parse.c File Reference

Parse a .lgo file. More...

#include "error.h"
#include "lblndx.h"
#include "misc.h"
#include "network.h"
#include "param.h"
#include "parse.h"
#include "parstore.h"
#include "ptrqueue.h"
#include "sampndx.h"
#include "strptrmap.h"
#include <assert.h>
#include <ctype.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

Macros
#define	CHECK_TOKEN(tok, orig)
	Abort if token is missing. More...
#define	ILLEGAL_INPUT(x, orig)
	Abort with an error message about illegal input. More...
#define	DUPLICATE_PAR(x, orig)
	Abort with an error message about duplicate parameter definition. More...
#define	ORDER_ERROR(word)
	Input statements out of order. More...

Functions
int	getDbl (double *x, char **next, const char *orig)
	Interpret token i as a double. More...
int	getULong (unsigned long *x, char **next, const char *orig)
	Interpret token i as an unsigned long integer. More...
int	getRange (double x[2], char **next, const char *orig)
	Read a range in form "[ 12, 34 ]". More...
void	parseParam (char *next, unsigned ptype, StrPtrMap *parmap, PtrQueue *fixedQ, PtrQueue *freeQ, PtrQueue *constrQ, Bounds *bnd, const char *orig)
	Parse a line of input defining a parameter. More...
void	parseSegment (char *next, StrPtrMap *popmap, SampNdx *sndx, LblNdx *lndx, ParStore *parstore, const char *orig)
	Parse a line describing a segment of the population tree. More...
void	parseDerive (char *next, StrPtrMap *popmap, ParStore *parstore, const char *orig)
	Parse a line of input describing a parent-offspring relationship between two nodes. More...
void	parseMix (char *next, StrPtrMap *popmap, ParStore *parstore, const char *orig)
	Parse a line of input describing gene flow. More...
static int	get_one_line (size_t n, char buff[n], FILE *fp)
PtrPair	mktree (FILE *fp, SampNdx *sndx, LblNdx *lndx, Bounds *bnd)
	Parse an input file in .lgo format. More...
int	countSegments (FILE *fp)
	Count the number of "segment" statements in input file.

Variables
tipId_t	union_all_samples

Detailed Description

Parse a .lgo file.

Author

Alan R. Rogers

Consider the following tree of populations:

 a-------|
         |ab--|
 b--|bb--|    |
    |         |abc--
    |c--------|

t = 0 1 3 5.5 inf

In this tree, a, b, c, bb, ab, and abc represent segments of the population tree. The input file begins with a series of "segment" statements that define each of the segments in the tree. The segment statements also provide the time (backwards from the present in generations) at which the segment starts, and the size, twoN, of the population. Optionally, it also provides the number of haploid samples observed in this segment of the tree.

The statements that follow the segment statements describe how the segments are connected. The "mix" statement is used when a segment originates as a mixture of two ancestral segments. The "derive" statement is used when a segment derives from a single ancestral segment.

No segment can have more than two "parents" or more than two "children".

Here is input that would generate the tree above:

time fixed T0=0 time free Tc=1 time free Tab=3 time free Tabc=5.5 twoN free twoNa=100 twoN fixed twoNb=123 twoN free twoNc=213.4 twoN fixed twoNbb=32.1 twoN free twoNab=222 twoN fixed twoNabc=1.2e2 mixFrac free Mc=0.8 segment a t=T0 twoN=twoNa samples=1 segment b t=T0 twoN=twoNb samples=2 segment c t=Tc twoN=twoNc samples=1 segment bb t=Tc twoN=twoNbb segment ab t=Tab twoN=twoNab segment abc t=Tabc twoN=twoNabc mix b from bb + Mc * c derive a from ab derive bb from ab derive ab from abc derive c from abc

Copyright

Copyright (c) 2016, Alan R. Rogers roger.nosp@m.s@an.nosp@m.thro..nosp@m.utah.nosp@m..edu. This file is released under the Internet Systems Consortium License, which can be found in file "LICENSE".

Macro Definition Documentation

CHECK_TOKEN

#define CHECK_TOKEN	(		tok,
			orig
	)

Value:

        {                                  \
        if((tok) == NULL) {                                        \
            fprintf(stderr, "%s:%d:"                               \
                    " input line incomplete in .lgo file.\n",      \
                    __FILE__,__LINE__);                            \
            fprintf(stderr,"  input: %s\n", (orig));               \
            exit(EXIT_FAILURE);                                    \
        }                                                          \
    }while(0);

Abort if token is missing.

DUPLICATE_PAR

#define DUPLICATE_PAR	(		x,
			orig
	)

Value:

        do{                                  \
        fprintf(stderr,"%s:%d: Duplicate parameter def: \"%s\"\n", \
                __FILE__,__LINE__, (x));                           \
        fprintf(stderr,"  input: %s\n", (orig));                   \
        exit(EXIT_FAILURE);                                        \
    }while(0)

Abort with an error message about duplicate parameter definition.

