void loadMgcStatus(struct sqlConnection *conn, char *mgcStatusTab, char *statusTblName)
/* load the mgcStatus or mgcFullStatus tables, return name loaded */
{
struct lineFile* inLf;
FILE *outFh;
char tmpFile[PATH_LEN];
gbVerbEnter(2, "loading %s", statusTblName);
/* uncompress to tmp file */
safef(tmpFile, sizeof(tmpFile), "%s/mgcStatus.%s.%d.tmp", workDir,
getHost(), getpid());
inLf = gzLineFileOpen(mgcStatusTab);
outFh = gzMustOpen(tmpFile, "w");
while (mgcStatusTblCopyRow(inLf, outFh))
continue;
gzClose(&outFh);
gzLineFileClose(&inLf);
mgcStatusTblCreate(conn, statusTblName);
sqlLoadTabFile(conn, tmpFile, statusTblName, SQL_TAB_FILE_ON_SERVER);
unlink(tmpFile);
gbVerbLeave(2, "loading %s", statusTblName);
}
FILE* gbMustOpenOutput(char* path)
/* Open an output file for atomic file creation. Create the directory if it
* doesn't exist. If the path has one of the compress extensions (.gz, .Z,
* .bz2) the output is compressed. File is opened under a tmp name until
* installed. */
{
char tmpPath[PATH_LEN];
gbGetOutputTmp(path, tmpPath);
gbMakeFileDirs(tmpPath);
return gzMustOpen(tmpPath, "w");
}
struct gbFa* gbFaOpen(char *fileName, char* mode)
/* open a fasta file for reading or writing. Uses atomic file
* creation for write.*/
{
struct gbFa *fa;
AllocVar(fa);
strcpy(fa->fileName, fileName);
if (!(sameString(mode, "r") || sameString(mode, "w") || sameString(mode, "a")))
errAbort("gbFaOpen only supports modes \"r\", \"w\" \"a\", got \"%s\"",
mode);
fa->mode[0] = mode[0];
if (mode[0] == 'w')
fa->fh = gbMustOpenOutput(fileName);
else
fa->fh = gzMustOpen(fileName, mode);
if (mode[0] == 'a')
{
/* line number is wrong on append */
fa->off = ftello(fa->fh);
if (fa->off < 0)
errnoAbort("can't get offset for append access for file: %s",
fileName);
}
fa->fhBuf = needMem(FA_STDIO_BUFSIZ);
setbuffer(fa->fh, fa->fhBuf, FA_STDIO_BUFSIZ);
/* setup buffers if read access */
if (mode[0] == 'r')
{
fa->headerCap = 256;
fa->headerBuf = needMem(fa->headerCap);
fa->seqCap = 1024;
fa->seqBuf = needMem(fa->seqCap);
}
return fa;
}
void oiDataOpen(boolean inclVersion, char *outFile)
/* open output file and set options. */
{
gInclVersion = inclVersion;
gOutOi = gzMustOpen(outFile, "w");
}
void extractAccFromGb(char *inName, char* outName, struct hash *accTbl)
/* Parse records of genBank file and print ones that match accession names.
* (yanked from gbOneAcc, changed to use stdio so we can access compressed). */
{
enum {maxHeadLines=20, headLineSize=256 };
char *headLines[maxHeadLines]; /* Store stuff between locus and accession. */
char line[headLineSize];
FILE *inFh;
FILE *outFh = NULL;
int lineNum = 0;
int i;
char* acc;
verbose(1, "copying from %s\n", inName);
inFh = gzMustOpen(inName, "r");
for (i=0; i<maxHeadLines; ++i)
headLines[i] = needMem(headLineSize);
while (TRUE)
{
boolean gotAcc = FALSE;
boolean gotMyAcc = FALSE;
int headLineCount = 0;
/* Seek to LOCUS */
for (;;)
{
if (!readData(inFh, inName, line, headLineSize, FALSE))
break;
lineNum++;
if (startsWith("LOCUS", line))
break;
}
if (feof(inFh))
break;
for (i=0; i<maxHeadLines; ++i)
{
++headLineCount;
strcpy(headLines[i], line);
readData(inFh, inName, line, headLineSize, TRUE);
lineNum++;
if (startsWith("ACCESSION", line))
{
gotAcc = TRUE;
break;
}
}
if (!gotAcc)
errAbort("LOCUS without ACCESSION in %d lines at line %d of %s",
maxHeadLines, lineNum, inName);
acc = lastWordInLine(line);
gotMyAcc = (hashLookup(accTbl, acc) != NULL);
if (gotMyAcc)
{
if (outFh == NULL)
outFh = gbMustOpenOutput(outName);
for (i=0; i<headLineCount; ++i)
{
fputs(headLines[i], outFh);
fputc('\n', outFh);
}
fputs(line, outFh);
fputc('\n', outFh);
}
for (;;)
{
readData(inFh, inName, line, headLineSize, TRUE);
lineNum++;
if (gotMyAcc)
{
fputs(line, outFh);
fputc('\n', outFh);
}
if (startsWith("//", line))
break;
}
if ((outFh != NULL) && ferror(outFh))
break; /* write error */
}
if (outFh != NULL)
gbOutputRename(outName, &outFh);
gzClose(&inFh);
}
