void docIdReport(char *database)
/* docIdReport - generate report from docIdSub table. */
{
struct sqlConnection *conn = sqlConnect(database);
struct docIdSub *docIdSub;
char query[10 * 1024];
struct sqlResult *sr;
char **row;
safef(query, sizeof query, "select * from %s", docIdTable);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
docIdSub = docIdSubLoad(row);
verbose(2, "ix %d\n", docIdSub->ix);
verbose(2, "submitDate %s\n", docIdSub->submitDate);
verbose(2, "md5sum %s\n", docIdSub->md5sum);
verbose(2, "valReport %s\n", docIdSub->valReport);
verbose(2, "metaData %s\n", docIdSub->metaData);
verbose(2, "submitPath %s\n", docIdSub->submitPath);
verbose(2, "submitter %s\n", docIdSub->submitter);
cgiDecode(docIdSub->metaData, docIdSub->metaData, strlen(docIdSub->metaData));
char *tempFile = "temp";
FILE *f = mustOpen(tempFile, "w");
fwrite(docIdSub->metaData, strlen(docIdSub->metaData), 1, f);
fclose(f);
boolean validated;
struct mdbObj *mdbObj = mdbObjsLoadFromFormattedFile(tempFile, &validated);
printf("%s %s %s %s %s\n", docIdDecorate(docIdSub->ix), mdbObjFindValue(mdbObj, "dataType"), mdbObjFindValue(mdbObj, "view"), mdbObjFindValue(mdbObj, "cell"), mdbObjFindValue(mdbObj, "lab"));
}
sqlFreeResult(&sr);
sqlDisconnect(&conn);
}
void expFind(char *assembly, char *file)
/* Find experiments in metaDb and output .ra file */
{
verbose(1, "Finding experiments in %s:%s\n", assembly, mdb);
struct sqlConnection *connMeta;
struct mdbObj *meta = NULL, *metas = NULL;
struct encodeExp *exp = NULL, *exps = NULL;
struct hash *oldExps, *newExps;
char *key;
int expNum = 0;
FILE *f = mustOpen(file, "w");
/* create hash of keys for existing experiments so we can distinguish new ones */
oldExps = expKeyHashFromTable(connExp, table);
newExps = hashNew(0);
/* read mdb objects from database */
connMeta = sqlConnect(assembly);
metas = mdbObjsQueryAll(connMeta, mdb);
verbose(2, "Found %d objects\n", slCount(metas));
/* order so that oldest have lowest ids */
mdbObjsSortOnVars(&metas, "dateSubmitted lab dataType cell");
/* create new experiments */
while ((meta = slPopHead(&metas)) != NULL)
{
if (!mdbObjIsEncode(meta))
continue;
if (composite != NULL && !mdbObjInComposite(meta, composite))
continue;
exp = encodeExpFromMdb(connMeta,assembly,meta);
if (exp == NULL)
continue;
key = encodeExpKey(exp);
verbose(3, "key: %s\n", key);
if (hashLookup(newExps, key) == NULL &&
hashLookup(oldExps, key) == NULL)
{
verbose(2, "Found new experiment - Date: %s Experiment %d: %s\n",
mdbObjFindValue(meta, "dateSubmitted"), ++expNum, key);
/* save new experiment */
hashAdd(newExps, key, NULL);
slAddHead(&exps, exp);
}
/* Skip other metas belonging to the same exp by:
struct mdbVar *edvs = mdbObjFindEncodeEdvs(connMeta,meta); // Can't use encodeExpVars(exp) because of "None" issues
assert(edvs != NULL);
char *expVars = slPairListToString(edvs,FALSE); // don't bother with quoting since edvs should not have spaces
struct mdbObj *mdbExpObjs = mdbObjsFilterByVars(&metas,expVars,TRUE,TRUE);
freeMem(expVars);
mdbVarsFree(&edvs); // If you want to do this, then encodeExpFromMdb() above should be replaced with encodeExpFromMdbVars()
mdbObjFree(&mdbExpObjs);
// Filtering destroyed sort order // NOTE: Given the re-sort, this may not prove much more efficient
mdbObjsSortOnVars(&metas, "dateSubmitted lab dataType cell");
*/
}
/* write out experiments in .ra format */
slReverse(&exps);
while ((exp = slPopHead(&exps)) != NULL)
{
exp->organism = organism;
encodeExpToRaFile(exp, f);
}
carefulClose(&f);
sqlDisconnect(&connMeta);
}
void doStandard(struct cart *theCart)
{
cart = theCart;
cartWebStart(cart, database, "ENCODE DCC Submissions");
struct sqlConnection *conn = sqlConnect(database);
struct docIdSub *docIdSub;
char query[10 * 1024];
struct sqlResult *sr;
char **row;
struct tempName tn;
trashDirFile(&tn, "docId", "meta", ".txt");
char *tempFile = tn.forCgi;
//printf("tempFile is %s\n<BR>", tempFile);
// <Data type> <Cell Type> <Key Metadata> <View>
printf("<table border=1><tr>");
printf("<th>dataType</th>");
printf("<th>cell type</th>");
printf("<th>metadata</th>");
printf("<th>view</th>");
printf("<th>fileType</th>");
printf("<th>file</th>");
printf("<th>lab</th>");
printf("<th>assembly</th>");
printf("<th>subId</th>");
printf("<th>val-report</th>");
printf("</tr>\n");
