static int myGetline(char **lineptr, int *n, FILE *stream)
{
char *line, *p;
int size, copy, len;
int chunkSize = 256 * sizeof(char);
if (*lineptr == NULL || *n < 2)
{
line = (char *)rax_realloc(*lineptr, chunkSize, FALSE);
if (line == NULL)
return -1;
*lineptr = line;
*n = chunkSize;
}
line = *lineptr;
size = *n;
copy = size;
p = line;
while(1)
{
while (--copy > 0)
{
register int c = getc(stream);
if (c == EOF)
goto lose;
else
{
*p++ = c;
if(c == '\n' || c == '\r')
goto win;
}
}
/* Need to enlarge the line buffer. */
len = p - line;
size *= 2;
line = rax_realloc (line, size, FALSE);
if (line == NULL)
goto lose;
*lineptr = line;
*n = size;
p = line + len;
copy = size - len;
}
lose:
if (p == *lineptr)
return -1;
/* Return a partial line since we got an error in the middle. */
win:
*p = '\0';
return p - *lineptr;
}
static void addInsertion(nodeptr p, double lh, insertions *ins)
{
if(ins->count < ins->maxCount)
{
ins->s[ins->count].lh = lh;
ins->s[ins->count].p = p;
ins->count = ins->count + 1;
}
else
{
ins->s = rax_realloc(ins->s, sizeof(scores) * ins->maxCount * 2);
ins->maxCount *= 2;
ins->s[ins->count].lh = lh;
ins->s[ins->count].p = p;
ins->count = ins->count + 1;
}
}
void parsePartitions(analdef *adef, rawdata *rdta, tree *tr)
{
FILE *f;
int numberOfModels = 0;
int nbytes = 0;
char *ch;
char *cc = (char *)NULL;
char **p_names;
int n, i, l;
int lower, upper, modulo;
char buf[256];
int **partitions;
int pairsCount;
int as, j;
int k;
f = myfopen(modelFileName, "rb");
while(myGetline(&cc, &nbytes, f) > -1)
{
if(!lineContainsOnlyWhiteChars(cc))
{
numberOfModels++;
}
if(cc)
rax_free(cc);
cc = (char *)NULL;
}
rewind(f);
p_names = (char **)rax_malloc(sizeof(char *) * numberOfModels);
partitions = (int **)rax_malloc(sizeof(int *) * numberOfModels);
tr->initialPartitionData = (pInfo*)rax_malloc(sizeof(pInfo) * numberOfModels);
for(i = 0; i < numberOfModels; i++)
{
tr->initialPartitionData[i].protModels = adef->proteinMatrix;
tr->initialPartitionData[i].usePredefinedProtFreqs = adef->protEmpiricalFreqs;
tr->initialPartitionData[i].optimizeBaseFrequencies = FALSE;
tr->initialPartitionData[i].dataType = -1;
}
for(i = 0; i < numberOfModels; i++)
partitions[i] = (int *)NULL;
i = 0;
while(myGetline(&cc, &nbytes, f) > -1)
{
if(!lineContainsOnlyWhiteChars(cc))
{
n = strlen(cc);
p_names[i] = (char *)rax_malloc(sizeof(char) * (n + 1));
strcpy(&(p_names[i][0]), cc);
i++;
}
if(cc)
rax_free(cc);
cc = (char *)NULL;
}
for(i = 0; i < numberOfModels; i++)
{
ch = p_names[i];
pairsCount = 0;
skipWhites(&ch);
if(*ch == '=')
{
printf("Identifier missing prior to '=' in %s\n", p_names[i]);
exit(-1);
}
analyzeIdentifier(&ch, i, tr);
ch++;
numberPairs:
pairsCount++;
partitions[i] = (int *)rax_realloc((void *)partitions[i], (1 + 3 * pairsCount) * sizeof(int), FALSE);
partitions[i][0] = pairsCount;
partitions[i][3 + 3 * (pairsCount - 1)] = -1;
skipWhites(&ch);
if(!isNum(*ch))
{
printf("%c Number expected in %s\n", *ch, p_names[i]);
exit(-1);
}
l = 0;
while(isNum(*ch))
{
/*printf("%c", *ch);*/
buf[l] = *ch;
ch++;
l++;
}
buf[l] = '\0';
lower = atoi(buf);
partitions[i][1 + 3 * (pairsCount - 1)] = lower;
skipWhites(&ch);
/* NEW */
if((*ch != '-') && (*ch != ','))
{
if(*ch == '\0' || *ch == '\n' || *ch == '\r')
{
upper = lower;
goto SINGLE_NUMBER;
}
else
{
printf("'-' or ',' expected in %s\n", p_names[i]);
exit(-1);
}
}
if(*ch == ',')
{
upper = lower;
goto SINGLE_NUMBER;
}
/* END NEW */
ch++;
skipWhites(&ch);
if(!isNum(*ch))
{
