AlpinoParser.cxx File Reference

#include "frog/AlpinoParser.h"
#include <string>
#include <iostream>
#include <ticcutils/XMLtools.h>
#include "ticcutils/StringOps.h"
#include "ticcutils/PrettyPrint.h"
#include "ticcutils/SocketBasics.h"
#include "ticcutils/FileUtils.h"
#include "frog/Frog-util.h"
#include "frog/FrogData.h"

Include dependency graph for AlpinoParser.cxx:

Macros
#define	LOG   *TiCC::Log(errLog)
#define	DBG   *TiCC::Dbg(dbgLog)

Functions
ostream &	operator<< (ostream &os, const dp_tree *node)
void	print_nodes (int indent, const dp_tree *store)
const dp_tree *	extract_hd (const dp_tree *node)
dp_tree *	parse_node (xmlNode *node)
dp_tree *	parse_nodes (xmlNode *node)
dp_tree *	resolve_mwus (dp_tree *in, int &compensate, int &restart, frog_data &fd)
dp_tree *	enumerate_top (dp_tree *in)
dp_tree *	resolve_mwus (dp_tree *in, frog_data &fd)
void	extract_dependencies (list< pair< const dp_tree *, const dp_tree * >> &result, const dp_tree *store, const dp_tree *top_root)
vector< parsrel >	extract (list< pair< const dp_tree *, const dp_tree * >> &l)
vector< parsrel >	extract_dp (xmlDoc *alp_doc, frog_data &fd)

Variables
const string	alpino_tagset = "http://ilk.uvt.nl/folia/sets/alpino-parse-nl"
const string	alpino_mwu_tagset = "http://ilk.uvt.nl/folia/sets/alpino-mwu-nl"

Macro Definition Documentation

DBG

#define DBG   *TiCC::Dbg(dbgLog)

LOG

#define LOG   *TiCC::Log(errLog)

Function Documentation

enumerate_top()

dp_tree* enumerate_top

(

dp_tree *

in

)

Enumerate the TOP nodes in the dp_tree

Parameters

in	the tree to enumerate

Returns

a pointer to the found top node

This also changes the linking order of the children under in.

These are NOT always sorted from low to high. We make it so and return a pointer to the new top

extract()

vector<parsrel> extract

(

list< pair< const dp_tree *, const dp_tree * >> &

l

)

convert a list of head-dependent pairs into a list of parsrel records

Parameters

l	a list of head-dependent pairs

Returns

a list of parsrel records. This to be able to use the same code as for the build-in non-alpino parser.

This function does a lot of trickery to handle special nodes.

extract_dependencies()

void extract_dependencies	(	list< pair< const dp_tree *, const dp_tree * >> &	result,
		const dp_tree *	store,
		const dp_tree *	top_root
	)

recursively extract all head-dependent pairs from store

Parameters

result	an aggregated list of all dependency pairs found
store	The tree to search through
top_root	The ROOT node of this sequence. For the topmost store it is 0

extract_dp()

vector<parsrel> extract_dp	(	xmlDoc *	alp_doc,
		frog_data &	fd
	)

extract a list of parsrel records from an Alpino XML file and resolve MWU's

Parameters

alp_doc	an Alpino XML document
fd	a frog_data structure to receive the MWU information

Returns

a list of parsrel records

extract_hd()

const dp_tree* extract_hd

(

const dp_tree *

node

)

search a dp_tree for a head node

Parameters

node	the top node to start searching

Returns

the dp_tree structure holding a head

a head is defined as a node with a rel value of 'hd', 'crd' ot 'cmp'

operator<<()

ostream& operator<<	(	ostream &	os,
		const dp_tree *	node
	)

print out one dp_tree structure node

Parameters

os	the output stream
node	the node to print

parse_node()

dp_tree* parse_node

(

xmlNode *

node

)

convert a singel node in an Alpino XML tree into a much simpler dp_tree node

Parameters

node	The Alpino XML node to parse

Returns

a dp_tree structure with the essential information from Alpino

parse_nodes()

dp_tree* parse_nodes

(

xmlNode *

node

)

recurively convert an Alpino XML tree into a much simpler dp_tree tree

Parameters

node	The Alpino XML tree to parse

Returns

A dp_tree node tree

an Alpino XML tree is quite complex. We try to simplify it and extract only what is needed.

print_nodes()

void print_nodes	(	int	indent,
		const dp_tree *	store
	)

recursively pretty print out a dp_tree tree to stderr

Parameters

indent	indentation level
store	the top node to print

resolve_mwus() [1/2]

dp_tree* resolve_mwus	(	dp_tree *	in,
		frog_data &	fd
	)

search for MWU's in the dp_tree and register them in the frog_data

Parameters

in	The dp_tree to analyze
fd	the frog_data to fill

Returns

pointer to the (modified) input.

resolve_mwus() [2/2]

dp_tree* resolve_mwus	(	dp_tree *	in,
		int &	compensate,
		int &	restart,
		frog_data &	fd
	)

extract MWU information into frog_data.

Parameters

in	the dp_tree to search through
compensate	complicated
restart	complicated
fd	the frog_data structure to register the MWU in.

Returns

the modified dp_tree

this function searches the dp_tree for MWUS (rel='mwp'). It will merge the found MWU words in a modified dp_tree AND register the start and finish indexes of the mwu in fd

The indices of the MWU's in dp_tree are not always consecutively, so we have to check that using restart.

Also, after modifying the tree, the indices are wrong by the size of the MWU. so we need to compensate for that too.

Variable Documentation

alpino_mwu_tagset

const string alpino_mwu_tagset = "http://ilk.uvt.nl/folia/sets/alpino-mwu-nl"

alpino_tagset

const string alpino_tagset = "http://ilk.uvt.nl/folia/sets/alpino-parse-nl"