From 52febe524767618b248bd8b3ba03b3f8a97a68b8 Mon Sep 17 00:00:00 2001
From: Tomas Koristka <koristo1@fit.cvut.cz>
Date: Sat, 28 Mar 2020 11:43:05 +0100
Subject: [PATCH] Changing inverted index to deque
---
querying/src/Computor.cpp | 59 ++++++++-----------
querying/src/Computor.h | 6 +-
querying/src/inverted_index/InputParser.cpp | 2 +-
querying/src/inverted_index/InvertedIndex.cpp | 32 ++++------
querying/src/inverted_index/InvertedIndex.h | 11 +---
querying/src/space/Query.cpp | 13 +++-
querying/src/space/Query.h | 6 +-
querying/src/space/Space.cpp | 2 +-
querying/src/space/Space.h | 2 +-
9 files changed, 60 insertions(+), 73 deletions(-)
diff --git a/querying/src/Computor.cpp b/querying/src/Computor.cpp
index 693500e..ebbd268 100644
--- a/querying/src/Computor.cpp
+++ b/querying/src/Computor.cpp
@@ -1,58 +1,49 @@
#include <iostream>
#include <climits>
+#include <utility>
#include "Computor.h"
#include "exceptions/Exceptions.h"
using namespace std;
-Computor::Computor(Space space, const Query &query) : space(space), query(query) {}
+Computor::Computor(Space space, Query query) : space(std::move(space)), query(std::move(query)) {}
map<int, double> Computor::compute() {
- exhaustedIndices = {};
- auto totalIndices = query.terms.size();
-
- int lowestDocumentID = nextID();
- space.forward(lowestDocumentID, query.terms);
map<int, double> results;
- while (exhaustedIndices.size() != totalIndices) {
- results.emplace(lowestDocumentID, computeForDocument(lowestDocumentID));
- lowestDocumentID = nextID();
- space.forward(lowestDocumentID, query.terms);
- }
-
- return results;
-}
+ availableTerms = query.termsKeyset;
+ while (!availableTerms.empty()) {
+ int ID = nextID(); //get lowest ID
+ double result = 0;
-double Computor::computeForDocument(int ID) {
- double result = 0;
+ for (const auto &term: availableTerms) /*Go through all the remaining terms*/ {
+ try {
+ double queryWeight = query.terms.at(term);
+ double documentWeight = space.getInvertedIndexByKey(term).getDocumentWeightByID(ID);
+ result += documentWeight * queryWeight;
+ }
+ catch (const IDNotFoundException &e) {//inverted index does not contain given ID
+ }
- for (const auto &[term, weight]: query.terms) {
- try {
- auto d = space.getInvertedIndexByKey(term).getDocumentWeightByID(ID);
- result += weight * d;
- }
-
- catch (EndOfIndexException &e) {
- exhaustedIndices.emplace(term); //add another exhausted inverted index
- }
- catch (IDNotFoundException &e) {
- continue;
+ catch (const EndOfIndexException &e) {
+ availableTerms.erase(term); //exhaust term
+ }
}
+ results[ID] = result;
}
- return result;
+ return results;
}
+int Computor::nextID() {
+ int lowestID = INT_MAX;
-int Computor::nextID() const {
- int nextID = INT_MAX;
+ for (const auto &queryRecord: query.terms)
+ lowestID = min(space[queryRecord.first].getLowestID(), lowestID);
- for (const auto &term: query.terms)
- nextID = min(space[term.first].getLowestID(), nextID);
-
- return nextID;
+ return lowestID;
}
+
diff --git a/querying/src/Computor.h b/querying/src/Computor.h
index 3615170..9a9fa86 100644
--- a/querying/src/Computor.h
+++ b/querying/src/Computor.h
@@ -14,15 +14,15 @@ private:
//Attributes--------------
Space space;
const Query query;
- std::set<std::string> exhaustedIndices;
+ std::set<std::string> availableTerms;
//Methods-----------------
- int nextID() const;
+ int nextID();
double computeForDocument(int ID);
public:
- Computor(Space space, const Query &query);
+ Computor(Space space, Query query);
std::map<int, double> compute();
};
diff --git a/querying/src/inverted_index/InputParser.cpp b/querying/src/inverted_index/InputParser.cpp
index f8d1e99..a26ac60 100644
--- a/querying/src/inverted_index/InputParser.cpp
+++ b/querying/src/inverted_index/InputParser.cpp
@@ -25,7 +25,7 @@ map<string, InvertedIndex> InputParser::getInvertedIndices() {
sort(weights.begin(), weights.end(),
[](const auto &a, const auto &b) { return a.getID() < b.getID(); });
- invertedIndices.insert({term, InvertedIndex(term, weights)});
+ invertedIndices.emplace(term, InvertedIndex(weights));
}
return invertedIndices;
diff --git a/querying/src/inverted_index/InvertedIndex.cpp b/querying/src/inverted_index/InvertedIndex.cpp
index 7a464a5..ed048ee 100644
--- a/querying/src/inverted_index/InvertedIndex.cpp
+++ b/querying/src/inverted_index/InvertedIndex.cpp
@@ -4,16 +4,21 @@
using namespace std;
-InvertedIndex::InvertedIndex(string term, vector<DocumentWeight> documentWeights)
- : position(0),
- term(move(term)),
- documentWeights(move(documentWeights)) {}
+InvertedIndex::InvertedIndex(deque<DocumentWeight> documentWeights)
+ : documentWeights(move(documentWeights)) {}
+
double InvertedIndex::getDocumentWeightByID(int ID) {
if (position + 1 == documentWeights.size())
throw EndOfIndexException();
+
+ while (documentWeights.front().getID() != ID) {
+ documentWeights.pop_front();
+ }
+ return documentWeights.front().getID();
+
for (auto it = documentWeights.begin() + position; it != documentWeights.end(); it++) {
int documentID = (*it).getID();
@@ -21,25 +26,8 @@ double InvertedIndex::getDocumentWeightByID(int ID) {
throw IDNotFoundException();
if (documentID == ID) //match!
