Rework list_builder.py

FAAAAAASTEEEEEEER

app_caches.py

@@ -76,7 +76,7 @@ def __app_cache_clone_or_update_mapped(data):
 def apps_cache_update_all(apps: dict[str, dict[str, Any]], parallel: int = 8) -> None:
     with Pool(processes=parallel) as pool:
         tasks = pool.imap_unordered(__app_cache_clone_or_update_mapped, apps.items())
-        for _ in tqdm.tqdm(tasks, total=len(apps.keys())):
+        for _ in tqdm.tqdm(tasks, total=len(apps.keys()), ascii=" ·#"):
             pass
 
 

appslib/xmpplogger.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python3
+
+import subprocess
+from shutil import which
+import logging
+import logging.handlers
+
+
+class XmppLogHandler(logging.Handler):
+    def __init__(self):
+        logging.Handler.__init__(self)
+        self.is_logging = False
+
+    def emit(self, record):
+        if which("sendxmpppy") is None:
+            return
+
+        msg = f"[Applist builder error] {record.msg}"
+        subprocess.call(["sendxmpppy", msg], stdout=subprocess.DEVNULL)
+
+    @classmethod
+    def add(cls, level=logging.ERROR):
+        if not logging.getLogger().handlers:
+            logging.basicConfig()
+
+        # create handler
+        handler = cls()
+        handler.setLevel(level)
+        # add the handler
+        logging.getLogger().handlers.append(handler)
+
+
+XmppLogHandler.add(logging.ERROR)

list_builder.py

@@ -3,27 +3,34 @@
 import copy
 import json
 import os
-import re
 import subprocess
-import sys
+import multiprocessing
-import time
-from collections import OrderedDict
 from pathlib import Path
-from shutil import which
+import time
-from typing import Any, Generator, TextIO
+import shutil
+from collections import OrderedDict
 
+import tqdm
+import logging
 import toml
 from git import Repo
 
+from app_caches import apps_cache_update_all, app_cache_folder  # pylint: disable=import-error
 from packaging_v2.convert_v1_manifest_to_v2_for_catalog import \
     convert_v1_manifest_to_v2_for_catalog  # pylint: disable=import-error
 
+from appslib.utils import (REPO_APPS_ROOT,  # pylint: disable=import-error
+                           get_catalog, git_repo_age)
+
+# Automatically enables error-to-xmpp
+import appslib.xmpplogger  # pylint: disable=import-error
+
+
 now = time.time()
 
-REPO_APPS_PATH = Path(__file__).parent.parent
 
 # Load categories and reformat the structure to have a list with an "id" key
-categories = toml.load((REPO_APPS_PATH / "categories.toml").open("r", encoding="utf-8"))
+categories = toml.load((REPO_APPS_ROOT / "categories.toml").open("r", encoding="utf-8"))
 for category_id, infos in categories.items():
     infos["id"] = category_id
     for subtag_id, subtag_infos in infos.get("subtags", {}).items():
@@ -33,13 +40,13 @@ for category_id, infos in categories.items():
 categories = list(categories.values())
 
 # (Same for antifeatures)
-antifeatures = toml.load((REPO_APPS_PATH / "antifeatures.toml").open("r", encoding="utf-8"))
+antifeatures = toml.load((REPO_APPS_ROOT / "antifeatures.toml").open("r", encoding="utf-8"))
 for antifeature_id, infos in antifeatures.items():
     infos["id"] = antifeature_id
 antifeatures = list(antifeatures.values())
 
 # Load the app catalog and filter out the non-working ones
-catalog = toml.load((REPO_APPS_PATH / "apps.toml").open("r", encoding="utf-8"))
+catalog = toml.load((REPO_APPS_ROOT / "apps.toml").open("r", encoding="utf-8"))
 catalog = {
     app: infos for app, infos in catalog.items() if infos.get("state") != "notworking"
 }
@@ -47,165 +54,55 @@ catalog = {
 my_env = os.environ.copy()
 my_env["GIT_TERMINAL_PROMPT"] = "0"
 
