Split ensure_git_rev_available out from git_fetch

[pinch] / pinch.py

diff --git a/pinch.py b/pinch.py
index dfdedac03258749d0f9a539cf533537abcba54a6..0f8f00a611390cff8aaca8e20ff0d1361a4a4274 100644 (file)
--- a/pinch.py
+++ b/pinch.py
@@ -1,3 +1,5 @@
+import argparse
+import configparser

 import filecmp
 import functools
 import hashlib
 import filecmp
 import functools
 import hashlib


@@ -17,7 +19,6 @@ from typing import (
     Iterable,
     List,
     NewType,
     Iterable,
     List,
     NewType,

-    Sequence,

     Tuple,
 )
 
     Tuple,
 )
 


@@ -25,19 +26,24 @@ Digest16 = NewType('Digest16', str)
 Digest32 = NewType('Digest32', str)
 
 
 Digest32 = NewType('Digest32', str)
 
 

-class InfoTableEntry(types.SimpleNamespace):
+class ChannelTableEntry(types.SimpleNamespace):
+    absolute_url: str

     digest: Digest16
     file: str
     size: int
     url: str
 
 
     digest: Digest16
     file: str
     size: int
     url: str
 
 

-class Info(types.SimpleNamespace):
+class Channel(types.SimpleNamespace):

     channel_html: bytes
     channel_html: bytes

+    channel_url: str

     forwarded_url: str
     forwarded_url: str

+    git_ref: str
+    git_repo: str

     git_revision: str
     git_revision: str

-    table: Dict[str, InfoTableEntry]
-    url: str
+    old_git_revision: str
+    release_name: str
+    table: Dict[str, ChannelTableEntry]

 
 
 class VerificationError(Exception):
 
 
 class VerificationError(Exception):


@@ -75,10 +81,7 @@ class Verification:
         self.result(True)
 
 
         self.result(True)
 
 

-def compare(a: str,
-            b: str) -> Tuple[Sequence[str],
-                             Sequence[str],
-                             Sequence[str]]:
+def compare(a: str, b: str) -> Tuple[List[str], List[str], List[str]]:

 
     def throw(error: OSError) -> None:
         raise error
 
     def throw(error: OSError) -> None:
         raise error


@@ -106,42 +109,52 @@ def compare(a: str,
     return filecmp.cmpfiles(a, b, files, shallow=False)
 
 
     return filecmp.cmpfiles(a, b, files, shallow=False)
 
 

-def fetch(v: Verification, channel_url: str) -> Info:
-    info = Info()
-    info.url = channel_url
+def fetch(v: Verification, channel: Channel) -> None:

     v.status('Fetching channel')
     v.status('Fetching channel')

-    request = urllib.request.urlopen(
-        'https://channels.nixos.org/nixos-20.03', timeout=10)
-    info.channel_html = request.read()
-    info.forwarded_url = request.geturl()
+    request = urllib.request.urlopen(channel.channel_url, timeout=10)
+    channel.channel_html = request.read()
+    channel.forwarded_url = request.geturl()

     v.result(request.status == 200)
     v.result(request.status == 200)

-    v.check('Got forwarded', info.url != info.forwarded_url)
-    return info
+    v.check('Got forwarded', channel.channel_url != channel.forwarded_url)

 
 
 
 

-def parse_table(v: Verification, info: Info) -> None:
+def parse_channel(v: Verification, channel: Channel) -> None:

     v.status('Parsing channel description as XML')
     v.status('Parsing channel description as XML')

-    d = xml.dom.minidom.parseString(info.channel_html)
+    d = xml.dom.minidom.parseString(channel.channel_html)

     v.ok()
 
     v.ok()
 

-    v.status('Extracting git commit')
+    v.status('Extracting release name:')
+    title_name = d.getElementsByTagName(
+        'title')[0].firstChild.nodeValue.split()[2]
+    h1_name = d.getElementsByTagName('h1')[0].firstChild.nodeValue.split()[2]
+    v.status(title_name)
+    v.result(title_name == h1_name)
+    channel.release_name = title_name
+
+    v.status('Extracting git commit:')

     git_commit_node = d.getElementsByTagName('tt')[0]
     git_commit_node = d.getElementsByTagName('tt')[0]

