workerd: multithread demon, to return html(inframe) with offers. Runs on background and listen unix socket(config.xml) ready to http requests from nginx.

Input
http variables:
	QUERY_STRING
	REMOTE_ADDR
	SCRIPT_NAME
	HTTP_COOKIE

and url variables:
	show(status) – show status
	ip(REMOTE_ADDR) - ip
	cookie_id(HTTP_COOKIE) – cookie id
	scr – informer id
	country – country for tests
	region – region for tests
	show(json) – html or json returns
	location – return to http response
	w - width
	h - height
	search – user search keywords
	context – site context keywords

Outputs:
html with offers selected by rating from:
cat db_dump/requests/01.sql

SELECT ofrs.id,
ofrs.guid,
ofrs.title,
ofrs.price,
ofrs.description,
ofrs.url,
ofrs.image,
ofrs.swf,
ofrs.campaignId,
ofrs.isOnClick,
ofrs.type,
CASE WHEN iret.rating IS NOT NULL
THEN iret.rating
ELSE ofrs.rating
END AS rating,
ofrs.retargeting,
ofrs.uniqueHits,
ofrs.height,
ofrs.width,
ca.social
FROM Offer AS ofrs
INNER JOIN Campaign AS ca ON ca.valid=1 AND ca.retargeting=0 AND ofrs.campaignId=ca.id
INNER JOIN (
		SELECT cn.id FROM CampaignNow AS cn
		%s
        INNER JOIN Campaign2Categories AS c2c ON cn.id=c2c.id_cam
        INNER JOIN Categories2Domain AS ct2d ON c2c.id_cat=ct2d.id_cat
                AND ct2d.id_dom=%lld
		EXCEPT
            SELECT c2d.id_cam AS id
            FROM Campaign2Domains AS c2d
            WHERE c2d.id_dom=%lld AND c2d.allowed=0
		EXCEPT
            SELECT c2a.id_cam AS id
            FROM Campaign2Accounts AS c2a
            WHERE c2a.id_acc=%lld AND c2a.allowed=0
        EXCEPT
            SELECT c2i.id_cam AS id
            FROM Campaign2Informer AS c2i
            WHERE c2i.id_inf=%lld AND c2i.allowed=0
) AS c ON ca.id=c.id
LEFT JOIN tmp%d%lld AS deph ON ofrs.id=deph.id
LEFT JOIN Informer2OfferRating AS iret ON iret.id_inf=%lld AND ofrs.id=iret.id_ofr
WHERE ofrs.valid=1
    AND deph.id IS NULL
;
where:
tmp%d%lld – user view offers history
CampaignNow - campaign allowed now by time
%s -
IF:

SELECT geo.id_cam FROM geoTargeting AS geo \
                INNER JOIN GeoLiteCity AS reg ON geo.id_geo = reg.locId AND reg.city='city by geo ip';"

RETURNS COUNT > 0

THEN:

INNER JOIN geoTargeting AS geo ON geo.id_cam=cn.id \
        INNER JOIN GeoLiteCity AS reg ON geo.id_geo = reg.locId AND reg.city='data from geo ip or url request'

ELSE:

INNER JOIN geoTargeting AS geo ON geo.id_cam=cn.id \
            INNER JOIN GeoLiteCity AS reg ON geo.id_geo = reg.locId AND(reg.country='UA' AND reg.city='')
ELSE:
EMPTY STRING

Core algorithm

each thread wait input, so it cann't get cpu resources, after parsing intput variables(class Param) and get geo data:
1) load informer:
cat db_dump/requests/02.sql
SELECT 	inf.id,
	inf.capacity,
	inf.bannersCss,
	inf.teasersCss,
	inf.domainId,
	inf.accountId
FROM Informer AS inf
WHERE inf.guid=@informerUid
LIMIT 1;
    <short_term>0</short_term>
    <long_term>0</long_term>
    <context>0</context>
    <search>1.2</search>
</range>
 if 0 – does not use in sphinx search and the don't make history request.
For fine tune sphinx(Context search by offers)

