From 3c0eec038f83adbac0a1eadea49362fd9566093c Mon Sep 17 00:00:00 2001
From: Nicholas Yang <parablank@gmail.com>
Date: Tue, 20 Mar 2018 15:02:38 -0400
Subject: [PATCH] seek lookup table in
---
CMakeLists.txt | 2 +-
constraintSolver/ISRWord.cpp | 26 ++++++++++++++++++++++---
constraintSolver/ISRWord.h | 11 ++++-------
constraintSolver/WordSeek.h | 7 +++++++
constraintSolver/tests/ISRWordTests.cpp | 1 -
5 files changed, 35 insertions(+), 12 deletions(-)
create mode 100644 constraintSolver/WordSeek.h
diff --git a/CMakeLists.txt b/CMakeLists.txt
index d0cadfc..7a341bb 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -54,7 +54,7 @@ add_executable(URLTEST shared/url.h shared/urlTest.cpp)
add_executable(search-engine search.cpp query/Query.cpp)
-add_executable(ISRWord-tests util/util.cpp constraintSolver/ISRWord.cpp constraintSolver/tests/ISRWordTests.cpp )
+add_executable(ISRWord-tests constraintSolver/ISRWord.cpp constraintSolver/tests/ISRWordTests.cpp )
find_package(OpenSSL REQUIRED)
diff --git a/constraintSolver/ISRWord.cpp b/constraintSolver/ISRWord.cpp
index a349730..639b151 100644
--- a/constraintSolver/ISRWord.cpp
+++ b/constraintSolver/ISRWord.cpp
@@ -2,9 +2,14 @@
// Created by Jake Close on 3/13/18.
//
+#include <string>
#include "ISRWord.h"
-using namespace std;
+size_t FileSize(int f) {
+ struct stat fileInfo;
+ fstat( f, &fileInfo);
+ return fileInfo.st_size;
+}
ISRWord::ISRWord(char* word) : term(word) {
getChunks();
@@ -24,15 +29,28 @@ vector<size_t> ISRWord::getSeekContents(string fileName) {
string word = "";
bool midWord = false;
bool midFind = false;
+ WordSeek wordDictionaryEntry;
if(memMap != MAP_FAILED) {
for(char* map = memMap; map < memMap + fileSize; map++) {
if(midFind && isalpha(*map)) {
break;
}
switch(*map) {
- case '\t':
+ if(midFind) {
+ case '<':
+ wordDictionaryEntry = WordSeek();
+ break;
+ case '>':
+ wordDictionaryEntry.seekOffset = stoll(word);
+ wordSeekLookupTable.push_back(wordDictionaryEntry);
+ break;
+ case ',':
+ wordDictionaryEntry.realLocation = stoll(word);
+ break;
+ }
case '\n':
case '\r':
+ case '\t':
case ' ':
if (midFind && word != "") {
contents.push_back(stoll(word));
@@ -151,6 +169,8 @@ Location ISRWord::next() {
//go to next chunk
Location ISRWord::seek( Location target ) {
-
+ if(!wordSeekLookupTable.empty()) {
+
+ }
}
diff --git a/constraintSolver/ISRWord.h b/constraintSolver/ISRWord.h
index e0eb8c8..eface93 100644
--- a/constraintSolver/ISRWord.h
+++ b/constraintSolver/ISRWord.h
@@ -5,6 +5,7 @@
#pragma once
//#include "ISR.h"
+#include <iostream>
#include <vector>
#include <fcntl.h>
#include <stdio.h>
@@ -13,18 +14,13 @@
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
+#include "WordSeek.h"
//#include "../util/util.h"
-
-size_t FileSize(int f) {
- struct stat fileInfo;
- fstat( f, &fileInfo);
- return fileInfo.st_size;
-}
-
using namespace std;
+
//Find occurrences of individual words
typedef size_t Location;
@@ -50,6 +46,7 @@ public:
char* term;
char* masterIndex;
vector<size_t> listOfChunks;
+ vector<WordSeek> wordSeekLookupTable;
size_t currentChunk;
char* currentMemMap;
diff --git a/constraintSolver/WordSeek.h b/constraintSolver/WordSeek.h
new file mode 100644
index 0000000..e0f44b9
--- /dev/null
+++ b/constraintSolver/WordSeek.h
@@ -0,0 +1,7 @@
+#pragma once
+
+class WordSeek {
+public:
+ ssize_t seekOffset;
+ size_t realLocation;
+};
\ No newline at end of file
diff --git a/constraintSolver/tests/ISRWordTests.cpp b/constraintSolver/tests/ISRWordTests.cpp
index d34c5ea..e9501ac 100644
--- a/constraintSolver/tests/ISRWordTests.cpp
+++ b/constraintSolver/tests/ISRWordTests.cpp
@@ -4,7 +4,6 @@
#include <iostream>
#include "../ISRWord.h"
-#include "../ISRWord.cpp"
using namespace std;
--
GitLab