Mirrors/privacore-open-source-search-engine
1
0
Fork 1
mirror of https://github.com/privacore/open-source-search-engine.git synced 2025-04-18 14:48:48 -04:00
Code Issues Packages Projects Releases Wiki Activity

Made spider status ¤defines into strongly typed enum

Browse Source
This commit is contained in:
Ivan Skytte Jørgensen 2017-09-26 13:29:36 +02:00
parent 0e9cec3133
commit d07ce5c4df
9 changed files with 44 additions and 40 deletions

6
Collectiondb.cpp
View File

@ -706,7 +706,7 @@ bool Collectiondb::resetColl2(collnum_t oldCollnum, collnum_t newCollnum) {
cr->m_spiderColl = NULL;
}
cr->m_spiderStatus = SP_INITIALIZING; // this is 0
cr->m_spiderStatus = spider_status_t::SP_INITIALIZING; // this is 0
//cr->m_spiderStatusMsg = NULL;
// so XmlDoc.cpp can detect if the collection was reset since it
@ -768,7 +768,7 @@ bool Collectiondb::resetColl2(collnum_t oldCollnum, collnum_t newCollnum) {
g_linkdb.getRdb()->deleteColl ( oldCollnum , newCollnum );
// reset crawl status too!
cr->m_spiderStatus = SP_INITIALIZING;
cr->m_spiderStatus = spider_status_t::SP_INITIALIZING;
// . set m_recs[oldCollnum] to NULL and remove from hash table
// . do after calls to deleteColl() above so it wont crash
@ -958,7 +958,7 @@ CollectionRec::CollectionRec() {
m_overflow = 0x12345678;
m_overflow2 = 0x12345678;
// the spiders are currently uninhibited i guess
m_spiderStatus = SP_INITIALIZING; // this is 0
m_spiderStatus = spider_status_t::SP_INITIALIZING; // this is 0
// inits for sortbydatetable
m_msg5 = NULL;
// JAB - track which regex parsers have been initialized

3
Collectiondb.h
View File

@ -8,6 +8,7 @@
#include <atomic>
#include "SafeBuf.h"
#include "rdbid_t.h"
#include "spider_status_t.h"
#include "GbMutex.h"
@ -236,7 +237,7 @@ public:
int32_t m_maxQueryTerms;
char m_spiderStatus;
spider_status_t m_spiderStatus;
//ranking settings
float m_sameLangWeight;

4
PageBasic.cpp
View File

@ -841,7 +841,7 @@ bool sendPageBasicStatus ( TcpSocket *socket , HttpRequest *hr ) {
// show stats
//
const char *crawlMsg;
int32_t crawlStatus = -1;
spider_status_t crawlStatus;
getSpiderStatusMsg ( cr , &crawlMsg, &crawlStatus );
sb.safePrintf(
@ -856,7 +856,7 @@ bool sendPageBasicStatus ( TcpSocket *socket , HttpRequest *hr ) {
"<td><b>Crawl Status Msg:</td>"
"<td>%s</td>"
"</tr>"
, crawlStatus
, (int)crawlStatus
, crawlMsg);
// print link to embed the code in their own site

6
PageCrawlBot.cpp
View File

@ -35,7 +35,7 @@
bool printCrawlDetails2 (SafeBuf *sb , CollectionRec *cx , char format ) {
const char *crawlMsg;
int32_t crawlStatus = -1;
spider_status_t crawlStatus;
getSpiderStatusMsg ( cx , &crawlMsg, &crawlStatus );
if ( format == FORMAT_JSON ) {
@ -43,7 +43,7 @@ bool printCrawlDetails2 (SafeBuf *sb , CollectionRec *cx , char format ) {
"\"response\":{\n"
"\t\"statusCode\":%" PRId32",\n"
"\t\"statusMsg\":\"%s\",\n"
, crawlStatus, crawlMsg);
, (int)crawlStatus, crawlMsg);
sb->safePrintf("\t\"processStartTime\":%" PRId64",\n", (g_process.m_processStartTime / 1000));
sb->safePrintf("\t\"currentTime\":%" PRIu32"\n", (uint32_t)getTimeGlobal() );
sb->safePrintf("\t}\n");
@ -51,7 +51,7 @@ bool printCrawlDetails2 (SafeBuf *sb , CollectionRec *cx , char format ) {
}
if ( format == FORMAT_XML ) {
sb->safePrintf("<response>\n\t<statusCode>%" PRId32"</statusCode>\n", crawlStatus);
sb->safePrintf("<response>\n\t<statusCode>%" PRId32"</statusCode>\n", (int)crawlStatus);
sb->safePrintf("\t<statusMsg><![CDATA[%s]]></statusMsg>\n", crawlMsg);
sb->safePrintf("\t<currentTime>%" PRIu32"</currentTime>\n", (uint32_t)getTimeGlobal() );
sb->safePrintf("\t<currentTimeUTC>%" PRIu32"</currentTimeUTC>\n", (uint32_t)getTimeGlobal() );