3) get all offers from request(Outputs)

4) wait until all histories were loaded and process rating correction in sphinx(class XXXSearcher) history with zero rating does not use. The daemon can correct rating by context search(sphinx) on selected offers(“Input”) by formula:

                float weight = sphinx_get_weight (res, i ) / 1000;

                pOffer->rating = pOffer->rating
                    + rate_from_config  * maxRating
                    + weight;

rate_from_config – config section:   <range></ range>
maxRating – max rating in offers
weight – sphinx  weight / 1000

The sphinx uses config.xml sphinx.fields describes:

 <sphinx>
	<field_name>field_rating</field_name>
</ sphinx>

field_name MUST be the same as in sphinx schema: /etc/sphinx/sphinx_driver.py
       <sphinx:schema>
        <sphinx:field name="title"/>
        <sphinx:field name="description"/>
        <sphinx:field name="keywords"/>
        <sphinx:field name="exactly_phrases"/>
        <sphinx:field name="phrases"/>
        <sphinx:field name="minuswords"/>
        <sphinx:attr name="stitle" type="string"/>
        <sphinx:attr name="sdescription" type="string"/>
        <sphinx:attr name="skeywords" type="string"/>
        <sphinx:attr name="sexactly_phrases" type="string"/>
        <sphinx:attr name="sphrases" type="string"/>
        <sphinx:attr name="guid" type="string"/>
        <sphinx:attr name="match" type="string"/>
        <sphinx:attr name="fguid" type="int"/>
        </sphinx:schema>

in section <index>worker-full</index> set sphinx index name: /etc/sphinx/sphinx_driver.py
mode == "worker-full"

The  mach_mode,  rank_mode, sort_mode see “Context search by offers”.

5) process RIS algorithm(main)
	input:
		const Offer::Map &items,
		unsigned outLen(capacity)
	output:
		Offer::Vector &RISResult


	b) teaser rating and banner check: if fount first banner(max rating) return
	c) turn on clean history flag if all offers are social
	d) if teaser vector size less then capacity
	e) remove social
	f) add teaser with company unique and rating > 0
	h) add teaser with company unique and with any rating
SYSLOG-NG
add that in /etc/syslog-ng/syslog-ng.conf:

template t_isots
{
    template("$R_ISODATE $MSG\n");
    template_escape(no);
};

destination getmyad { file("/var/log/workerd.log" template(t_isots)); };
filter f_getmyad { program("getmyad"); };
log { source(src); filter(f_getmyad); destination(getmyad); };

NGINX
the sample nginx configuration:

        upstream fastcgi_backend {
                server unix:/tmp/workerd.sock;
                server unix:/tmp/workerd1.sock;
                server unix:/tmp/workerd2.sock;
        }

        server {
                listen 0.0.0.0;
                server_name yottos.com;
                access_log off;#/var/log/nginx/localhost.access_log main;
                error_log /var/log/nginx/localhost.error_log info;
                root /var/www/localhost/htdocs;
                keepalive_timeout 1ms;

                location /adshow.fcgi {
                        send_timeout 100ms;
                        fastcgi_cache off;
                        fastcgi_pass fastcgi_backend;
                        fastcgi_buffer_size 24k;
                        fastcgi_buffers 2 24k;
                        fastcgi_busy_buffers_size 24k;
                        fastcgi_keep_conn on;
                        include fastcgi_params;
                }
        }

DATABASES
At start(if not configured with --enable-debug) forks in background and get data from mongo database described in config.xml section:
<mongo>
    <main>
        <!--ip:port-->
        <host>213.186.121.76:27018</host>
        <!--the database name-->
        <db>getmyad_db</db>
        <!--replica set name-->
        <set></set>
       <!-- enable slave-->
        <slave>true</slave>
    </main>
</mongo>

