From e5cd3e1e02607b2e8499bb4e343fe810afe46b1c Mon Sep 17 00:00:00 2001
From: Kevin Morris <kevr@0cost.org>
Date: Tue, 2 Nov 2021 00:33:34 -0700
Subject: [PATCH] fix(mkpkglists): add all RPC type=search fields to
packages-meta-v1
The SQL logic in this file for package metadata now exactly
reflects RPC's search logic, without searching for specific
packages.
Two command line arguments are available:
--extended | Include License, Keywords and Groups fields
--all | Include License, Keywords, Groups, dependencies and
relations
When one of these arguments are passed, the script will create
a packages-meta-ext-v1.json.gz, configured via packagesmetaextfile.
Done this way so that there are multiple options for archive;
the roughly 2.4MB basic metadata (-meta-v1) and the
~4-7MB (--extended) / 9.8MB (--all) metadata (-meta-ext-v1).
Archive JSON is in the following format: line-separated package objects
enclosed in a list:
[
{...},
{...},
{...}
]
This allows users to stream the archive and process packages
line-by-line, during the local storage process.
Signed-off-by: Kevin Morris <kevr@0cost.org>
---
.gitlab-ci.yml | 2 +-
INSTALL | 2 +-
aurweb/scripts/mkpkglists.py | 241 +++++++++++++++++++++++++++++------
conf/config.defaults | 1 +
test/setup.sh | 2 +
5 files changed, 209 insertions(+), 39 deletions(-)
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index aff18a839..ce3740827 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -12,7 +12,7 @@ before_script:
python-pygit2 python-srcinfo python-bleach python-markdown
python-sqlalchemy python-alembic python-pytest python-werkzeug
python-pytest-tap python-fastapi hypercorn nginx python-authlib
- python-itsdangerous python-httpx
+ python-itsdangerous python-httpx python-orjson
test:
script:
diff --git a/INSTALL b/INSTALL
index b161edd2d..b0989f48e 100644
--- a/INSTALL
+++ b/INSTALL
@@ -50,7 +50,7 @@ read the instructions below.
# pacman -S python-mysql-connector python-pygit2 python-srcinfo python-sqlalchemy \
python-bleach python-markdown python-alembic python-jinja \
python-itsdangerous python-authlib python-httpx hypercorn \
- composer
+ composer python-orjson
# python3 setup.py install
4a) Install `composer` dependencies while inside of aurweb's root:
diff --git a/aurweb/scripts/mkpkglists.py b/aurweb/scripts/mkpkglists.py
index 646d08650..16fe752e1 100755
--- a/aurweb/scripts/mkpkglists.py
+++ b/aurweb/scripts/mkpkglists.py
@@ -1,12 +1,35 @@
#!/usr/bin/env python3
+"""
+Produces package, package base and user archives for the AUR
+database.
+
+Archives:
+
+ packages.gz | A line-separated list of package names
+ packages-meta-v1.json | A type=search RPC-formatted JSON dataset
+ packages-meta-ext-v1.json | An --extended or --all archive
+ pkgbase.gz | A line-separated list of package base names
+ users.gz | A line-separated list of user names
+
+This script takes an optional argument: --extended or --all. Based
+on the following, right-hand side fields are added to each item.
+
+ --extended | License, Keywords, Groups
+ --all | License, Keywords, Groups, relations and dependencies
+
+"""
import datetime
import gzip
-import json
import os
+import sys
+from collections import defaultdict
+from decimal import Decimal
from typing import Tuple
+import orjson
+
import aurweb.config
import aurweb.db
@@ -20,6 +43,7 @@ def state_path(archive: str) -> str:
packagesfile = aurweb.config.get('mkpkglists', 'packagesfile')
packagesmetafile = aurweb.config.get('mkpkglists', 'packagesmetafile')
+packagesmetaextfile = aurweb.config.get('mkpkglists', 'packagesmetaextfile')
packages_state = state_path(packagesfile)
pkgbasefile = aurweb.config.get('mkpkglists', 'pkgbasefile')
@@ -53,6 +77,142 @@ def update_state(state: str, update_time: int) -> None:
f.write(str(update_time))
+TYPE_MAP = {
+ "depends": "Depends",
+ "makedepends": "MakeDepends",
+ "checkdepends": "CheckDepends",
+ "optdepends": "OptDepends",
+ "conflicts": "Conflicts",
+ "provides": "Provides",
+ "replaces": "Replaces",
+}
+
+
+def get_extended_dict(query: str):
+ """
+ Produce data in the form in a single bulk SQL query:
+
+ {
+ <integer_package_id>: {
+ "Depends": [...],
+ "Conflicts": [...],
+ "License": [...]
+ }
+ }
+
+ The caller can then use this data to populate a dataset of packages.
