isr word test working again

constraintSolver/ISREndDoc.cpp

@@ -11,11 +11,11 @@ ISREndDoc::ISREndDoc() {
 
 DocumentEnding ISREndDoc::next() {
     if(memMap == nullptr) {
-        string fileName = util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/index" + to_string(currentChunk) + ".txt";
+        string fileName = util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/" + to_string(currentChunk) + ".txt";
         currentFile = open(fileName.c_str(), O_RDONLY);
-        vector<size_t> contents = getSeekContents();
+        MMDiskHashTable de(util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/" + to_string(currentChunk) + "-seek.txt", 30, 8);
         memMap = (char*) mmap(nullptr, util::FileSize(currentFile), PROT_READ, MAP_PRIVATE, currentFile, 0);
-        memMap += contents[0];
+        memMap += stoll(de.find("=docEnding"));
     }
     string currentOne;
     for(char* map = memMap; map < memMap + util::FileSize(currentFile); map++) {
@@ -24,7 +24,7 @@ DocumentEnding ISREndDoc::next() {
             memMap = nullptr;
             return DocumentEnding();
         }
-        if(currentChunk == 8) {
+        if(currentChunk == 5) {
             DocumentEnding a = DocumentEnding();
             a.url = "aaa";
             return a;
@@ -64,45 +64,3 @@ DocumentEnding ISREndDoc::next() {
 DocumentEnding ISREndDoc::getCurrentDoc() {
     return currentDoc;
 }
-
-vector<size_t> ISREndDoc::getSeekContents() {
-    string fileName = util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/index" + to_string(currentChunk) + "-seek.txt";
-    int file = open(fileName.c_str(), O_RDONLY);
-    ssize_t fileSize = util::FileSize(file);
-    vector<size_t> contents;
-
-
-    char* memMap = (char*) mmap(nullptr, fileSize, PROT_READ, MAP_PRIVATE, file, 0);
-    // char* memMap = util::getFileMap(fileName);
-    string word = "";
-    bool midWord = false;
-    bool midFind = false;
-    if(memMap != MAP_FAILED) {
-        for(char* map = memMap; map < memMap + fileSize; map++) {
-            if(midFind && isalpha(*map)) {
-                break;
-            }
-            switch(*map) {
-                case '\n':
-                case '\r':
-                case '\t':
-                case ' ':
-                    if (midFind && word != "") {
-                        contents.push_back(stoll(word));
-                        word = "";
-                    } else if (midWord) {
-                        midWord = false;
-                        if(word == "=docEnding") {
-                            midFind = true;
-                        }
-                        word = "";
-                    }
-                    break;
-                default:
-                    word += *map;
-                    midWord = true;
-            }
-        }
-    }
-    return contents;
-}
\ No newline at end of file

constraintSolver/ISREndDoc.h

@@ -17,6 +17,7 @@
 #include "WordSeek.h"
 #include "../util/util.h"
 #include "../indexer/DocumentEnding.h"
+#include "../DataStructures/DiskHashTable/MMDiskHashTable.h"
 
 // Find occurrences of document ends.
 

constraintSolver/ISRWord.cpp

@@ -141,26 +141,26 @@ void ISRWord::getWordSeek() {
 //if so, set location to that big chunk
 //go to next chunk
 Location ISRWord::Seek( Location target ) {
-//    if(!wordSeekLookupTable.empty()) {
-//        auto best = wordSeekLookupTable.front();
-//        for(auto entry : wordSeekLookupTable) {
-//            if(entry.realLocation < target) {
-//                best = entry;
-//            } else {
-//                string currentChunkFileLocation = util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/index" + to_string(listOfChunks[currentChunk]) + ".txt";
-//                int currentChunkFile = open(currentChunkFileLocation.c_str(), O_RDONLY);
-//                ssize_t currentChunkFileSize = FileSize(currentChunkFile);
-//                currentMemMap = (char*) mmap(nullptr, currentChunkFileSize, PROT_READ, MAP_PRIVATE, currentChunkFile, 0);
-//                currentMemMap += best.seekOffset;
-//                currentLocation = best.realLocation;
-//                return best.realLocation;
-//            }
-//        }
-//    } else {
-//        while(Next() <= target) {
-//        }
-//        return currentLocation;
-//    }
+    if(!wordSeekLookupTable.empty()) {
+        auto best = wordSeekLookupTable.front();
+        for(auto entry : wordSeekLookupTable) {
+            if(entry.realLocation < target) {
+                best = entry;
+            } else {
+                string currentChunkFileLocation = util::GetCurrentWorkingDir() + "/constraintSolver/index-test-files/twitter/" + to_string(listOfChunks[currentChunk]) + ".txt";
+                int currentChunkFile = open(currentChunkFileLocation.c_str(), O_RDONLY);
+                ssize_t currentChunkFileSize = FileSize(currentChunkFile);
+                currentMemMap = (char*) mmap(nullptr, currentChunkFileSize, PROT_READ, MAP_PRIVATE, currentChunkFile, 0);
+                currentMemMap += best.seekOffset;
+                currentLocation = best.realLocation;
+                return best.realLocation;
+            }
+        }
+    } else {
+        while(Next() <= target) {
+        }
+        return currentLocation;
+    }
 }
 
 

constraintSolver/tests/ISRWordTests.cpp

@@ -14,7 +14,7 @@ using namespace std;
 int main ( )
 {
     char* query = "iphone";
-    ISRWord queryWord("fsfasfas");
+    ISRWord queryWord("nike");
     ISREndDoc endDocs;
     vector<size_t> locations;
     vector<DocumentEnding> docEnds;
@@ -23,22 +23,19 @@ int main ( )
     while(queryWord.getCurrentLocation() != MAX_Location) {
         locations.push_back(queryWord.Next());
     }
-    for(auto loc : locations) {
-        cout << loc << endl;
+    while(endDocs.next().url != "aaa")
+        {
+        for(auto locs : locations)
+            {
+            if(locs < endDocs.getCurrentDoc().docEndPosition &&
+               locs >= (endDocs.getCurrentDoc().docEndPosition - endDocs.getCurrentDoc().docNumWords)) {
+                urls.insert(endDocs.getCurrentDoc().url);
+            }
+        }
+
+        }
+    for(auto urrl : urls) {
+        cout << urrl << endl;
     }
-//    while(endDocs.next().url != "aaa")
-//        {
-//        for(auto locs : locations)
-//            {
-//            if(locs < endDocs.getCurrentDoc().docEndPosition &&
-//               locs >= (endDocs.getCurrentDoc().docEndPosition - endDocs.getCurrentDoc().docNumWords)) {
-//                urls.insert(endDocs.getCurrentDoc().url);
-//            }
-//        }
-//
-//        }
-//    for(auto urrl : urls) {
-//        cout << urrl << endl;
-//    }
     return 0;
 }
\ No newline at end of file