ILLEGAL_INPUT

#define ILLEGAL_INPUT	(		x,
			orig
	)

Value:

        do{                               \
        fprintf(stderr,"%s:%d: Illegal input: \"%s\"\n",        \
                __FILE__,__LINE__, (x));                        \
        fprintf(stderr,"  input: %s\n", (orig));                \
        exit(EXIT_FAILURE);                                     \
    }while(0)

Abort with an error message about illegal input.

ORDER_ERROR

#define ORDER_ERROR

(

word

)

Value:

        do{                                           \
        fprintf(stderr,"%s:%d: Order error at \"%s\" in .lgo file.\n",  \
                __FILE__,__LINE__, (word));                             \
        fprintf(stderr,"Parameter definitions should come first\n"      \
                "   (twoN, time, mixFrac, and param),\n"                \
                "   then \"segment\" statements, and finally \"mix\"\n" \
                "   and \"derive\" statements.\n");                     \
        exit(EXIT_FAILURE);                                             \
    }while(0)

Input statements out of order.

Function Documentation

getDbl()

int getDbl	(	double *	x,
		char **	next,
		const char *	orig
	)

Interpret token i as a double.

Parameters

[out]	x	points to variable into which double value will be placed
[in,out]	next	points to unparsed portion of input line

getRange()

int getRange	(	double	x[2],
		char **	next,
		const char *	orig
	)

Read a range in form "[ 12, 34 ]".

Return 0 on success or 1 if range is not present. Abort if first character is "[" but the rest of the string is not interpretable as a range.

Referenced by parseParam().

getULong()

int getULong	(	unsigned long *	x,
		char **	next,
		const char *	orig
	)

Interpret token i as an unsigned long integer.

Parameters

[out]	x	points to variable into which value will be placed
[in,out]	next	points to unparsed portion of input line integer.

Referenced by parseSegment().

mktree()

PtrPair mktree	(	FILE *	fp,
		SampNdx *	sndx,
		LblNdx *	lndx,
		Bounds *	bnd
	)

Parse an input file in .lgo format.

Parameters

[in,out]	fp	input file pointer
[in,out]	sndx	associates the index of each sample with the node to which it belongs.
[in,out]	lndx	associated index of each sample with its name
[out]	parstore	structure that maintains info about parameters
[in]	bnd	the bounds of each type of parameter

Check the sanity of each node and make sure there is only one root.

References StrPtrMap_new().

parseDerive()

void parseDerive	(	char *	next,
		StrPtrMap *	popmap,
		ParStore *	parstore,
		const char *	orig
	)

Parse a line of input describing a parent-offspring relationship between two nodes.

Parameters

[in]	next	unparsed portion of input line
[in,out]	popmap	associates names of segments with pointers to them.

References CHECK_TOKEN, nextWhitesepToken(), stripWhiteSpace(), and StrPtrMap_get().

parseMix()

void parseMix	(	char *	next,
		StrPtrMap *	popmap,
		ParStore *	parstore,
		const char *	orig
	)

Parse a line of input describing gene flow.

Parameters

[in,out]	next	unparsed portion of input line
[in,out]	popmap	associates names of segments with pointers to them.
[out]	parstore	structure that maintains info about parameters

References CHECK_TOKEN, nextWhitesepToken(), ParStore_getIndex(), stripWhiteSpace(), and StrPtrMap_get().

parseParam()

void parseParam	(	char *	next,
		unsigned	ptype,
		StrPtrMap *	parmap,
		PtrQueue *	fixedQ,
		PtrQueue *	freeQ,
		PtrQueue *	constrQ,
		Bounds *	bnd,
		const char *	orig
	)

Parse a line of input defining a parameter.

Parameters

[in,out]	next	points to unparsed portion of input line
[in]	ptype	TWON, TIME, or MIXFRAC
[out]	parstore	structure that maintains info about parameters
[in]	bnd	the bounds of each type of parameter
[in]	orig	original input line

References CHECK_TOKEN, getRange(), and nextWhitesepToken().

parseSegment()

void parseSegment	(	char *	next,
		StrPtrMap *	popmap,
		SampNdx *	sndx,
		LblNdx *	lndx,
		ParStore *	parstore,
		const char *	orig
	)

Parse a line describing a segment of the population tree.

Parameters

[in,out]	next	pointer to unparsed portion of input line
[in,out]	popmap	associates names of segments with pointers to them.
[in,out]	sndx	associates the index of each sample with the node to which it belongs.
[in,out]	lndx	associated index of each sample with its name
[out]	parstore	structure that maintains info about parameters

References CHECK_TOKEN, getULong(), LblNdx_addSamples(), nextWhitesepToken(), ParStore_getIndex(), SampNdx_addSamples(), stripWhiteSpace(), and StrPtrMap_insert().