Add update hook template

This adds a script that can be used as an update hook to check all
commits for validity and to regenerate the package details page before
updating a named ref.

Signed-off-by: Lukas Fleischer <archlinux@cryptocrack.de>

conf/config.proto

@@ -36,4 +36,5 @@ ssh-options = no-port-forwarding,no-X11-forwarding,no-pty
 [serve]
 repo-base = /pub/git/
 repo-regex = [a-z0-9][a-z0-9.+_-]*$
+git-update-hook = /srv/http/aur/scripts/git-integration/git-update.py
 git-shell-cmd = /usr/bin/git-shell

scripts/git-integration/aurinfo.py

@@ -0,0 +1,204 @@
+#!/usr/bin/env python
+
+from copy import copy, deepcopy
+import pprint
+import sys
+
+class Attr(object):
+    def __init__(self, name, is_multivalued=False, allow_arch_extensions=False):
+        self.name = name
+        self.is_multivalued = is_multivalued
+        self.allow_arch_extensions = allow_arch_extensions
+
+PKGBUILD_ATTRIBUTES = {
+    'arch':         Attr('arch',            True),
+    'backup':       Attr('backup',          True),
+    'changelog':    Attr('changelog',       False),
+    'checkdepends': Attr('checkdepends',    True),
+    'conflicts':    Attr('conflicts',       True, True),
+    'depends':      Attr('depends',         True, True),
+    'epoch':        Attr('epoch',           False),
+    'groups':       Attr('groups',          True),
+    'install':      Attr('install',         False),
+    'license':      Attr('license',         True),
+    'makedepends':  Attr('makedepends',     True, True),
+    'md5sums':      Attr('md5sums',         True, True),
+    'noextract':    Attr('noextract',       True),
+    'optdepends':   Attr('optdepends',      True, True),
+    'options':      Attr('options',         True),
+    'pkgname':      Attr('pkgname',         False),
+    'pkgrel':       Attr('pkgrel',          False),
+    'pkgver':       Attr('pkgver',          False),
+    'provides':     Attr('provides',        True, True),
+    'replaces':     Attr('replaces',        True, True),
+    'sha1sums':     Attr('sha1sums',        True, True),
+    'sha224sums':   Attr('sha224sums',      True, True),
+    'sha256sums':   Attr('sha256sums',      True, True),
+    'sha384sums':   Attr('sha384sums',      True, True),
+    'sha512sums':   Attr('sha512sums',      True, True),
+    'source':       Attr('source',          True, True),
+    'url':          Attr('url',             False),
+    'validpgpkeys': Attr('validpgpkeys',    True),
+}
+
+def find_attr(attrname):
+    # exact match
+    attr = PKGBUILD_ATTRIBUTES.get(attrname, None)
+    if attr:
+        return attr
+
+    # prefix match
+    # XXX: this could break in the future if PKGBUILD(5) ever
+    # introduces a key which is a subset of another.
+    for k in PKGBUILD_ATTRIBUTES.keys():
+        if attrname.startswith(k + '_'):
+            return PKGBUILD_ATTRIBUTES[k]
+
+def IsMultiValued(attrname):
+    attr = find_attr(attrname)
+    return attr and attr.is_multivalued
+
+class AurInfo(object):
+    def __init__(self):
+        self._pkgbase = {}
+        self._packages = {}
+
+    def GetPackageNames(self):
+        return self._packages.keys()
+
+    def GetMergedPackage(self, pkgname):
+        package = deepcopy(self._pkgbase)
+        package['pkgname'] = pkgname
+        for k, v in self._packages.get(pkgname).items():
+            package[k] = deepcopy(v)
+        return package
+
+    def AddPackage(self, pkgname):
+        self._packages[pkgname] = {}
+        return self._packages[pkgname]
+
+    def SetPkgbase(self, pkgbasename):
+        self._pkgbase = {'pkgname' : pkgbasename}
+        return self._pkgbase
+
+
