def parse_risk_file(risk_file, bus_risks):
risk_f = open(risk_file, 'r')
for line in risk_f:
line = line.strip()
if not line:
continue
dev, risk = line.split('=')
dev = safe_author_name(dev)
bus_risks[dev] = float(risk)
risk_f.close()
def parse_risk_file(risk_file, bus_risks):
risk_f = open(risk_file, 'r')
for line in risk_f:
line = line.strip()
if not line:
continue
dev, risk = line.split('=')
dev = safe_author_name(dev)
bus_risks[dev] = float(risk)
risk_f.close()
def parse_experience(log):
"""
Parse the dev experience from the git log.
"""
# list of tuple of shape [(dev, lines_add, lines_removed), ...]
exp = []
# entry lines were zero separated with -z
entry_lines = log.split('\0')
current_entry = []
for entry_line in entry_lines:
if not entry_line.strip():
# blank entry line marks the end of an entry, we're ready to process
local_entry = current_entry
current_entry = []
if len(local_entry) < 2:
print >> sys.stderr, "Weird entry, cannot parse: %s\n-----" % '\n'.join(
local_entry)
continue
author, changes = local_entry[:2]
author = safe_author_name(author)
try:
changes_split = re.split(r'\s+', changes)
# this can be two fields if there were file renames
# detected, in which case the file names are on the
# following entry lines, or three fields (third being
# the filename) if there were no file renames
lines_added, lines_removed = changes_split[:2]
lines_added = int(lines_added)
lines_removed = int(lines_removed)
# don't record revisions that don't have any removed or
# added lines...they mean nothing to our algorithm
if lines_added or lines_removed:
exp.append((author, lines_added, lines_removed))
except ValueError:
print >> sys.stderr, "Weird entry, cannot parse: %s\n-----" % '\n'.join(
local_entry)
continue
else:
# continue to aggregate the entry
lines = entry_line.split('\n')
current_entry.extend([line.strip() for line in lines])
# we need the oldest log entries first.
exp.reverse()
return exp
def parse_experience(log):
"""
Parse the dev experience from the git log.
"""
# list of tuple of shape [(dev, lines_add, lines_removed), ...]
exp = []
# entry lines were zero separated with -z
entry_lines = log.split('\0')
current_entry = []
for entry_line in entry_lines:
if not entry_line.strip():
# blank entry line marks the end of an entry, we're ready to process
local_entry = current_entry
current_entry = []
if len(local_entry) < 2:
print >> sys.stderr, "Weird entry, cannot parse: %s\n-----" % '\n'.join(local_entry)
continue
author, changes = local_entry[:2]
author = safe_author_name(author)
try:
changes_split = re.split(r'\s+', changes)
# this can be two fields if there were file renames
# detected, in which case the file names are on the
# following entry lines, or three fields (third being
# the filename) if there were no file renames
lines_added, lines_removed = changes_split[:2]
lines_added = int(lines_added)
lines_removed = int(lines_removed)
# don't record revisions that don't have any removed or
# added lines...they mean nothing to our algorithm
if lines_added or lines_removed:
exp.append((author, lines_added, lines_removed))
except ValueError:
print >> sys.stderr, "Weird entry, cannot parse: %s\n-----" % '\n'.join(local_entry)
continue
else:
# continue to aggregate the entry
lines = entry_line.split('\n')
current_entry.extend([line.strip() for line in lines])
# we need the oldest log entries first.
exp.reverse()
return exp
def parse_dev_experience(f, client, repo_root):
"""
f: a path relative to repo_root for a file from whose log we want
to parse dev experience.
client: the pysvn client
repo_root: the root of the svn repository
"""
# a list of tuples of form (dev, added_lines, deleted_lines), each
# representing one commit
dev_experience = []
# a list of tuples with the paths / revisions we want to run diffs
# on to reconstruct dev experience
comps_to_make = []
# since the name of the file can change through its history due to
# moves, we need to keep the most recent one we're looking for
fname_to_follow = f
added_line_re = re.compile(r'^\+')
# strict_node_history=False: follow copies
#
# discover_changed_paths: make the data about copying available in the changed_paths field
for log in client.log("%s%s" % (repo_root, f),
strict_node_history=False,
discover_changed_paths=True):
cp = log.changed_paths
# even though we are only asking for the log of a single file,
# svn gives us back all changed paths for that revision, so we
# have to look for the right one
for c in cp:
if fname_to_follow == c.path:
# since we're going back in time with the log process,
# a copyfrom_path means we need to follow the old file
# from now on.
if c.copyfrom_path:
fname_to_follow = c.copyfrom_path
comps_to_make.append((c.path, log.revision, log.author))
break
# our logic needs oldest logs first
comps_to_make.reverse()
# for the first revision, every line is attributed to the first
# author as an added line
txt = client.cat("%s%s" % (repo_root, comps_to_make[0][0]),
comps_to_make[0][1])
exp = txt.count('\n')
if not txt.endswith('\n'):
exp += 1
dev_experience.append((comps_to_make[0][2], exp, 0))
# for all the other entries, we must diff between revisions to
# find the number and kind of changes
for i in range(len(comps_to_make) - 1):
old_path = "%s%s" % (repo_root, comps_to_make[i][0])
old_rev = comps_to_make[i][1]
new_path = "%s%s" % (repo_root, comps_to_make[i + 1][0])
new_rev = comps_to_make[i + 1][1]
author = comps_to_make[i + 1][2]
try:
diff = client.diff('.',
old_path,
revision1=old_rev,
url_or_path2=new_path,
revision2=new_rev,
diff_options=['-w'])
diff = diff.split('\n')
ind_dbl_ats = 0
for i, line in enumerate(diff):
if line.startswith('@@'):
ind_dbl_ats = i
break
added = 0
removed = 0
for line in diff[ind_dbl_ats:]:
if line.startswith('+'):
added += 1
if line.startswith('-'):
removed += 1
dev_experience.append((safe_author_name(author), added, removed))
except:
# on one occasion I saw a non-binary item that existed in
# the filesystem with svn ls but errored out with a diff
# against that revision. Note the error and proceed.
print >> sys.stderr, "Error diffing %s %s and %s %s: " % \
(old_path, str(old_rev), new_path, str(new_rev)), sys.exc_info()[0]
return dev_experience
def parse_dev_experience(f, client, repo_root):
"""
f: a path relative to repo_root for a file from whose log we want
to parse dev experience.
client: the pysvn client
repo_root: the root of the svn repository
"""
# a list of tuples of form (dev, added_lines, deleted_lines), each
# representing one commit
dev_experience = []
# a list of tuples with the paths / revisions we want to run diffs
# on to reconstruct dev experience
comps_to_make = []
# since the name of the file can change through its history due to
# moves, we need to keep the most recent one we're looking for
fname_to_follow = f
added_line_re = re.compile(r'^\+')
# strict_node_history=False: follow copies
#
# discover_changed_paths: make the data about copying available in the changed_paths field
for log in client.log("%s%s" %(repo_root, f), strict_node_history=False, discover_changed_paths=True):
cp = log.changed_paths
# even though we are only asking for the log of a single file,
# svn gives us back all changed paths for that revision, so we
# have to look for the right one
for c in cp:
if fname_to_follow == c.path:
# since we're going back in time with the log process,
# a copyfrom_path means we need to follow the old file
# from now on.
if c.copyfrom_path:
fname_to_follow = c.copyfrom_path
comps_to_make.append((c.path, log.revision, log.author))
break
# our logic needs oldest logs first
comps_to_make.reverse()
# for the first revision, every line is attributed to the first
# author as an added line
txt = client.cat("%s%s" % (repo_root, comps_to_make[0][0]),
comps_to_make[0][1])
exp = txt.count('\n')
if not txt.endswith('\n'):
exp += 1
dev_experience.append((comps_to_make[0][2], exp, 0))
# for all the other entries, we must diff between revisions to
# find the number and kind of changes
for i in range(len(comps_to_make) - 1):
old_path = "%s%s" % (repo_root, comps_to_make[i][0])
old_rev = comps_to_make[i][1]
new_path = "%s%s" % (repo_root, comps_to_make[i + 1][0])
new_rev = comps_to_make[i + 1][1]
author = comps_to_make[i + 1][2]
try:
diff = client.diff('.',
old_path,
revision1=old_rev,
url_or_path2=new_path,
revision2=new_rev,
diff_options=['-w'])
diff = diff.split('\n')
ind_dbl_ats = 0
for i, line in enumerate(diff):
if line.startswith('@@'):
ind_dbl_ats = i
break
added = 0
removed = 0
for line in diff[ind_dbl_ats:]:
if line.startswith('+'):
added += 1
if line.startswith('-'):
removed += 1
dev_experience.append((safe_author_name(author), added, removed))
except:
# on one occasion I saw a non-binary item that existed in
# the filesystem with svn ls but errored out with a diff
# against that revision. Note the error and proceed.
print >> sys.stderr, "Error diffing %s %s and %s %s: " % \
(old_path, str(old_rev), new_path, str(new_rev)), sys.exc_info()[0]
return dev_experience
