input.c File Reference

Read the a row-by-row collection of data samples from a file. More...

#include <config.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <ctype.h>
#include <inttypes.h>
#include <assert.h>
#include <errno.h>
#include <libxml/tree.h>
#include <libxml/parser.h>
#include <libxml/valid.h>
#include <jwsc/base/error.h>
#include <jwsc/vector/vector.h>
#include <jwsc/matrix/matrix.h>
#include <jwsc/matblock/matblock.h>
#include <xml.h>
#include <haplo_groups.h>
#include <options.h>
#include <input.h>

Go to the source code of this file.

Defines
#define	MAX_COL_SIZE   256
Functions
static uint32_t	read_block_into_buffer (char *buf, uint32_t size, FILE *fp)
	Reads a block of characters off a file stream into a buffer.
static Error *	read_file_into_buffer (char **buf_out, uint32_t *buf_len_out, const char *fname)
	Opens a file stream and reads in a file into a string buffer.
static uint8_t	skip_to_csv_column (const char **buf_in_out, uint32_t buf_len, uint32_t col)
	Skips a buffer pointer ahead to the specified CSV column.
static const char *	read_csv_column (const char *src, uint32_t src_len, char *dst, uint32_t dst_len)
	Reads the current CSV column from a source buffer into a destination and returns a pointer to the next column.
static uint8_t	skip_to_txt_column (const char **buf_in_out, uint32_t buf_len, uint32_t col)
	Skips a buffer pointer ahead to the specified text column.
static const char *	read_txt_column (const char *src, uint32_t src_len, char *dst, uint32_t dst_len)
	Reads the current text column from a source buffer into a destination and returns a pointer to the next column.
static uint8_t	skip_line (const char **buf_in_out, const char *buf, uint32_t buf_len)
static Error *	get_num_samples_from_txt_csv (uint32_t *num_samples_out, const char *file_buf, uint32_t file_len)
	Counts the number of samples (rows) in a file buffer.
static Error *	read_ids_from_txt_csv (Matblock_u8 **ids_out, const char *file_buf, uint32_t file_len, uint32_t num_samples, uint8_t aux)
	Reads the sample IDs from a file buffer.
static Error *	read_labels_from_txt_csv (Vector_u32 **labels_out, const char *file_buf, uint32_t file_len, uint32_t num_samples, uint8_t aux)
	Reads the haplo group labels from a file buffer.
static Error *	read_markers_from_txt_csv (Matrix_i32 **markers_out, const char *file_buf, uint32_t file_len, uint32_t num_samples, uint8_t aux)
	Reads the marker data from a file.
static Error *	read_input_from_txt_csv (Matblock_u8 **ids_out, Vector_u32 **labels_out, Matrix_i32 **markers_out, const char *file_buf, uint32_t file_len, uint8_t aux)
	Reads file data from a text or csv file.
static uint32_t	get_num_samples_from_xml (xmlDoc *doc)
	Counts the number of samples (rows) in an xml document.
static Error *	read_ids_from_xml (Matblock_u8 **ids_out, xmlDoc *doc, uint32_t num_samples, uint8_t aux)
	Reads the sample ids from an xml document.
static Error *	read_labels_from_xml (Vector_u32 **labels_out, xmlDoc *doc, uint32_t num_samples, uint8_t aux)
	Reads the sample labels from an xml document.
static Error *	read_markers_from_xml (Matrix_i32 **markers_out, xmlDoc *doc, uint32_t num_samples, uint8_t aux)
	Reads the sample markers from an xml document.
static Error *	read_input_from_xml (Matblock_u8 **ids_out, Vector_u32 **labels_out, Matrix_i32 **markers_out, const char *file_buf, uint32_t file_len, const char *fname, uint8_t aux)
	Reads the input data structures from an xml file in a buffer.
Error *	read_input (Matblock_u8 **ids_out, Vector_u32 **labels_out, Matrix_i32 **markers_out, const char *fname)
	Reads the IDs, haplo labels and markers from a file.
Error *	read_aux_input (Matblock_u8 **ids_out, Vector_u32 **labels_out, Matrix_i32 **markers_out, const char *fname)
	Reads the IDs, haplo labels and markers from an auxiliary file.
Error *	impute_marker_from_parent_of_haplogroup_index (uint32_t haplo_group_index, uint32_t marker_no, uint32_t sample_no, Matrix_i32 *marker_sums, Matrix_i32 *imp_markers)
Error *	impute_missing_markers_from_avg (const Vector_u32 *imp_labels, Matrix_i32 *imp_markers, const Vector_u32 *src_labels, const Matrix_i32 *src_markers)
	Imputes missing marker values (zero) with their label-dependent mean value.
Error *	impute_missing_markers_from_nn (Matrix_i32 *imp_markers, const Matrix_i32 *src_markers)
	Imputes missing marker values (zero) from nearest neighbor interpolation.