The database will be loaded into sqlite database file(may be file on tmpfs for debug or control proposes, check write access from user):
<dbpath>/run/mem.db</dbpath>
 or memory:
<dbpath>:memory:</dbpath>
The database dump
db_dump/requests/02.sql
db_dump/requests/01.sql
db_dump/requests/04.sql
db_dump/requests/03.sql
db_dump/UA_RU
db_dump/tables/11_GeoCountries.sql
db_dump/tables/00_Campaign.sql
db_dump/tables/17_Retargeting.sql
db_dump/tables/12_GeoRerions.sql
db_dump/tables/20_GeoLiteCity.sql
db_dump/tables/06_Accounts.sql
db_dump/tables/02_Offer.sql
db_dump/tables/07_Campaign2Accounts.sql
db_dump/tables/15_CampaignNow.sql
db_dump/tables/16_Informer.sql
db_dump/tables/18_Informer2OfferRating.sql
db_dump/tables/10_CronCampaign.sql
db_dump/tables/05_Campaign2Informer.sql
db_dump/tables/14_regionTargeting.sql
db_dump/tables/08_Domains.sql
db_dump/tables/19_Campaign2Categories.sql
db_dump/tables/04_Categories2Domain.sql
db_dump/tables/13_geoTargeting.sql
db_dump/tables/03_Categories.sql
db_dump/tables/09_Campaign2Domains.sql
db_dump/tables/04_Categories2Informer.sql
db_dump/view/00_Campaign2GeoRerions.sql
db_dump/view/03_Campaign2Acnts.sql
db_dump/view/02_Campaign2Doms.sql
db_dump/view/04_Campaign2Infs.sql
db_dump/post/01.sql
db_dump/schema/CampaignNow.sql

loaded from
<db_dump_path>/etc/worker/db_dump</db_dump_path>
all files from sub directories sort by name and load into database.

THREADS
Threads number: server/children

SPHINX
To match search
<sphinx>
 <!--host>213.186.121.76</host-->
 <host>127.0.0.1</host>
 <port>9312</port>
 <index>worker</index>
 <!--all from /usr/include/sphinxclient.h-->
 <mach_mode>SPH_MATCH_EXTENDED2</mach_mode>
 <rank_mode>SPH_RANK_SPH04</rank_mode>
 <sort_mode>SPH_SORT_RELEVANCE</sort_mode>
 <fields>
 <title>80</title>
 <description>30</description>
 <!--keywords>70</keywords-->
 <!--exactly_phrases>100</exactly_phrases-->
 <!--phrases>90</phrases-->
 <!--minuswords>100</minuswords-->
 </fields>
 <min_offres_process>10</min_offres_process><!-- 20 default value -->
</sphinx>

The programm can make sphinx requests by:
    user search,
    page context,
    short term history,
    long term history;

The requests were shown in log file when in config.xml set root->log->sphinx turned on 1.

Jun 10 13:13:07 srv-2 workerd-dev[22003]: shpinx: request for: context : (@title bind | @title linux | @title windows | @title apache | @title nginx | @title mysql | @t
itle сервер | @title установка | @title настройка | @title mongodb | @title mssql | @title iptables | @title hyper)|(@description bind | @description linux | @descripti
on windows | @description apache | @description nginx | @description mysql | @description сервер | @description установка | @description настройка | @description mongod
b | @description mssql | @description iptables | @description hyper)|(@keywords bind | @keywords linux | @keywords windows | @keywords apache | @keywords nginx | @keywo
rds mysql | @keywords сервер | @keywords установка | @keywords настройка | @keywords mongodb | @keywords mssql | @keywords iptables | @keywords hyper)|(@exactly_phrases
 bind | @exactly_phrases linux | @exactly_phrases windows | @exactly_phrases apache | @exactly_phrases nginx | @exactly_phrases mysql | @exactly_phrases сервер | @exact
ly_phrases установка | @exactly_phrases настройка | @exactly_phrases mongodb | @exactly_phrases mssql | @exactly_phrases iptables | @exactly_phrases hyper)|(@phrases bi
nd | @phrases linux | @phrases windows | @phrases apache | @phrases nginx | @phrases mysql | @phrases сервер | @phrases установка | @phrases настройка | @phrases mongod
b | @phrases mssql | @phrases iptables | @phrases hyper)|(@minuswords bind | @minuswords linux | @minuswords windows | @minuswords apache | @minuswords nginx | @minuswo
rds mysql | @minuswords сервер | @minuswords установка | @minuswords настройка | @minuswords mongodb | @minuswords mssql | @minuswords iptables | @minuswords hyper)