+
+ output = produce_base_output_data()
+ data = get_extended_dict(query)
+ for i in range(len(output)):
+ package_id = output[i].get("ID")
+ output[i].update(data.get(package_id))
+ """
+
+ conn = aurweb.db.Connection()
+
+ cursor = conn.execute(query)
+
+ data = defaultdict(lambda: defaultdict(list))
+
+ for result in cursor.fetchall():
+
+ pkgid = result[0]
+ key = TYPE_MAP.get(result[1])
+ output = result[2]
+ if result[3]:
+ output += result[3]
+
+ # In all cases, we have at least an empty License list.
+ if "License" not in data[pkgid]:
+ data[pkgid]["License"] = []
+
+ # In all cases, we have at least an empty Keywords list.
+ if "Keywords" not in data[pkgid]:
+ data[pkgid]["Keywords"] = []
+
+ if "Groups" not in data[pkgid]:
+ data[pkgid]["Groups"] = []
+
+ data[pkgid][key].append(output)
+
+ conn.close()
+ return data
+
+
+def get_limited_extended_fields():
+ # Returns: [ID, Type, Name, Cond]
+ query = """
+ SELECT PackageGroups.PackageID AS ID, 'Groups' AS Type,
+ Groups.Name, '' AS Cond
+ FROM Groups INNER JOIN PackageGroups
+ ON PackageGroups.GroupID = Groups.ID
+ UNION SELECT PackageLicenses.PackageID AS ID, 'License' AS Type,
+ Licenses.Name, '' as Cond
+ FROM Licenses INNER JOIN PackageLicenses
+ ON PackageLicenses.LicenseID = Licenses.ID
+ UNION SELECT Packages.ID AS ID, 'Keywords' AS Type,
+ PackageKeywords.Keyword AS Name, '' as Cond
+ FROM PackageKeywords
+ INNER JOIN Packages ON Packages.PackageBaseID = PackageKeywords.PackageBaseID
+ """
+ return get_extended_dict(query)
+
+
+def get_extended_fields():
+ # Returns: [ID, Type, Name, Cond]
+ query = """
+ SELECT PackageDepends.PackageID AS ID, DependencyTypes.Name AS Type,
+ PackageDepends.DepName AS Name, PackageDepends.DepCondition AS Cond
+ FROM PackageDepends
+ LEFT JOIN DependencyTypes
+ ON DependencyTypes.ID = PackageDepends.DepTypeID
+ UNION SELECT PackageRelations.PackageID AS ID, RelationTypes.Name AS Type,
+ PackageRelations.RelName AS Name,
+ PackageRelations.RelCondition AS Cond
+ FROM PackageRelations
+ LEFT JOIN RelationTypes
+ ON RelationTypes.ID = PackageRelations.RelTypeID
+ UNION SELECT PackageGroups.PackageID AS ID, 'Groups' AS Type,
+ Groups.Name, '' AS Cond
+ FROM Groups
+ INNER JOIN PackageGroups ON PackageGroups.GroupID = Groups.ID
+ UNION SELECT PackageLicenses.PackageID AS ID, 'License' AS Type,
+ Licenses.Name, '' as Cond
+ FROM Licenses
+ INNER JOIN PackageLicenses ON PackageLicenses.LicenseID = Licenses.ID
+ UNION SELECT Packages.ID AS ID, 'Keywords' AS Type,
+ PackageKeywords.Keyword AS Name, '' as Cond
+ FROM PackageKeywords
+ INNER JOIN Packages ON Packages.PackageBaseID = PackageKeywords.PackageBaseID
+ """
+ return get_extended_dict(query)
+
+
+EXTENDED_FIELD_HANDLERS = {
+ "--extended": get_limited_extended_fields,
+ "--all": get_extended_fields
+}
+
+
+def is_decimal(column):
+ """ Check if an SQL column is of decimal.Decimal type. """
+ if isinstance(column, Decimal):
+ return float(column)
+ return column
+
+
+def write_archive(archive: str, output: list):
+ with gzip.open(archive, "wb") as f:
+ f.write(b"[\n")
+ for i, item in enumerate(output):
+ f.write(orjson.dumps(item))
+ if i < len(output) - 1:
+ f.write(b",")
+ f.write(b"\n")
+ f.write(b"]")
+
+
def main():
conn = aurweb.db.Connection()
@@ -64,47 +224,52 @@ def main():
updated, update_time = should_update(packages_state, "Packages")
if not updated:
print("Updating Packages...")
- columns = ("Packages.ID, PackageBaseID, Packages.Name, "
- "Version, Description, URL")
+
+ # Query columns; copied from RPC.
