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 edwMakeContaminationQa(int startId, int endId)
/* edwMakeContaminationQa - Screen for contaminants by aligning against contaminant genomes.. */
{
/* Make list with all files in ID range */
struct sqlConnection *conn = edwConnectReadWrite();
struct edwFile *ef, *efList = edwFileLoadIdRange(conn, startId, endId);
for (ef = efList; ef != NULL; ef = ef->next)
{
doContaminationQa(conn, ef);
}
}
void edwAddQaContamTarget(char *assemblyName)
/* edwAddQaContamTarget - Add a new contamination target to warehouse.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
char query[256 + PATH_LEN];
sqlSafef(query, sizeof(query), "select id from edwAssembly where name='%s'", assemblyName);
int assemblyId = sqlQuickNum(conn, query);
if (assemblyId == 0)
errAbort("Assembly %s doesn't exist in warehouse. Typo or time for edwAddAssembly?",
assemblyName);
sqlSafef(query, sizeof(query), "insert edwQaContamTarget(assemblyId) values(%d)", assemblyId);
sqlUpdate(conn, query);
printf("Added target %s\n", assemblyName);
}
void edwFixGtfBigBed(char *how)
/* edwFixGtfBigBed - In original import the .gtf.bigBed files were bad about half the time. Cricket
* caught this because a bunch of them ended up with the same md5 sum. This program regenerates them
* all. */
{
doReal = sameString(how, "real");
struct sqlConnection *conn = edwConnectReadWrite();
struct edwFile *redoEf, *redoList =
edwFileLoadByQuery(conn, "select * from edwFile where submitFileName like '%.gtf.bigBed'");
for (redoEf = redoList; redoEf != NULL; redoEf = redoEf->next)
{
redoOne(conn, redoEf);
}
}
void doMiddle()
/* doMiddle - put up middle part of web page, not including http and html headers/footers */
{
printf("<FORM ACTION=\"../cgi-bin/edwWebDeprecate\" METHOD=GET>\n");
struct sqlConnection *conn = edwConnectReadWrite(edwDatabase);
userEmail = edwGetEmailAndVerify();
if (userEmail == NULL)
logIn();
else if (cgiVarExists("fileList") && cgiVarExists("reason"))
tryToDeprecate(conn);
else
getFileListAndReason(conn);
printf("</FORM>");
}
void edwRunOnIds(char *program, char *queryString)
/* edwRunOnIds - Run a edw command line program (one that takes startId endId as it's two parameters) for a range of ids,
* putting it on edwJob queue. */
{
struct sqlConnection *conn = edwConnectReadWrite();
struct slName *id, *idList = sqlQuickList(conn, queryString);
for (id = idList; id != NULL; id = id->next)
{
char query[512];
sqlSafef(query, sizeof(query), "insert into %s (commandLine) values ('%s %s %s')",
runTable, program, id->name, id->name);
sqlUpdate(conn, query);
}
}
void edwMakeRepeatQa(int startFileId, int endFileId)
/* edwMakeRepeatQa - Figure out what proportion of things align to repeats.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
struct edwFile *ef, *efList = edwFileAllIntactBetween(conn, startFileId, endFileId);
for (ef = efList; ef != NULL; ef = ef->next)
{
struct edwValidFile *vf = edwValidFileFromFileId(conn, ef->id);
if (vf != NULL)
{
if (sameString(vf->format, "fastq"))
fastqRepeatQa(conn, ef, vf);
}
}
sqlDisconnect(&conn);
}
void edwCreateNewUser(char *email)
/* Create new user, checking that user does not already exist. */
{
/* Now make sure user is not already in user table. */
struct sqlConnection *conn = edwConnectReadWrite();
struct dyString *query = dyStringNew(0);
sqlDyStringPrintf(query, "select count(*) from edwUser where email = '%s'", email);
if (sqlQuickNum(conn, query->string) > 0)
errAbort("User %s already exists", email);
/* Do database insert. */
dyStringClear(query);
sqlDyStringPrintf(query, "insert into edwUser (email) values('%s')", email);
