# dictionary login -> [list of roles]
rolepacks = {}
def read_role_body(header, raw, curline):
login = raw[header['who']]
raw_role = raw[header['role']].split(lib.tsep)
roles = []
for str_role in raw_role:
if str_role == '':
break
raw_role = str_role.split(lib.fsep)
newpack = lib.infopack()
newpack.data = raw_role[1]
newpack.when = int(raw_role[0])
roles.append(newpack)
rolepacks[login] = roles
lib.read_file(sys.argv[2], lib.empty_header, read_role_body)
# 2) link to level4_assigning
def link_header(header, raw, curline):
print curline[:-1] + lib.sep + sys.argv[5] + '_role'
def link_body(header, raw, curline):
login = raw[header[sys.argv[3]]]
when = int(raw[header[sys.argv[4]]])
# no one should remain 'unknow'
role = 'unknow'
if rolepacks.has_key(login):
index = lib.binary_search(rolepacks[login], lib.func_time, when)
if index >= 0:
role = rolepacks[login][index].data
global cntMaxExp login = raw[header['login']] if login == actor: # cntN cntN = GetValueByTimeFromStr(raw[header['contacter_num']], actionTime) # cntMaxExp cntMaxExp = GetValueByTimeFromStr(raw[header['max_contecter_exp']], actionTime) return 1 def ReadCommentRecord(header, raw, curline): global cmtN login = raw[header['login']] if login == actor: # cmtN cmtN = GetValueByTimeFromStr(raw[header['cmt_num']], actionTime) return 1 lib.read_file(sys.argv[1], lib.empty_header, ReadProductAssignmentRecord) lib.read_file(sys.argv[2], lib.empty_header, ReadLoginProductAssignmentRecord) lib.read_file(sys.argv[3], lib.empty_header, ReadLoginPeerRecord) lib.read_file(sys.argv[4], lib.empty_header, ReadCommentRecord) print str(nPrdCorAssi) + ',' + str(nPrdAssi) + ',' + str(nLogPrdCorAssi) + ',' + str(nLogPrdAssi) + ',' + str(cntMaxExp) + ',' + str(cntN) + ',' + str(cmtN)
# exp raw_exps = raw[header['max_contecter_exp']].split(lib.tsep) exp_packs = [] for raw_exp in raw_exps: if raw_exp == '': continue raw_pack = raw_exp.split(lib.fsep) exp = int(raw_pack[0]) time = int(raw_pack[1]) newpack = lib.infopack() newpack.data = exp newpack.when = time exp_packs.append(newpack) con_exp[login] = exp_packs lib.read_file(sys.argv[2], lib.empty_header, read_contact_info) # link to target file def link_header(header, raw, curline): print curline[:-1] + lib.sep + 'cnt_num' + lib.sep + 'cnt_max_exp' def link_body(header, raw, curline): global con_num, con_exp, con_dev login = raw[header['login']] when = int(raw[header['when']]) # num num = 0 if con_num.has_key(login): num_index = lib.binary_search(con_num[login], lib.func_time, when)
global alias
global use_email
who = raw[header['who']]
when = int(raw[header['when']])
if use_email:
name = who.split('@')[0]
if not alias.has_key(name):
alias[name] = {}
alias[name][who] = 1
if not actors.has_key(who):
actors[who] = actor()
actors[who].nonself_since = when
lib.read_file(sys.argv[2], lib.empty_header, read_reporter_body)
# read maintainer namelist
def read_maint_body(header, raw, curline):
global actors
global alias
global use_email
who = raw[header['who']]
if use_email:
name = who.split('@')[0]
if not alias.has_key(name):
alias[name] = {}
alias[name][who] = 1
if not actors.has_key(who):
global bug_final_time
if bug_id in bug_fixed:
# no need to care about it
return
if llevel4_type == '0':
final_time = int((raw[header['product']].split(lib.tsep)[-1]).split(lib.fsep)[1])
bug_final_time[bug_id] = final_time
elif llevel4_type == '1':
# get time for fields in field_set
bug_final_time[bug_id] = {}
for field in field_set:
final_time = int((raw[header[field]].split(lib.tsep)[-1]).split(lib.fsep)[1])
bug_final_time[bug_id][field] = final_time
# read llevel3
lib.read_file(sys.argv[3], lib.empty_header, read_llevel3_body)
# bug_id -> the first time of verifying the final action
bug_verifying_time = {}
# read activity level2
def read_level2(header, raw, curline):
global bug_resolving_time
global bug_confirmed
global bug_fixed
global bug_final_time
bug_id = raw[header['bug_id']]
# we ignore some bugs, because they don't need this information
if (not bug_resolving_time.has_key(bug_id)) or (bug_id in bug_confirmed) or (bug_id in bug_fixed) or (not bug_final_time.has_key(bug_id)):
if bug_id != cur_bug_id:
cur_bug_id = bug_id
cur_set = set()
login = raw[header['who']]
when = int(raw[header['when']])
month_aligned_when = lib.month_align(when)
cur_set.add(login)
if not logins.has_key(login):
logins[login] = {}
if not logins[login].has_key(month_aligned_when):
logins[login][month_aligned_when] = contacters()
logins[login][month_aligned_when].logins = cur_set.union(logins[login][month_aligned_when].logins)
lib.read_file(sys.argv[1], lib.empty_header, read_lcmt_level2_body)
# step 2) read login general experience
# dictionary login -> [list of <exp,when> tuples]
login_exp = {}
def read_body(header, raw, curline):
who = raw[header['who']]
exps = raw[header['exp']].split(lib.tsep)
explist = []
for str_exp in exps:
if str_exp == '':
break
raw_exp = str_exp.split(lib.fsep)
newpack = lib.infopack()
newpack.data = int(raw_exp[0])
newpack.when = int(raw_exp[1])
#!/usr/bin/python # Author: Xie Jialiang # Date: 2013-05-23 # Function: link correct or not to level4_info # Parameters: 1) level4_info # Output stream: linked level4_info import sys import leon_lib as lib # read target file def read_llevel4_info_header(header, raw, curline): print curline[:-1] + lib.sep + 'correct' def read_llevel4_info_body(header, raw, curline): new = raw[header['new']].lower() final = raw[header['final']].lower() print curline[:-1] + lib.sep + str(new == final) lib.read_file(sys.argv[1], read_llevel4_info_header, read_llevel4_info_body)
'target_milestone':'Target Milestone', 'version':'Version', 'blocked':'Blocks', 'dependson':'Depends on', 'hardware':'Hardware', 'flag':'Flags', 'target_milestone':'Target Milestone'} def read_head(header, raw, curline): print curline[:-1] def read_body(header, raw, curline): bug_id = raw[header['bug_id']] who = raw[header['who']] when = raw[header['when']] what = raw[header['what']] old = raw[header['old']] new = raw[header['new']] if uniform_dic.has_key(what): what = uniform_dic[what] print bug_id + ';;' + who + ';' + when + ';' + what + ';' + old + ';' + new lib.read_file(sys.argv[1], read_head, read_body)
def read_login_comment_number(header, raw, curline):
global login_cmt
login = raw[header['login']]
raw_cmt = raw[header['cmt_num']].split(lib.tsep)
cmt_num = []
for str_num in raw_cmt:
raw_num = str_num.split(lib.fsep)
newpack = lib.infopack()
newpack.data = int(raw_num[0])
newpack.when = int(raw_num[1])
cmt_num.append(newpack)
login_cmt[login] = cmt_num
lib.read_file(sys.argv[2], lib.empty_header, read_login_comment_number)
# link to target file
def link_header(header, raw, curline):
print curline[:-1] + lib.sep + 'cmt_num'
def link_body(header, raw, curline):
global login_cmt
login = raw[header['login']]
when = int(raw[header['when']])
num = 0
if login_cmt.has_key(login):
#!/usr/bin/python
# Author: Xie Jialiang
# Date: 2013-05-23
# Function: link correct or not to level4_info
# Parameters: 1) level4_info
# Output stream: linked level4_info
import sys
import leon_lib as lib
# read target file
def read_llevel4_info_header(header, raw, curline):
print curline[:-1] + lib.sep + 'correct'
def read_llevel4_info_body(header, raw, curline):
new = raw[header['new']].lower()
final = raw[header['final']].lower()
print curline[:-1] + lib.sep + str(new == final)
lib.read_file(sys.argv[1], read_llevel4_info_header, read_llevel4_info_body)
resolved_time = lib.resolved_time_checker()
resolved_time.read(sys.argv[4])
# read final product
bug_final_product = {}
def read_final_product_body(header, raw, curline):
global bug_final_product
bug_id = raw[header['bug_id']]
final_product = (raw[header['product']].split(lib.tsep)[-1]).split(
lib.fsep)[0]
bug_final_product[bug_id] = final_product
lib.read_file(sys.argv[4], lib.empty_header, read_final_product_body)
# add to dict
def add_to_dict(dict, a, b, c):
if not dict.has_key(a):
dict[a] = {}
if not dict[a].has_key(b):
dict[a][b] = {}
if not dict[a][b].has_key(c):
dict[a][b][c] = 0
dict[a][b][c] += 1
logPrdAssi = {}
logPrdNotAssi = {}
# read product convertion
convertion = lib.product_convertion()
convertion.read_product_convertion(sys.argv[5])
# read leon_level3
resolved_time = lib.resolved_time_checker()
resolved_time.read(sys.argv[4])
# read final product
bug_final_product = {}
def read_final_product_body(header, raw, curline):
global bug_final_product
bug_id = raw[header['bug_id']]
final_product = (raw[header['product']].split(lib.tsep)[-1]).split(lib.fsep)[0]
bug_final_product[bug_id] = final_product
lib.read_file(sys.argv[4], lib.empty_header, read_final_product_body)
# add to dict
def add_to_dict(dict, a, b, c):
if not dict.has_key(a):
dict[a] = {}
if not dict[a].has_key(b):
dict[a][b] = {}
if not dict[a][b].has_key(c):
dict[a][b][c] = 0
dict[a][b][c] += 1
logPrdAssi = {}
logPrdNotAssi = {}
def read_triage_body(header, raw, curline):
login_cmt = {}
def read_login_comment_number(header, raw, curline):
global login_cmt
login = raw[header['login']]
raw_cmt = raw[header['cmt_num']].split(lib.tsep)
cmt_num = []
for str_num in raw_cmt:
raw_num = str_num.split(lib.fsep)
newpack = lib.infopack()
newpack.data = int(raw_num[0])
newpack.when = int(raw_num[1])
cmt_num.append(newpack)
login_cmt[login] = cmt_num
lib.read_file(sys.argv[2], lib.empty_header, read_login_comment_number)
# link to target file
def link_header(header, raw, curline):
print curline[:-1] + lib.sep + 'cmt_num'
def link_body(header, raw, curline):
global login_cmt
login = raw[header['login']]
when = int(raw[header['when']])
num = 0
if login_cmt.has_key(login):
index = lib.binary_search(login_cmt[login], lib.func_time, when)
if index >= 0:
login = raw[header['login']]
if login == actor:
# cmtN
cmtN = GetValueByTimeFromStr(raw[header['cmt_num']], actionTime)
return 1
def ReadRoleRecord(header, raw, curline):
global role
login = raw[header['who']]
if login == actor:
role = GetValueByTimeFromStr(raw[header['role']], actionTime)
lib.read_file(sys.argv[1], lib.empty_header, ReadProductAssignmentRecord)
lib.read_file(sys.argv[2], lib.empty_header, ReadLoginProductAssignmentRecord)
lib.read_file(sys.argv[3], lib.empty_header, ReadLoginPeerRecord)
lib.read_file(sys.argv[4], lib.empty_header, ReadCommentRecord)
lib.read_file(sys.argv[5], lib.empty_header, ReadRoleRecord)