+ columns = ("Packages.ID, Packages.Name, "
+ "PackageBases.ID AS PackageBaseID, "
+ "PackageBases.Name AS PackageBase, "
+ "Version, Description, URL, NumVotes, "
+ "Popularity, OutOfDateTS AS OutOfDate, "
+ "Users.UserName AS Maintainer, "
+ "SubmittedTS AS FirstSubmitted, "
+ "ModifiedTS AS LastModified")
+
+ # Perform query.
cur = conn.execute(f"SELECT {columns} FROM Packages "
- "INNER JOIN PackageBases "
+ "LEFT JOIN PackageBases "
"ON PackageBases.ID = Packages.PackageBaseID "
+ "LEFT JOIN Users "
+ "ON PackageBases.MaintainerUID = Users.ID "
"WHERE PackageBases.PackagerUID IS NOT NULL")
- # Store JSON-data in `output`, which can be reused for the
- # more basic packagesfile generation afterward.
- output = dict()
- with gzip.open(packagesmetafile, "wt") as f:
- """ The output "data" json key points to a list of dictionaries,
- each representing a single result, filled with column names as
- keys and column values as values.
-
- Example:
- {
- "data": [
- {
- "ID": 123,
- "Name": "package_name",
- "PackageBaseID": 234,
- "Version": "0.1.1",
- "Description": "Some description...",
- "URL": "https://some.url"
- },
- ...
- ]
- }
- """
- output = [{
- column[0]: result[i]
+ # Produce packages-meta-v1.json.gz
+ output = list()
+ snapshot_uri = aurweb.config.get("options", "snapshot_uri")
+ for result in cur.fetchall():
+ item = {
+ column[0]: is_decimal(result[i])
for i, column in enumerate(cur.description)
- } for result in cur.fetchall()]
- json.dump({
- "warning": ("This is a experimental! It can be removed "
- "or modified without warning!"),
- "data": output
- }, f)
-
- with gzip.open(packagesfile, "w") as f:
+ }
+ item["URLPath"] = snapshot_uri % item.get("Name")
+ output.append(item)
+
+ write_archive(packagesmetafile, output)
+
+ # Produce packages-meta-ext-v1.json.gz
+ if len(sys.argv) > 1 and sys.argv[1] in EXTENDED_FIELD_HANDLERS:
+ f = EXTENDED_FIELD_HANDLERS.get(sys.argv[1])
+ data = f()
+
+ default_ = {"Groups": [], "License": [], "Keywords": []}
+ for i in range(len(output)):
+ data_ = data.get(output[i].get("ID"), default_)
+ output[i].update(data_)
+
+ write_archive(packagesmetaextfile, output)
+
+ # Produce packages.gz
+ with gzip.open(packagesfile, "wb") as f:
f.write(bytes(pkglist_header + "\n", "UTF-8"))
f.writelines([
bytes(x.get("Name") + "\n", "UTF-8")
@@ -118,6 +283,7 @@ def main():
updated, update_time = should_update(pkgbases_state, "PackageBases")
if not updated:
print("Updating PackageBases...")
+ # Produce pkgbase.gz
with gzip.open(pkgbasefile, "w") as f:
f.write(bytes(pkgbaselist_header + "\n", "UTF-8"))
cur = conn.execute("SELECT Name FROM PackageBases " +
@@ -130,6 +296,7 @@ def main():
updated, update_time = should_update(users_state, "Users")
if not updated:
print("Updating Users...")
+ # Produce users.gz
with gzip.open(userfile, "w") as f:
f.write(bytes(userlist_header + "\n", "UTF-8"))
cur = conn.execute("SELECT UserName FROM Users")
diff --git a/conf/config.defaults b/conf/config.defaults
index 36ea02efb..a04f21bcc 100644
--- a/conf/config.defaults
+++ b/conf/config.defaults
@@ -93,5 +93,6 @@ server = ftp://mirrors.kernel.org/archlinux/%s/os/x86_64
[mkpkglists]
packagesfile = /srv/http/aurweb/web/html/packages.gz
packagesmetafile = /srv/http/aurweb/web/html/packages-meta-v1.json.gz
+packagesmetaextfile = /srv/http/aurweb/web/html/packages-meta-ext-v1.json.gz
pkgbasefile = /srv/http/aurweb/web/html/pkgbase.gz
userfile = /srv/http/aurweb/web/html/users.gz
diff --git a/test/setup.sh b/test/setup.sh
index 24bb5f48a..f74cd1b7d 100644
--- a/test/setup.sh
+++ b/test/setup.sh
@@ -31,6 +31,7 @@ enable-maintenance = 0
maintenance-exceptions = 127.0.0.1
commit_uri = https://aur.archlinux.org/cgit/aur.git/log/?h=%s&id=%s
localedir = $TOPLEVEL/web/locale/
+snapshot_uri = /cgit/aur.git/snapshot/%s.tar.gz
[notifications]
notify-cmd = $NOTIFY
@@ -62,6 +63,7 @@ server = file://$(pwd)/remote/
[mkpkglists]
packagesfile = packages.gz
packagesmetafile = packages-meta-v1.json.gz
+packagesmetaextfile = packages-meta-ext-v1.json.gz
pkgbasefile = pkgbase.gz
userfile = users.gz
EOF
--
GitLab