sqlUpdate(conn, query->string);
sqlDisconnect(&conn);
}
void edwFixTargetSeq(char *when)
/* edwFixTargetSeq - Fill in new fields about target seq to edwBamFile and edwAssembly.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
struct edwAssembly *as, *asList = edwAssemblyLoadByQuery(conn, "select * from edwAssembly");
char query[512];
for (as = asList; as != NULL; as = as->next)
{
char *twoBitFileName = edwPathForFileId(conn, as->twoBitId);
struct twoBitFile *tbf = twoBitOpen(twoBitFileName);
safef(query, sizeof(query), "update edwAssembly set seqCount=%u where id=%u",
tbf->seqCount, as->id);
sqlUpdate(conn, query);
freez(&twoBitFileName);
twoBitClose(&tbf);
}
edwAssemblyFreeList(&asList);
struct edwBamFile *bam, *bamList = edwBamFileLoadByQuery(conn, "select * from edwBamFile");
for (bam = bamList; bam != NULL; bam = bam->next)
{
char *fileName = edwPathForFileId(conn, bam->fileId);
samfile_t *sf = samopen(fileName, "rb", NULL);
if (sf == NULL)
errnoAbort("Couldn't open %s.\n", fileName);
bam_header_t *head = sf->header;
if (head == NULL)
errAbort("Aborting ... Bad BAM header in file: %s", fileName);
/* Sum up some target sizes. */
long long targetBaseCount = 0; /* Total size of all bases in target seq */
int i;
for (i=0; i<head->n_targets; ++i)
targetBaseCount += head->target_len[i];
safef(query, sizeof(query),
"update edwBamFile set targetBaseCount=%lld,targetSeqCount=%u where id=%u",
targetBaseCount, (unsigned)head->n_targets, bam->id);
sqlUpdate(conn, query);
samclose(sf);
freez(&fileName);
}
}
void doMiddle()
/* doMiddle - put up middle part of web page, not including http and html headers/footers */
{
pushWarnHandler(localWarn);
printf("<FORM ACTION=\"../cgi-bin/edwWebSubmit\" METHOD=GET>\n");
struct sqlConnection *conn = edwConnectReadWrite(edwDatabase);
userEmail = edwGetEmailAndVerify();
if (userEmail == NULL)
logIn();
else if (cgiVarExists(stopButtonName))
stopUpload(conn);
else if (cgiVarExists("submitUrl"))
submitUrl(conn);
else if (cgiVarExists("monitor"))
monitorSubmission(conn);
else
getUrl(conn);
printf("</FORM>");
}
void loadEapDb(char *dir)
/* Load up EAP portion of database from tab file directory. */
{
struct sqlConnection *conn = edwConnectReadWrite();
char *tables[] = {"eapJob", "eapSoftware", "eapSwVersion", "eapStep", "eapStepSoftware", "eapStepVersion",
"eapStepSwVersion", "eapAnalysis", "eapInput", "eapOutput",};
int i;
for (i=0; i<ArraySize(tables); ++i)
{
char *table = tables[i];
/* Make up tab separated file name and ask database to load files into the table. */
char tabName[PATH_LEN];
safef(tabName, PATH_LEN, "%s/%s.tab", dir, table);
char query[2*PATH_MAX];
sqlSafef(query, sizeof(query), "load data local infile '%s' into table %s", tabName, table);
verbose(2, "%s\n", query);
sqlUpdate(conn, query);
}
}
void edwChangeFormat(char *format, int idCount, char *idStrings[])
/* edwChangeFormat - Change format and force a revalidation for a file.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
/* Convert ascii id's to valid file ids so we catch errors early. */
long long ids[idCount];
struct edwValidFile *vfs[idCount];
int i;
for (i=0; i<idCount; ++i)
{
long long id = ids[i] = sqlLongLong(idStrings[i]);
struct edwValidFile *vf = vfs[i] = edwValidFileFromFileId(conn, id);
if (vf == NULL)
errAbort("%lld is not a fileId in the edwValidFile table", id);
}
/* Loop through each file and change format. */
for (i=0; i<idCount; ++i)
{
changeFormat(conn, vfs[i], format);
}
sqlDisconnect(&conn);
}
void doMiddle()
/* Write what goes between BODY and /BODY */
{
pushWarnHandler(localWarn);
if (!cgiServerHttpsIsOn())
usage();
struct sqlConnection *conn = edwConnectReadWrite();