-    info.git_commit = git_commit_node.firstChild.nodeValue
+    channel.git_revision = git_commit_node.firstChild.nodeValue
+    v.status(channel.git_revision)

     v.ok()
     v.status('Verifying git commit label')
     v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ')
 
     v.status('Parsing table')
     v.ok()
     v.status('Verifying git commit label')
     v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ')
 
     v.status('Parsing table')

-    info.table = {}
+    channel.table = {}

     for row in d.getElementsByTagName('tr')[1:]:
         name = row.childNodes[0].firstChild.firstChild.nodeValue
         url = row.childNodes[0].firstChild.getAttribute('href')
         size = int(row.childNodes[1].firstChild.nodeValue)
         digest = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue)
     for row in d.getElementsByTagName('tr')[1:]:
         name = row.childNodes[0].firstChild.firstChild.nodeValue
         url = row.childNodes[0].firstChild.getAttribute('href')
         size = int(row.childNodes[1].firstChild.nodeValue)
         digest = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue)

-        info.table[name] = InfoTableEntry(url=url, digest=digest, size=size)
+        channel.table[name] = ChannelTableEntry(
+            url=url, digest=digest, size=size)

     v.ok()
 
 
     v.ok()
 
 

+def digest_string(s: bytes) -> Digest16:
+    return Digest16(hashlib.sha256(s).hexdigest())
+
+

 def digest_file(filename: str) -> Digest16:
     hasher = hashlib.sha256()
     with open(filename, 'rb') as f:
 def digest_file(filename: str) -> Digest16:
     hasher = hashlib.sha256()
     with open(filename, 'rb') as f:


@@ -185,33 +198,269 @@ def fetch_with_nix_prefetch_url(
     return path
 
 
     return path
 
 

-def fetch_resources(v: Verification, info: Info) -> None:
+def fetch_resources(v: Verification, channel: Channel) -> None:

     for resource in ['git-revision', 'nixexprs.tar.xz']:
     for resource in ['git-revision', 'nixexprs.tar.xz']:

-        fields = info.table[resource]
-        url = urllib.parse.urljoin(info.forwarded_url, fields.url)
-        fields.file = fetch_with_nix_prefetch_url(v, url, fields.digest)
+        fields = channel.table[resource]
+        fields.absolute_url = urllib.parse.urljoin(
+            channel.forwarded_url, fields.url)
+        fields.file = fetch_with_nix_prefetch_url(
+            v, fields.absolute_url, fields.digest)

     v.status('Verifying git commit on main page matches git commit in table')
     v.result(
         open(
     v.status('Verifying git commit on main page matches git commit in table')
     v.result(
         open(

-            info.table['git-revision'].file).read(999) == info.git_commit)
+            channel.table['git-revision'].file).read(999) == channel.git_revision)
+
+
+def git_cachedir(git_repo: str) -> str:
+    # TODO: Consider using pyxdg to find this path.
+    return os.path.expanduser(
+        '~/.cache/nix-pin-channel/git/%s' %
+        digest_string(
+            git_repo.encode()))
+
+
+def verify_git_ancestry(v: Verification, channel: Channel) -> None:
+    cachedir = git_cachedir(channel.git_repo)
+    v.status('Verifying rev is an ancestor of ref')
+    process = subprocess.run(['git',
+                              '-C',
+                              cachedir,
+                              'merge-base',
+                              '--is-ancestor',
+                              channel.git_revision,
+                              channel.git_ref])
+    v.result(process.returncode == 0)

 
 