8
PageSpider.cpp
View File

@ -224,10 +224,10 @@ static bool printList ( State11 *st ) {
static bool generatePageHTML(CollectionRec *cr, SafeBuf *sb, const SafeBuf *doledbbuf) {
// print reason why spiders are not active for this collection
int32_t tmp2;
spider_status_t tmp2;
const char *crawlMsg;
getSpiderStatusMsg ( cr , &crawlMsg, &tmp2 );
if ( crawlMsg && tmp2 != SP_INITIALIZING )
if ( crawlMsg && tmp2 != spider_status_t::SP_INITIALIZING )
sb->safePrintf("<table cellpadding=5 style=\"max-width:600px\" border=0>"
"<tr>"
"<td>"
@ -464,11 +464,11 @@ static bool generatePageHTML(CollectionRec *cr, SafeBuf *sb, const SafeBuf *dole
static bool generatePageJSON(CollectionRec *cr, SafeBuf *sb, const SafeBuf *doledbbuf) {
sb->safePrintf("{\n\"response\": {\n");
int32_t crawlStatus;
spider_status_t crawlStatus;
const char *crawlMsg;
getSpiderStatusMsg ( cr , &crawlMsg , &crawlStatus );
sb->safePrintf("\t\"statusCode\": %d,\n", crawlStatus);
sb->safePrintf("\t\"statusCode\": %d,\n", (int)crawlStatus);
sb->safePrintf("\t\"statusMsg\": \"%s\",\n", crawlMsg);
sb->safePrintf("\t\"spiderCount\": %d,\n", g_spiderLoop.getNumSpidersOut());

26
Spider.cpp
View File

@ -2785,34 +2785,34 @@ void dedupSpiderdbList ( RdbList *list ) {
void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, int32_t *status) {
void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, spider_status_t *status) {
if ( ! g_conf.m_spideringEnabled ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "Spidering disabled in master controls. You can turn it back on there.";
return;
}
if ( g_conf.m_readOnlyMode ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "In read-only mode. Spidering off.";
return;
}
if ( g_dailyMerge.m_mergeMode ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "Daily merge engaged, spidering paused.";
return;
}
if ( g_repairMode ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "In repair mode, spidering paused.";
return;
}
// do not spider until collections/parms in sync with host #0
if ( ! g_parms.inSyncWithHost0() ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "Parms not in sync with host #0, spidering paused";
return;
}
@ -2820,25 +2820,25 @@ void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, int32_t *stat
// don't spider if not all hosts are up, or they do not all
// have the same hosts.conf.
if ( g_hostdb.hostsConfInDisagreement() ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "Hosts.conf discrepancy, spidering paused.";
return;
}
if ( ! cx->m_spideringEnabled ) {
*status = SP_PAUSED;
*status = spider_status_t::SP_PAUSED;
*msg = "Spidering disabled in spider controls.";
return;
}
if ( cx->m_spiderStatus == SP_INITIALIZING ) {
*status = SP_INITIALIZING;
if ( cx->m_spiderStatus == spider_status_t::SP_INITIALIZING ) {
*status = spider_status_t::SP_INITIALIZING;
*msg = "Job is initializing.";
return;
}
if ( ! g_conf.m_spideringEnabled ) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "All crawling temporarily paused by root administrator for maintenance.";
return;
}
@ -2847,14 +2847,14 @@ void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, int32_t *stat
// host's counts tallied into it, which could make a difference on
// whether we have exceed a maxtocrawl limit or some such, so wait...
if (g_hostdb.hasDeadHost()) {
*status = SP_ADMIN_PAUSED;
*status = spider_status_t::SP_ADMIN_PAUSED;
*msg = "All crawling temporarily paused because a shard is down.";
return;
}
// otherwise in progress?
*status = SP_INPROGRESS;
*status = spider_status_t::SP_INPROGRESS;
*msg = "Spider is in progress.";
}

15
Spider.h
View File

@ -8,6 +8,7 @@
#include "Rdb.h"
#include "Titledb.h" //DOCID_MASK
#include "hash.h"
#include "spider_status_t.h"
class RdbList;
@ -35,20 +36,8 @@ class SafeBuf;
// we reduce this below if the spiderdb is smaller.
#define MAX_WINNER_NODES 2000
// . values for CollectionRec::m_spiderStatus
// . reasons why crawl is not happening
#define SP_INITIALIZING 0
//#define SP_UNUSED_1 1
//#define SP_UNUSED_2 2
//#define SP_UNUSED_3 3
//#define SP_UNUSED_4 4
//#define SP_UNUSED_5 5
#define SP_PAUSED 6 // user paused spider
#define SP_INPROGRESS 7 // it is going on!
#define SP_ADMIN_PAUSED 8 // g_conf.m_spideringEnabled = false
//#define SP_UNUSED_9 9
void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, int32_t *status);
void getSpiderStatusMsg(const CollectionRec *cx, const char **msg, spider_status_t *status);

2
SpiderLoop.cpp
View File

@ -907,7 +907,7 @@ skipDoledbRec:
}
// reset reason why crawl is not running, because we basically are now
cr->m_spiderStatus = SP_INPROGRESS; // this is 7
cr->m_spiderStatus = spider_status_t::SP_INPROGRESS;
// be sure to save state so we do not re-send emails
cr->setNeedsSave();

14
spider_status_t.h Normal file
View File

@ -0,0 +1,14 @@
#ifndef COLLECTION_SPIDER_STATUS_T_H_
#define COLLECTION_SPIDER_STATUS_T_H_
// . values for CollectionRec::m_spiderStatus
// . reasons why crawl is not happening
enum class spider_status_t : char {
SP_INITIALIZING = 0,
SP_PAUSED = 6, // user paused spider
SP_INPROGRESS = 7, // it is going on!
SP_ADMIN_PAUSED = 8, // g_conf.m_spideringEnabled = false
};
#endif