printf("<FORM ACTION=\"edwWebRegisterScript\" METHOD=POST>\n");
printf("<B>Register Script with ENCODE Data Warehouse</B><BR>\n");
#ifdef SOON
uglyf("HTTP_AUTHENTICATION: '%s'<BR>\n", getenv("HTTP_AUTHENTICATION"));
uglyf("HTTP_AUTHORIZATION: '%s'<BR>\n", getenv("HTTP_AUTHORIZATION"));
dumpEnv(mainEnv);
#endif
if (userEmail == NULL)
{
printf("Please sign in:");
printf("<INPUT TYPE=BUTTON NAME=\"signIn\" VALUE=\"sign in\" id=\"signin\">");
}
else if (cgiVarExists("description"))
{
struct edwUser *user = edwUserFromEmail(conn, userEmail);
if (user == NULL)
edwWarnUnregisteredUser(userEmail);
else
{
char password[HEXED_32_SIZE];
edwRandomHexed32(password);
char babyName[HEXED_32_SIZE];
edwRandomBabble(babyName, sizeof(babyName));
edwRegisterScript(conn, user, babyName, password, cgiString("description"));
printf("Script now registered.<BR>\n");
printf("The script user name is %s.<BR>\n", babyName);
printf("The script password is %s.<BR>\n", password);
printf("Please save the script user name and password somewhere. ");
puts("Please pass these two and the URL");
puts(" of your validated manifest file (validated.txt) to our server to submit data.");
puts("Construct a URL of the form:<BR>");
printf("<PRE>https://encodedcc.sdsc.edu/cgi-bin/edwScriptSubmit"
"?user=%s&password=%s&url=%s\n</PRE>",
babyName, password,
cgiEncode("http://your.host.edu/your_dir/validated.txt"));
puts("That is pass the CGI encoded variables user, password, and url to the ");
puts("web services CGI at");
puts("https://encodedcc.sdsc.edu/cgi-bin/edwScriptSubmit. ");
puts("You can use the http://encodedcc.sdsc.edu/cgi-bin/edwWebBrowse site to ");
puts("monitor your submission interactively. Please contact your wrangler if you ");
puts("have any questions.<BR>");
cgiMakeButton("submit", "Register another script");
}
printf(" ");
edwPrintLogOutButton();
}
else
{
struct edwUser *user = edwUserFromEmail(conn, userEmail);
edwPrintLogOutButton();
if (user == NULL)
edwWarnUnregisteredUser(userEmail);
else
{
printf("%s is authorized to register a new script<BR>\n", userEmail);
printf("<BR>Script description:\n");
cgiMakeTextVar("description", NULL, 80);
cgiMakeSubmitButton();
}
}
printf("</FORM>\n");
}
void edwSubmit(char *submitUrl, char *email)
/* edwSubmit - Submit URL with validated.txt to warehouse. */
{
/* Parse out url a little into submitDir and submitFile */
char *lastSlash = strrchr(submitUrl, '/');
if (lastSlash == NULL)
errAbort("%s is not a valid URL - it has no '/' in it.", submitUrl);
char *submitFile = lastSlash+1;
int submitDirSize = submitFile - submitUrl;
char submitDir[submitDirSize+1];
memcpy(submitDir, submitUrl, submitDirSize);
submitDir[submitDirSize] = 0; // Add trailing zero
/* Make sure user has access. */
struct sqlConnection *conn = edwConnectReadWrite();
struct edwUser *user = edwMustGetUserFromEmail(conn, email);
int userId = user->id;
/* See if we are already running on same submission. If so council patience and quit. */
notOverlappingSelf(conn, submitUrl);
/* Make a submit record. */
int submitId = makeNewEmptySubmitRecord(conn, submitUrl, userId);
/* The next errCatch block will fill these in if all goes well. */
struct submitFileRow *sfrList = NULL, *oldList = NULL, *newList = NULL;
int oldCount = 0;
long long oldBytes = 0, newBytes = 0, byteCount = 0;
/* Start catching errors from here and writing them in submitId. If we don't
* throw we'll end up having a list of all files in the submit in sfrList. */
struct errCatch *errCatch = errCatchNew();
char query[1024];
if (errCatchStart(errCatch))
{
/* Make sure they got a bit of space, enough for a reasonable submit file.
