2 HCE project, Python bindings, Distributed Crawler application. 3 Event objects functional tests. 6 @author bgv bgv.hce@gmail.com 7 @link: http://hierarchical-cluster-engine.com/ 8 @copyright: Copyright © 2013-2014 IOIX Ukraine 9 @license: http://hierarchical-cluster-engine.com/license/ 17 from datetime
import datetime
23 if __name__ ==
"__main__":
25 TEST_TITLE_OBJECT =
" object:\n" 32 print TEST_TITLE + site.__class__.__name__ + TEST_TITLE_OBJECT, vars(site)
37 dc.EventObjects.SiteUpdate.UPDATE_TYPE_OVERWRITE)
41 su.cDate = datetime.now()
44 su.description =
"Test update" 45 su.urls = [
"http://localhost/"]
47 su.properties = {
"PROCESS_CTYPES":
"text/plain",
"STORE_HTTP_REQUEST":
"1",
"STORE_HTTP_HEADERS":
"1"}
48 su.state = dc.EventObjects.Site.STATE_DISABLED
51 su.maxResources = 2000
53 su.maxResourceSize = 20000000
54 su.requestDelay = 200000000
55 su.httpTimeout = 2000000000
56 su.errorMask = 20000000000
57 su.errors = 200000000000
59 print TEST_TITLE + su.__class__.__name__ + TEST_TITLE_OBJECT, vars(su)
64 dc.EventObjects.SiteUpdate.UPDATE_TYPE_APPEND)
67 su.cDate =
"2012-12-12 12:12:12" 70 su.description =
"Test update" 71 su.urls = [
"http://localhost/"]
73 su.properties = {
"PROCESS_CTYPES":
"text/plain",
"STORE_HTTP_REQUEST":
"1",
"STORE_HTTP_HEADERS":
"1"}
74 su.state = dc.EventObjects.Site.STATE_DISABLED
77 su.maxResources = 2000
79 su.maxResourceSize = 20000000
80 su.requestDelay = 200000000
81 su.httpTimeout = 2000000000
82 su.errorMask = 20000000000
83 su.errors = 200000000000
85 print TEST_TITLE + su.__class__.__name__ + TEST_TITLE_OBJECT, vars(su)
91 print TEST_TITLE + siteStatus.__class__.__name__ + TEST_TITLE_OBJECT, vars(siteStatus)
92 print siteStatus.toJSON()
96 print TEST_TITLE + sd.__class__.__name__ + TEST_TITLE_OBJECT, vars(sd)
101 print TEST_TITLE + sc.__class__.__name__ + TEST_TITLE_OBJECT, vars(sc)
106 print TEST_TITLE + sf.__class__.__name__ + TEST_TITLE_OBJECT, vars(sf)
110 url =
dc.EventObjects.URL(
"b85ab149a528bd0a024fa0f43e80b5fc",
"http://127.0.0.1/_site_random_generator.php?a=1")
111 print TEST_TITLE + url.__class__.__name__ + TEST_TITLE_OBJECT, vars(url)
116 us.urlType = dc.EventObjects.URLStatus.URL_TYPE_MD5
117 print TEST_TITLE + us.__class__.__name__ + TEST_TITLE_OBJECT, vars(us)
123 print TEST_TITLE + bi1.__class__.__name__ + TEST_TITLE_OBJECT, vars(bi1)
129 print TEST_TITLE + b.__class__.__name__ + TEST_TITLE_OBJECT, vars(b)
134 print TEST_TITLE + uf.__class__.__name__ + TEST_TITLE_OBJECT, vars(uf)
137 print TEST_TITLE + uf.__class__.__name__ + TEST_TITLE_OBJECT, vars(uf)
140 uf.urlsCriterions[dc.EventObjects.URLFetch.CRITERION_WHERE] = \
141 "Status=7 AND Crawled>0 AND Processed>0 AND CDate BETWEEN '2014-06-28 00:00:01' AND '2014-06-28 23:59:59'" 142 print TEST_TITLE + uf.__class__.__name__ + TEST_TITLE_OBJECT, vars(uf)
147 print TEST_TITLE + uu.__class__.__name__ + TEST_TITLE_OBJECT, vars(uu)
150 urlType=dc.EventObjects.URLStatus.URL_TYPE_MD5,
151 stateField=dc.EventObjects.URL.STATE_ENABLED,
152 statusField=dc.EventObjects.URL.STATUS_NEW)
153 print TEST_TITLE + uu.__class__.__name__ + TEST_TITLE_OBJECT, vars(uu)
158 print TEST_TITLE + ud.__class__.__name__ + TEST_TITLE_OBJECT, vars(ud)
161 urlType=dc.EventObjects.URLStatus.URL_TYPE_MD5)
162 print TEST_TITLE + ud.__class__.__name__ + TEST_TITLE_OBJECT, vars(ud)
167 print TEST_TITLE + uc.__class__.__name__ + TEST_TITLE_OBJECT, vars(uc)
169 uc =
dc.EventObjects.URLCleanup(
"3463463463463463", urlString=
"235325634634263", urlType=dc.EventObjects.URLStatus.URL_TYPE_MD5,
170 stateField=dc.EventObjects.URL.STATE_DISABLED)
171 print TEST_TITLE + uc.__class__.__name__ + TEST_TITLE_OBJECT, vars(uc)
176 print TEST_TITLE + ucr.__class__.__name__ + TEST_TITLE_OBJECT, vars(ucr)
179 dc.EventObjects.URLContentRequest.CONTENT_TYPE_PROCESSED +
180 dc.EventObjects.URLContentRequest.CONTENT_TYPE_RAW_ALL)
181 print TEST_TITLE + ucr.__class__.__name__ + TEST_TITLE_OBJECT, vars(ucr)
185 uf.urlsCriterions[dc.EventObjects.URLFetch.CRITERION_WHERE] = \
186 "Status=7 AND Crawled>0 AND Processed>0 AND CDate BETWEEN '2014-06-28 00:00:01' AND '2014-06-28 23:59:59'" 187 uf.sitesCriterions[dc.EventObjects.URLFetch.CRITERION_WHERE] = \
189 uf.algorithm = dc.EventObjects.URLFetch.PROPORTIONAL_ALGORITHM
192 print TEST_TITLE + ucr.__class__.__name__ + TEST_TITLE_OBJECT, vars(ucr)
200 print TEST_TITLE + ucr.__class__.__name__ + TEST_TITLE_OBJECT, vars(ucr)
208 print TEST_TITLE + cr.__class__.__name__ + TEST_TITLE_OBJECT, vars(cr)
GeneralResponse event object, represents general state response for multipurpose usage.