safef(query, sizeof query, "select * from %s", docIdTable);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
docIdSub = docIdSubLoad(row);
verbose(2, "ix %d\n", docIdSub->ix);
verbose(2, "submitDate %s\n", docIdSub->submitDate);
verbose(2, "md5sum %s\n", docIdSub->md5sum);
verbose(2, "valReport %s\n", docIdSub->valReport);
verbose(2, "metaData %s\n", docIdSub->metaData);
verbose(2, "submitPath %s\n", docIdSub->submitPath);
verbose(2, "submitter %s\n", docIdSub->submitter);
cgiDecode(docIdSub->metaData, docIdSub->metaData, strlen(docIdSub->metaData));
//printf("tempFile %s\n", tempFile);
FILE *f = mustOpen(tempFile, "w");
fwrite(docIdSub->metaData, strlen(docIdSub->metaData), 1, f);
fclose(f);
boolean validated;
struct mdbObj *mdbObj = mdbObjsLoadFromFormattedFile(tempFile, &validated);
unlink(tempFile);
// <Data type> <Cell Type> <Key Metadata> <View>
char *docIdType = mdbObjFindValue(mdbObj, "type");
char *docIdComposite = mdbObjFindValue(mdbObj, "composite");
char buffer[10 * 1024];
safef(buffer, sizeof buffer, "%d", docIdSub->ix);
if (sameString(database, "encpipeline_beta"))
docIdDir = docIdDirBeta;
printf("<tr>");
printf("<td>%s</td> ", mdbObjFindValue(mdbObj, "dataType"));
printf("<td>%s</td> ", mdbObjFindValue(mdbObj, "cell"));
struct dyString *str = newDyString(100);
addValue(str, mdbObjFindValue(mdbObj, "antibody"));
addValue(str, mdbObjFindValue(mdbObj, "treatment"));
addValue(str, mdbObjFindValue(mdbObj, "rnaExtract"));
addValue(str, mdbObjFindValue(mdbObj, "localization"));
printf("<td>%s<a href=docIdView?docId=%s&db=%s&meta=\"\"> ...</a></td>", str->string,buffer, database);
freeDyString(&str);
printf("<td>%s</td> ", mdbObjFindValue(mdbObj, "view"));
printf("<td>%s</td> ", mdbObjFindValue(mdbObj, "type"));
printf("<td><a href=%s> %s</a></td>",
docIdGetPath(buffer, docIdDir, docIdType, NULL) ,
docIdDecorate(docIdComposite,docIdSub->ix));
char *lab = mdbObjFindValue(mdbObj, "lab");
char *subId = mdbObjFindValue(mdbObj, "subId");
printf("<td><a href=docIdView?docId=%s&db=%s&lab=\"%s\"> %s</a></td>",buffer, database, subId, lab);
printf("<td>%s</td> ", mdbObjFindValue(mdbObj, "assembly"));
printf("<td>%s</td> ", subId);
printf("<td><a href=docIdView?docId=%s&db=%s&report=\"\"> report</a></td>", buffer, database);
printf("</tr>\n");
}
printf("</table>");
sqlFreeResult(&sr);
sqlDisconnect(&conn);
cartWebEnd();
}
void docIdTidy(char *database, char *docIdDir)
/* docIdTidy - tidy up the docId library by compressing and md5suming where appropriate. */
{
char query[10 * 1024];
struct sqlResult *sr;
char **row;
struct sqlConnection *conn = sqlConnect(database);
struct tempName tn;
trashDirFile(&tn, "docId", "meta", ".txt");
char *tempFile = tn.forCgi;
struct toDoList *toDoList = NULL;
safef(query, sizeof query, "select * from %s", docIdTable);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
struct docIdSub *docIdSub = docIdSubLoad(row);
cgiDecode(docIdSub->metaData, docIdSub->metaData, strlen(docIdSub->metaData));
FILE *f = mustOpen(tempFile, "w");
fwrite(docIdSub->metaData, strlen(docIdSub->metaData), 1, f);
fclose(f);
boolean validated;
//printf("metadata is %s\n", docIdSub->metaData);
struct mdbObj *mdbObj = mdbObjsLoadFromFormattedFile(tempFile, &validated);
unlink(tempFile);
char *docIdType = mdbObjFindValue(mdbObj, "type");
char buffer[10 * 1024];
safef(buffer, sizeof buffer, "%d", docIdSub->ix);
char *path = docIdGetPath(buffer, docIdDir, docIdType, NULL);
//docIdDecorate(docIdSub->ix));
printf("path %s\n", path);
struct toDoList *toDoItem = NULL;
if (!fileIsCompressed(path))
{
if (toDoItem == NULL) AllocVar(toDoItem);
printf("foo\n");
toDoItem->needs |= NEEDS_COMPRESSION;
toDoItem->path = path;
toDoItem->docId = docIdSub->ix;
}
printf("docId %d md5sum %s valReport %s\n",docIdSub->ix, docIdSub->md5sum, docIdSub->valReport);
//if (docIdSub->md5sum == NULL)
if (sameString(docIdSub->md5sum, ""))
{
printf("mdsum\n");
if (toDoItem == NULL) AllocVar(toDoItem);
toDoItem->needs |= NEEDS_MD5SUM;
toDoItem->path = path;
toDoItem->docId = docIdSub->ix;
}
//if (docIdSub->valReport == NULL)
if (sameString(docIdSub->valReport,""))
{
printf("report\n");
if (toDoItem == NULL) AllocVar(toDoItem);
toDoItem->needs |= NEEDS_REPORT;
toDoItem->path = path;
toDoItem->docId = docIdSub->ix;
}
if (toDoItem)
slAddHead(&toDoList, toDoItem);
}
sqlFreeResult(&sr);
doCompression(toDoList);
doReports(toDoList);
doMd5Summing(toDoList);
}
