kiwix/kiwix-tools
1
0
Fork 0
mirror of https://github.com/kiwix/kiwix-tools.git synced 2025-09-22 19:38:53 -04:00
Code Issues Packages Projects Releases Wiki Activity

+ introduction of the search engine

Browse Source
This commit is contained in:
kelson42 2010-01-23 14:11:52 +00:00
parent 9ac478aca7
commit 474c4e9a7a
1 changed files with 84 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

119
src/server/kiwix-serve.cpp
View File

@ -16,6 +16,7 @@
#include <pthread.h> #include <pthread.h>
#include <regex.h> #include <regex.h>
#include <kiwix/reader.h> #include <kiwix/reader.h>
#include <kiwix/searcher.h>
using namespace std; using namespace std;
@ -33,6 +34,8 @@ static const string HTMLScripts = " \
margin-right: 5px; \n \ margin-right: 5px; \n \
padding: 5px; \n \ padding: 5px; \n \
font-weight: bold; \n \ font-weight: bold; \n \
font-size: 14px; \n \
height: min; \n \
background: #FFFFFF; \n \ background: #FFFFFF; \n \
visibility: hidden; \n \ visibility: hidden; \n \
z-index: 100; \n \ z-index: 100; \n \
@ -99,13 +102,14 @@ else if (document.getElementById) \n \
"; ";
static const string HTMLDiv = " \ static const string HTMLDiv = " \
<div id=\"topbar\"> \n \ <div id=\"topbar\">Search<form method=\"GET\" action=\"/search\"><input type=\"textbox\" name=\"pattern\" /><input type=\"submit\" value=\"Go\" /></form></div> \n \
Search <input type=\"textbox\" />\n \
</div> \n \
"; ";
static kiwix::Reader* reader; static kiwix::Reader* reader;
static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER; static kiwix::Searcher* searcher;
static pthread_mutex_t readerLock = PTHREAD_MUTEX_INITIALIZER;
static pthread_mutex_t searcherLock = PTHREAD_MUTEX_INITIALIZER;
static bool hasSearchIndex = false;
static void appendToFirstOccurence(string &content, const string regex, const string &replacement) { static void appendToFirstOccurence(string &content, const string regex, const string &replacement) {
regmatch_t matchs[1]; regmatch_t matchs[1];
@ -140,34 +144,62 @@ static int accessHandlerCallback(void *cls,
return MHD_YES; return MHD_YES;
} }
/* Prepare the variable */ /* Prepare the variables */
struct MHD_Response *response; struct MHD_Response *response;
string content = ""; string content = "";
string mimeType = ""; string mimeType = "";
unsigned int contentLength = 0; unsigned int contentLength = 0;
/* Mutex lock */ if (!strcmp(url, "/search")) {
pthread_mutex_lock(&lock); const char* pattern = MHD_lookup_connection_value(connection, MHD_GET_ARGUMENT_KIND, "pattern");
std::string urlStr;
std::string titleStr;
unsigned int scoreInt;
char scoreStr[4];
/* Mutex lock */
pthread_mutex_lock(&searcherLock);
searcher->search(pattern, 30);
content = "<html><head><title>Kiwix search results</title></head><body><h1>Results</h1><hr/><ol>\n";
while (searcher->getNextResult(urlStr, titleStr, scoreInt)) {
sprintf(scoreStr, "%d", scoreInt);
content += "<li><a href=\"" + urlStr + "\">" + titleStr+ "</a> (" + scoreStr + "%)</li>\n";
/* Load the article from the ZIM file */
cout << "Loading '" << url << "'... " << endl;
try {
reader->getContent(url, content, contentLength, mimeType);
cout << "content size: " << contentLength << endl;
cout << "mimeType: " << mimeType << endl;
/* Rewrite the content */
if (mimeType == "text/html") {
appendToFirstOccurence(content, "<head>", HTMLScripts);
appendToFirstOccurence(content, "<body[^>]*>", HTMLDiv);
contentLength = content.size();
} }
} catch (const std::exception& e) { content += "</ol></body></html>\n";
std::cerr << e.what() << std::endl;
mimeType ="text/html";
contentLength = content.size();
/* Mutex unlock */
pthread_mutex_unlock(&searcherLock);
} else {
/* Mutex Lock */
pthread_mutex_lock(&readerLock);
/* Load the article from the ZIM file */
cout << "Loading '" << url << "'... " << endl;
try {
reader->getContent(url, content, contentLength, mimeType);
cout << "content size: " << contentLength << endl;
cout << "mimeType: " << mimeType << endl;
} catch (const std::exception& e) {
std::cerr << e.what() << std::endl;
}
/* Mutex unlock */
pthread_mutex_unlock(&readerLock);
} }
/* Mutex unlock */ /* Rewrite the content (add the search box) */
pthread_mutex_unlock(&lock); if (mimeType == "text/html") {
appendToFirstOccurence(content, "<head>", HTMLScripts);
appendToFirstOccurence(content, "<body[^>]*>", HTMLDiv);
contentLength = content.size();
}
/* clear context pointer */ /* clear context pointer */
*ptr = NULL; *ptr = NULL;
@ -201,8 +233,35 @@ int main(int argc, char **argv) {
string zimPath = (argv[1]); string zimPath = (argv[1]);
int port = atoi(argv[2]); int port = atoi(argv[2]);
string indexPath = (argc>3 ? argv[3] : "");
void *page; void *page;
/* Instanciate the ZIM file handler */
try {
reader = new kiwix::Reader(zimPath);
} catch (...) {
cout << "Unable to open the ZIM file '" << zimPath << "'." << endl;
exit(1);
}
/* Instanciate the ZIM index (if necessary) */
if (indexPath != "") {
try {
searcher = new kiwix::Searcher(indexPath);
hasSearchIndex = true;
} catch (...) {
cout << "Unable to open the search index '" << zimPath << "'." << endl;
exit(1);
}
} else {
hasSearchIndex = false;
}
/* Mutex init */
pthread_mutex_init(&readerLock, NULL);
pthread_mutex_init(&searcherLock, NULL);
/* Start the HTTP daemon */ /* Start the HTTP daemon */
daemon = MHD_start_daemon(MHD_USE_THREAD_PER_CONNECTION, daemon = MHD_start_daemon(MHD_USE_THREAD_PER_CONNECTION,
port, port,
@ -217,17 +276,6 @@ int main(int argc, char **argv) {
exit(1); exit(1);
} }
/* Instanciate the ZIM file handler */
try {
reader = new kiwix::Reader(zimPath);
} catch (...) {
cout << "Unable to open the ZIM file '" << zimPath << "'." << endl;
exit(1);
}
/* Mutex init */
pthread_mutex_init(&lock, NULL);
/* Run endless */ /* Run endless */
while (42) sleep(1); while (42) sleep(1);
@ -235,7 +283,8 @@ int main(int argc, char **argv) {
MHD_stop_daemon(daemon); MHD_stop_daemon(daemon);
/* Mutex destroy */ /* Mutex destroy */
pthread_mutex_destroy(&lock); pthread_mutex_destroy(&readerLock);
pthread_mutex_destroy(&searcherLock);
exit(0); exit(0);
} }