#include <ctype.h>
#include "../cl/globals.h"
#include "../cl/cl.h"
#include "../cl/corpus.h"
#include "../cl/attributes.h"

Data Structures
struct	SAttRegion
	Represents a single s-attribuite region and its annotation. More...

Macros
#define	MAX_ATTRS 1024
	Maximum number of attributes that can be printed. More...

#define	MAX_PRINT_VALUES 1024
	Maximum number of attributes whose "surrounding values" can be printed in matchlist mode. More...

Typedefs
typedef enum _output_modes	OutputMode

Enumerations
enum	_output_modes { StandardMode, LispMode, EncodeMode, ConclineMode, XMLMode }

Functions
void	decode_cleanup (int error_code)
	Cleans up memory prior to an error-prompted exit. More...

void	decode_usage (int exit_code)
	Prints a usage message and exits the program. More...

int	is_num (char *s)
	Check whether a string represents a number. More...

const char *	decode_string_escape (const char *s)
	Escapes a string according to the currently active global mode. More...

void	decode_print_xml_declaration (void)
	Prints an XML declaration, using character set specification obtained from the global corpus variable. More...

void	decode_sort_s_att_regions (void)
	Sorts s_att_regions[MAX_ATTRS] in ascending 'nested' order, using sar_sort_index[] (which is automatically initialised). More...

int	decode_attribute_is_in_list (Attribute attr, Attribute *att_list, int att_list_size)
	Determines whether or not a given Attribute is in an array of Attributes. More...

int	decode_add_attribute (Attribute *attr)
	Adds a specified Attribute to the global print_list array. More...

void	decode_verify_print_value_list (void)
	Check the context of the global printValues array, to check that no s-attribute in it is declared more in the main print_list_index as well. More...

void	decode_print_surrounding_s_att_values (int position)
	Prints a starting tag for each s-attribute. More...

void	decode_print_token_sequence (int start_position, int end_position, Attribute *context)
	Prints out the requested attributes for a sequence of tokens (or a single token if end_position == -1). More...

int	main (int argc, char **argv)
	Main function for cwb-decode. More...

Variables
char *	progname = NULL

char *	registry_directory = NULL

char *	corpus_id = NULL

Corpus *	corpus = NULL

Attribute *	print_list [MAX_ATTRS]
	array of attributes selected by user for printing More...

int	print_list_index = 0
	Number of atts added to print_list (so far); used with less-than, = top limit for scrolling that array. More...

SAttRegion	s_att_regions [MAX_ATTRS]

int	sar_sort_index [MAX_ATTRS]
	index used for bubble-sorting list of regions More...

int	N_sar = 0
	number of regions currently in list (may change for each token printed) More...

Attribute *	printValues [MAX_PRINT_VALUES]
	List of s-attributes whose values are to be printed. More...

int	printValuesIndex = 0
	Number of atts added to printValues (so far); used with less-than, = top limit for scrolling that array. More...

int	first_token
	cpos of token to begin output at More...

int	last
	cpos of token to end output at (inclusive; ie this one gets printed!) More...

int	maxlast
	maximum ending cpos (deduced from size of p-attribute); More...

int	printnum = 0
	whether or not token numbers are to be printed (-n option) More...

OutputMode	mode = StandardMode
	global variable for overall output mode More...

int	xml_compatible = 0
	xml-style, for (cwb-encode -x ...); EncodeMode only, selected by -Cx More...

Macro Definition Documentation

#define MAX_ATTRS 1024

Maximum number of attributes that can be printed.

Referenced by decode_add_attribute().

#define MAX_PRINT_VALUES 1024

Maximum number of attributes whose "surrounding values" can be printed in matchlist mode.

Referenced by main().

Typedef Documentation

typedef enum _output_modes OutputMode

Enumeration Type Documentation

enum _output_modes

Enumerator
StandardMode
LispMode
EncodeMode
ConclineMode
XMLMode

Function Documentation

int decode_add_attribute ( Attribute * attr )

Adds a specified Attribute to the global print_list array.

Aborts the program if that array is already full.

Returns: Boolean.

References _Attribute::any, corpus_id, decode_attribute_is_in_list(), decode_cleanup(), MAX_ATTRS, and print_list_index.

Referenced by main().

int decode_attribute_is_in_list	(	Attribute *	attr,
		Attribute **	att_list,
		int	att_list_size
	)

Determines whether or not a given Attribute is in an array of Attributes.