+class StderrECatcher(object):
+    def Catch(self, lineno, error):
+        print('ERROR[%d]: %s' % (lineno, error), file=sys.stderr)
+
+
+class CollectionECatcher(object):
+    def __init__(self):
+        self._errors = []
+
+    def Catch(self, lineno, error):
+        self._errors.append((lineno, error))
+
+    def HasErrors(self):
+        return len(self._errors) > 0
+
+    def Errors(self):
+        return copy(self._errors)
+
+
+def ParseAurinfoFromIterable(iterable, ecatcher=None):
+    aurinfo = AurInfo()
+
+    if ecatcher is None:
+        ecatcher = StderrECatcher()
+
+    current_package = None
+    lineno = 0
+
+    for line in iterable:
+        lineno += 1
+
+        if line.startswith('#'):
+            continue
+
+        if not line.strip():
+            # end of package
+            current_package = None
+            continue
+
+        if not line.startswith('\t'):
+            # start of new package
+            try:
+                key, value = map(str.strip, line.split('=', 1))
+            except ValueError:
+                ecatcher.Catch(lineno, 'unexpected header format in section=%s' %
+                    current_package['pkgname'])
+                continue
+
+            if key == 'pkgbase':
+                current_package = aurinfo.SetPkgbase(value)
+            else:
+                current_package = aurinfo.AddPackage(value)
+        else:
+            # package attribute
+            if current_package is None:
+                ecatcher.Catch(lineno, 'package attribute found outside of '
+                               'a package section')
+                continue
+
+            try:
+                key, value = map(str.strip, line.split('=', 1))
+            except ValueError:
+                ecatcher.Catch(lineno, 'unexpected attribute format in '
+                               'section=%s' % current_package['pkgname'])
+
+            if IsMultiValued(key):
+                if not current_package.get(key):
+                    current_package[key] = []
+                if value:
+                    current_package[key].append(value)
+            else:
+                if not current_package.get(key):
+                    current_package[key] = value
+                else:
+                    ecatcher.Catch(lineno, 'overwriting attribute '
+                                   '%s: %s -> %s' % (key, current_package[key],
+                                                     value))
+
+    return aurinfo
+
+
+def ParseAurinfo(filename='.AURINFO', ecatcher=None):
+    with open(filename) as f:
+        return ParseAurinfoFromIterable(f, ecatcher)
+
+
+def ValidateAurinfo(filename='.AURINFO'):
+    ecatcher = CollectionECatcher()
+    ParseAurinfo(filename, ecatcher)
+    errors = ecatcher.Errors()
+    for error in errors:
+        print('error on line %d: %s' % error, file=sys.stderr)
+    return not errors
+
+
+if __name__ == '__main__':
+    pp = pprint.PrettyPrinter(indent=4)
+
+    if len(sys.argv) == 1:
+        print('error: not enough arguments')
+        sys.exit(1)
+    elif len(sys.argv) == 2:
+        action = sys.argv[1]
+        filename = '.AURINFO'
+    else:
+        action, filename = sys.argv[1:3]
+
+    if action == 'parse':
+        aurinfo = ParseAurinfo()
+        for pkgname in aurinfo.GetPackageNames():
+            print(">>> merged package: %s" % pkgname)
+            pp.pprint(aurinfo.GetMergedPackage(pkgname))
+            print()
+    elif action == 'validate':
+        sys.exit(not ValidateAurinfo(filename))
+    else:
+        print('unknown action: %s' % action)
+        sys.exit(1)
+
+# vim: set et ts=4 sw=4:

