From 5debff88a2eb295fe25da8764e4cdcb41f244275 Mon Sep 17 00:00:00 2001
From: minda <minda@kalafior.(none)>
Date: Wed, 18 Jan 2012 10:08:02 +0100
Subject: [PATCH] Changes in poliqarp library - new possibility to take all
sentence
---
poliqarp-library/sakura/poliqarp.c | 2 +-
poliqarp-library/sakura/poliqarp.h | 2 ++
poliqarp-library/sakura/query.c | 2 ++
poliqarp-library/sakura/value.h | 2 +-
poliqarp/c2pqtest.cpp | 2 +-
poliqarp/pqclient.cpp | 5 +++--
poliqarp/pqclient.h | 2 +-
poliqarp/pqreader.cpp | 17 ++++++++++++-----
poliqarp/pqreader.h | 6 +++++-
9 files changed, 28 insertions(+), 12 deletions(-)
diff --git a/poliqarp-library/sakura/poliqarp.c b/poliqarp-library/sakura/poliqarp.c
index 7d1a5c5..123876a 100644
--- a/poliqarp-library/sakura/poliqarp.c
+++ b/poliqarp-library/sakura/poliqarp.c
@@ -31,7 +31,7 @@
#define POLIQARP_MAJOR_VERSION 1
#define POLIQARP_MINOR_VERSION 3
-#define POLIQARP_REVISION_NUMBER 11
+#define POLIQARP_REVISION_NUMBER 12
#define POLIQARP_LIBRARY_NAME "sakura"
const int poliqarp_major_version = POLIQARP_MAJOR_VERSION;
diff --git a/poliqarp-library/sakura/poliqarp.h b/poliqarp-library/sakura/poliqarp.h
index a779d67..f5926ae 100644
--- a/poliqarp-library/sakura/poliqarp.h
+++ b/poliqarp-library/sakura/poliqarp.h
@@ -376,6 +376,8 @@ int poliqarp_resize_match_buffer(struct poliqarp_match_buffer *buffer,
struct poliqarp_match {
size_t start; /** Offset of the first segment that belongs to this
match. */
+ size_t withinEnd;
+ size_t withinStart;
size_t end; /** Offset of one-past-end segment in this match. */
size_t focus; /** Offset (relative to start of corpus) of focus point. */
size_t document; /** Document identifier associated with this match. */
diff --git a/poliqarp-library/sakura/query.c b/poliqarp-library/sakura/query.c
index 8413553..57943f1 100644
--- a/poliqarp-library/sakura/query.c
+++ b/poliqarp-library/sakura/query.c
@@ -683,6 +683,8 @@ look_breakout:
match.start = ctx.m_start;
assert(ctx.m_end != (size_t)-1);
match.end = ctx.m_end;
+ match.withinStart = ctx.subdocument.corpus_low;
+ match.withinEnd = ctx.subdocument.corpus_high;
match.focus = ctx.m_focus == (size_t)-1 ? ctx.m_start : ctx.m_focus;
match.document = corpus->document.current - 1;
pthread_mutex_lock(&result->mutex);
diff --git a/poliqarp-library/sakura/value.h b/poliqarp-library/sakura/value.h
index 7775847..62455b2 100644
--- a/poliqarp-library/sakura/value.h
+++ b/poliqarp-library/sakura/value.h
@@ -229,7 +229,7 @@ found_true:
* @param pos Position for which the value is calculated.
*/
static inline bool poliqarp_value_eval(const struct poliqarp_value *this,
- const struct poliqarp_corpus *corpus,
+ const struct poliqarp_corpus * corpus,
const struct poliqarp_binary_segment *pos)
{
switch (this->domain) {
diff --git a/poliqarp/c2pqtest.cpp b/poliqarp/c2pqtest.cpp
index 29a1114..3715675 100644
--- a/poliqarp/c2pqtest.cpp
+++ b/poliqarp/c2pqtest.cpp
@@ -18,7 +18,7 @@ int main(int argc, char** argv)
//while (Corpus2::Token* t = pqc.get_next_focus_token()) {
// writer->write_token_dispose(t);
//}
- while (Corpus2::Sentence::Ptr s = pqc.get_next_match_sequence()) {
+ while (Corpus2::Sentence::Ptr s = pqc.get_next_match_sequence(false)) {
writer->write_sentence(*s);
}
}
diff --git a/poliqarp/pqclient.cpp b/poliqarp/pqclient.cpp
index 494b4cd..90f1e91 100644
--- a/poliqarp/pqclient.cpp
+++ b/poliqarp/pqclient.cpp
@@ -115,11 +115,12 @@ Token* PoliqarpClient::get_next_focus_token()
}
};
-Sentence::Ptr PoliqarpClient::get_next_match_sequence()
+Sentence::Ptr PoliqarpClient::get_next_match_sequence(bool getWholeSentence)
{
poliqarp_match match;
if (next_match(match)) {
- return get_token_range(match.start, match.end);
+ if(getWholeSentence) return get_token_range(match.withinStart, match.withinEnd);
+ else return get_token_range(match.start, match.end);
} else {
return Sentence::Ptr();
}
diff --git a/poliqarp/pqclient.h b/poliqarp/pqclient.h
index 0420849..266186e 100644
--- a/poliqarp/pqclient.h
+++ b/poliqarp/pqclient.h
@@ -31,7 +31,7 @@ public:
bool next_match(poliqarp_match& match);
Token* get_next_focus_token();
- Sentence::Ptr get_next_match_sequence();
+ Sentence::Ptr get_next_match_sequence(bool getWholeSentence);
boost::shared_ptr<Chunk> get_next_document();
Token* get_token(size_t pos);
diff --git a/poliqarp/pqreader.cpp b/poliqarp/pqreader.cpp
index 6334b4f..2fb4c83 100644
--- a/poliqarp/pqreader.cpp
+++ b/poliqarp/pqreader.cpp
@@ -20,7 +20,7 @@ bool PoliqarpReader::registered = TokenReader::register_path_reader<PoliqarpRead
PoliqarpReader::PoliqarpReader(const Tagset &tagset, const std::string &filename)
: TokenReader(tagset), pq_(new PoliqarpClient(tagset, filename)),
- executed_(false), mode_(PQ_SENTENCES)
+ executed_(false), mode_(PQ_SENTENCES), getWholeSentence(false)
{
pq_->compile_query("[]+ within s");
}
@@ -33,7 +33,7 @@ void PoliqarpReader::set_query(const std::string &query)
{
pq_->compile_query(query);
mode_ = PQ_MANUAL;
- executed_ = false;
+ executed_ = false;
}
void PoliqarpReader::execute()
@@ -50,8 +50,8 @@ Token* PoliqarpReader::get_next_token()
Sentence::Ptr PoliqarpReader::get_next_sentence()
{
- if (!executed_) execute();
- return pq_->get_next_match_sequence();
+ if (!executed_) execute();
+ return pq_->get_next_match_sequence(getWholeSentence);
}
boost::shared_ptr<Chunk> PoliqarpReader::get_next_chunk()
@@ -74,11 +74,18 @@ void PoliqarpReader::set_option(const std::string &option)
pq_->compile_query("[]");
mode_ = PQ_TOKENS;
executed_ = false;
- } else {
+ } else if (option == "getWholeSentence") {
+ setGetWholeSentence(true);
+ }else {
TokenReader::set_option(option);
}
}
+void PoliqarpReader::setGetWholeSentence(bool getWholeSentence_)
+{
+ getWholeSentence = getWholeSentence_;
+}
+
std::string PoliqarpReader::get_option(const std::string& option) const
{
return TokenReader::get_option(option);
diff --git a/poliqarp/pqreader.h b/poliqarp/pqreader.h
index a8fb40b..e7dc398 100644
--- a/poliqarp/pqreader.h
+++ b/poliqarp/pqreader.h
@@ -40,6 +40,8 @@ public:
boost::shared_ptr<Chunk> get_next_chunk();
+ void setGetWholeSentence(bool getWholeSentence_);
+
void set_option(const std::string& option);
std::string get_option(const std::string& option) const;
@@ -51,9 +53,11 @@ protected:
boost::scoped_ptr<PoliqarpClient> pq_;
- bool executed_;
+ bool executed_;
PQ_MODE mode_;
+
+ bool getWholeSentence;
};
} /* end ns Corpus2 */
--
GitLab