struct slName *stringToSlNames(char *string)
/* Convert string to a list of slNames separated by
* white space, but allowing multiple words in quotes.
* Quotes if any are stripped. */
{
struct slName *list = NULL, *name;
char *dupe = cloneString(string);
char c, *s = dupe, *e;
for (;;)
{
if ((s = skipLeadingSpaces(s)) == NULL)
break;
if ((c = *s) == 0)
break;
if (c == '\'' || c == '"')
{
if (!parseQuotedString(s, s, &e))
errAbort("missing closing %c in %s", c, string);
}
else
{
e = skipToSpaces(s);
if (e != NULL) *e++ = 0;
}
name = slNameNew(s);
slAddHead(&list, name);
s = e;
}
freeMem(dupe);
slReverse(&list);
return list;
}
struct hash *readInfoFile(char *fileName)
/* Read in 'info' file into hash keyed by gene field and
* containing a partial sanger22extra. */
{
struct hash *infoHash = newHash(0);
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *line, *type, *val;
struct dyString *dy = newDyString(512);
struct sanger22extra *sx = NULL;
while (lineFileNext(lf, &line, NULL))
{
line = skipLeadingSpaces(line);
if (line[0] == 0)
{
finishRecord(sx, dy);
sx = NULL;
}
else
{
if (sx == NULL)
AllocVar(sx);
type = nextWord(&line);
val = strchr(line, '"');
if (val == NULL)
errAbort("No quoted value line %d of %s\n", lf->lineIx, lf->fileName);
parseQuotedString(val, val, &line);
if (sameString("Gene", type))
{
sx->name = cloneString(val);
hashAdd(infoHash, val, sx);
}
else if (sameString("Remark", type))
{
dyStringAppend(dy, val);
dyStringAppend(dy, ". ");
}
}
}
finishRecord(sx, dy);
lineFileClose(&lf);
return infoHash;
}
struct hash *hashThisEqThatLine(char *line, int lineIx, boolean firstStartsWithLetter)
/* Return a symbol table from a line of form:
* 1-this1=val1 2-this='quoted val2' var3="another val"
* If firstStartsWithLetter is true, then the left side of the equals must start with
* a letter. */
{
char *dupe = cloneString(line);
char *s = dupe, c;
char *var, *val;
struct hash *hash = newHash(8);
for (;;)
{
if ((var = skipLeadingSpaces(s)) == NULL)
break;
if ((c = *var) == 0)
break;
if (firstStartsWithLetter && !isalpha(c))
errAbort("line %d of custom input: variable needs to start with letter '%s'", lineIx, var);
val = strchr(var, '=');
if (val == NULL)
{
errAbort("line %d of var %s in custom input: %s \n missing = in var/val pair", lineIx, var, line);
}
*val++ = 0;
c = *val;
if (c == '\'' || c == '"')
{
if (!parseQuotedString(val, val, &s))
errAbort("line %d of input: missing closing %c", lineIx, c);
}
else
{
s = skipToSpaces(val);
if (s != NULL) *s++ = 0;
}
hashAdd(hash, var, cloneString(val));
}
freez(&dupe);
return hash;
}
char *findVal(struct lineFile *lf, char *group, char *key)
/* Return value that matches key in group or NULL. */
{
char *s, *var, *val;
static char buf[512];
if (strlen(group) >= sizeof(buf))
errAbort("Line too long line %d of %s", lf->lineIx, lf->fileName);
strcpy(buf, group);
s = buf;
for (;;)
{
var = nextWord(&s);
if (var == NULL)
return "";
s = skipLeadingSpaces(s);
if (s == NULL || s[0] == 0)
errAbort("Unmatched key/val pair in group line %d of %s", lf->lineIx, lf->fileName);
val = s;
if (s[0] == '\'' || s[0] == '"')
{
if (!parseQuotedString(val, val, &s))