printf("%c Number expected in %s\n", *ch, p_names[i]);
exit(-1);
}
l = 0;
while(isNum(*ch))
{
buf[l] = *ch;
ch++;
l++;
}
buf[l] = '\0';
upper = atoi(buf);
SINGLE_NUMBER:
partitions[i][2 + 3 * (pairsCount - 1)] = upper;
if(upper < lower)
{
printf("Upper bound %d smaller than lower bound %d for this partition: %s\n", upper, lower, p_names[i]);
exit(-1);
}
skipWhites(&ch);
if(*ch == '\0' || *ch == '\n' || *ch == '\r') /* PC-LINEBREAK*/
{
goto parsed;
}
if(*ch == ',')
{
ch++;
goto numberPairs;
}
if(*ch == '\\')
{
ch++;
skipWhites(&ch);
if(!isNum(*ch))
{
printf("%c Number expected in %s\n", *ch, p_names[i]);
exit(-1);
}
if(adef->compressPatterns == FALSE)
{
printf("\nError: You are not allowed to use interleaved partitions, that is, assign non-contiguous sites\n");
printf("to the same partition model, when pattern compression is disabled via the -H flag,\n");
printf("or when pattern compression is disabled implicitely by some other option that requires it!\n\n");
exit(-1);
}
l = 0;
while(isNum(*ch))
{
buf[l] = *ch;
ch++;
l++;
}
buf[l] = '\0';
modulo = atoi(buf);
partitions[i][3 + 3 * (pairsCount - 1)] = modulo;
skipWhites(&ch);
if(*ch == '\0' || *ch == '\n' || *ch == '\r')
{
goto parsed;
}
if(*ch == ',')
{
ch++;
goto numberPairs;
}
}
if(*ch == '/')
{
printf("\nRAxML detected the character \"/\" in your partition file.\n");
printf("Did you mean to write something similar to this: \"DNA, p1=1-100\\3\" ?\n");
printf("It's actually a backslash, not a slash, the program will exit now with an error!\n\n");
}
else
{
printf("\nRAxML detected the character \"%c\" in your partition file,\n", *ch);
printf("while it does not belong there!\n");
printf("\nAre you sure that your partition file complies with the RAxML partition file format?\n");
printf("\nActually reading the manual, does indeed do help a lot\n\n");
printf("The program will exit now with an error!\n\n");
}
printf("The problematic line in your partition file is this one here:\n\n");
printf("%s\n\n", p_names[i]);
assert(0);
parsed:
;
}
fclose(f);
/*********************************************************************************************************************/
for(i = 0; i <= rdta->sites; i++)
tr->model[i] = -1;
for(i = 0; i < numberOfModels; i++)
{
as = partitions[i][0];
for(j = 0; j < as; j++)
{
lower = partitions[i][1 + j * 3];
upper = partitions[i][2 + j * 3];
modulo = partitions[i][3 + j * 3];
if(modulo == -1)
{
for(k = lower; k <= upper; k++)
setModel(i, k, tr->model);
}
else
{
for(k = lower; k <= upper; k += modulo)
{
if(k <= rdta->sites)
setModel(i, k, tr->model);
}
}
}
}
for(i = 1; i < rdta->sites + 1; i++)
{
if(tr->model[i] == -1)
{
printf("ERROR: Alignment Position %d has not been assigned any model\n", i);
exit(-1);
}
}
for(i = 0; i < numberOfModels; i++)
{
rax_free(partitions[i]);
rax_free(p_names[i]);
}
rax_free(partitions);
rax_free(p_names);
tr->NumberOfModels = numberOfModels;
if(adef->perGeneBranchLengths)
{
if(tr->NumberOfModels > NUM_BRANCHES)
{
printf("You are trying to use %d partitioned models for an individual per-gene branch length estimate.\n", tr->NumberOfModels);
printf("Currently only %d are allowed to improve efficiency.\n", NUM_BRANCHES);
printf("\n");
printf("In order to change this please replace the line \"#define NUM_BRANCHES %d\" in file \"axml.h\" \n", NUM_BRANCHES);
printf("by \"#define NUM_BRANCHES %d\" and then re-compile RAxML.\n", tr->NumberOfModels);
exit(-1);
}
else
{
tr->multiBranch = 1;
tr->numBranches = tr->NumberOfModels;
}
}
}
