def get_repo_commit_stats (gh, reponame):
"""Get statistics for the logged-in user's commits in the named repository.
The `reponame` should look something like "pkgw/bibtools". Returns a Holder.
"""
from inifile import Holder
repo = gh.get_repo (reponame)
res = Holder (commits=0, lines=0)
latest = None
for c in repo.get_commits (author=gh.get_user ()):
res.commits += 1
# I want to count the total lines committed, but this requires
# fetching the full commit information for each commit, which is slow
# and blows up my GitHup API rate limit. TODO: figure out alternative
# metric?
### XXX res.lines += c.stats.total
if latest is None:
latest = c.commit.committer.date
else:
latest = max (latest, c.commit.committer.date)
res.latest_date = latest
return res
def setup_processing (render, datadir):
context = Holder ()
context.render = render
context.items = list (load (datadir))
context.pubs = [i for i in context.items if i.section == 'pub']
context.pubgroups = partition_pubs (context.pubs)
context.props = [i for i in context.items if i.section == 'prop']
context.time_allocs = compute_time_allocations (context.props)
context.cur_formatter = None
context.my_abbrev_name = None
commands = {}
commands['CITESTATS'] = cmd_cite_stats
commands['FORMAT'] = cmd_format
commands['MYABBREVNAME'] = cmd_my_abbrev_name
commands['PUBLIST'] = cmd_pub_list
commands['TALLOCLIST'] = cmd_talloc_list
commands['RMISCLIST'] = cmd_rev_misc_list
commands['RMISCLIST_IF'] = cmd_rev_misc_list_if
commands['RMISCLIST_IF_NOT'] = cmd_rev_misc_list_if_not
commands['TODAY.'] = cmd_today
return context, commands
def get_repo_impact_stats (gh, reponame):
"""Get statistics that try to get at the impact/popularity of a particular
repository. The `reponame` should look something like "pkgw/bibtools".
Returns a Holder.
"""
from inifile import Holder
from time import sleep
repo = gh.get_repo (reponame)
res = Holder ()
res.description = repo.description # OK this isn't impact but it's handy
# It can take GitHub a little while to compute the 'stats' items, in which
# case the relevant binding functions will return None. We make these
# requests first, then make other ones; the hope is that by the time we
# come back and retry, the stats will have been computed.
def retry (func, first_try):
if first_try is not None:
return first_try
for i in xrange (5):
result = func ()
if result is not None:
return result
sleep (1)
raise Exception ('function %r took too long' % func)
contrib = repo.get_stats_contributors ()
res.commits = github_list_size (repo.get_commits ()) # this counts commits on main branch
res.forks = github_list_size (repo.get_forks ())
res.stars = github_list_size (repo.get_stargazers ())
res.contributors = github_list_size (retry (repo.get_stats_contributors, contrib))
return res
def partition_pubs (pubs):
groups = Holder ()
groups.all = []
groups.refereed = []
groups.refpreprint = []
groups.non_refereed = []
groups.all_formal = []
groups.all_non_refereed = []
groups.informal = []
for pub in pubs:
refereed = (pub.refereed == 'y')
refpreprint = (pub.get ('refpreprint', 'n') == 'y')
formal = (pub.get ('informal', 'n') == 'n')
# we assume refereed implies formal.
groups.all.append (pub)
if formal:
groups.all_formal.append (pub)
if refereed:
groups.refereed.append (pub)
elif refpreprint:
groups.refpreprint.append (pub)
else:
groups.all_non_refereed.append (pub)
if formal:
groups.non_refereed.append (pub)
else:
groups.informal.append (pub)
groups.all_rev = groups.all[::-1]
groups.refereed_rev = groups.refereed[::-1]
groups.refpreprint_rev = groups.refpreprint[::-1]
groups.non_refereed_rev = groups.non_refereed[::-1]
groups.informal_rev = groups.informal[::-1]
return groups
def compute_cite_stats (pubs):
"""Compute an h-index and other stats from the known publications."""
from time import gmtime
stats = Holder ()
stats.refpubs = 0
stats.refcites = 0
stats.reffirstauth = 0
cites = []
dates = []
for pub in pubs:
if pub.refereed == 'y':
stats.refpubs += 1
if int (pub.mypos) == 1:
stats.reffirstauth += 1
citeinfo = parse_ads_cites (pub)
if citeinfo is None:
continue
if citeinfo.cites < 1:
continue
cites.append (citeinfo.cites)
dates.append (citeinfo.lastupdate)
if pub.refereed == 'y':
stats.refcites += citeinfo.cites
if not len (cites):
stats.meddate = 0
stats.hindex = 0
else:
ranked = sorted (cites, reverse=True)
index = 0
while index < len (ranked) and ranked[index] >= index + 1:
index += 1
dates = sorted (dates)
stats.meddate = dates[len (dates) // 2]
stats.hindex = index
stats.year, stats.month, stats.day = gmtime (stats.meddate)[:3]
stats.monthstr = months[stats.month - 1]
stats.italich = MupItalics ('h')
stats.adslink = MupLink ('http://labs.adsabs.harvard.edu/adsabs', 'ADS')
return stats
def setup_processing(render, datadir):
context = Holder()
context.render = render
context.items = list(load(datadir))
context.pubs = [i for i in context.items if i.section == "pub"]
context.pubgroups = partition_pubs(context.pubs)
context.props = [i for i in context.items if i.section == "prop"]
context.time_allocs = compute_time_allocations(context.props)
context.repos = process_repositories(context.items)
context.cite_stats = compute_cite_stats(context.pubgroups.all_formal)
context.repo_stats = compute_repo_stats(context.repos)
context.talk_stats = summarize_talks([i for i in context.items if i.section == "talk"])
context.engagement_stats = summarize_engagement([i for i in context.items if i.section == "engagement"])
context.cur_formatter = None
context.my_abbrev_name = None
commands = {}
commands["BEGIN_SUBST"] = cmd_begin_subst
commands["FORMAT"] = cmd_format
commands["MYABBREVNAME"] = cmd_my_abbrev_name
commands["PUBLIST"] = cmd_pub_list
commands["TALLOCLIST"] = cmd_talloc_list
commands["RMISCLIST"] = cmd_rev_misc_list
commands["RMISCLIST_IF"] = cmd_rev_misc_list_if
commands["RMISCLIST_IF_NOT"] = cmd_rev_misc_list_if_not
commands["RREPOLIST"] = cmd_rev_repo_list
commands["TODAY."] = cmd_today
return context, commands
def partition_pubs(pubs):
groups = Holder()
groups.all = []
groups.refereed = []
groups.refpreprint = []
groups.non_refereed = []
groups.all_formal = []
groups.all_non_refereed = []
groups.informal = []
for pub in pubs:
refereed = pub.refereed == "y"
refpreprint = pub.get("refpreprint", "n") == "y"
formal = pub.get("informal", "n") == "n"
# we assume refereed implies formal.
groups.all.append(pub)
if formal:
groups.all_formal.append(pub)
if refereed:
groups.refereed.append(pub)
elif refpreprint:
groups.refpreprint.append(pub)
else:
groups.all_non_refereed.append(pub)
if formal:
groups.non_refereed.append(pub)
else:
groups.informal.append(pub)
groups.all_rev = groups.all[::-1]
groups.refereed_rev = groups.refereed[::-1]
groups.refpreprint_rev = groups.refpreprint[::-1]
groups.non_refereed_rev = groups.non_refereed[::-1]
groups.informal_rev = groups.informal[::-1]
return groups
