
{
GtUword nextLocate = BWTSeqTerminatorPos(bwtSeq);
if (suffixArray.longest.defined &&
suffixArray.longest.valueunsignedlong != nextLocate)
{
gt_error_set(err, "terminator/0-rotation position mismatch "GT_WU""
" vs. "GT_WU"", suffixArray.longest.valueunsignedlong,
nextLocate);
retval = VERIFY_BWTSEQ_TERMPOS_ERROR;
break;
}
if ((checkFlags & VERIFY_BWTSEQ_LFMAPWALK)
&& (bwtSeq->featureToggles & BWTReversiblySorted))
{
GtUword i = seqLen;
/* handle first symbol specially because the encseq
* will not return the terminator symbol */
{
Symbol sym = BWTSeqGetSym(bwtSeq, nextLocate);
if (sym != UNDEFBWTCHAR)
{
gt_error_set(err, "symbol mismatch at position "GT_WU": "
"%d vs. reference symbol %d", i - 1, (int)sym,
(int)UNDEFBWTCHAR);
retval = VERIFY_BWTSEQ_LFMAPWALK_ERROR;
break;
}
--i;
nextLocate = BWTSeqLFMap(bwtSeq, nextLocate, &extBits);
}
while (i > 0)
{
Symbol symRef =
gt_encseq_get_encoded_char(suffixArray.encseq,
--i,
suffixArray.readmode);
Symbol symCmp = BWTSeqGetSym(bwtSeq, nextLocate);
if (symCmp != symRef)
{
gt_error_set(err, "symbol mismatch at position "GT_WU": "
"%d vs. reference symbol %d", i, symCmp, symRef);
retval = VERIFY_BWTSEQ_LFMAPWALK_ERROR;
break;
}
nextLocate = BWTSeqLFMap(bwtSeq, nextLocate, &extBits);
}
if (retval != VERIFY_BWTSEQ_NO_ERROR)
break;
}
else if ((checkFlags & VERIFY_BWTSEQ_LFMAPWALK)
&& !(bwtSeq->featureToggles & BWTReversiblySorted))
{
gt_error_set(err, "requested complete backwards regeneration in index"
" without regeneration capability");
retval = VERIFY_BWTSEQ_LFMAPWALK_IMP_ERROR;
break;
}
}
if (checkFlags & VERIFY_BWTSEQ_CONTEXT)
{
BWTSeqContextRetriever *bwtSeqCR =
gt_BWTSeqCRLoad(bwtSeq, projectName, CTX_MAP_ILOG_AUTOSIZE);
if (!bwtSeqCR)
{
gt_error_set(err, "cannot load BWT sequence context access table"
" for project %s", projectName);
retval = VERIFY_BWTSEQ_CONTEXT_LOADFAIL;
break;
}
fputs("Checking context regeneration.\n", stderr);
{
GtUword i, start, subSeqLen,
maxSubSeqLen = MIN(MAX(MIN_CONTEXT_LEN, seqLen/CONTEXT_FRACTION),
MAX_CONTEXT_LEN),
numTries = MIN(MAX_NUM_CONTEXT_CHECKS,
MAX(2, seqLen/CONTEXT_INTERVAL));
Symbol *contextBuf = gt_malloc(sizeof (Symbol) * MAX_CONTEXT_LEN);
GtEncseqReader *esr =
gt_encseq_create_reader_with_readmode(suffixArray.encseq,
suffixArray.readmode,
0);
for (i = 0; i < numTries && retval == VERIFY_BWTSEQ_NO_ERROR; ++i)
{
GtUword j, end, inSubSeqLen;
subSeqLen = random()%maxSubSeqLen + 1;
start = random()%(seqLen - subSeqLen + 1);
end = start + subSeqLen;
inSubSeqLen = subSeqLen - ((end==seqLen)?1:0);
gt_BWTSeqCRAccessSubseq(bwtSeqCR, start, subSeqLen, contextBuf);
gt_encseq_reader_reinit_with_readmode(esr, suffixArray.encseq,
suffixArray.readmode, start);
for (j = 0; j < inSubSeqLen; ++j)
{
Symbol symRef = gt_encseq_reader_next_encoded_char(esr);
Symbol symCmp = contextBuf[j];
if (symCmp != symRef)
{
gt_error_set(err, "symbol mismatch at position "GT_WU": "
"%d vs. reference symbol %d", start + j, (int)symCmp,
(int)symRef);
retval = VERIFY_BWTSEQ_CONTEXT_SYMFAIL;
break;
}
}
while (j < subSeqLen)
{
Symbol symRef = UNDEFBWTCHAR;
Symbol symCmp = contextBuf[j];
if (symCmp != symRef)
{
gt_error_set(err, "symbol mismatch at position "GT_WU": "
"%d vs. reference symbol %d", start + j, (int)symCmp,
(int)symRef);
retval = VERIFY_BWTSEQ_CONTEXT_SYMFAIL;
break;
}
++j;
}
}
if (retval == VERIFY_BWTSEQ_NO_ERROR)
fputs("Context regeneration completed successfully.\n", stderr);
gt_encseq_reader_delete(esr);
gt_free(contextBuf);
}
gt_deleteBWTSeqCR(bwtSeqCR);
}
} while (0);
if (suffixArrayIsInitialized) gt_freesuffixarray(&suffixArray);
if (extBitsAreInitialized) destructExtBitsRetrieval(&extBits);
return retval;
}