---

Detailed Description

Read the a row-by-row collection of data samples from a file.

Author:

Joseph Schlecht

License:

Creative Commons BY-NC-SA 3.0

Definition in file input.c.

---

Define Documentation

#define MAX_COL_SIZE   256

Definition at line 75 of file input.c.

---

Function Documentation

static uint32_t read_block_into_buffer	(	char *	buf,
		uint32_t	size,
		FILE *	fp
	)		[static]

Reads a block of characters off a file stream into a buffer.

Assumes that size space is available in buf.

Definition at line 83 of file input.c.

static Error* read_file_into_buffer	(	char **	buf_out,
		uint32_t *	buf_len_out,
		const char *	fname
	)		[static]

Opens a file stream and reads in a file into a string buffer.

Definition at line 104 of file input.c.

static uint8_t skip_to_csv_column	(	const char **	buf_in_out,
		uint32_t	buf_len,
		uint32_t	col
	)		[static]

Skips a buffer pointer ahead to the specified CSV column.

Skips ahead col columns from the current column. Stops if a newline character is encountered.

Parameters:

buf_in_out	Buffer pointer to skip to the specified column.
buf_len	Number of buffer elements to potentially skip through.
col	Column to skip to.

Returns:

0 if the buffer does not have enough columns to support the operation; 1 otherwise.

Definition at line 175 of file input.c.

static const char* read_csv_column	(	const char *	src,
		uint32_t	src_len,
		char *	dst,
		uint32_t	dst_len
	)		[static]

Reads the current CSV column from a source buffer into a destination and returns a pointer to the next column.

Parameters:

src	Buffer to read from.
src_len	Length of the source buffer.
dst	Buffer to read into.
dst_len	Length of the destination buffer.

Returns:

A pointer to the next CSV column, or NULL if there isn't one.

Definition at line 224 of file input.c.

static uint8_t skip_to_txt_column	(	const char **	buf_in_out,
		uint32_t	buf_len,
		uint32_t	col
	)		[static]

Skips a buffer pointer ahead to the specified text column.

Skips ahead col columns from the current column. Stops if a newline character is encountered.

Parameters:

buf_in_out	Buffer pointer to skip to the specified column.
buf_len	Number of buffer elements to potentially skip through.
col	Column to skip to.

Returns:

0 if the buffer does not have enough columns to support the operation; 1 otherwise.

Definition at line 276 of file input.c.

static const char* read_txt_column	(	const char *	src,
		uint32_t	src_len,
		char *	dst,
		uint32_t	dst_len
	)		[static]

Reads the current text column from a source buffer into a destination and returns a pointer to the next column.

Parameters:

src	Buffer to read from.
src_len	Length of the source buffer.
dst	Buffer to read into.
dst_len	Length of the destination buffer.

Returns:

A pointer to the next text column, or NULL if there isn't one.

Definition at line 330 of file input.c.

static uint8_t skip_line	(	const char **	buf_in_out,
		const char *	buf,
		uint32_t	buf_len
	)		[static]

Skips the file pointer ahead, one position past the newline character '\n'.

Returns:

0 if *buf_in_out points beyond the buffer after skipping a line; 1 otherwise.

Definition at line 375 of file input.c.

static Error* get_num_samples_from_txt_csv	(	uint32_t *	num_samples_out,
		const char *	file_buf,
		uint32_t	file_len
	)		[static]

Counts the number of samples (rows) in a file buffer.

Definition at line 399 of file input.c.

static Error* read_ids_from_txt_csv	(	Matblock_u8 **	ids_out,
		const char *	file_buf,
		uint32_t	file_len,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the sample IDs from a file buffer.

Column count begins with 1. If id_cols is zero, this function is a NOOP.

Parameters:

ids_out	Result parameter.
file_buf	File buffer containing the data file.
file_len	Length of the file buffer.
num_samples	Number of samples (rows) in the file buffer.
aux	Whether this is an auxilary file.

Definition at line 442 of file input.c.

static Error* read_labels_from_txt_csv	(	Vector_u32 **	labels_out,
		const char *	file_buf,
		uint32_t	file_len,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the haplo group labels from a file buffer.

Column count begins with 1. If label_col is set to zero, this function is a NOOP.

Parameters:

labels_out	Result parameter.
file_buf	File buffer containing the data file.
file_len	Length of the file buffer.
num_samples	Number of samples (rows) in the file buffer.
aux	Whether this is an auxilary file.

Definition at line 534 of file input.c.

static Error* read_markers_from_txt_csv	(	Matrix_i32 **	markers_out,
		const char *	file_buf,
		uint32_t	file_len,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the marker data from a file.