* We do this here just because we can make error message more informative. */
long long diskFreeSpace = freeSpaceOnFileSystem(edwRootDir);
if (diskFreeSpace < 4*1024*1024)
errAbort("No space left in warehouse!");
/* Open remote submission file. This is most likely where we will fail. */
int hostId=0, submitDirId = 0;
long long startUploadTime = edwNow();
int remoteFd = edwOpenAndRecordInDir(conn, submitDir, submitFile, submitUrl,
&hostId, &submitDirId);
/* Copy to local temp file. */
char tempSubmitFile[PATH_LEN];
fetchFdToTempFile(remoteFd, tempSubmitFile);
mustCloseFd(&remoteFd);
long long endUploadTime = edwNow();
/* Calculate MD5 sum, and see if we already have such a file. */
char *md5 = md5HexForFile(tempSubmitFile);
int fileId = findFileGivenMd5AndSubmitDir(conn, md5, submitDirId);
/* If we already have it, then delete temp file, otherwise put file in file table. */
char submitLocalPath[PATH_LEN];
if (fileId != 0)
{
remove(tempSubmitFile);
char submitRelativePath[PATH_LEN];
sqlSafef(query, sizeof(query), "select edwFileName from edwFile where id=%d", fileId);
sqlNeedQuickQuery(conn, query, submitRelativePath, sizeof(submitRelativePath));
safef(submitLocalPath, sizeof(submitLocalPath), "%s%s", edwRootDir, submitRelativePath);
}
else
{
/* Looks like it's the first time we've seen this submission file, so
* save the file itself. We'll get to the records inside the file in a bit. */
fileId = makeNewEmptyFileRecord(conn, submitId, submitDirId, submitFile, 0);
/* Get file/path names for submission file inside warehouse. */
char edwFile[PATH_LEN];
edwMakeFileNameAndPath(fileId, submitFile, edwFile, submitLocalPath);
/* Move file to final resting place and get update time and size from local file system. */
mustRename(tempSubmitFile, submitLocalPath);
time_t updateTime = fileModTime(submitLocalPath);
off_t size = fileSize(submitLocalPath);
/* Update file table which now should be complete including updateTime. */
sqlSafef(query, sizeof(query),
"update edwFile set "
" updateTime=%lld, size=%lld, md5='%s', edwFileName='%s',"
" startUploadTime=%lld, endUploadTime=%lld"
" where id=%u\n",
(long long)updateTime, (long long)size, md5, edwFile,
startUploadTime, endUploadTime, fileId);
sqlUpdate(conn, query);
}
/* By now there is a submit file on the local file system. We parse it out. */
edwParseSubmitFile(conn, submitLocalPath, submitUrl, &sfrList);
/* Save our progress so far to submit table. */
sqlSafef(query, sizeof(query),
"update edwSubmit"
" set submitFileId=%lld, submitDirId=%lld, fileCount=%d where id=%d",
(long long)fileId, (long long)submitDirId, slCount(sfrList), submitId);
sqlUpdate(conn, query);
/* Weed out files we already have. */
struct submitFileRow *sfr, *sfrNext;
for (sfr = sfrList; sfr != NULL; sfr = sfrNext)
{
sfrNext = sfr->next;
struct edwFile *bf = sfr->file;
long long fileId;
if ((fileId = edwGotFile(conn, submitDir, bf->submitFileName, bf->md5, bf->size)) >= 0)
{
++oldCount;
oldBytes += bf->size;
sfr->md5MatchFileId = fileId;
