/**
* Compress data from fd_source into fd_dest.
*
* @param fd_source Source file descriptor
* @param fd_dest Destination file descriptor
* @return Number of bytes readed from fd_source or -1 on error
*/
static off_t gzCompress(void *cancel_cookie, int fd_source, int fd_dest)
{
char buf[BUF_SIZE];
int wr;
int rd;
gzFile fd_gz = NULL;
off_t size = 0;
int dup_fd;
dup_fd = dup(fd_dest);
if (dup_fd == -1)
{
return (off_t) FAIL;
}
fd_gz = gzdopen(dup_fd, COMPRESSLEVEL_BACKGROUND);
if (fd_gz == NULL)
{
file_close(&dup_fd);
return (off_t) FAIL;
}
while ((rd = read(fd_source, buf, sizeof(buf))) > 0)
{
size += rd;
wr = gzwrite(fd_gz, buf, rd);
if(wr == -1)
{
gzClose(fd_gz);
return (off_t) FAIL;
}
if (compress_testcancel(cancel_cookie))
{
break;
}
}
if (rd < 0)
{
gzClose(fd_gz);
return (off_t) FAIL;
}
gzClose(fd_gz);
return size;
}
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);
}
void gzifstream::close()
{
if (f)
{
gzClose();
fclose(f);
f = NULL;
}
}
/**
* Decompress data from fd_source into fd_dest.
*
* @param fd_source Source file descriptor
* @param fd_dest Destination file descriptor
* @return Number of bytes written to fd_dest or (off_t)-1 on error
*/
static off_t gzDecompress(int fd_source, int fd_dest)
{
int wr;
int rd;
char buf[BUF_SIZE];
off_t size = 0;
gzFile fd_gz;
int dup_fd;
dup_fd = dup(fd_source);
if (dup_fd == -1)
{
return (off_t) FAIL;
}
fd_gz = gzdopen(dup_fd, "rb");
if (fd_gz == NULL)
{
file_close(&dup_fd);
return (off_t) FAIL;
}
while ((rd = gzread(fd_gz, buf, BUF_SIZE)) > 0)
{
wr = write(fd_dest, buf, rd);
if(wr == -1)
{
gzClose(fd_gz);
return (off_t) FAIL;
}
size += wr;
}
if (rd == -1)
{
gzClose(fd_gz);
return (off_t) FAIL;
}
gzClose(fd_gz);
return size;
}
void gbOutputRemove(char* path, FILE** fhPtr)
/* Remove an output file instead of installing it. If fh is not NULL, the file
* is also closed. */
{
char tmpPath[PATH_LEN];
gbGetOutputTmp(path, tmpPath);
if (fhPtr != NULL)
gzClose(fhPtr);
unlink(tmpPath);
}
void gzofstream::close()
{
if (f)
{
bool write_error = !gzClose();
fclose(f);
f = NULL;
if (write_error)
throw FileWriteError(gz_filename);
}
}
void gbOutputRename(char* path, FILE** fhPtr)
/* Install an output file create by gbOutputOpen. If fh is not NULL, the file
* is also closed. If closed separately, must use gzClose. */
{
char tmpPath[PATH_LEN];
gbGetOutputTmp(path, tmpPath);
if (fhPtr != NULL)
gzClose(fhPtr);
if (!sameString(path, tmpPath))
{
if (rename(tmpPath, path) < 0)
errnoAbort("renaming %s to %s", tmpPath, path);
}
}
void gbFaClose(struct gbFa **faPtr)
/* close a fasta file, check for any undetected I/O errors. */
{
struct gbFa *fa = *faPtr;
if (fa != NULL)
{
if (ferror(fa->fh))
errAbort("%s error on %s", ((fa->mode[0] != 'r') ? "write" : "read"),
fa->fileName);
if (fa->mode[0] == 'w')
gbOutputRename(fa->fileName, &fa->fh);
else
gzClose(&fa->fh);
freez(&fa->headerBuf);
freez(&fa->seqBuf);
freez(&fa->fhBuf);
freez(faPtr); /* NULLs var */
}
}
void oiDataClose()
/* close the output file */
{
gzClose(&gOutOi);
}
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);
}