Jun 10 13:13:07 srv-2 workerd-dev[22003]: shpinx: request for: short term history : ((@exactly_phrases "<<мано <<керам <<сковор <<плюс <<придбати <<cковор <<нтернет <<м
агазин <<shop <<замовити "~1) | (@title "мано керам сковор плюс придбати cковор нтернет магазин shop замовити"/3)| (@description "мано керам сковор плюс придбати cковор
 нтернет магазин shop замовити"/3) | (@keywords  мано | керам | сковор | плюс | придбати | cковор | нтернет | магазин | shop | замовити ) | (@phrases "мано керам сковор
 плюс придбати cковор нтернет магазин shop замовити"~5))

When offer match the request it rating recalculated:

Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx:  matches:#14 doc_id=21965, weight=280500 by:  guidint=1695190717  rating=721.218  match=broadmatch sphrases=  skeywords=интернет магазин топшоп sexactly_phrases=
Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx:  matches:#8 doc_id=22903, weight=420500 by:  guidint=3157471387  rating=409.096  match=broadmatch sphrases=  skeywords=магазин topshop sexactly_phrases=
Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx:  matches:#25 doc_id=22968, weight=280500 by:  guidint=1981191986  rating=1676.64  match=broadmatch sphrases=  skeywords=интернет магазин шланги для полива sexactly_phrases=
~
Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx: offer id: 1695190717 old rating: 625.812 new: 22045.8 weight: 420 match: broadmatch
Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx: offer id: 3157471387 old rating: 610.511 new: 22030.5 weight: 420 match: broadmatch
Jun 10 13:13:07 srv-2 workerd-dev[22003]: sphinx: offer id: 1981191986 old rating: 2787.85 new: 24207.8 weight: 420 match: broadmatch

by formula:
new offer rating =
    old offer rating +
    request rate[user search, page context,short term history,long term history] * max offers rating
    + weight;

ALL
Please read and edit config.xml to fine tune.

The configuration options by groups(mongo/main/db means xpath:
<mongo><main><db/></main></mongo>):

Main configuration:
 server_ip: ip uses to redirect script
 redirect_script: the redirect script name
 geocity_path: geo city database file path
 shortterm_expire: time to live keys in shortterm redis database in seconds
	views_expire: time to live keys in  user view redis database in seconds
	context_expire: time to live keys in site contex  redis database in seconds

The  re-targeting
	retargeting_by_persents: how match max re-targeting offers can be in result
	retargeting_by_time: re-targeting user view history time to live
	retargeting_unique_by_campaign: re-targeting unique by campaign

The range coefficients for shpinx, 0- off, 1-max:
	range/short_term: short term hostory
	range/ long_term: long term history
	range/context: page context
	range/search: user search keywords

The source database
	 {$db} in main,log
	main – the main database(read only),
	log – log store

	mongo/{$db}/host: host:port
	mongo/{$db}/db: the database name
	mongo/{$db}/set: Specifies the name of the replica set, if the mongod is a member of a replica set
       	mongo/{$db}/slave: to use slaves data

