diff --git a/CMakeLists.txt b/CMakeLists.txt
index da93d71d69a335ab1aff9fb620c7ec7506691bd1..f7ade6a771f95a12b32de43b34ed6a28c6421f66 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -150,6 +150,16 @@ add_executable(ISROR-tests
util/stringProcessing.cpp
util/Stemmer.cpp )
+add_executable(ISRAnd-tests
+ util/util.cpp
+ constraintSolver/ISR.cpp
+ constraintSolver/ISRWord.cpp
+ constraintSolver/ISRAnd.cpp
+ constraintSolver/tests/ISROrTests.cpp
+ constraintSolver/ISREndDoc.cpp
+ util/stringProcessing.cpp
+ util/Stemmer.cpp )
+
find_package(OpenSSL REQUIRED)
diff --git a/constraintSolver/ISR.h b/constraintSolver/ISR.h
index b8096879a1506e6e17e43237571b0050f000b906..0f9307606f810ad7b926c1fef39a773f7705dec6 100644
--- a/constraintSolver/ISR.h
+++ b/constraintSolver/ISR.h
@@ -12,8 +12,10 @@
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
+#include <limits>
typedef size_t Location; // Location 0 is the null location.
+const Location MAX_Location = std::numeric_limits<unsigned>::max();
class ISR
{
diff --git a/constraintSolver/ISRAnd.cpp b/constraintSolver/ISRAnd.cpp
index 82ff1555c9a6baf146e496f09afe815af242bd74..ff50caf07e1d05abd3afe8fbee85517b28c2ca11 100644
--- a/constraintSolver/ISRAnd.cpp
+++ b/constraintSolver/ISRAnd.cpp
@@ -3,3 +3,115 @@
//
#include "ISRAnd.h"
+#include <cassert>
+
+ISRAnd::ISRAnd ( vector<ISR * > InputTerms ) : Terms( InputTerms )
+ {
+
+ assert(InputTerms.size() > 1);
+
+ for (auto currentTerm : InputTerms)
+ {
+ currentTerm->First();
+ Location currentLocation = currentTerm->currentLocation;
+ if (currentLocation < nearestStartLocation) {
+ nearestTerm = currentTerm;
+ nearestStartLocation = currentLocation;
+
+ }
+ if (currentLocation > nearestEndLocation) {
+ nearestEndLocation = currentLocation;
+ }
+ ++NumberOfTerms;
+ currentTerm++;
+
+ }
+ }
+
+Location ISRAnd::First()
+ {
+ //Fixme?
+ Location first = MAX_Location;
+
+ for(auto isr : Terms)
+ {
+ Location temp = isr->First();
+ if (temp < first)
+ {
+ first = temp;
+ }
+ }
+
+ //fixme should this return the nearest location of one subterm or the nearest location all the terms match?
+ currentLocation = Seek(first);
+ return currentLocation;
+ }
+
+
+Location ISRAnd::Next ( )
+ {
+ return Seek( nearestStartLocation);
+ }
+
+Location ISRAnd::NextDocument()
+ {
+ return Seek(GetEndDocument());
+ }
+
+Location ISRAnd::Seek ( Location target )
+ {
+ //Todo
+ // 1. Seek all the ISRs to the first occurrence beginning at
+ // the target location.
+ // 2. Move the document end ISR to just past the furthest
+ // word, then calculate the document begin location.
+ // 3. Seek all the other terms to past the document begin.
+ // 4. If any term is past the document end, return to
+ // step 2.
+ // 5. If any ISR reaches the end, there is no match.
+
+ Location nearest;
+ Location furthest = 0;
+
+ while(furthest != MAX_Location) {
+
+ nearest = MAX_Location;
+
+ //find nearest & furthest ISR
+ for (auto isr : Terms) {
+ Location temp = isr->Seek(target);
+ if (temp == MAX_Location) return MAX_Location;
+ if (temp > furthest) furthest = temp;
+ if (temp < nearest) nearest = temp;
+ }
+
+ //Get Document of the furthest ISR
+ DocumentEnd->Seek(target);
+
+ //set next target to be starting location of document
+ //Location lastDocStart = DocumentEnd->GetStart()?;
+ Location lastDocStart = 0;
+
+ if(nearest >= lastDocStart)
+ {
+ //MATCH
+ //Does it matter for 'AND' what location we return? first/last in document?
+ //Should we return an ISR as well?
+ return nearest;
+ } else{
+ //No Match yet, keep searching
+ target = lastDocStart;
+ }
+
+ }
+
+ }
+
+Location ISRAnd::GetEndDocument()
+ {
+ //What does currentLocation hold? When is it updated?
+ return DocumentEnd->Seek(currentLocation);
+ }
+
+
+
diff --git a/constraintSolver/ISRAnd.h b/constraintSolver/ISRAnd.h
index f899588143fd8f0317d2a73575ccb23b43ed7f8c..60921e31e1fa4f239da0df2f9df92a1f630b01ab 100644
--- a/constraintSolver/ISRAnd.h
+++ b/constraintSolver/ISRAnd.h
@@ -5,37 +5,31 @@
#pragma once
#include "ISR.h"
+#include <vector>
-
+using namespace std;
//Find occurrences of all child ISRs within a single document
-
class ISRAnd : ISR
{
public:
- ISR **Terms;
+ vector<ISR*>Terms;
unsigned NumberOfTerms;
- Post *Seek ( Location target )
- {
- // 1. Seek all the ISRs to the first occurrence beginning at
- // the target location.
