/**
* Return the confidence with with this algorithm can load the file
* @param line the line to parse
* @param cols The vector to parse into
* @param expectedCommas The number of expected commas in the line
* @param isOldTBL boolean to deal with new and old TBL formats.
* @returns An integer specifying how many columns were parsed into.
* @throws std::length_error if anything other than 17 columns (or 16
* cell-delimiting commas) is found when loading an old Refl TBL. A
* length_error will be thrown for new TBL formats if there are less column
* headings than expected commas.
*/
size_t LoadTBL::getCells(std::string line, std::vector<std::string> &cols,
size_t expectedCommas, bool isOldTBL) const {
// first check the number of commas in the line.
size_t found = countCommas(line);
if (isOldTBL) {
if (found == expectedCommas) {
// If there are 16 that simplifies things and i can get boost to do the
// hard
// work
boost::split(cols, line, boost::is_any_of(","),
boost::token_compress_off);
} else if (found < expectedCommas) {
// less than 16 means the line isn't properly formatted. So Throw
std::string message =
"A line must contain " + std::to_string(expectedCommas) +
" cell-delimiting commas. Found " + std::to_string(found) + ".";
throw std::length_error(message);
} else {
// More than 16 will need further checks as more is only ok when pairs of
// quotes surround a comma, meaning it isn't a delimiter
std::vector<std::vector<size_t>> quoteBounds;
findQuotePairs(line, quoteBounds);
// if we didn't find any quotes, then there are too many commas and we
// definitely have too many delimiters
if (quoteBounds.empty()) {
std::string message =
"A line must contain " + std::to_string(expectedCommas) +
" cell-delimiting commas. Found " + std::to_string(found) + ".";
throw std::length_error(message);
}
// now go through and split it up manually. Throw if we find ourselves in
// a
// positon where we'd add a 18th value to the vector
csvParse(line, cols, quoteBounds, expectedCommas);
}
} else {
std::vector<std::vector<size_t>> quoteBounds;
findQuotePairs(line, quoteBounds);
csvParse(line, cols, quoteBounds, expectedCommas);
if (cols.size() > expectedCommas) {
for (size_t i = expectedCommas + 1; i < cols.size(); i++) {
cols[expectedCommas].append(
boost::lexical_cast<std::string>("," + cols[i]));
}
} else if (cols.size() < expectedCommas) {
std::string message =
"A line must contain " + std::to_string(expectedCommas) +
" cell-delimiting commas. Found " + std::to_string(found) + ".";
throw std::length_error(message);
}
}
return cols.size();
}
struct slName *tagFindValList(struct tagStanza *stanza, char *tag)
/* Read in tag as a list. Do a slFreeList on this when done.
* Returns NULL if no value */
{
char *val = tagFindVal(stanza, tag);
return csvParse(val);
}
struct slName *tagMustFindValList(struct tagStanza *stanza, char *tag)
/* Find tag or die trying, and return it as parsed out list */
{
char *val = tagMustFindVal(stanza, tag);
return csvParse(val);
}
void rWriteJson(FILE *f, struct tagStorm *storm, struct tagStanza *stanza,
struct ttjSubObj *obj, struct ttjSubObj *labeledObj, struct hash *schemaHash,
struct dyString *scratch)
/* Write out json object recursively */
{
boolean isArray = allDigitNames(obj->children);
struct ttjSubObj *field;
if (isArray)
{
fprintf(f, "[");
for (field = obj->children; field != NULL; field = field->next)
{
if (field != obj->children) // Only write comma separators after the first one
fprintf(f, ",");
rWriteJson(f, storm, stanza, field, labeledObj, schemaHash, scratch);
}
fprintf(f, "]");
}
else
{
fprintf(f, "{");
boolean firstOut = TRUE;
/* Figure out if we need to attach a core object and do so. The figuring bit is
* frankly clunky. */
char *objType = labeledObj->name;
if (sameString(objType, "submitter") || sameString(objType, "contributors"))
objType = "contact";
else if (sameString(objType, "publications"))
objType = "publication";
else if (sameString(objType, "protocol")) // protocol is actually just protocol_id
