ustring lastChar (const gchar *s) {
return lastChar (_(s));
}
bool analyseWordCheckSpecialConsonants (ustringArrayT word,
ustring consonant) {
return lastChar (word[0]).lowercase () == _(consonant[0]) &&
firstChar (word[1]).lowercase () == _(consonant[1]);
}
ustring lastChar (string s) {
return lastChar (_(s));
}
/*
* The command isprint() is used to determine printable
* characters.
*
* @param s Input string
* @return Returns a copy of the string, stripped
* of leading and trailing white space
*/
std::string stripws(const std::string &s) {
int ifirst = firstChar(s);
int ilast = lastChar(s);
return s.substr(ifirst, ilast - ifirst + 1);
}
std::size_t ZLUnicodeUtil::lastChar(Ucs4Char &ch, const std::string &utf8String) {
return lastChar(ch, utf8String.data() + utf8String.length());
}
void readFinfFiles(char *gsDir)
/* Read in .finf files and save info in cloneHash/cloneList. */
{
struct lineFile *lf;
struct clone *clone = NULL;
struct endInfo *end;
char fileName[512];
int i;
char *words[7];
char lastClone[64];
char cloneName[64];
int gsInfoCount = 0;
struct frag *frag;
boolean isFin;
char *s, *e;
strcpy(lastClone, "");
for (i=0; i<ArraySize(gsFiles); ++i)
{
isFin = (i <= 0);
sprintf(fileName, "%s/%s", gsDir, gsFiles[i]);
printf("Reading info from %s\n", fileName);
lf = lineFileOpen(fileName, TRUE);
while (lineFileRow(lf, words))
{
if (!sameString(words[1], lastClone))
{
struct clone *oldClone;
strcpy(lastClone, words[1]);
strcpy(cloneName, words[1]);
AllocVar(clone);
s = strchr(cloneName, '.');
if (s == NULL)
errAbort("Bad clone name format line %d of %s\n", lf->lineIx, lf->fileName);
if (strlen(s) >= sizeof(clone->version))
errAbort("Bad clone name format line %d of %s\n", lf->lineIx, lf->fileName);
strcpy(clone->version, s);
chopSuffix(cloneName);
clone->size = atoi(words[3]);
if ((oldClone = hashFindVal(cloneHash, cloneName)) != NULL)
{
if (isFin && clone->size == oldClone->size && sameString(clone->version, oldClone->version))
warn("Apparently benign duplication of %s line %d of %s", cloneName, lf->lineIx, lf->fileName);
else
warn("%s duplicated line %d of %s (size %d oldSize %d)", cloneName, lf->lineIx, lf->fileName,
clone->size, oldClone->size);
}
hashAddSaveName(cloneHash, cloneName, clone, &clone->name);
clone->isFin = isFin;
slAddHead(&cloneList, clone);
}
frag = newFrag(words[0], lf);
slAddTail(&clone->fragList, frag);
++clone->fragCount;
if (!clone->isFin && !sameString(words[6], "?") && !sameString(words[6], "i")
&& !sameString(words[6], "w"))
{
char *s = strchr(words[0], '~');
char c;
if (s == NULL)
errAbort("Expecting ~ in fragment name line %d of %s\n", lf->lineIx, lf->fileName);
++s;
AllocVar(end);
end->contig = cloneString(s);
subChar(s, '.', '_');
end->text = cloneString(words[6]);
c = lastChar(end->text);
if (!(c == 'L' || c == 'R'))
c = '?';
end->lr = c;
slAddHead(&clone->gsList, end);
++gsInfoCount;
}
}
lineFileClose(&lf);
}
printf("Found %d ends in %d clones\n", gsInfoCount, slCount(cloneList));
}
void addBacEndInfo(char *spFile)
/* Add BAC end info from Shiaw-Pyng's file to clones in cloneHash. */
{
struct lineFile *lf = lineFileOpen(spFile, TRUE);
char *line;
int lineSize, wordCount;
int spCount = 0;
char *words[16];
while (lineFileNext(lf, &line, &lineSize))
{
char *s, *e, c;
struct clone *clone;
struct endInfo *end;
char *firstWord;
char *contig;
if (line[0] == '#')
continue;
wordCount = chopLine(line, words);
if (wordCount == 0)
continue;
firstWord = words[0];
s = strchr(firstWord, '.');
if (s == NULL)
errAbort("Expecting dot line %d of %s\n", lf->lineIx, lf->fileName);
*s++ = 0;
if ((clone = hashFindVal(cloneHash, firstWord)) == NULL)
{
warn("%s in %s but not .finf files", firstWord, spFile);
continue;
}
if (!startsWith("Contig", s))
errAbort("Expecting .Contig line %d of %s\n", lf->lineIx, lf->fileName);
s += 6;
contig = s;
if (wordCount == 1)
{
/* Older style - just one word. */
e = strrchr(contig, '.');
if (e == NULL)
errAbort("Expecting last dot line %d of %s\n", lf->lineIx, lf->fileName);
*e++ = 0;
AllocVar(end);
subChar(s, '.', '_');
end->contig = cloneString(contig);
end->text = cloneString(e);
c = lastChar(end->text);
if (!(c == 'L' || c == 'R'))
c = '?';
end->lr = c;
slAddHead(&clone->spList, end);
++spCount;
}
else if (wordCount == 15)
{
/* Newer style - 15 words. */
if (!sameWord(words[11], "total_repeats"))
{
AllocVar(end);
end->contig = cloneString(contig);
end->text = cloneString(words[2]);
c = words[3][0];
if (!(c == 'L' || c == 'R'))
c = '?';
end->lr = c;
slAddHead(&clone->spList, end);
++spCount;
}
}
else
{
lineFileExpectWords(lf, 15, wordCount);
}
}
lineFileClose(&lf);
printf("Info on %d ends in %s\n", spCount, spFile);
}