+    if hasattr(channel, 'old_git_revision'):
+        v.status(
+            'Verifying rev is an ancestor of previous rev %s' %
+            channel.old_git_revision)
+        process = subprocess.run(['git',
+                                  '-C',
+                                  cachedir,
+                                  'merge-base',
+                                  '--is-ancestor',
+                                  channel.old_git_revision,
+                                  channel.git_revision])
+        v.result(process.returncode == 0)
+
+
+def git_fetch(v: Verification, channel: Channel) -> None:
+    # It would be nice if we could share the nix git cache, but as of the time
+    # of writing it is transitioning from gitv2 (deprecated) to gitv3 (not ready
+    # yet), and trying to straddle them both is too far into nix implementation
+    # details for my comfort.  So we re-implement here half of nix.fetchGit.
+    # :(
+
+    cachedir = git_cachedir(channel.git_repo)
+    if not os.path.exists(cachedir):
+        v.status("Initializing git repo")
+        process = subprocess.run(
+            ['git', 'init', '--bare', cachedir])
+        v.result(process.returncode == 0)
+
+    v.status('Fetching ref "%s" from %s' % (channel.git_ref, channel.git_repo))
+    # We don't use --force here because we want to abort and freak out if forced
+    # updates are happening.
+    process = subprocess.run(['git',
+                              '-C',
+                              cachedir,
+                              'fetch',
+                              channel.git_repo,
+                              '%s:%s' % (channel.git_ref,
+                                         channel.git_ref)])
+    v.result(process.returncode == 0)

 
 

-def extract_channel(v: Verification, info: Info) -> None:
-    with tempfile.TemporaryDirectory() as d:
-        v.status('Extracting nixexprs.tar.xz')
-        shutil.unpack_archive(info.table['nixexprs.tar.xz'].file, d)
-        v.ok()
-        v.status('Removing temporary directory')
+    if hasattr(channel, 'git_revision'):
+        v.status('Verifying that fetch retrieved this rev')
+        process = subprocess.run(
+            ['git', '-C', cachedir, 'cat-file', '-e', channel.git_revision])
+        v.result(process.returncode == 0)
+    else:
+        channel.git_revision = open(
+            os.path.join(
+                cachedir,
+                'refs',
+                'heads',
+                channel.git_ref)).read(999).strip()
+
+    verify_git_ancestry(v, channel)
+
+
+def ensure_git_rev_available(v: Verification, channel: Channel) -> None:
+    cachedir = git_cachedir(channel.git_repo)
+    if os.path.exists(cachedir):
+        v.status('Checking if we already have this rev:')
+        process = subprocess.run(
+            ['git', '-C', cachedir, 'cat-file', '-e', channel.git_revision])
+        if process.returncode == 0:
+            v.status('yes')
+        if process.returncode == 1:
+            v.status('no')
+        v.result(process.returncode == 0 or process.returncode == 1)
+        if process.returncode == 0:
+            verify_git_ancestry(v, channel)
+            return
+    git_fetch(v, channel)
+
+
+def compare_tarball_and_git(
+        v: Verification,
+        channel: Channel,
+        channel_contents: str,
+        git_contents: str) -> None:
+    v.status('Comparing channel tarball with git checkout')
+    match, mismatch, errors = compare(os.path.join(
+        channel_contents, channel.release_name), git_contents)
+    v.ok()
+    v.check('%d files match' % len(match), len(match) > 0)
+    v.check('%d files differ' % len(mismatch), len(mismatch) == 0)
+    expected_errors = [
+        '.git-revision',
+        '.version-suffix',
+        'nixpkgs',
+        'programs.sqlite',
+        'svn-revision']
+    benign_errors = []
+    for ee in expected_errors:
+        if ee in errors:
+            errors.remove(ee)
+            benign_errors.append(ee)
+    v.check(
+        '%d unexpected incomparable files' %
+        len(errors),
+        len(errors) == 0)
+    v.check(
+        '(%d of %d expected incomparable files)' %
+        (len(benign_errors),
+         len(expected_errors)),
+        len(benign_errors) == len(expected_errors))
+
+
+def extract_tarball(v: Verification, channel: Channel, dest: str) -> None:
+    v.status('Extracting tarball %s' %
+             channel.table['nixexprs.tar.xz'].file)
+    shutil.unpack_archive(
+        channel.table['nixexprs.tar.xz'].file,
+        dest)

     v.ok()
 
 
     v.ok()
 
 