print str(nPrdCorAssi) + ',' + str(nPrdAssi) + ',' + str(
nLogPrdCorAssi) + ',' + str(nLogPrdAssi) + ',' + str(
cntMaxExp) + ',' + str(cntN) + ',' + str(cmtN) + ',' + '\"' + str(
role) + '\"'
raw_exps = raw[header['max_contecter_exp']].split(lib.tsep)
exp_packs = []
for raw_exp in raw_exps:
if raw_exp == '':
continue
raw_pack = raw_exp.split(lib.fsep)
exp = int(raw_pack[0])
time = int(raw_pack[1])
newpack = lib.infopack()
newpack.data = exp
newpack.when = time
exp_packs.append(newpack)
con_exp[login] = exp_packs
lib.read_file(sys.argv[2], lib.empty_header, read_contact_info)
# link to target file
def link_header(header, raw, curline):
print curline[:-1] + lib.sep + 'cnt_num' + lib.sep + 'cnt_max_exp'
def link_body(header, raw, curline):
global con_num, con_exp, con_dev
login = raw[header['login']]
when = int(raw[header['when']])
# num
num = 0
return
if llevel4_type == '0':
final_time = int(
(raw[header['product']].split(lib.tsep)[-1]).split(lib.fsep)[1])
bug_final_time[bug_id] = final_time
elif llevel4_type == '1':
# get time for fields in field_set
bug_final_time[bug_id] = {}
for field in field_set:
final_time = int(
(raw[header[field]].split(lib.tsep)[-1]).split(lib.fsep)[1])
bug_final_time[bug_id][field] = final_time
# read llevel3
lib.read_file(sys.argv[3], lib.empty_header, read_llevel3_body)
# bug_id -> the first time of verifying the final action
bug_verifying_time = {}
# read activity level2
def read_level2(header, raw, curline):
global bug_resolving_time
global bug_confirmed
global bug_fixed
global bug_final_time
bug_id = raw[header['bug_id']]
# we ignore some bugs, because they don't need this information
def read_role_body(header, raw, curline):
login = raw[header['who']]
raw_role = raw[header['role']].split(lib.tsep)
roles = []
for str_role in raw_role:
if str_role == '':
break
raw_role = str_role.split(lib.fsep)
newpack = lib.infopack()
newpack.data = raw_role[0]
newpack.when = int(raw_role[1])
roles.append(newpack)
rolepacks[login] = roles
lib.read_file(sys.argv[2], lib.empty_header, read_role_body)
# 2) link to level4_assigning
def link_header(header, raw, curline):
print curline[:-1] + lib.sep + sys.argv[5] + '_role'
def link_body(header, raw, curline):
login = raw[header[sys.argv[3]]]
when = int(raw[header[sys.argv[4]]])
# no one should remain 'unknow'
role = 'unknow'
if rolepacks.has_key(login):
index = lib.binary_search(rolepacks[login], lib.func_time, when)
info_packs[bug_id].os = value elif key == 'priority': info_packs[bug_id].priority = value elif key == 'product': info_packs[bug_id].product = value elif key == 'reporter': info_packs[bug_id].reporter = value elif key == 'resolution': info_packs[bug_id].resolution = value elif key == 'target_milestone': info_packs[bug_id].target_milestone = value elif key == 'version': info_packs[bug_id].version = value # read linfo_level2 lib.read_file(sys.argv[1], lib.empty_header, read_linfo_level2_body) # 2) read activity level2 class act_pack: def act_pack(self): self.time = 0 self.old = '' self.new = '' self.who = '' class issue: def __init__(self): # basic info self.bug_id = '' self.reporter = ''
if not n_tri[field][login].has_key(product):
n_tri[field][login][product] = {}
if not n_tri[field][login][product].has_key(valid_t):
n_tri[field][login][product][valid_t] = 0
n_tri[field][login][product][valid_t] += 1
# n_cor_tri
if is_correct:
if not n_cor_tri[field].has_key(login):
n_cor_tri[field][login] = {}
if not n_cor_tri[field][login].has_key(product):
