commands: add resource stats to spack url stats (#13205)
`spack url stats` now also looks at packages' resources when outputting
statistics.
Example:
```
$ spack url stats
==> URL stats for 3531 packages:
--------------------------------------------------------------
stat versions % resources %
--------------------------------------------------------------
url 8335 89.3% 339 89.0%
schemes
https 6489 69.5% 93 24.4%
ftp 32 0.3% 8 2.1%
http 1763 18.9% 237 62.2%
file 51 0.5% 1 0.3%
checksums
md5 26 0.3% 0 0.0%
sha256 8306 89.0% 336 88.2%
no checksum 3 0.0% 3 0.8%
--------------------------------------------------------------
go 1 0.0% 0 0.0%
--------------------------------------------------------------
hg 7 0.1% 0 0.0%
--------------------------------------------------------------
no code 4 0.0% 0 0.0%
--------------------------------------------------------------
svn 4 0.0% 16 4.2%
--------------------------------------------------------------
git 981 10.5% 26 6.8%
branch 442 4.7% 4 1.0%
commit 362 3.9% 14 3.7%
no ref 36 0.4% 2 0.5%
tag 141 1.5% 6 1.6%
--------------------------------------------------------------
```
This commit is contained in:
Todd Gamblin
GitHub
parent
8a1e5f68f4
commit
385d2bf3fa
2 changed files with 92 additions and 64 deletions
|
|
@ -239,80 +239,104 @@ def url_summary(args):
|
||||||
|
|
||||||
|
|
||||||
def url_stats(args):
|
def url_stats(args):
|
||||||
stats = {} # stats about fetchers in packages.
|
class UrlStats(object):
|
||||||
nvers = 0 # total number of versions
|
def __init__(self):
|
||||||
npkgs = 0 # total number of packages
|
self.total = 0
|
||||||
|
self.schemes = defaultdict(lambda: 0)
|
||||||
|
self.checksums = defaultdict(lambda: 0)
|
||||||
|
self.url_type = defaultdict(lambda: 0)
|
||||||
|
self.git_type = defaultdict(lambda: 0)
|
||||||
|
|
||||||
def inc(fstype, category, attr=None):
|
def add(self, fetcher):
|
||||||
"""Increment statistics in the stats dict."""
|
self.total += 1
|
||||||
categories = stats.setdefault(fstype, {})
|
|
||||||
if attr:
|
|
||||||
cat_stats = categories.setdefault(category, {})
|
|
||||||
val = cat_stats.setdefault(attr, 0)
|
|
||||||
stats[fstype][category][attr] = val + 1
|
|
||||||
else:
|
|
||||||
val = categories.setdefault(category, 0)
|
|
||||||
stats[fstype][category] = val + 1
|
|
||||||
|
|
||||||
# over all packages
|
url_type = fetcher.url_attr
|
||||||
for pkg in spack.repo.path.all_packages():
|
self.url_type[url_type or 'no code'] += 1
|
||||||
npkgs += 1
|
|
||||||
|
|
||||||
if not pkg.has_code:
|
if url_type == 'url':
|
||||||
for _ in pkg.versions:
|
digest = getattr(fetcher, 'digest', None)
|
||||||
inc('No code', 'total')
|
if digest:
|
||||||
nvers += 1
|
algo = crypto.hash_algo_for_digest(digest)
|
||||||
continue
|
|
||||||
|
|
||||||
# look at each version
|
|
||||||
for v, args in pkg.versions.items():
|
|
||||||
# figure out what type of fetcher it is
|
|
||||||
fetcher = fs.for_package_version(pkg, v)
|
|
||||||
nvers += 1
|
|
||||||
|
|
||||||
fstype = fetcher.url_attr
|
|
||||||
inc(fstype, 'total')
|
|
||||||
|
|
||||||
# put some special stats in for particular types of fetchers.
|
|
||||||
if fstype == 'git':
|
|
||||||
if 'commit' in args:
|
|
||||||
inc('git', 'security', 'commit')
|
|
||||||
else:
|
else:
|
||||||
inc('git', 'security', 'no commit')
|
algo = 'no checksum'
|
||||||
elif fstype == 'url':
|
self.checksums[algo] += 1
|
||||||
for h in crypto.hashes:
|
|
||||||
if h in args:
|
|
||||||
inc('url', 'checksums', h)
|
|
||||||
break
|
|
||||||
else:
|
|
||||||
if 'checksum' in args:
|
|
||||||
h = crypto.hash_algo_for_digest(args['checksum'])
|
|
||||||
inc('url', 'checksums', h)
|
|
||||||
else:
|
|
||||||
inc('url', 'checksums', 'no checksum')