-def main() -> None:
+def git_checkout(v: Verification, channel: Channel, dest: str) -> None:
+    v.status('Checking out corresponding git revision')
+    git = subprocess.Popen(['git',
+                            '-C',
+                            git_cachedir(channel.git_repo),
+                            'archive',
+                            channel.git_revision],
+                           stdout=subprocess.PIPE)
+    tar = subprocess.Popen(
+        ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout)
+    git.stdout.close()
+    tar.wait()
+    git.wait()
+    v.result(git.returncode == 0 and tar.returncode == 0)
+
+
+def check_channel_metadata(
+        v: Verification,
+        channel: Channel,
+        channel_contents: str) -> None:
+    v.status('Verifying git commit in channel tarball')
+    v.result(
+        open(
+            os.path.join(
+                channel_contents,
+                channel.release_name,
+                '.git-revision')).read(999) == channel.git_revision)
+
+    v.status(
+        'Verifying version-suffix is a suffix of release name %s:' %
+        channel.release_name)
+    version_suffix = open(
+        os.path.join(
+            channel_contents,
+            channel.release_name,
+            '.version-suffix')).read(999)
+    v.status(version_suffix)
+    v.result(channel.release_name.endswith(version_suffix))
+
+
+def check_channel_contents(v: Verification, channel: Channel) -> None:
+    with tempfile.TemporaryDirectory() as channel_contents, \
+            tempfile.TemporaryDirectory() as git_contents:
+
+        extract_tarball(v, channel, channel_contents)
+        check_channel_metadata(v, channel, channel_contents)
+
+        git_checkout(v, channel, git_contents)
+
+        compare_tarball_and_git(v, channel, channel_contents, git_contents)
+
+        v.status('Removing temporary directories')
+    v.ok()
+
+
+def pin_channel(v: Verification, channel: Channel) -> None:
+    fetch(v, channel)
+    parse_channel(v, channel)
+    fetch_resources(v, channel)
+    ensure_git_rev_available(v, channel)
+    check_channel_contents(v, channel)
+
+
+def git_revision_name(v: Verification, channel: Channel) -> str:
+    v.status('Getting commit date')
+    process = subprocess.run(['git',
+                              '-C',
+                              git_cachedir(channel.git_repo),
+                              'lo',
+                              '-n1',
+                              '--format=%ct-%h',
+                              '--abbrev=11',
+                              channel.git_revision],
+                             capture_output=True)
+    v.result(process.returncode == 0 and process.stdout != '')
+    return '%s-%s' % (os.path.basename(channel.git_repo),
+                      process.stdout.decode().strip())
+
+
+def make_channel(conf: configparser.SectionProxy) -> Channel:
+    channel = Channel(**dict(conf.items()))
+    if hasattr(channel, 'git_revision'):
+        channel.old_git_revision = channel.git_revision
+        del channel.git_revision
+    return channel
+
+
+def pin(args: argparse.Namespace) -> None:

     v = Verification()
     v = Verification()

-    info = fetch(v, 'https://channels.nixos.org/nixos-20.03')
-    parse_table(v, info)
-    fetch_resources(v, info)
-    extract_channel(v, info)
-    print(info)
+    config = configparser.ConfigParser()
+    config.read_file(open(args.channels_file), args.channels_file)
+    for section in config.sections():
+        channel = make_channel(config[section])
+        if 'channel_url' in config[section]:
+            pin_channel(v, channel)
+            config[section]['name'] = channel.release_name
+            config[section]['tarball_url'] = channel.table['nixexprs.tar.xz'].absolute_url
+            config[section]['tarball_sha256'] = channel.table['nixexprs.tar.xz'].digest
+        else:
+            git_fetch(v, channel)
+            config[section]['name'] = git_revision_name(v, channel)
+        config[section]['git_revision'] = channel.git_revision
+
+    with open(args.channels_file, 'w') as configfile:
+        config.write(configfile)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(prog='pinch')
+    subparsers = parser.add_subparsers(dest='mode', required=True)
+    parser_pin = subparsers.add_parser('pin')
+    parser_pin.add_argument('channels_file', type=str)
+    parser_pin.set_defaults(func=pin)
+    args = parser.parse_args()
+    args.func(args)

 
 
 main()
 
 
 main()