- // 2. Move the document end ISR to just past the furthest
- // word, then calculate the document begin location.
- // 3. Seek all the other terms to past the document begin.
- // 4. If any term is past the document end, return to
- // step 2.
- // 5. If any ISR reaches the end, there is no match.
- }
+ ISRAnd ( vector<ISR * > InputTerms );
+
+ Location First ( ) override;
+ Location Next ( ) override;
+ Location NextDocument ( ) override;
+ Location Seek ( Location target ) override;
+ Location GetEndDocument ( ) override;
+ Location GetCurrentLocation();
- Post *Next ( )
- {
- return Seek( nearestStartLocation + 1 );
- }
private:
- unsigned nearestTerm, farthestTerm;
+ ISR *nearestTerm;
+ // nearStartLocation and nearestEndLocation are// the start and end of the nearestTerm.
Location nearestStartLocation, nearestEndLocation;
};
diff --git a/constraintSolver/ISROr.cpp b/constraintSolver/ISROr.cpp
index e765d1e081b581c694106561affda796aba553be..d3282a1727df2f678268f217f5dfa7bd74ee41c3 100644
--- a/constraintSolver/ISROr.cpp
+++ b/constraintSolver/ISROr.cpp
@@ -5,6 +5,28 @@
#include "ISROr.h"
+ISROr::ISROr ( vector<ISR * > InputTerms ) : Terms( InputTerms )
+ {
+
+ for (auto currentTerm : InputTerms)
+ {
+ currentTerm->First();
+ Location currentLocation = currentTerm->currentLocation;
+ if (currentLocation < nearestStartLocation) {
+ nearestTerm = currentTerm;
+ nearestStartLocation = currentLocation;
+
+ }
+ if (currentLocation > nearestEndLocation) {
+ nearestEndLocation = currentLocation;
+ }
+ ++NumberOfTerms;
+ currentTerm++;
+
+ }
+ }
+
+
Location ISROr::GetStartLocation ( )
{
return nearestStartLocation;
diff --git a/constraintSolver/ISROr.h b/constraintSolver/ISROr.h
index 29d3b0936de276ca52e1e7257cf7146a99cf961f..657c3fee37363b7665b2f43101688b26227bd38c 100644
--- a/constraintSolver/ISROr.h
+++ b/constraintSolver/ISROr.h
@@ -15,6 +15,8 @@ public:
vector<ISR*>Terms;
unsigned NumberOfTerms;
+ ISROr ( vector<ISR * > InputTerms );
+
Location First ( ) override;
Location Next ( ) override;
Location NextDocument ( ) override;
@@ -27,30 +29,6 @@ public:
Location GetEndLocation ( );
- ISROr ( vector<ISR * > InputTerms ) : Terms( InputTerms )
- {
-
- for(auto currentTerm : InputTerms)
- {
- currentTerm->First( );
- Location currentLocation = currentTerm->currentLocation;
- if ( currentLocation < nearestStartLocation )
- {
- nearestTerm = currentTerm;
- nearestStartLocation = currentLocation;
-
- }
- if ( currentLocation > nearestEndLocation )
- {
- nearestEndLocation = currentLocation;
- }
- ++NumberOfTerms;
- currentTerm++;
-
- }
-
- }
-
private:
ISR *nearestTerm;
// nearStartLocation and nearestEndLocation are// the start and end of the nearestTerm.
diff --git a/constraintSolver/tests/ISROrTests.cpp b/constraintSolver/tests/ISROrTests.cpp
index 2db362b0a3e4f036aeaf68dc8b941e8e6d4ace7b..3f982cb10d120ae34b2372232797e8439e5349db 100644
--- a/constraintSolver/tests/ISROrTests.cpp
+++ b/constraintSolver/tests/ISROrTests.cpp
@@ -24,7 +24,7 @@ int main ( )
vector<size_t> locations;
vector<DocumentEnding> docEnds;
set<string> urls;
- while(queryOr->GetCurrentLocation() != 9999999999999) {
+ while(queryOr->GetCurrentLocation() != MAX_Location) {
locations.push_back(queryOr->Next());
}
while(endDocs.next().url != "aaa")
diff --git a/constraintSolver/tests/ISRWordTests.cpp b/constraintSolver/tests/ISRWordTests.cpp
index 37d4f8356b75dc3dfd4916485f4ebd20a33bfa96..58b4cf50bacb04b1fc6ed363b1eeee725de1a111 100644
--- a/constraintSolver/tests/ISRWordTests.cpp
+++ b/constraintSolver/tests/ISRWordTests.cpp
@@ -14,13 +14,13 @@ using namespace std;
int main ( )
{
char* query = "iphone";
- ISRWord queryWord(query);
+ char* query;
+ ISRWord queryWord("aare");
ISREndDoc endDocs;
vector<size_t> locations;
vector<DocumentEnding> docEnds;
set<string> urls;
- locations.push_back(queryWord.getCurrentLocation());
- while(queryWord.getCurrentLocation() != 9999999999999) {
+ while(queryWord.getCurrentLocation() != MAX_Location) {
locations.push_back(queryWord.Next());
}
for(auto loc : locations) {