Parameters

attr	The attribute to look for.
att_list	Pointer to the first member of the array (i.e. array name).
att_list_size	Upper bound of the array (the last member the function checks is attlist[attlist_size-1]).

Returns: Boolean.

Referenced by decode_add_attribute(), and decode_verify_print_value_list().

void decode_cleanup ( int error_code )

Cleans up memory prior to an error-prompted exit.

Parameters

error_code Value to be returned by the program when it exits.

References cl_delete_corpus().

Referenced by decode_add_attribute(), decode_print_token_sequence(), decode_usage(), and main().

void decode_print_surrounding_s_att_values ( int position )

Prints a starting tag for each s-attribute.

References _Attribute::any, cl_cpos2struc(), cl_struc2str(), ConclineMode, decode_string_escape(), EncodeMode, LispMode, mode, printValuesIndex, StandardMode, and XMLMode.

Referenced by main().

void decode_print_token_sequence	(	int	start_position,
		int	end_position,
		Attribute *	context
	)

Prints out the requested attributes for a sequence of tokens (or a single token if end_position == -1).

If the -c flag was used (and, thus, the context parameter is not NULL), then the sequence is extended to the entire s-attribute region (in matchlist mode).

References SAttRegion::annot, _Attribute::any, ATT_ALIGN, ATT_DYN, ATT_POS, ATT_STRUC, CDA_OK, cl_alg2cpos(), cl_cpos2alg(), cl_cpos2str(), cl_cpos2struc(), cl_cpos2struc2cpos(), cl_errno, cl_error(), cl_struc2cpos(), cl_struc2str(), cl_struc_values(), ConclineMode, decode_cleanup(), decode_sort_s_att_regions(), decode_string_escape(), EncodeMode, SAttRegion::end, LispMode, mode, N_sar, SAttRegion::name, print_list_index, printnum, sar_sort_index, StandardMode, SAttRegion::start, _Attribute::type, and XMLMode.

Referenced by main().

void decode_print_xml_declaration ( void )

Prints an XML declaration, using character set specification obtained from the global corpus variable.

References arabic, charset, cl_corpus_charset(), cyrillic, greek, hebrew, latin1, latin2, latin3, latin4, latin5, latin6, latin7, latin8, latin9, unknown_charset, and utf8.

Referenced by main().

void decode_sort_s_att_regions ( void )

Sorts s_att_regions[MAX_ATTRS] in ascending 'nested' order, using sar_sort_index[] (which is automatically initialised).

Since only regions which begin or end at the current token are considered, such an ordering is always possible; without knowing the current token, we sort by end position descending, then by start position ascending, which gives us:

first the regions corresponding to start tags, beginning with the 'largest' region
then the regions corresponding to end tags, again beginning with the 'largest' region

The function uses bubble sort in order to retain the existing order of identical regions.

References SAttRegion::end, N_sar, sar_sort_index, and SAttRegion::start.

Referenced by decode_print_token_sequence().

const char* decode_string_escape ( const char * s )

Escapes a string according to the currently active global mode.

In XMLMode, this function converts the string to an encoded XML string; all 'critical' characters are replaced by entity references, and C0 control characters are replaced with blanks. (This also happens in other modes - i.e. compact - if the global xml_compatible variable is true.)

In LispMode, it converts the string to a Lisp string with the required escapes (probably!)

In any other mode, it does nothing, and just returns the argument pointer.

It is safe to use this function without checking for a NULL argument, as NULLs will just be returned as NULLs.

Warning: returns pointer to static internal buffer of fixed size; in particular, don't use it twice in a single argument list!

See also: EncodeMode

Parameters

s	String to encode.

Returns: Pointer to encoded string in static internal buffer; or, the argument s iff the mode is not one that requires any encoding. If the argument is NULL, NULL is returned.

References CL_MAX_LINE_LENGTH, LispMode, mode, xml_compatible, and XMLMode.

Referenced by decode_print_surrounding_s_att_values(), and decode_print_token_sequence().

void decode_usage ( int exit_code )

Prints a usage message and exits the program.

Parameters

exit_code Value to be returned by the program when it exits.

References decode_cleanup(), progname, and VERSION.

Referenced by main().

void decode_verify_print_value_list ( void )

Check the context of the global printValues array, to check that no s-attribute in it is declared more in the main print_list_index as well.

If an attribute is found to be declared in nboth, a warning is printed.

References corpus_id, decode_attribute_is_in_list(), print_list_index, and printValuesIndex.