|
|
||||||
|
|
||||||
# parse out the URL scheme (https/http/ftp/etc.)
|
# parse out the URL scheme (https/http/ftp/etc.)
|
||||||
urlinfo = urlparse(fetcher.url)
|
urlinfo = urlparse(fetcher.url)
|
||||||
inc('url', 'schemes', urlinfo.scheme)
|
self.schemes[urlinfo.scheme] += 1
|
||||||
|
|
||||||
|
elif url_type == 'git':
|
||||||
|
if getattr(fetcher, 'commit', None):
|
||||||
|
self.git_type['commit'] += 1
|
||||||
|
elif getattr(fetcher, 'branch', None):
|
||||||
|
self.git_type['branch'] += 1
|
||||||
|
elif getattr(fetcher, 'tag', None):
|
||||||
|
self.git_type['tag'] += 1
|
||||||
|
else:
|
||||||
|
self.git_type['no ref'] += 1
|
||||||
|
|
||||||
|
npkgs = 0
|
||||||
|
version_stats = UrlStats()
|
||||||
|
resource_stats = UrlStats()
|
||||||
|
|
||||||
|
for pkg in spack.repo.path.all_packages():
|
||||||
|
npkgs += 1
|
||||||
|
|
||||||
|
for v, args in pkg.versions.items():
|
||||||
|
fetcher = fs.for_package_version(pkg, v)
|
||||||
|
version_stats.add(fetcher)
|
||||||
|
|
||||||
|
for _, resources in pkg.resources.items():
|
||||||
|
for resource in resources:
|
||||||
|
resource_stats.add(resource.fetcher)
|
||||||
|
|
||||||
# print a nice summary table
|
# print a nice summary table
|
||||||
tty.msg("%d total versions for %d packages:" % (nvers, npkgs))
|
tty.msg("URL stats for %d packages:" % npkgs)
|
||||||
line_width = 36
|
|
||||||
print("-" * line_width)
|
|
||||||
for fetcher, fetcher_stats in sorted(stats.items(), reverse=True):
|
|
||||||
fs_total = fetcher_stats['total']
|
|
||||||
fs_pct = float(fs_total) / nvers * 100
|
|
||||||
print("%-22s%5d%8.1f%%" % (fetcher, fs_total, fs_pct))
|
|
||||||
|
|
||||||
for category, cat_stats in sorted(fetcher_stats.items(), reverse=True):
|
def print_line():
|
||||||
if category == 'total':
|
print("-" * 62)
|
||||||
continue
|
|
||||||
print(" %s" % category)
|
|
||||||
|
|
||||||
for name, number in sorted(cat_stats.items(), reverse=True):
|
def print_stat(indent, name, stat_name=None):
|
||||||
pct = float(number) / fs_total * 100
|
width = 20 - indent
|
||||||
print(" %-18s%5d%8.1f%%" % (name, number, pct))
|
fmt = " " * indent
|
||||||
print("-" * line_width)
|
fmt += "%%-%ds" % width
|
||||||
|
if stat_name is None:
|
||||||
|
print(fmt % name)
|
||||||
|
else:
|
||||||
|
fmt += "%12d%8.1f%%%12d%8.1f%%"
|
||||||
|
v = getattr(version_stats, stat_name).get(name, 0)
|
||||||
|
r = getattr(resource_stats, stat_name).get(name, 0)
|
||||||
|
print(fmt % (name,
|
||||||
|
v, v / version_stats.total * 100,
|
||||||
|
r, r / resource_stats.total * 100))
|
||||||
|
|
||||||
|
print_line()
|
||||||
|
print("%-20s%12s%9s%12s%9s" % ("stat", "versions", "%", "resources", "%"))
|
||||||
|
print_line()
|
||||||
|
print_stat(0, "url", "url_type")
|
||||||
|
|
||||||
|
print_stat(4, "schemes")
|
||||||
|
schemes = set(version_stats.schemes) | set(resource_stats.schemes)
|
||||||
|
for scheme in schemes:
|
||||||
|
print_stat(8, scheme, "schemes")
|
||||||
|
|
||||||
|
print_stat(4, "checksums")
|
||||||
|
checksums = set(version_stats.checksums) | set(resource_stats.checksums)
|
||||||
|
for checksum in checksums:
|
||||||
|
print_stat(8, checksum, "checksums")
|
||||||
|
print_line()
|
||||||
|
|
||||||
|
types = set(version_stats.url_type) | set(resource_stats.url_type)
|
||||||
|
types -= set(["url", "git"])
|
||||||
|
for url_type in sorted(types):
|
||||||
|
print_stat(0, url_type, "url_type")
|
||||||
|
print_line()
|
||||||
|
|
||||||
|
print_stat(0, "git", "url_type")
|
||||||
|
git_types = set(version_stats.git_type) | set(resource_stats.git_type)
|
||||||
|
for git_type in sorted(git_types):
|
||||||
|
print_stat(4, git_type, "git_type")
|
||||||
|
print_line()
|
||||||
|
|
||||||
|
|
||||||
def print_name_and_version(url):
|
def print_name_and_version(url):
|
||||||
|
|
|
||||||
|
|
@ -131,4 +131,8 @@ def test_url_stats(capfd):
|
||||||
output = url('stats')
|
output = url('stats')
|
||||||
npkgs = '%d packages' % len(spack.repo.all_package_names())
|
npkgs = '%d packages' % len(spack.repo.all_package_names())
|
||||||
assert npkgs in output
|
assert npkgs in output
|
||||||
assert 'total versions' in output
|
assert 'url' in output
|
||||||
|
assert 'git' in output
|
||||||
|
assert 'schemes' in output
|
||||||
|
assert 'versions' in output
|
||||||
|
assert 'resources' in output
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue