From 94aca274de6022940db8c483307b1fe103d68553 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Mon, 29 Nov 2010 13:20:07 +0100
Subject: [PATCH] sentence loading in wcclparser
---
wcclparser/main.cpp | 67 +++++++++++++++++++++++++++------------------
1 file changed, 40 insertions(+), 27 deletions(-)
diff --git a/wcclparser/main.cpp b/wcclparser/main.cpp
index f1ddd3e..aed4a47 100644
--- a/wcclparser/main.cpp
+++ b/wcclparser/main.cpp
@@ -1,4 +1,5 @@
#include <cstdlib>
+#include <fstream>
#include <libwccl/values/strset.h>
#include <libwccl/parser/Parser.h>
@@ -6,6 +7,7 @@
#include <boost/bind.hpp>
#include <boost/program_options.hpp>
+#include <libcorpus2/io/xcesreader.h>
#include <antlr/MismatchedTokenException.hpp>
// ----------------------------------------------------------------------------
@@ -86,7 +88,7 @@ void libedit_read_loop(boost::function<bool (const std::string&)>& line_cb)
}
#endif
-bool process_line(const std::string& line, Parser& parser)
+bool process_line(const std::string& line, Parser& parser, Wccl::SentenceContext& sc)
{
if (line.empty() || line == "exit" || line == "quit") {
return true;
@@ -97,10 +99,6 @@ bool process_line(const std::string& line, Parser& parser)
boost::shared_ptr<const Wccl::Value> retVal;
boost::shared_ptr<ANTLRParserResultBase> retOp;
- boost::shared_ptr<Corpus2::Sentence> sentence;
- sentence.reset(new Corpus2::Sentence);
- Wccl::SentenceContext sc(sentence);
-
try {
retOp = parser.parseAnyOperator(line);
if (retOp) {
@@ -138,6 +136,8 @@ int main(int argc, char** argv)
{
std::string tagset_load = "kipi";
std::string query = "";
+ std::string sentence_load = "";
+ std::string position = "0";
bool quiet = false;
using boost::program_options::value;
@@ -145,6 +145,10 @@ int main(int argc, char** argv)
desc.add_options()
("tagset,t", value(&tagset_load),
"Tagset to use\n")
+ ("sentence,s", value(&sentence_load),
+ "Sentence to load (XCES)\n")
+ ("position,p", value(&position),
+ "Position in the sentence to use, 'all' iterates through the sentence\n")
("query,Q", value(&query),
"Query to run (disables interactive mode)\n")
("quiet,q", value(&quiet)->zero_tokens(),
@@ -171,31 +175,40 @@ int main(int argc, char** argv)
}
try {
- Corpus2::get_named_tagset(tagset_load);
- } catch (Corpus2::FileNotFound& e) {
- std::cerr << e.info() << std::endl;
- return 2;
- }
- const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_load);
-
- Parser parser(tagset);
-
- if (!query.empty()) {
- process_line(query, parser);
- return 0;
- }
-
- if (clear_screen()) {
- //
- }
-
- boost::function<bool (const std::string&)> f;
- f = boost::bind(&process_line, _1, boost::ref(parser));
+ const Corpus2::Tagset& tagset = Corpus2::get_named_tagset(tagset_load);
+ boost::shared_ptr<Corpus2::Sentence> sentence;
+ if (sentence_load.empty()) {
+ sentence.reset(new Corpus2::Sentence);
+ } else {
+ std::ifstream ifs(sentence_load.c_str());
+ if (ifs.good()) {
+ Corpus2::XcesReader reader(tagset, ifs, false);
+ sentence.reset(reader.get_next_sentence());
+ std::cerr << "Sentence loaded, " << sentence->size()
+ << " tokens.\n";
+ } else {
+ throw Wccl::FileNotFound(sentence_load, "", "Sentence");
+ }
+ }
+ Wccl::SentenceContext sc(sentence);
+ int pos = atoi(position.c_str());
+ sc.set_position(pos);
+ Parser parser(tagset);
+ if (!query.empty()) {
+ process_line(query, parser, sc);
+ return 0;
+ }
+ boost::function<bool (const std::string&)> f;
+ f = boost::bind(&process_line, _1, boost::ref(parser), boost::ref(sc));
#ifdef HAVE_LIBEDIT
- libedit_read_loop(f);
+ libedit_read_loop(f);
#else
- std_read_loop(f);
+ std_read_loop(f);
#endif
+ } catch (PwrNlp::PwrNlpError& e) {
+ std::cerr << e.info() << std::endl;
+ return 2;
+ }
return 0;
}
--
GitLab