- {
- position = it - documentWeights.begin() + 1; //update position
return (*it).getWeight();
- }
- }
-
- /*
-
- for (size_t i = position; position < documentWeights.size(); ++i) {
- int documentID = documentWeights[i].getWeight();
-
- if (documentID < ID)
- throw IDNotFoundException();
- if (documentID == ID) {
- position = i + 1;
- return documentWeights[i];
- }
}
-*/
throw IDNotFoundException();
}
@@ -49,7 +37,7 @@ const DocumentWeight &InvertedIndex::operator[](size_t i) {
}
int InvertedIndex::getLowestID() const {
- return documentWeights[position].getID();
+ return documentWeights.front().getID();
}
void InvertedIndex::forward(int ID) {
diff --git a/querying/src/inverted_index/InvertedIndex.h b/querying/src/inverted_index/InvertedIndex.h
index 1b9dee3..677715e 100644
--- a/querying/src/inverted_index/InvertedIndex.h
+++ b/querying/src/inverted_index/InvertedIndex.h
@@ -3,7 +3,7 @@
#include <string>
-#include <vector>
+#include <deque>
#include "../DocumentWeight.h"
/**
@@ -14,19 +14,16 @@
class InvertedIndex {
private:
//Attributes-------------
- size_t position;
- std::string term;/**<Term identifier */
- std::vector<DocumentWeight> documentWeights; /**<Inverted index list of documents & their weights in which the term appears */
+ std::deque<DocumentWeight> documentWeights; /**<Inverted index list of documents & their weights in which the term appears */
public:
//Methods-----------------
/**
* Constructor
- * @param term Term identifier
* @param documentWeights Inverted index list of documents in which the term appears, and their weights
*/
- InvertedIndex(std::string term, std::vector<DocumentWeight> documentWeights);
+ InvertedIndex(std::deque<DocumentWeight> documentWeights);
/**
* @brief Finds the @ref DocumentWeight object with a given @ref DocumentWeight::ID in @ref documentWeights
@@ -42,8 +39,6 @@ public:
int getLowestID() const;
- void forward(int ID);
-
const DocumentWeight &operator[](size_t i);
};
diff --git a/querying/src/space/Query.cpp b/querying/src/space/Query.cpp
index 7c3eac2..6d92bfe 100644
--- a/querying/src/space/Query.cpp
+++ b/querying/src/space/Query.cpp
@@ -6,5 +6,14 @@
using namespace std;
-Query::Query(std::map<std::string, double> terms, const double threshold)
- : terms(std::move(terms)), threshold(threshold) {}
+Query::Query(std::map<std::string, double> terms, double threshold)
+ : terms(std::move(terms)),
+ termsKeyset(getKeyset(terms)),
+ threshold(threshold) {}
+
+set<string> Query::getKeyset(const map<std::string, double> &terms) {
+ set<string> dummy;
+ for (const auto &term: terms)
+ dummy.emplace(term.first);
+ return dummy;
+}
diff --git a/querying/src/space/Query.h b/querying/src/space/Query.h
index 0b23e77..bda25cd 100644
--- a/querying/src/space/Query.h
+++ b/querying/src/space/Query.h
@@ -3,6 +3,7 @@
#include <map>
+#include <set>
#include <vector>
#include <string>
#include "Space.h"
@@ -14,6 +15,7 @@ class Query {
public:
//Attributes-------------
const std::map<std::string, double> terms; /**<Terms and their weights in the query */
+ const std::set<std::string> termsKeyset; /**<Keyset of terms */
const double threshold; /**<Acceptable result threshold */
//Methods----------------
@@ -22,7 +24,9 @@ public:
* @param terms Terms and their weights in the query
* @param threshold Acceptable result threshold
*/
- explicit Query(std::map<std::string, double> terms, const double threshold);
+ explicit Query(std::map<std::string, double> terms, double threshold);
+
+ static std::set<std::string> getKeyset(const std::map<std::string, double> &terms);
};
diff --git a/querying/src/space/Space.cpp b/querying/src/space/Space.cpp
index 727dfdb..f1c8fc2 100644
--- a/querying/src/space/Space.cpp
+++ b/querying/src/space/Space.cpp
@@ -17,7 +17,7 @@ InvertedIndex &Space::getInvertedIndexByKey(const string &key) {
return terms.at(key);
}
-void Space::forward(int ID, const std::map<std::string, double> q) {
+void Space::forward(int ID, const std::map<std::string, double> &q) {
for (const auto &term: q)
terms.at(term.first).forward(ID);
}
diff --git a/querying/src/space/Space.h b/querying/src/space/Space.h
index 641355e..783bd0b 100644
--- a/querying/src/space/Space.h
+++ b/querying/src/space/Space.h
@@ -32,7 +32,7 @@ public:
const InvertedIndex &operator[](const std::string &key) const;
- void forward(int ID, const std::map<std::string, double> q);
+ void forward(int ID, const std::map<std::string, double> &q);
};
--
GitLab