# # Small event handler to automatically open URLs and file # bug reports at a bugzilla of your choiche # # This class requires python2.4 because of the urllib2 usage # def seppuku_login(opener, login, user, password): """ We need to post to query.cgi with the parameters Bugzilla_login and Bugzilla_password and will scan the resulting page then @param opened = cookie enabled urllib2 opener @param login = http://bugzilla.openmoko.org/cgi-bin/bugzilla/query.cgi? @param user = Your username @param password = Your password """ import urllib param = urllib.urlencode( {"GoAheadAndLogIn" : 1, "Bugzilla_login" : user, "Bugzilla_password" : password } ) result = opener.open(login + param) if result.code != 200: return False txt = result.read() if not 'Log out' in txt: return False return True def seppuku_find_bug_report_old(): from HTMLParser import HTMLParser class BugQueryExtractor(HTMLParser): STATE_NONE = 0 STATE_FOUND_TR = 1 STATE_FOUND_NUMBER = 2 STATE_FOUND_PRIO = 3 STATE_FOUND_PRIO2 = 4 STATE_FOUND_NAME = 5 STATE_FOUND_PLATFORM = 6 STATE_FOUND_STATUS = 7 STATE_FOUND_WHATEVER = 8 # I don't know this field STATE_FOUND_DESCRIPTION =9 def __init__(self): HTMLParser.__init__(self) self.state = self.STATE_NONE self.bugs = [] def handle_starttag(self, tag, attr): if self.state == self.STATE_NONE and tag.lower() == "tr": if len(attr) == 1 and attr[0] == ('class', 'bz_normal bz_P2 '): self.state = self.STATE_FOUND_TR elif self.state == self.STATE_FOUND_TR and tag.lower() == "td": self.state += 1 def handle_endtag(self, tag): if tag.lower() == "tr": if self.state != self.STATE_NONE: self.bugs.append( (self.bug,self.status) ) self.state = self.STATE_NONE if self.state > 1 and tag.lower() == "td": self.state += 1 def handle_data(self,data): data = data.strip() # skip garbage if len(data) == 0: return if self.state == self.STATE_FOUND_NUMBER: self.bug = data elif self.state == self.STATE_FOUND_STATUS: self.status = data def result(self): return self.bugs return BugQueryExtractor() def seppuku_find_bug_report(opener, query, product, component, bugname): """ Find a bug report with the sane name and return the bug id and the status. @param opener = urllib2 opener @param query = e.g. https://bugzilla.openmoko.org/cgi-bin/bugzilla/query.cgi? @param product = search for this product @param component = search for this component @param bugname = the bug to search for https://bugzilla.openmoko.org/cgi-bin/bugzilla/buglist.cgi?short_desc_type=substring&short_desc=manual+test+bug&product=OpenMoko&emailreporter2=1&emailtype2=substring&email2=freyther%40yahoo.com but it does not support ctype=csv... """ result = opener.open("%(query)s?product=%(product)s&component=%(component)s&short_desc_type=substring&short_desc=%(bugname)s" % vars()) if result.code != 200: raise "Can not query the bugzilla at all" txt = result.read() scanner = seppuku_find_bug_report_old() scanner.feed(txt) if len(scanner.result()) == 0: return (False,None) else: # silently pick the first result (number,status) = scanner.result()[0] return (not status in ["CLOS", "RESO", "VERI"],number) def seppuku_reopen_bug(opener, file, product, component, bug_number, bugname, text): """ Reopen a bug report and append to the comment Same as with opening a new report, some bits need to be inside the url http://bugzilla.openmoko.org/cgi-bin/bugzilla/process_bug.cgi?id=239&bug_file_loc=http%3A%2F%2F&version=2007&longdesclength=2&product=OpenMoko&component=autobuilds&comment=bla&priority=P2&bug_severity=normal&op_sys=Linux&rep_platform=Neo1973&knob=reopen&target_milestone=Phase+0&short_desc=foo """ import urllib, urllib2 param = urllib.urlencode( { "product" : product, "component" : component, "longdesclength" : 