void tagStormQueryMain(char *query)
/* tagStormQuery - Find stanzas in tag storm based on SQL-like query.. */
{
/* Get parsed out query */
struct lineFile *lf = lineFileOnString("query", TRUE, cloneString(query));
struct rqlStatement *rql = rqlStatementParse(lf);
int stormCount = slCount(rql->tableList);
if (stormCount != 1)
errAbort("Can only handle one tag storm file in query, got %d", stormCount);
char *tagsFileName = rql->tableList->name;
/* Read in tags */
struct tagStorm *tags = tagStormFromFile(tagsFileName);
/* Expand any field names with wildcards. */
struct slName *allFieldList = tagStormFieldList(tags);
rql->fieldList = wildExpandList(allFieldList, rql->fieldList, TRUE);
/* Traverse tree applying query */
struct lm *lm = lmInit(0);
doSelect = sameWord(rql->command, "select");
traverse(tags, tags->forest, rql, lm);
tagStormFree(&tags);
if (sameWord(rql->command, "count"))
printf("%d\n", matchCount);
}
void cdwQuery(char *rqlQuery)
/* cdwQuery - Get list of tagged files.. */
{
/* Turn rqlQuery string into a parsed out rqlStatement. */
struct rqlStatement *rql = rqlStatementParseString(rqlQuery);
/* Load tags from database */
struct sqlConnection *conn = cdwConnect();
struct tagStorm *tags = cdwTagStorm(conn);
/* Get list of all tag types in tree and use it to expand wildcards in the query
* field list. */
struct slName *allFieldList = tagStormFieldList(tags);
slSort(&allFieldList, slNameCmp);
rql->fieldList = wildExpandList(allFieldList, rql->fieldList, TRUE);
/* Output header row in tab case */
if (sameString(clOut, "tab"))
{
char before = '#';
struct slName *field;
for (field = rql->fieldList; field != NULL; field = field->next)
{
printf("%c%s", before, field->name);
before = '\t';
}
printf("\n");
}
/* Traverse tag tree outputting when rql statement matches in select case, just
* updateing count in count case. */
doSelect = sameWord(rql->command, "select");
struct lm *lm = lmInit(0);
traverse(0, tags, tags->forest, rql, lm);
if (sameWord(rql->command, "count"))
printf("%d\n", matchCount);
/* Clean up and go home. */
tagStormFree(&tags);
}
void hcaStormToBundles(char *inTags, char *dataUrl, char *schemaFile, char *outDir)
/* hcaStormToBundles - Convert a HCA formatted tagStorm to a directory full of bundles.. */
{
/* Check that have full path name for dataFileDir */
if (sameString("urls", dataUrl))
gUrls = TRUE;
else if (!stringIn("://", dataUrl))
errAbort("data file directory must be a url.");
/* Load up schema and put it in hash */
struct tagSchema *schemaList = tagSchemaFromFile(schemaFile);
struct hash *schemaHash = tagSchemaHash(schemaList);
/* Load up tagStorm get leaf list */
struct tagStorm *storm = tagStormFromFile(inTags);
struct tagStanzaRef *refList = tagStormListLeaves(storm);
verbose(1, "Got %d leaf nodes in %s\n", slCount(refList), inTags);
/* Add in assay.sample_id as just a dupe of sample.id */
dupeValToNewTag(storm, storm->forest, "sample.id", "assay.sample_id");
dupeValToNewTag(storm, storm->forest, "project.id", "sample.project_id");
addMissingUuids(storm, "assay.seq.ena_experiment", "assay.id", FALSE);
addMissingUuids(storm, "assay.seq.sra_experiment", "assay.id", FALSE);
/* Do some figuring based on all fields available of what objects to make */
struct slName *allFields = tagStormFieldList(storm);
verbose(1, "Got %d fields in %s\n", slCount(allFields), inTags);
struct slName *topLevelList = ttjUniqToDotList(allFields, NULL, 0);
verbose(1, "Got %d top level objects\n", slCount(topLevelList));
/* Make list of objects */
struct slName *topEl;
struct ttjSubObj *objList = NULL;
for (topEl = topLevelList; topEl != NULL; topEl = topEl->next)
{
verbose(1, " %s\n", topEl->name);
struct ttjSubObj *obj = ttjMakeSubObj(allFields, topEl->name, topEl->name);
slAddHead(&objList, obj);
}
/* Loop through stanzas making bundles */
struct tagStanzaRef *ref;
int bundleIx = 0;
makeDirsOnPath(outDir);
for (ref = refList; ref != NULL; ref = ref->next)
{
/* Fetch stanza and comma-separated list of files. */
struct tagStanza *stanza = ref->stanza;
char *fileCsv = tagFindVal(stanza, "assay.seq.files");
if (fileCsv == NULL)
errAbort("Stanza without a files tag. Stanza starts line %d of %s",
stanza->startLineIx, inTags);
/* Make subdirectory for bundle */
++bundleIx;
char bundleDir[PATH_LEN];
safef(bundleDir, sizeof(bundleDir), "%s/bundle%d", outDir, bundleIx);
makeDir(bundleDir);
/* Make symbolic link of all files */
char localUrl[PATH_LEN*2];
if (gUrls)
{
struct slName *fileList = tagMustFindValList(stanza, "assay.seq.files");
splitPath(fileList->name, localUrl, NULL, NULL);
dataUrl = localUrl;
slFreeList(&fileList);
}
makeBundleJson(storm, bundleDir, stanza, objList, dataUrl, schemaHash);
}
verbose(1, "wrote json files into %s/bundle* dirs\n", outDir);
}
