void edwCorrectFileTags(char *tabFileName)
/* edwCorrectFileTags - Use this to correct tags in the edwFile table and corresponding fields
* in the edwValidFile table without forcing a validateManifest rerun or a reupload.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
char *requiredFields[] = {"accession",};
char *forbiddenFields[] = {"md5_sum", "size", "valid_key", "file_name"};
struct fieldedTable *table = fieldedTableFromTabFile(tabFileName, tabFileName,
requiredFields, ArraySize(requiredFields));
checkForbiddenFields(table, forbiddenFields, ArraySize(forbiddenFields));
int accessionIx = stringArrayIx("accession", table->fields, table->fieldCount);
struct fieldedRow *fr;
for (fr = table->rowList; fr != NULL; fr = fr->next)
{
char *acc = fr->row[accessionIx];
long long id = edwNeedFileIdForLicensePlate(conn, acc);
struct edwFile *ef = edwFileFromId(conn, id);
int i;
char *tags = ef->tags;
for (i=0; i<table->fieldCount; ++i)
{
if (i != accessionIx)
tags = cgiStringNewValForVar(tags, table->fields[i], fr->row[i]);
}
edwFileResetTags(conn, ef, tags);
edwFileFree(&ef);
}
}
void changeFormat(struct sqlConnection *conn, struct edwValidFile *vf, char *format)
/* Set up vf to change format. */
{
struct edwFile *ef = edwFileFromId(conn, vf->fileId);
char *newTags = cgiStringNewValForVar(ef->tags, tagToChange, format);
edwFileResetTags(conn, ef, newTags);
edwFileFree(&ef);
}
static int handleOldFileTags(struct sqlConnection *conn, struct submitFileRow *sfrList,
boolean update)
/* Check metadata on files mentioned in manifest that by MD5 sum we already have in
* warehouse. We may want to update metadata on these. This returns the number
* of files with tags updated. */
{
struct submitFileRow *sfr;
int updateCount = 0;
for (sfr = sfrList; sfr != NULL; sfr = sfr->next)
{
struct edwFile *newFile = sfr->file;
struct edwFile *oldFile = edwFileFromId(conn, sfr->md5MatchFileId);
verbose(2, "looking at old file %s (%s)\n", oldFile->submitFileName, newFile->submitFileName);
struct cgiDictionary *newTags = cgiDictionaryFromEncodedString(newFile->tags);
struct cgiDictionary *oldTags = cgiDictionaryFromEncodedString(oldFile->tags);
boolean updateName = !sameString(oldFile->submitFileName, newFile->submitFileName);
boolean updateTags = !cgiDictionarySame(oldTags, newTags);
if (updateName)
{
if (!update)
errAbort("%s already uploaded with name %s. Please use the 'update' option if you "
"want to give it a new name.",
newFile->submitFileName, oldFile->submitFileName);
updateSubmitName(conn, oldFile->id, newFile->submitFileName);
}
if (updateTags)
{
if (!update)
{
char *name="", *oldVal="", *newVal="";
cgiDictionaryFirstDiff(oldTags, newTags, &name, &oldVal, &newVal);
errAbort("%s is duplicate of %s in warehouse, but %s column went from %s to %s.\n"
"Please use the 'update' option if you are meaning to update the information\n"
"associated with this file and try again if this is intentional.",
newFile->submitFileName, oldFile->edwFileName,
name, oldVal, newVal);
}
edwFileResetTags(conn, oldFile, newFile->tags, TRUE);
}
if (updateTags || updateName)
++updateCount;
cgiDictionaryFree(&oldTags);
cgiDictionaryFree(&newTags);
}
return updateCount;
}
void redoOne(struct sqlConnection *conn, struct edwFile *redoEf)
/* Redo one file. */
{
/* Figure out submit file name of the gtf file. */
char gtfFileName[PATH_LEN];
strcpy(gtfFileName, redoEf->submitFileName);
chopSuffix(gtfFileName);
strcat(gtfFileName, ".gz");
/* Get edwFile record for gtf file. */
char query[PATH_LEN+64];
safef(query, sizeof(query), "select * from edwFile where submitFileName='%s'", gtfFileName);
struct edwFile *sourceEf = edwFileLoadByQuery(conn, query);
assert(slCount(sourceEf) == 1);
/* Get UCSC database */
safef(query, sizeof(query), "select ucscDb from edwValidFile where fileId=%u", sourceEf->id);
char ucscDb[64] = "";
sqlQuickQuery(conn, query, ucscDb, sizeof(ucscDb));
assert(ucscDb[0] != 0);
/* Remake the big bed file. */
char sourceFileName[PATH_LEN], destFileName[PATH_LEN];
safef(sourceFileName, sizeof(sourceFileName), "%s%s", edwRootDir, sourceEf->edwFileName);
safef(destFileName, sizeof(destFileName), "%s%s", edwRootDir, redoEf->edwFileName);
makeGtfBigBed(ucscDb, sourceFileName, destFileName);
/* Recalculate size and md5 sum and validation key. */
char *md5 = md5HexForFile(destFileName);
long long size = fileSize(destFileName);
char *validKey = encode3CalcValidationKey(md5, size);
/* Issue command to update md5 in database. */
char command[2*PATH_LEN];
safef(command, sizeof(command),
"hgsql -e 'update edwFile set md5=\"%s\" where id=%u' encodeDataWarehouse", md5, redoEf->id);
doSystem(command);
/* Issue command to update tags in database. */
char *newTags = cgiStringNewValForVar(redoEf->tags, "valid_key", validKey);
if (doReal)
{
edwFileResetTags(conn, redoEf, newTags);
}
}