Column count begins with 1.

Parameters:

markers_out	Result parameter.
file_buf	File buffer containing the data file.
file_len	Length of the file buffer.
num_samples	Number of samples (rows) in the file buffer.
aux	Whether this is an auxilary file.

Definition at line 620 of file input.c.

static Error* read_input_from_txt_csv	(	Matblock_u8 **	ids_out,
		Vector_u32 **	labels_out,
		Matrix_i32 **	markers_out,
		const char *	file_buf,
		uint32_t	file_len,
		uint8_t	aux
	)		[static]

Reads file data from a text or csv file.

Definition at line 716 of file input.c.

static uint32_t get_num_samples_from_xml

(

xmlDoc *

doc

)

[static]

Counts the number of samples (rows) in an xml document.

Definition at line 756 of file input.c.

static Error* read_ids_from_xml	(	Matblock_u8 **	ids_out,
		xmlDoc *	doc,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the sample ids from an xml document.

Definition at line 778 of file input.c.

static Error* read_labels_from_xml	(	Vector_u32 **	labels_out,
		xmlDoc *	doc,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the sample labels from an xml document.

Definition at line 841 of file input.c.

static Error* read_markers_from_xml	(	Matrix_i32 **	markers_out,
		xmlDoc *	doc,
		uint32_t	num_samples,
		uint8_t	aux
	)		[static]

Reads the sample markers from an xml document.

Definition at line 903 of file input.c.

static Error* read_input_from_xml	(	Matblock_u8 **	ids_out,
		Vector_u32 **	labels_out,
		Matrix_i32 **	markers_out,
		const char *	file_buf,
		uint32_t	file_len,
		const char *	fname,
		uint8_t	aux
	)		[static]

Reads the input data structures from an xml file in a buffer.

Definition at line 974 of file input.c.

Error* read_input	(	Matblock_u8 **	ids_out,
		Vector_u32 **	labels_out,
		Matrix_i32 **	markers_out,
		const char *	fname
	)

Reads the IDs, haplo labels and markers from a file.

Parameters:

ids_out	Result parameter. If opts.id_cols is zero *ids_out will be NULL.
labels_out	Result parameter. If opts.labels_col is zero *labels_out will be NULL.
markers_out	Result parameter. Will always be defined, unless there was an error reading the file.
fname	Data file to read.

Definition at line 1054 of file input.c.

Error* read_aux_input	(	Matblock_u8 **	ids_out,
		Vector_u32 **	labels_out,
		Matrix_i32 **	markers_out,
		const char *	fname
	)

Reads the IDs, haplo labels and markers from an auxiliary file.

Parameters:

ids_out	Result parameter. If opts.id_cols is zero *ids_out will be NULL.
labels_out	Result parameter. If opts.labels_col is zero *labels_out will be NULL.
markers_out	Result parameter. Will always be defined, unless there was an error reading the file.
fname	Data file to read.

Definition at line 1109 of file input.c.

Error* impute_marker_from_parent_of_haplogroup_index	(	uint32_t	haplo_group_index,
		uint32_t	marker_no,
		uint32_t	sample_no,
		Matrix_i32 *	marker_sums,
		Matrix_i32 *	imp_markers
	)

Parameters:

haplo_group_index	Index of the haplogroup we look to impute with ancestor marker average.
marker_no	Index of the missing marker to impute.
sample_no	Number of sample being processed from the input file.
marker_sums	Matrix containing the computed average for each of the haplogroups.
imp_markers	vector of markers from the source sample to be imputed.

Note:

This function will keep looking for a parent with a value different than zero, and if doesn't find any it fails, and we will have o remove this sample

Definition at line 1167 of file input.c.

Error* impute_missing_markers_from_avg	(	const Vector_u32 *	imp_labels,
		Matrix_i32 *	imp_markers,
		const Vector_u32 *	src_labels,
		const Matrix_i32 *	src_markers
	)

Imputes missing marker values (zero) with their label-dependent mean value.

Parameters:

imp_labels	Labels to use for selecting samples to impute markers from.
imp_markers	Marker values of zero are imputed.
src_labels	Labels to use for selecting samples to impute markers from.
src_markers	Markers to use for imputing values.

Note:

The imputed and source sets can be the same.

Definition at line 1208 of file input.c.

Error* impute_missing_markers_from_nn	(	Matrix_i32 *	imp_markers,
		const Matrix_i32 *	src_markers
	)

Imputes missing marker values (zero) from nearest neighbor interpolation.

Parameters:

imp_markers	Marker values of zero are imputed.
src_markers	Markers to use for imputing values.

Note:

The imputed and source sets can be the same.

Definition at line 1301 of file input.c.