errAbort("Unmatched quote line %d of %s", lf->lineIx, lf->fileName);
}
else
{
int end;
val = nextWord(&s);
end = strlen(val) - 1;
if (val[end] == ';')
val[end] = 0;
}
s = skipLeadingSpaces(s);
if (s != NULL && s[0] == ';')
s += 1;
if (sameString(key, var))
{
subChar(val, '\t', ' ');
return val;
}
}
}
struct hash *hashVarLine(char *line, int lineIx)
/* Return a symbol table from a line of form:
* var1=val1 var2='quoted val2' var3="another val" */
{
char *dupe = cloneString(line);
char *s = dupe, c;
char *var, *val;
struct hash *hash = newHash(8);
for (;;)
{
if ((var = skipLeadingSpaces(s)) == NULL)
break;
if ((c = *var) == 0)
break;
if (!isalpha(c))
errAbort("line %d of custom input: variable needs to start with letter '%s'", lineIx, var);
val = strchr(var, '=');
if (val == NULL)
{
errAbort("line %d of var %s in custom input: %s \n missing = in var/val pair", lineIx, var, line);
}
*val++ = 0;
c = *val;
if (c == '\'' || c == '"')
{
if (!parseQuotedString(val, val, &s))
errAbort("line %d of input: missing closing %c", lineIx, c);
}
else
{
s = skipToSpaces(val);
if (s != NULL) *s++ = 0;
}
hashAdd(hash, var, cloneString(val));
}
freez(&dupe);
return hash;
}
char *nextQuotedWord(char **pLine)
/* Generalization of nextWord. Returns next quoted
* string or if no quotes next word. Updates *pLine
* to point past word that is returned. Does not return
* quotes. */
{
char *line, c;
line = skipLeadingSpaces(*pLine);
if (line == NULL || line[0] == 0)
return NULL;
c = *line;
if (c == '"' || c == '\'')
{
if (!parseQuotedString(line, line, pLine))
return NULL;
return line;
}
else
{
return nextWord(pLine);
}
}
/*
Parses the data string to make sure it is valid
*/
char *parseData(char *data) {
int isQuoted = quotedString(data);
if(isQuoted == -1) {
printf("Error: Data string not closed\n");
return NULL;
}
char *newData = NULL;
char *temp = NULL;
if(isQuoted) {
temp = removeQuotes(data);
newData = parseQuotedString(temp);
return newData;
}else {
newData = parseString(data);
if(newData == NULL) {
return NULL;
}
temp = (char *) malloc((strlen(newData) + 1) * sizeof(char));
strncpy(temp, newData, strlen(newData) + 1);
return temp;
}
}
static void readQuotedString(char *fileName, int lineIx, char *in, char *out, char **retNext)
/* Parse quoted string and abort on error. */
{
if (!parseQuotedString(in, out, retNext))
errAbort("Line %d of %s\n", lineIx, fileName);
}
struct hash *makeProbeBed(char *inGff, char *outBed)
/* Convert probe location GFF file to BED. */
{
struct lineFile *lf = lineFileOpen(inGff, TRUE);
char *row[9];
struct hash *hash = newHash(16);
FILE *f = mustOpen(outBed, "w");
while (lineFileNextRowTab(lf, row, ArraySize(row)))
{
int chromIx = romanToArabicChrom(row[0], lf);
int start = lineFileNeedNum(lf, row, 3) - 1;
int end = lineFileNeedNum(lf, row, 4);
char *s = row[8];
char *probe, *orf, *note;
char *boundAt = "Bound at ";
struct tfBinding *tfbList = NULL, *tfb;
if (!startsWith("Probe ", s))
errAbort("Expecting 9th column to start with 'Probe ' line %d of %s",
lf->lineIx, lf->fileName);
probe = nextWord(&s);
orf = nextWord(&s);
chopOff(orf, ';');
note = nextWord(&s);
if (!sameWord("Note", note))
errAbort("Expecting 'note' in 9th column line %d of %s",