objType = "string";
else if (sameString(objType, "protocols")) // but protocols array is protocol
objType = "protocol";
else if (sameString(objType, "umi_barcode"))
objType = "barcode";
if (objNeedsCore(objType))
printCore(f, objType, &firstOut);
for (field = obj->children; field != NULL; field = field->next)
{
char *fieldName = field->name;
if (field->children != NULL)
{
/* Look for funny characteristics_ as these are largely up to user. */
if (startsWith("characteristics_", field->name))
errAbort("No '.' allowed in field name after characteristics_ in %s",
field->children->fullName);
/* If actually have data in this stanza write our field. */
if (prefixDotInStanza(field->fullName, stanza, scratch))
{
writeJsonTag(f, fieldName, &firstOut);
rWriteJson(f, storm, stanza, field, field, schemaHash, scratch);
}
}
else
{
char *val = tagFindVal(stanza, field->fullName);
if (val != NULL)
{
boolean isNum = FALSE;
char *schemaName = tagSchemaFigureArrayName(field->fullName, scratch);
struct tagSchema *schema = hashFindVal(schemaHash, schemaName);
if (schema != NULL)
isNum = (schema->type == '#' || schema->type == '%');
if (sameString(fieldName, "files"))
{
writeJsonTag(f, "lanes", &firstOut);
writeLaneArray(f, stanza, val);
}
else
{
boolean isArray = FALSE;
writeJsonTag(f, fieldName, &firstOut);
if (schema != NULL)
isArray = schema->isArray;
struct slName *list = csvParse(val);
if (isArray)
fputc('[', f);
else
{
if (list->next != NULL) // more than one element
errAbort("Multiple vals for scalar tag %s in stanza starting line %d of %s",
field->fullName, stanza->startLineIx, storm->fileName);
}
struct slName *el;
for (el = list; el != NULL; el = el->next)
{
writeJsonVal(f, el->name, isNum);
if (el->next != NULL)
fputc(',', f);
}
if (isArray)
fputc(']', f);
slFreeList(&list);
}
}
}
}
fprintf(f, "}");
}
}
void writeLaneArray(FILE *f, struct tagStanza *stanza, char *csvList)
/* Write out an array of file objects base on file names in csvList */
{
struct slName *list = csvParse(csvList), *file;
/* Figure out number of files per lanes. We'll take the lane number for the
* file names if available, but if not we'll assume list is sorted and will
* put the appropriate number of files in each lane. */
int laneCounter = 1;
int filesPerLane = 1;
int curFileInLane = 0;
char *pairedEnds = tagMustFindVal(stanza, "assay.seq.paired_ends");
if (!sameString(pairedEnds, "no"))
filesPerLane = 2;
/* First pass, make a list of lanes */
struct laneFiles *laneList = NULL, *lane;
for (file = list; file != NULL; file = file->next)
{
/* Figure out laneIx, from file name if possible, otherwise by counting */
char *fileName = file->name;
int laneIx = laneFromFileName(fileName);
if (laneIx == 0)
laneIx = laneCounter;
/* Update laneCounter */
if (++curFileInLane >= filesPerLane)
{
++laneCounter;
curFileInLane = 0;
}
/* Find lane in laneList, make new lane if it's not there. */
lane = laneFilesFind(laneList, laneIx);
if (lane == NULL)
{
AllocVar(lane);
lane->laneIx = laneIx;
slAddHead(&laneList, lane);
}
slNameAddTail(&lane->fileList, fileName);
}
slReverse(&laneList);
slSort(&laneList, laneFilesCmp);
/* Now make a lane array and go through lane list */
boolean firstOut = TRUE;
fputc('[', f);
for (lane = laneList; lane != NULL; lane = lane->next)
{
/* Write comma between lane objects */
if (firstOut)
firstOut = FALSE;
else
fputc(',', f);
/* Write lane object starting with lane index*/
fputc('{', f);
fprintf(f, "\"%s\": %d", "number", lane->laneIx);
/* Rest of lane fields are based on files we contain */
for (file = lane->fileList; file != NULL; file = file->next)
{
/* Calculate type */
char *fileName = file->name;
char *type = NULL;
if (sameString(pairedEnds, "no"))
{
type = "r1";
}
else if (sameString(pairedEnds, "yes"))
{
int end = endFromFileName(fileName);
if (end == 1)
type = "r1";
else
type = "r2";
}
else
errAbort("Unrecognized paired_ends %s", pairedEnds);
fprintf(f, ",\"%s\":", type);
writeJsonVal(f, fileName, FALSE);
}
fputc('}', f);
}
fputc(']', f);
slFreeList(&list);
}