scripts/git-integration/git-serve.py

@@ -18,6 +18,7 @@ aur_db_pass = config.get('database', 'password')
 
 repo_base_path = config.get('serve', 'repo-base')
 repo_regex = config.get('serve', 'repo-regex')
+git_update_hook = config.get('serve', 'git-update-hook')
 git_shell_cmd = config.get('serve', 'git-shell-cmd')
 
 def repo_path_validate(path):
@@ -60,6 +61,7 @@ def setup_repo(repo, user):
 
     repo_path = repo_base_path + '/' + repo + '.git/'
     pygit2.init_repository(repo_path, True)
+    os.symlink(git_update_hook, repo_path + 'hooks/update')
 
 def check_permissions(pkgbase, user):
     db = mysql.connector.connect(host=aur_db_host, user=aur_db_user,

scripts/git-integration/git-update.py

@@ -0,0 +1,212 @@
+#!/usr/bin/python3
+
+from copy import copy, deepcopy
+import configparser
+import mysql.connector
+import os
+import pygit2
+import re
+import sys
+
+import aurinfo
+
+config = configparser.RawConfigParser()
+config.read(os.path.dirname(os.path.realpath(__file__)) + "/../../conf/config")
+
+aur_db_host = config.get('database', 'host')
+aur_db_name = config.get('database', 'name')
+aur_db_user = config.get('database', 'user')
+aur_db_pass = config.get('database', 'password')
+
+def save_srcinfo(srcinfo, db, cur, user):
+    # Obtain package base ID and previous maintainer.
+    pkgbase = srcinfo._pkgbase['pkgname']
+    cur.execute("SELECT ID, MaintainerUID FROM PackageBases "
+                "WHERE Name = %s", [pkgbase])
+    (pkgbase_id, maintainer_uid) = cur.fetchone()
+    was_orphan = not maintainer_uid
+
+    # Obtain the user ID of the new maintainer.
+    cur.execute("SELECT ID FROM Users WHERE Username = %s", [user])
+    user_id = int(cur.fetchone()[0])
+
+    # Update package base details and delete current packages.
+    cur.execute("UPDATE PackageBases SET ModifiedTS = UNIX_TIMESTAMP(), " +
+                "MaintainerUID = %s, PackagerUID = %s, " +
+                "OutOfDateTS = NULL WHERE ID = %s",
+                [user_id, user_id, pkgbase_id])
+    cur.execute("DELETE FROM Packages WHERE PackageBaseID = %s",
+                [pkgbase_id])
+
+    for pkgname in srcinfo.GetPackageNames():
+        pkginfo = srcinfo.GetMergedPackage(pkgname)
+
+        if 'epoch' in pkginfo and pkginfo['epoch'] > 0:
+            ver = '%d:%s-%s' % (pkginfo['epoch'], pkginfo['pkgver'],
+                                pkginfo['pkgrel'])
+        else:
+            ver = '%s-%s' % (pkginfo['pkgver'], pkginfo['pkgrel'])
+
+        # Create a new package.
+        cur.execute("INSERT INTO Packages (PackageBaseID, Name, " +
+                    "Version, Description, URL) " +
+                    "VALUES (%s, %s, %s, %s, %s)",
+                    [pkgbase_id, pkginfo['pkgname'], ver,
+                     pkginfo['pkgdesc'], pkginfo['url']])
+        db.commit()
+        pkgid = cur.lastrowid
+
+        # Add package sources.
+        for source in pkginfo['source']:
+            cur.execute("INSERT INTO PackageSources (PackageID, Source) " +
+                        "VALUES (%s, %s)", [pkgid, source])
+
+        # Add package dependencies.
+        for deptype in ('depends', 'makedepends',
+                        'checkdepends', 'optdepends'):
+            if not deptype in pkginfo:
+                continue
+            cur.execute("SELECT ID FROM DependencyTypes WHERE Name = %s",
+                        [deptype])
+            deptypeid = cur.fetchone()[0]
+            for dep in pkginfo[deptype]:
+                depname = re.sub(r'(<|=|>).*', '', dep)
+                depcond = dep[len(depname):]
+                cur.execute("INSERT INTO PackageDepends (PackageID, " +
+                            "DepTypeID, DepName, DepCondition) " +
+                            "VALUES (%s, %s, %s, %s)", [pkgid, deptypeid,
+                                                        depname, depcond])
+
+        # Add package relations (conflicts, provides, replaces).
+        for reltype in ('conflicts', 'provides', 'replaces'):
+            if not reltype in pkginfo:
+                continue
+            cur.execute("SELECT ID FROM RelationTypes WHERE Name = %s",
+                        [reltype])
+            reltypeid = cur.fetchone()[0]
+            for rel in pkginfo[reltype]:
+                relname = re.sub(r'(<|=|>).*', '', rel)
+                relcond = rel[len(relname):]
+                cur.execute("INSERT INTO PackageRelations (PackageID, " +
+                            "RelTypeID, RelName, RelCondition) " +
+                            "VALUES (%s, %s, %s, %s)", [pkgid, reltypeid,