n_cor_tri[field][login][product] = {}
if not n_cor_tri[field][login][product].has_key(valid_t):
n_cor_tri[field][login][product][valid_t] = 0
n_cor_tri[field][login][product][valid_t] += 1
lib.read_file(sys.argv[1], lib.empty_header, read_body)
# output
print 'login' + lib.sep + 'product' + lib.sep + 'field' + lib.sep + 'n_tri' + lib.sep + 'n_cor_tri'
for field in n_tri:
for login in n_tri[field]:
for product in n_tri[field][login]:
# n_tri
str_tri = ''
timelist = sorted(n_tri[field][login][product].keys())
count = 0
for time in timelist:
count += n_tri[field][login][product][time]
str_tri += str(count) + lib.fsep + str(time) + lib.tsep
if str_tri.__len__() > 0:
str_tri = str_tri[:-lib.tsep.__len__()]
bugID = raw[header['bug_id']]
if bugID != targetID:
# if seen target and it is not target now, skip the rest
if seenTarget:
return -1
else:
return 0
seenTarget = True
field = raw[header['field']]
if curDict.has_key(field):
curDict[field] = raw[header['what']]
# read linfo_level2
lib.read_file(sys.argv[1], lib.empty_header, read_linfo_level2_body)
if not seenTarget:
print 'False' + sep + 'who' + sep + 'when' + sep + 'noSuchBug' + 'False' + sep + 'who' + sep + 'when' + sep + 'noSuchBug'
sys.exit(1)
if curDict['status'] == 'RESOLVED' or curDict[
'status'] == 'VERIFIED' or curDict['status'] == 'CLOSED':
# print 'False,who,when,' + curDict['status']
# sys.exit(1)
pass
# prepare read activity to find out who@when set the current value
actList = []
seenTarget = False
global use_email
who = raw[header['who']]
when = int(raw[header['when']])
if use_email:
name = who.split('@')[0]
if not alias.has_key(name):
alias[name] = {}
alias[name][who] = 1
if not actors.has_key(who):
actors[who] = actor()
actors[who].nonself_since = when
lib.read_file(sys.argv[2], lib.empty_header, read_reporter_body)
# read maintainer namelist
def read_maint_body(header, raw, curline):
global actors
global alias
global use_email
who = raw[header['who']]
if use_email:
name = who.split('@')[0]
if not alias.has_key(name):
alias[name] = {}
alias[name][who] = 1
# get issue reporter
# dictionary bug_id -> reporter
issue_reporter = {}
def read_body(header, raw, curline):
global issue_reporter
bug_id = raw[header['bug_id']]
field = raw[header['field']]
what = raw[header['what']]
if field == 'reporter':
issue_reporter[bug_id] = what
lib.read_file(sys.argv[1], lib.empty_header, read_body)
# only reporter
# dictionary login -> the first time she touched others issues, -1 means never
touch_others = {}
def read_act_body(header, raw, curline):
global touch_others
global issue_reporter
issue = raw[header['bug_id']]
actor = raw[header['who']]
if not touch_others.has_key(actor):
# the first time we meet the gay, assume it only touch its own issues
touch_others[actor] = -1
'bug_severity': 'Severity',
'bug_status': 'Status',
'target_milestone': 'Target Milestone',
'version': 'Version',
'blocked': 'Blocks',
'dependson': 'Depends on',
'hardware': 'Hardware',
'flag': 'Flags',
'target_milestone': 'Target Milestone'
}
def read_head(header, raw, curline):
print curline[:-1]
def read_body(header, raw, curline):
bug_id = raw[header['bug_id']]
who = raw[header['who']]
when = raw[header['when']]
what = raw[header['what']]
old = raw[header['old']]
new = raw[header['new']]
if uniform_dic.has_key(what):
what = uniform_dic[what]
print bug_id + ';;' + who + ';' + when + ';' + what + ';' + old + ';' + new
lib.read_file(sys.argv[1], read_head, read_body)