Redis servers
	 {$db} in  short_term_history, long_term_history, user_view_history
	short_term_history – short term history,
	long_term_history – long term history
	user_view_history  – user view history

	redis/{$db}/host: host
	redis/{$db}/port: port

Context search by offers:
Sphinx server:
    sphinx/host: host(ip address)
    sphinx/port: port
    sphinx/index: index name
    sphinx/mach_mode:
	SPH_MATCH_ALL Match all query words (default mode).
	SPH_MATCH_ANY Match any of query words.
	SPH_MATCH_PHRASE Match query as a phrase, requiring perfect match.
	SPH_MATCH_BOOLEAN Match query as a boolean expression.
	SPH_MATCH_EXTENDED Match query as an expression in Sphinx internal query language.
	SPH_MATCH_FULLSCAN Enables fullscan.
	SPH_MATCH_EXTENDED2 The same as SPH_MATCH_EXTENDED plus ranking and quorum searching support.

 sphinx/rank_mode:
	SPH_RANK_PROXIMITY_BM25 = sum(lcs*user_weight)*1000+bm25
	SPH_RANK_BM25 = bm25
	SPH_RANK_NONE = 1
	SPH_RANK_WORDCOUNT = sum(hit_count*user_weight)
	SPH_RANK_PROXIMITY = sum(lcs*user_weight)
	SPH_RANK_MATCHANY = sum((word_count+(lcs-1)*max_lcs)*user_weight)
	SPH_RANK_FIELDMASK = field_mask
	SPH_RANK_SPH04 = sum((4*lcs+2*(min_hit_pos==1)+exact_hit)*user_weight)*1000+bm25

 sphinx/sort_mode:
	SPH_SORT_RELEVANCE Sort by relevance in descending order (best matches first).
	SPH_SORT_ATTR_DESC Sort by an attribute in descending order (bigger attribute values first).
	SPH_SORT_ATTR_ASC Sort by an attribute in ascending order (smaller attribute values first).
	SPH_SORT_TIME_SEGMENTS Sort by time segments (last hour/day/week/month) in descending order, and then by relevance in descending order.
	SPH_SORT_EXTENDED Sort by SQL-like combination of columns in ASC/DESC order.
	SPH_SORT_EXPR Sort by an arithmetic expression.

 sphinx/fields{title,description,keywords,exactly_phrases,phrases,minuswords}: search fields with weight

Server configuration:
	server/socket_path: unix socket path fast cgi, the same path must in nginx.conf:
	        upstream fastcgi_backend {
              	  	server unix:/tmp/workerd.sock;
                	server unix:/tmp/workerd1.sock;
                	server unix:/tmp/workerd2.sock;
       		}

 server/children: threads number
 server/mq_path: message queue server user and password
 server/dbpath sqlite database file path
 server/db_dump_path sqlite database dump(without data, only schema)
 server/db_geo_csv geo ip database file path
 server/template_teaser teaser template file path:
 %1%	    CSS
 %2%	    JSON
 %3%	    offers count
 %4%	    next link
 server/template_banner the banner template path
 server/swfobject swfobject.js file
 server/geoGity geo city file path
 server/cookie_name coockie name
 server/cookie_domain: coockie domain name
 server/cookie_path coockie path
 server/lock_file lock file path
 server/pid_file pid file path
 server/user: run as user
 server/time_update: campaign by time update interval

 install:
 #aclocal -I m4 && autoconf && automake
 #./configure
 #make

TESTING
start:
nginx, redis[1-6], searchd, rabbitmq, syslog-ng, mongodb

compile debug version:
make clean
./configure ./configure --with-debug
make
./workerd

in default browser:
http://yottos.com/adshow.fcgi?scr=00122a9a-9f60-11e1-a3a7-00163e0301c1&location=http%3A%2F%2Fhelpers.com.ua%2F&w=1855&h=1110&search=%D0%B5%D0%B4%D0%B0%20%D0%BA%D0%B0%D0%B7%D0%B8%D0%BD%D0%BE&context=bind%20linux%20windows%20apache%20nginx%20mysql%20%D1%81%D0%B5%D1%80%D0%B2%D0%B5%D1%80%20%D1%83%D1%81%D1%82%D0%B0%D0%BD%D0%BE%D0%B2%D0%BA%D0%B0%20%D0%BD%D0%B0%D1%81%D1%82%D1%80%D0%BE%D0%B9%D0%BA%D0%B0%20mongodb%20mssql%20iptables%20hyper&rand=105921

where
src – informer guid
alex@hp ~/Projects/worker $ echo "select guid from informer limit 1;"|sqlite3 /run/alex/mem.db
00122a9a-9f60-11e1-a3a7-00163e0301c1
search – words by space(%20)(rating.search in the config.xml must be more then 0)

output:
alex@hp ~/Projects/worker $ ./workerd
Wed Mar 12 16:50:56.338 starting new replica set monitor for replica set vsrv with seed of 213.186.121.76:27018
Wed Mar 12 16:50:56.364 successfully connected to seed 213.186.121.76:27018 for replica set vsrv
Wed Mar 12 16:50:56.390 changing hosts to { 0: "213.186.121.76:27018", 1: "212.113.32.147:27018", 2: "212.113.32.146:27018", 3: "212.113.32.167:27018", 4: "212.113.33.66:27018", 5: "213.186.121.199:27018" } from vsrv/
Wed Mar 12 16:50:56.390 trying to add new host 212.113.32.146:27018 to replica set vsrv
Wed Mar 12 16:50:56.401 successfully connected to new host 212.113.32.146:27018 in replica set vsrv
Wed Mar 12 16:50:56.401 trying to add new host 212.113.32.147:27018 to replica set vsrv
Wed Mar 12 16:50:56.414 successfully connected to new host 212.113.32.147:27018 in replica set vsrv
Wed Mar 12 16:50:56.414 trying to add new host 212.113.32.167:27018 to replica set vsrv
Wed Mar 12 16:50:56.425 successfully connected to new host 212.113.32.167:27018 in replica set vsrv
Wed Mar 12 16:50:56.425 trying to add new host 212.113.33.66:27018 to replica set vsrv
Wed Mar 12 16:50:56.444 successfully connected to new host 212.113.33.66:27018 in replica set vsrv
Wed Mar 12 16:50:56.444 trying to add new host 213.186.121.199:27018 to replica set vsrv
Wed Mar 12 16:50:56.461 successfully connected to new host 213.186.121.199:27018 in replica set vsrv
Wed Mar 12 16:50:56.461 trying to add new host 213.186.121.76:27018 to replica set vsrv
Wed Mar 12 16:50:56.488 successfully connected to new host 213.186.121.76:27018 in replica set vsrv
Wed Mar 12 16:50:56.648 Primary for replica set vsrv changed to 213.186.121.199:27018
Wed Mar 12 16:50:56.893 replica set monitor for replica set vsrv started, address is vsrv/212.113.32.146:27018,212.113.32.147:27018,212.113.32.167:27018,212.113.33.66:27018,213.186.121.199:27018,213.186.121.76:27018
Wed Mar 12 16:50:56.893 [ReplicaSetMonitorWatcher] starting
SELECT ofrs.id,
ofrs.guid,
ofrs.title,
ofrs.price,
ofrs.description,
ofrs.url,
ofrs.image,
ofrs.swf,
ofrs.campaignId,
ofrs.isOnClick,
ofrs.type,
CASE WHEN iret.rating IS NOT NULL
THEN iret.rating
ELSE ofrs.rating
END AS rating,
ofrs.retargeting,
ofrs.uniqueHits,
ofrs.height,
ofrs.width,
ca.social
FROM Offer AS ofrs
INNER JOIN Campaign AS ca ON ca.valid=1 AND ca.retargeting=0 AND ofrs.campaignId=ca.id
INNER JOIN (
		SELECT cn.id FROM CampaignNow AS cn

