def companiesData(period, startdate, enddate, identities_db, destdir, bots, npeople):
companies = SCM.companies_name_wo_affs(bots, startdate, enddate)
companies = companies['name']
createJSON(companies, destdir+"/scm-companies.json")
for company in companies:
company_name = "'"+ company+ "'"
print (company_name)
evol_data = SCM.GetSCMEvolutionaryData(period, startdate, enddate, identities_db, ["company", company_name])
evol_data = completePeriodIds(evol_data)
createJSON(evol_data, destdir+"/"+company+"-scm-com-evolutionary.json")
agg = SCM.GetSCMStaticData(period, startdate, enddate, identities_db, ["company", company_name])
createJSON(agg, destdir+"/"+company+"-scm-com-static.json")
top_authors = SCM.company_top_authors(company_name, startdate, enddate, npeople)
createJSON(top_authors, destdir+"/"+company+"-scm-com-top-authors.json", False)
for i in [2006,2009,2012]:
data = SCM.company_top_authors_year(company_name, i, npeople)
createJSON(data, destdir+"/"+company+"-scm-top-authors_"+str(i)+".json", False)
commits = SCM.GetCommitsSummaryCompanies(period, startdate, enddate, opts.identities_db, 10)
createJSON (commits, destdir+"/scm-companies-commits-summary.json")
def countriesData(period, startdate, enddate, identities_db, destdir):
countries = SCM.scm_countries_names(identities_db,startdate, enddate)
countries = countries['name']
createJSON(countries, destdir+"/scm-countries.json")
for country in countries:
print (country)
country_name = "'"+country+"'"
evol_data = SCM.GetSCMEvolutionaryData(period, startdate, enddate, identities_db, ["country", country_name])
evol_data = completePeriodIds(evol_data)
createJSON (evol_data, destdir+"/"+country+"-scm-cou-evolutionary.json")
agg = SCM.GetSCMStaticData(period, startdate, enddate, identities_db, ["country", country_name])
createJSON (agg, destdir+"/"+country+"-scm-cou-static.json")
def aggData(period, startdate, enddate, identities_db, destdir):
# data = dataFrame2Dict(vizr.GetSCMStaticData(period, startdate, enddate, identities_db))
data = SCM.GetSCMStaticData(period, startdate, enddate, identities_db, None)
agg = data
static_url = SCM.StaticURL()
agg = dict(agg.items() + static_url.items())
if ('companies' in reports):
data = SCM.evol_info_data_companies (startdate, enddate)
agg = dict(agg.items() + data.items())
if ('countries' in reports):
data = SCM.evol_info_data_countries (startdate, enddate)
agg = dict(agg.items() + data.items())
if ('domains' in reports):
data = SCM.evol_info_data_domains (startdate, enddate)
agg = dict(agg.items() + data.items())
data = SCM.GetCodeCommunityStructure(period, startdate, enddate, identities_db)
agg = dict(agg.items() + data.items())
# TODO: repeated data
# data = dataFrame2Dict(vizr.GetDiffCommitsDays(period, enddate, 365))
# agg = dict(agg.items() + data.items())
# Tendencies
for i in [7,30,365]:
data = SCM.GetDiffCommitsDays(period, enddate, identities_db, i)
agg = dict(agg.items() + data.items())
data = SCM.GetDiffAuthorsDays(period, enddate, identities_db, i)
agg = dict(agg.items() + data.items())
data = SCM.GetDiffFilesDays(period, enddate, identities_db, i)
agg = dict(agg.items() + data.items())
data = SCM.GetDiffLinesDays(period, enddate, identities_db, i)
agg = dict(agg.items() + data.items())
# Last Activity: to be removed
for i in [7,14,30,60,90,180,365,730]:
data = SCM.last_activity(i)
agg = dict(agg.items() + data.items())
# Fields with wrong data in R
skip_fields = ['percentage_removed_lines_30','percentage_added_lines_30','diff_netadded_lines_30','diff_netremoved_lines_30']
createJSON (agg, destdir+"/scm-static.json", True, skip_fields)
def domainsData(period, startdate, enddate, identities_db, destdir):
domains = SCM.scm_domains_names(identities_db,startdate, enddate)
domains = domains['name']
createJSON(domains, destdir+"/scm-domains.json")
# Some R ts are wrong
bad_R_json_domains = ['gerrit','gmx','emsenhuber','bitergia']
for domain in domains :
domain_name = "'"+domain+"'"
print (domain_name)
evol_data = SCM.GetSCMEvolutionaryData(period, startdate, enddate, identities_db, ["domain", domain_name])
evol_data = completePeriodIds(evol_data)
if domain in bad_R_json_domains:
createJSON(evol_data, destdir+"/"+domain+"-scm-dom-evolutionary.json", False)
else:
createJSON(evol_data, destdir+"/"+domain+"-scm-dom-evolutionary.json")
agg = SCM.GetSCMStaticData(period, startdate, enddate, identities_db, ["domain", domain_name])
createJSON(agg, destdir+ "/"+domain+"-scm-dom-static.json")
def reposData(period, startdate, enddate, identities_db, destdir, conf):
# repos = dataFrame2Dict(vizr.repos_name(startdate, enddate))
repos = SCM.repos_name(startdate, enddate)
repos = repos['name']
if not isinstance(repos, (list)):
repos = [repos]
createJSON(repos, destdir+"/scm-repos.json", False)
else:
createJSON(repos, destdir+"/scm-repos.json")
for repo in repos :
repo_name = "'"+ repo+ "'"
print (repo_name)
evol_data = SCM.GetSCMEvolutionaryData(period, startdate, enddate, identities_db, ["repository", repo_name])
evol_data = completePeriodIds(evol_data)
createJSON(evol_data, destdir+"/"+repo+"-scm-rep-evolutionary.json")
agg = SCM.GetSCMStaticData(period, startdate, enddate, identities_db, ["repository", repo_name])
createJSON(agg, destdir+"/"+repo+"-scm-rep-static.json")