lf->lineIx, lf->fileName);
s = skipLeadingSpaces(s);
if (!parseQuotedString(s, s, NULL))
errAbort("Expecting quoted string in 9th column line %d of %s",
lf->lineIx, lf->fileName);
if (startsWith("Bad Probe", s))
continue;
else if (startsWith("Not bound", s))
{
/* Ok, we do nothing. */
}
else if (startsWith(boundAt, s))
{
while (s != NULL && startsWith(boundAt, s))
{
char *word, *by;
double binding;
s += strlen(boundAt);
word = nextWord(&s);
binding = atof(word);
by = nextWord(&s);
if (!sameString("by:", by))
errAbort("Expecting by: line %d of %s", lf->lineIx, lf->fileName);
while ((word = nextWord(&s)) != NULL)
{
char lastChar = 0, *e;
e = word + strlen(word) - 1;
lastChar = *e;
if (lastChar == ';' || lastChar == ',')
*e = 0;
AllocVar(tfb);
tfb->binding = binding;
tfb->tf = cloneString(word);
slAddHead(&tfbList, tfb);
if (lastChar == ';')
break;
}
s = skipLeadingSpaces(s);
}
slReverse(&tfbList);
}
else
{
errAbort("Expecting %s in note line %d of %s", boundAt,
lf->lineIx, lf->fileName);
}
fprintf(f, "chr%d\t%d\t%d\t", chromIx+1, start, end);
fprintf(f, "%s\t%d\t", orf, slCount(tfbList));
for (tfb = tfbList; tfb != NULL; tfb = tfb->next)
fprintf(f, "%s,", tfb->tf);
fprintf(f, "\t");
for (tfb = tfbList; tfb != NULL; tfb = tfb->next)
fprintf(f, "%4.3f,", tfb->binding);
fprintf(f, "\n");
hashAdd(hash, orf, NULL);
}
lineFileClose(&lf);
carefulClose(&f);
return hash;
}
/*
Takes a string path and collapses it to find the
absolute path without "." or ".."
*/
node *collapseFilePath(char *path) {
int isQuoted = quotedString(path);
if(isQuoted == -1) {
printf("Error: File Path string not closed\n");
return NULL;
}
node *currentNode = NULL;
char *checkedStr = NULL;
if(!isQuoted) {
currentNode = getCurrentPath();
checkedStr = parseString(path);
if(checkedStr == NULL) {
freePathReverse(currentNode);
return NULL;
}
char *temp = (char *) malloc(strlen(checkedStr) + 8);
if(temp == NULL) {
printf("ERROR: No memory left to append uni to file path\n");
freePathReverse(currentNode);
return NULL;
}
strncpy(temp, checkedStr, strlen(checkedStr) + 1);
strncat(temp, ".jg3538", 8);
currentNode = insertNode(currentNode, temp);
free(temp);
}else {
if(path[1] != '/') {
currentNode = getCurrentPath();
}
path = removeQuotes(path);
char temp[strlen(path)+1];
strncpy(temp, path, strlen(path)+1);
char *base = basename(temp);
if(strncmp(base, "..", 3) == 0 && strlen(base) == 2) {
path = (char *) realloc(path, strlen(path) + 9);
if(path == NULL) {
printf("ERROR: No memory left to append uni to file path\n");
freePathReverse(currentNode);
return NULL;
}
strncat(path, "/.jg3538", 8);
}else {
path = (char *) realloc(path, strlen(path) + 8);
if(path == NULL) {
printf("ERROR: No memory left to append uni to file path\n");
freePathReverse(currentNode);
return NULL;
}
strncat(path, ".jg3538", 8);
}
char *token = strtok(path, "/");
while(token != NULL) {
checkedStr = parseQuotedString(token);
if(checkedStr == NULL) {
freePathReverse(currentNode);
free(path);
return NULL;
}
currentNode = insertNode(currentNode, checkedStr);
free(checkedStr);
token = strtok(NULL, "/");
}
free(path);
}
while(currentNode->prev != NULL) {
currentNode = currentNode->prev;
}
return currentNode;
}