slAddHead(&oldList, sfr);
}
else
slAddHead(&newList, sfr);
byteCount += bf->size;
}
sfrList = NULL;
slReverse(&newList);
slReverse(&oldList);
/* Update database with oldFile count. */
sqlSafef(query, sizeof(query),
"update edwSubmit set oldFiles=%d,oldBytes=%lld,byteCount=%lld where id=%u",
oldCount, oldBytes, byteCount, submitId);
sqlUpdate(conn, query);
/* Deal with old files. This may throw an error. We do it before downloading new
* files since we want to fail fast if we are going to fail. */
int updateCount = handleOldFileTags(conn, oldList, doUpdate);
sqlSafef(query, sizeof(query),
"update edwSubmit set metaChangeCount=%d where id=%u", updateCount, submitId);
sqlUpdate(conn, query);
}
errCatchEnd(errCatch);
if (errCatch->gotError)
{
handleSubmitError(conn, submitId, errCatch->message->string);
/* The handleSubmitError will keep on throwing. */
}
errCatchFree(&errCatch);
/* Go through list attempting to load the files if we don't already have them. */
struct submitFileRow *sfr;
for (sfr = newList; sfr != NULL; sfr = sfr->next)
{
if (edwSubmitShouldStop(conn, submitId))
break;
struct edwFile *bf = sfr->file;
int submitUrlSize = strlen(submitDir) + strlen(bf->submitFileName) + 1;
char submitUrl[submitUrlSize];
safef(submitUrl, submitUrlSize, "%s%s", submitDir, bf->submitFileName);
if (edwGotFile(conn, submitDir, bf->submitFileName, bf->md5, bf->size)<0)
{
/* We can't get a ID for this file. There's two possible reasons -
* either somebody is in the middle of fetching it or nobody's started.
* If somebody is in the middle of fetching it, assume they died
* if they took more than an hour, and start up another fetch.
* So here we fetch unless somebody else is fetching recently. */
if (edwGettingFile(conn, submitDir, bf->submitFileName) < 0)
{
verbose(1, "Fetching %s\n", bf->submitFileName);
getSubmittedFile(conn, bf, submitDir, submitUrl, submitId);
newBytes += bf->size;
sqlSafef(query, sizeof(query),
"update edwSubmit set newFiles=newFiles+1,newBytes=%lld where id=%d",
newBytes, submitId);
sqlUpdate(conn, query);
}
}
else
{
verbose(2, "Already got %s\n", bf->submitFileName);
sqlSafef(query, sizeof(query), "update edwSubmit set oldFiles=oldFiles+1 where id=%d",
submitId);
sqlUpdate(conn, query);
}
if (sfr->replacesFile != 0)
{
/* What happens when the replacement doesn't validate? */
verbose(2, "Replacing %s with %s\n", sfr->replaces, bf->submitFileName);
sqlSafef(query, sizeof(query),
"update edwFile set replacedBy=%u, deprecated='%s' where id=%u",
bf->id, sfr->replaceReason, sfr->replacesFile);
sqlUpdate(conn, query);
}
}
/* If we made it here, update submit endUploadTime */
sqlSafef(query, sizeof(query),
"update edwSubmit set endUploadTime=%lld where id=%d",
edwNow(), submitId);
sqlUpdate(conn, query);
/* Get a real submission record and then set things up so mail user when all done. */
struct edwSubmit *submit = edwSubmitFromId(conn, submitId);
sqlDisconnect(&conn); // We'll be waiting a while so free connection
waitForValidationAndSendEmail(submit, email);
}