+                                                        relname, relcond])
+
+        # Add package licenses.
+        if 'license' in pkginfo:
+            for license in pkginfo['license']:
+                cur.execute("SELECT ID FROM Licenses WHERE Name = %s",
+                            [license])
+                if cur.rowcount == 1:
+                    licenseid = cur.fetchone()[0]
+                else:
+                    cur.execute("INSERT INTO Licenses (Name) VALUES (%s)",
+                                [license])
+                    db.commit()
+                    licenseid = cur.lastrowid
+                cur.execute("INSERT INTO PackageLicenses (PackageID, " +
+                            "LicenseID) VALUES (%s, %s)",
+                            [pkgid, licenseid])
+
+        # Add package groups.
+        if 'groups' in pkginfo:
+            for group in pkginfo['groups']:
+                cur.execute("SELECT ID FROM Groups WHERE Name = %s",
+                            [group])
+                if cur.rowcount == 1:
+                    groupid = cur.fetchone()[0]
+                else:
+                    cur.execute("INSERT INTO Groups (Name) VALUES (%s)",
+                                [group])
+                    db.commit()
+                    groupid = cur.lastrowid
+                cur.execute("INSERT INTO PackageGroups (PackageID, "
+                            "GroupID) VALUES (%s, %s)", [pkgid, groupid])
+
+    # Add user to notification list on adoption.
+    if was_orphan:
+        cur.execute("INSERT INTO CommentNotify (PackageBaseID, UserID) " +
+                    "VALUES (%s, %s)", [pkgbase_id, user_id])
+
+    db.commit()
+
+def die(msg):
+    sys.stderr.write("error: %s\n" % (msg))
+    exit(1)
+
+def die_commit(msg, commit):
+    sys.stderr.write("error: The following error " +
+                     "occurred when parsing commit\n")
+    sys.stderr.write("error: %s:\n" % (commit))
+    sys.stderr.write("error: %s\n" % (msg))
+    exit(1)
+
+if len(sys.argv) != 4:
+    die("invalid arguments")
+
+refname = sys.argv[1]
+sha1_old = sys.argv[2]
+sha1_new = sys.argv[3]
+
+user = os.environ.get("AUR_USER")
+pkgbase = os.environ.get("AUR_PKGBASE")
+git_dir = os.environ.get("AUR_GIT_DIR")
+
+if refname != "refs/heads/master":
+    die("pushing to a branch other than master is restricted")
+
+repo = pygit2.Repository(git_dir)
+walker = repo.walk(sha1_new, pygit2.GIT_SORT_TOPOLOGICAL)
+if sha1_old != "0000000000000000000000000000000000000000":
+    walker.hide(sha1_old)
+
+for commit in walker:
+    if not '.SRCINFO' in commit.tree:
+        die_commit("missing .SRCINFO", commit.id)
+
+    for treeobj in commit.tree:
+        if repo[treeobj.id].size > 100000:
+            die_commit("maximum blob size (100kB) exceeded", commit.id)
+
+    srcinfo_raw = repo[commit.tree['.SRCINFO'].id].data.decode()
+    srcinfo_raw = srcinfo_raw.split('\n')
+    ecatcher = aurinfo.CollectionECatcher()
+    srcinfo = aurinfo.ParseAurinfoFromIterable(srcinfo_raw, ecatcher)
+    errors = ecatcher.Errors()
+    if errors:
+        sys.stderr.write("error: The following errors occurred "
+                         "when parsing .SRCINFO in commit\n")
+        sys.stderr.write("error: %s:\n" % (commit.id))
+        for error in errors:
+            sys.stderr.write("error: line %d: %s\n" % error)
+        exit(1)
+
+    srcinfo_pkgbase = srcinfo._pkgbase['pkgname']
+    if srcinfo_pkgbase != pkgbase:
+        die_commit('invalid pkgbase: %s' % (srcinfo_pkgbase), commit.id)
+
+    for pkgname in srcinfo.GetPackageNames():
+        pkginfo = srcinfo.GetMergedPackage(pkgname)
+
+        if not re.match(r'[a-z0-9][a-z0-9\.+_-]*$', pkginfo['pkgname']):
+            die_commit('invalid package name: %s' % (pkginfo['pkgname']),
+                       commit.id)
+
+        if not re.match(r'(?:http|ftp)s?://.*', pkginfo['url']):
+            die_commit('invalid URL: %s' % (pkginfo['url']), commit.id)
+
+        for field in ('pkgname', 'pkgdesc', 'url'):
+            if len(pkginfo[field]) > 255:
+                die_commit('%s field too long: %s' % (field, pkginfo[field]),
+                           commit.id)
+
+srcinfo_raw = repo[repo[sha1_new].tree['.SRCINFO'].id].data.decode()
+srcinfo_raw = srcinfo_raw.split('\n')
+srcinfo = aurinfo.ParseAurinfoFromIterable(srcinfo_raw)
+
+db = mysql.connector.connect(host=aur_db_host, user=aur_db_user,
+                             passwd=aur_db_pass, db=aur_db_name,
+                             buffered=True)
+cur = db.cursor()
+save_srcinfo(srcinfo, db, cur, user)
+db.close()