From df54b988af6d1d3c4b930a0a666505e16cd0e93a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Luk=C3=A1=C5=A1=20Paukert?= <paukeluk@fit.cvut.cz>
Date: Fri, 10 Apr 2020 18:27:22 +0200
Subject: [PATCH] Improved getting query
---
querying/src/calculation/Space.cpp | 12 ------------
querying/src/calculation/Space.h | 8 --------
querying/src/database/Database.cpp | 13 +++++++------
querying/src/database/Database.h | 6 +++---
querying/src/ui/MainPage.cpp | 2 +-
5 files changed, 11 insertions(+), 30 deletions(-)
diff --git a/querying/src/calculation/Space.cpp b/querying/src/calculation/Space.cpp
index 77c608d..6b05dbf 100644
--- a/querying/src/calculation/Space.cpp
+++ b/querying/src/calculation/Space.cpp
@@ -17,15 +17,3 @@ const InvertedIndex &Space::operator[](const string &key) const {
InvertedIndex &Space::getInvertedIndexByKey(const string &key) {
return terms.at(key);
}
-
-const map<string, double> Space::getTermsAndWeightsByID(Database & database, int document_id) {
- map<string, double> terms;
- vector<string> dummy = database.getTermsByDocumentID(document_id);
-
- for (const string & term : dummy) {
- InvertedIndex tmp = getInvertedIndexByKey(term);
- terms[term] = tmp.getDocumentWeightByID(document_id);
- }
-
- return terms;
-}
diff --git a/querying/src/calculation/Space.h b/querying/src/calculation/Space.h
index d3e643a..f7eeadb 100644
--- a/querying/src/calculation/Space.h
+++ b/querying/src/calculation/Space.h
@@ -3,7 +3,6 @@
#include <string>
-#include <set>
#include <map>
#include "InvertedIndex.h"
@@ -35,13 +34,6 @@ public:
*/
InvertedIndex &getInvertedIndexByKey(const std::string &key);
- /**
- * @brief Finds all terms in DB which occurs in specific document
- * @param database, document_id Instance of DB connection and document_id to process
- * @return Map with terms as keys and weights as their values
- */
- const std::map<std::string, double> getTermsAndWeightsByID(Database & database, int document_id);
-
/**
* @brief Gets an element from @ref Space::terms with key @ref key
* @param key Key of the element to find
diff --git a/querying/src/database/Database.cpp b/querying/src/database/Database.cpp
index 26887a4..6439c7f 100644
--- a/querying/src/database/Database.cpp
+++ b/querying/src/database/Database.cpp
@@ -24,17 +24,18 @@ Document Database::getDocumentByID(int id) {
return {id, query.getColumn(0)};
}
-vector<string> Database::getTermsByDocumentID(int document_id) {
- vector<string> terms;
- SQLite::Statement query(db, "SELECT Term.value FROM Term "
+map<string, double> Database::getTermsAndWightsByDocumentID(int document_id) {
+ map<string, double> termsAndWeights;
+ SQLite::Statement query(db, "SELECT Term.value, TermDocumentOccurrence.weight FROM Term "
"JOIN TermDocumentOccurrence ON Term.id = TermDocumentOccurrence.Term_id "
"WHERE TermDocumentOccurrence.Document_id = :id");
query.bind(":id", document_id);
- while(query.executeStep())
- terms.emplace_back(query.getColumn(0));
+ while(query.executeStep()) {
+ termsAndWeights[query.getColumn(0)] = query.getColumn(1);
+ }
- return terms;
+ return termsAndWeights;
}
map<int, double> Database::getVectorSizes() {
diff --git a/querying/src/database/Database.h b/querying/src/database/Database.h
index 8e759c6..2bb4a66 100644
--- a/querying/src/database/Database.h
+++ b/querying/src/database/Database.h
@@ -31,11 +31,11 @@ public:
Document getDocumentByID(int id);
/**
- * @brief Finds all terms from specified document
+ * @brief Finds all terms and their weights from specified document
* @param document_id to process
- * @return Vector with strings which are in specified document
+ * @return Map with terms as keys and weights as their values
*/
- std::vector<std::string> getTermsByDocumentID(int document_id);
+ std::map<std::string, double> getTermsAndWightsByDocumentID(int document_id);
/**
* @brief Computes size of vector for every document in database
diff --git a/querying/src/ui/MainPage.cpp b/querying/src/ui/MainPage.cpp
index 716e4b7..03075ca 100644
--- a/querying/src/ui/MainPage.cpp
+++ b/querying/src/ui/MainPage.cpp
@@ -75,7 +75,7 @@ void MainPage::displayDetail(Space space, Wt::WContainerWidget * container, int
Document document = database.getDocumentByID(document_id);
// threshold je nyni nastaven na -1 --> ve vysledku budou i uplne rozdilne dokumenty
- Query query(space.getTermsAndWeightsByID(database, document.id), -1);
+ Query query(database.getTermsAndWightsByDocumentID(document_id), -1);
auto result = Computor(space, query).compute(database);
// dodelat proklikavani na zobrazene podobne dokumenty
--
GitLab