void gapFileToTable(struct sqlConnection *conn, char *gapFileName,
char *gapTableName)
/* Build a single gap table from a single gap file. */
{
struct lineFile *lf = lineFileOpen(gapFileName, TRUE);
char tabFileName[256];
FILE *tabFile = NULL;
char *words[16];
int wordCount;
safef(tabFileName, sizeof(tabFileName), "%s.tab", gapTableName);
tabFile = mustOpen(tabFileName, "w");
while ((wordCount = lineFileChop(lf, words)) > 0)
{
if (wordCount < 5)
errAbort("Short line %d of %s", lf->lineIx, lf->fileName);
if (words[4][0] == 'N' || words[4][0] == 'U')
{
int len = strlen(words[0]);
if (len > maxChromNameSize)
{
maxChromNameSize = len;
if (maxChromNameSize > 254)
errAbort("ERROR: chrom name size is over 254(%d) characters: "
"'%s'", maxChromNameSize, words[0]);
}
struct agpGap gap;
agpGapStaticLoad(words, &gap);
gap.chromStart -= 1;
fprintf(tabFile, "%u\t", hFindBin(gap.chromStart, gap.chromEnd));
agpGapTabOut(&gap, tabFile);
}
}
lineFileClose(&lf);
fclose(tabFile);
if (! noLoad)
{
struct dyString *ds = newDyString(2048);
if (unsplit)
sqlDyStringPrintf(ds, createGapUnsplit, gapTableName,
maxChromNameSize, maxChromNameSize);
else
sqlDyStringPrintf(ds, createGapSplit, gapTableName);
char query[1024];
sqlRemakeTable(conn, gapTableName, ds->string);
sqlSafef(query, sizeof(query), "LOAD data local infile '%s' into table %s",
tabFileName, gapTableName);
sqlUpdate(conn, query);
remove(tabFileName);
freeDyString(&ds);
}
}
void splitAgp(char *agpName, char *goldFileName, char *gapFileName)
/* Split up agp file into gold and gap files. */
{
struct lineFile *lf;
char *words[16];
int wordCount;
FILE *goldTab, *gapTab;
/* Scan through .agp file splitting it into gold
* and gap components. */
goldTab = mustOpen(goldFileName, "w");
gapTab = mustOpen(gapFileName, "w");
lf = lineFileOpen(agpName, TRUE);
while ((wordCount = lineFileChop(lf, words)) > 0)
{
int start, end;
if (wordCount < 5)
errAbort("Short line %d of %s", lf->lineIx, lf->fileName);
int len = strlen(words[0]);
if (len > maxChromNameSize)
{
maxChromNameSize = len;
if (maxChromNameSize > 254)
errAbort("ERROR: chrom name size is over 254(%d) characters: "
"'%s'", maxChromNameSize, words[0]);
}
start = sqlUnsigned(words[1])-1;
end = sqlUnsigned(words[2]);
if (words[4][0] == 'N' || words[4][0] == 'U')
{
struct agpGap gap;
agpGapStaticLoad(words, &gap);
gap.chromStart -= 1;
fprintf(gapTab, "%u\t", hFindBin(start, end));
agpGapTabOut(&gap, gapTab);
verbose(3,"#GAP\t%s:%d-%d\n", gap.chrom, gap.chromStart, gap.chromEnd);
}
else
{
struct agpFrag gold;
agpFragStaticLoad(words, &gold);
agpFragValidate(&gold);
len = strlen(words[5]);
if (len > maxFragNameSize)
{
maxFragNameSize = len;
if (maxFragNameSize > 254)
errAbort("ERROR: fragment name size is over 254(%d) "
"characters: '%s'", maxFragNameSize, words[5]);
}
// file is 1-based. agpFragLoad() now assumes 0-based.
// and agpFragTabOut() will assume 1-based, but we will load
// the generated file straight into the database, so
// subtract 2:
gold.chromStart -= 2;
gold.fragStart -= 2;
fprintf(goldTab, "%u\t", hFindBin(start, end));
agpFragTabOut(&gold, goldTab);
}
}
lineFileClose(&lf);
carefulClose(&goldTab);
carefulClose(&gapTab);
}