2, "short_desc" : bugname, "knob" : "reopen", "id" : bug_number, "comment" : text } ) try: result = opener.open( file + param ) except urllib2.HTTPError, e: print e.geturl() print e.info() return False if result.code != 200: return False else: return True def seppuku_file_bug(opener, file, product, component, bugname, text): """ Create a completely new bug report http://bugzilla.openmoko.org/cgi-bin/bugzilla/post_bug.cgi?bug_file_loc=http%3A%2F%2F&version=2007&product=OpenMoko&component=autobuilds&short_desc=foo&comment=bla&priority=P2&bug_severity=normal&op_sys=Linux&rep_platform=Neo1973 You are forced to add some default values to the bugzilla query and stop with '&' @param opener urllib2 opener @param file The url used to file a bug report @param product Product @param component Component @param bugname Name of the to be created bug @param text Text """ import urllib,urllib2 param = urllib.urlencode( { "product" : product, "component" : component, "short_desc" : bugname, "comment" : text } ) try: result = opener.open( file + param ) except urllib2.HTTPError, e: print e.geturl() print e.info() raise e return False if result.code != 200: return False else: return True addhandler seppuku_eventhandler python seppuku_eventhandler() { """ Report task failures to the bugzilla and succeeded builds to the box """ from bb.event import NotHandled, getName from bb import data, mkdirhier, build import bb, os, glob bb.note( "Ran" ) try: import urllib2, cookielib except: bb.note("Failed to import the cookielib and urllib2, make sure to use python2.4") return NotHandled event = e data = e.data name = getName(event) if name == "PkgFailed": if not bb.data.getVar('SEPPUKU_AUTOBUILD', data, True) == "0": build.exec_task('do_clean', data) elif name == "TaskFailed" or name == "NoProvider": cj = cookielib.CookieJar() opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj)) login = bb.data.getVar("SEPPUKU_LOGIN", data, True) query = bb.data.getVar("SEPPUKU_QUERY", data, True) newbug = bb.data.getVar("SEPPUKU_NEWREPORT", data, True) reopen = bb.data.getVar("SEPPUKU_ADDCOMMENT", data, True) user = bb.data.getVar("SEPPUKU_USER", data, True) passw = bb.data.getVar("SEPPUKU_PASS", data, True) product = bb.data.getVar("SEPPUKU_PRODUCT", data, True) component = bb.data.getVar("SEPPUKU_COMPONENT", data, True) if not seppuku_login(opener, login, user, passw): bb.note("Login to bugzilla failed") return NotHandled else: print "Logged into the box" if name == "TaskFailed": bugname = "%(package)s-%(pv)s-%(pr)s-%(task)s" % { "package" : bb.data.getVar("PN", data, True), "pv" : bb.data.getVar("PV", data, True), "pr" : bb.data.getVar("PR", data, True), "task" : e.task } log_file = glob.glob("%s/log.%s.*" % (bb.data.getVar('T', event.data, True), event.task)) if len(log_file) != 0: to_file = bb.data.getVar('TINDER_LOG', event.data, True) text = "".join(open(log_file[0], 'r').readlines()) elif name == "NoProvider": bugname = "noprovider for %s runtime: %s" % (event.getItem, event.getisRuntime) text = "Please fix it" else: assert False (bug_open, bug_number) = seppuku_find_bug_report(opener, query, product, component, bugname) bb.note("Bug is open: %s and bug number: %s" % (bug_open, bug_number)) # The bug is present and still open, no need to attach an error log if bug_number and bug_open: bb.note("The bug is known as '%s'" % bug_number) return NotHandled if bug_number and not bug_open: if not seppuku_reopen_bug(opener, reopen, product, component, bug_number, bugname, text): bb.note("Failed to reopen the bug report") elif not seppuku_file_bug(opener, newbug, product, component, bugname, text): bb.note("Filing a bugreport failed") return NotHandled }