        INNER JOIN Campaign2Categories AS c2c ON cn.id=c2c.id_cam
        INNER JOIN Categories2Domain AS ct2d ON c2c.id_cat=ct2d.id_cat
                AND ct2d.id_dom=324
		EXCEPT
            SELECT c2d.id_cam AS id
            FROM Campaign2Domains AS c2d
            WHERE c2d.id_dom=324 AND c2d.allowed=0
		EXCEPT
            SELECT c2a.id_cam AS id
            FROM Campaign2Accounts AS c2a
            WHERE c2a.id_acc=324 AND c2a.allowed=0
        EXCEPT
            SELECT c2i.id_cam AS id
            FROM Campaign2Informer AS c2i
            WHERE c2i.id_inf=3399645854 AND c2i.allowed=0
) AS c ON ca.id=c.id
LEFT JOIN tmp19507140160618098432 AS deph ON ofrs.id=deph.id
LEFT JOIN Informer2OfferRating AS iret ON iret.id_inf=3399645854 AND ofrs.id=iret.id_ofr
WHERE ofrs.valid=1
    AND deph.id IS NULL
;

SELECT ofrs.id,
ofrs.guid,
ofrs.title,
ofrs.price,
ofrs.description,
ofrs.url,
ofrs.image,
ofrs.swf,
ofrs.campaignId,
ofrs.isOnClick,
ofrs.type,
CASE WHEN iret.rating IS NOT NULL
THEN iret.rating
ELSE ofrs.rating
END AS rating,
ofrs.retargeting,
ofrs.uniqueHits,
ofrs.height,
ofrs.width,
ca.social
FROM Offer AS ofrs
INNER JOIN Campaign AS ca ON ca.valid=1 AND ca.retargeting=0 AND ofrs.campaignId=ca.id
INNER JOIN (
		SELECT cn.id FROM CampaignNow AS cn

        INNER JOIN Campaign2Categories AS c2c ON cn.id=c2c.id_cam
        INNER JOIN Categories2Domain AS ct2d ON c2c.id_cat=ct2d.id_cat
                AND ct2d.id_dom=324
		EXCEPT
            SELECT c2d.id_cam AS id
            FROM Campaign2Domains AS c2d
            WHERE c2d.id_dom=324 AND c2d.allowed=0
		EXCEPT
            SELECT c2a.id_cam AS id
            FROM Campaign2Accounts AS c2a
            WHERE c2a.id_acc=324 AND c2a.allowed=0
        EXCEPT
            SELECT c2i.id_cam AS id
            FROM Campaign2Informer AS c2i
            WHERE c2i.id_inf=3399645854 AND c2i.allowed=0
) AS c ON ca.id=c.id
LEFT JOIN tmp19507140160618098432 AS deph ON ofrs.id=deph.id
LEFT JOIN Informer2OfferRating AS iret ON iret.id_inf=3399645854 AND ofrs.id=iret.id_ofr
WHERE ofrs.valid=1
    AND deph.id IS NULL
;

query0: (@title еда | @title казино)|(@description еда | @description казино)
retrieved 2 of 2 matches
Query stats:
	'=еда' found 2 times in 2 documents
	'еда*' found 2 times in 2 documents
	'ед' found 4 times in 4 documents
	'=казино' found 9 times in 7 documents
	'казино*' found 9 times in 7 documents
	'казин' found 9 times in 7 documents

Matches:
1. doc_id=315, weight=120695, guidint=1379263440, rating=3993.646484, match=nomatch, sphrases=, skeywords=, sexactly_phrases=
2. doc_id=268, weight=120660, guidint=175701722, rating=0.000000, match=nomatch, sphrases=, skeywords=, sexactly_phrases=



good luck:)