-(REPO_APPS_PATH / ".apps_cache").mkdir(exist_ok=True)
+(REPO_APPS_ROOT / "builds").mkdir(exist_ok=True)
-(REPO_APPS_PATH / "builds").mkdir(exist_ok=True)
-
-
-def error(msg: str) -> None:
-    msg = "[Applist builder error] " + msg
-    if which("sendxmpppy") is not None:
-        subprocess.call(["sendxmpppy", msg], stdout=open(os.devnull, "wb"))
-    print(msg + "\n")
-
-
-# Progress bar helper, stolen from https://stackoverflow.com/a/34482761
-def progressbar(it: list[Any], prefix: str = "", size: int = 60, file: TextIO = sys.stdout
-                ) -> Generator[Any, None, None]:
-    count = len(it)
-
-    def show(j, name=""):
-        name += "          "
-        x = int(size * j / count)
-        file.write(
-            "%s[%s%s] %i/%i %s\r" % (prefix, "#" * x, "." * (size - x), j, count, name)
-        )
-        file.flush()
-
-    show(0)
-    for i, item in enumerate(it):
-        yield item
-        show(i + 1, item[0])
-    file.write("\n")
-    file.flush()
-
-
-###################################
-# App git clones cache management #
-###################################
-
-
-def app_cache_folder(app: str) -> Path:
-    return REPO_APPS_PATH / ".apps_cache" / app
-
-
-def refresh_all_caches() -> None:
-    for app, infos in progressbar(sorted(catalog.items()), "Updating git clones: ", 40):
-        app = app.lower()
-        if not app_cache_folder(app).exists():
-            try:
-                init_cache(app, infos)
-            except Exception as e:
-                error("Failed to init cache for %s" % app)
-        else:
-            try:
-                refresh_cache(app, infos)
-            except Exception as e:
-                error("Failed to not refresh cache for %s: %s" % (app, e))
-                raise e
-
-
-def init_cache(app: str, infos: dict[str, str]) -> None:
-    git_depths = {
-        "notworking": 5,
-        "inprogress": 20,
-        "default": 40,
-    }
-
-    Repo.clone_from(
-        infos["url"],
-        to_path=app_cache_folder(app),
-        depth=git_depths.get(infos["state"], git_depths["default"]),
-        single_branch=True, branch=infos.get("branch", "master"),
-    )
-
-
-def git_repo_age(path: Path) -> bool | int:
-    fetch_head = path / ".git" / "FETCH_HEAD"
-    if fetch_head.exists():
-        return int(time.time() - fetch_head.stat().st_mtime)
-    return False
-
-
-def refresh_cache(app: str, infos: dict[str, str]) -> None:
-    app_path = app_cache_folder(app)
-
-    # Don't refresh if already refreshed during last hour
-    age = git_repo_age(app_path)
-    if age is not False and age < 3600:
-        return
-
-    try:
-        repo = Repo(app_path)
-
-        repo.remote("origin").set_url(infos["url"])
-
-        branch = infos.get("branch", "master")
-        if repo.active_branch != branch:
-            all_branches = [str(b) for b in repo.branches]
-            if branch in all_branches:
-                repo.git.checkout(branch, "--force")
-            else:
-                repo.git.remote("set-branches", "--add", "origin", branch)
-                repo.remote("origin").fetch(f"{branch}:{branch}")
-
-        repo.remote("origin").fetch(refspec=branch, force=True)
-        repo.git.reset("--hard", f"origin/{branch}")
-    except:
-        # Sometimes there are tmp issue such that the refresh cache ..
-        # we don't trigger an error unless the cache hasnt been updated since more than 24 hours
-        age = git_repo_age(app_path)
-        if age is not False and age < 24 * 3600:
-            pass
-        else:
-            raise
 
 
 ################################
 # Actual list build management #
 ################################
 
+def __build_app_dict(data):
+    name, info = data
+    try:
+        return name, build_app_dict(name, info)
+    except Exception as err:
+        logging.error("Error while updating %s: %s", name, err)
 
-def build_catalog():
 
+def build_base_catalog():
     result_dict = {}
 
-    for app, infos in progressbar(sorted(catalog.items()), "Processing: ", 40):
+    with multiprocessing.Pool(processes=multiprocessing.cpu_count()) as pool:
+        tasks = pool.imap(__build_app_dict, catalog.items())
 
-        app = app.lower()
+        for result in tqdm.tqdm(tasks, total=len(catalog.keys()), ascii=" ·#"):
+            assert result is not None
+            name, info = result
+            result_dict[name] = info
 
-        try:
+    return result_dict
-            app_dict = build_app_dict(app, infos)
-        except Exception as e:
-            error("Processing %s failed: %s" % (app, str(e)))
-            continue
 
-        result_dict[app_dict["id"]] = app_dict
 
-    #############################
+def write_catalog_v2(base_catalog, target_dir: Path) -> None:
-    # Current catalog API v2    #
+    result_dict_with_manifest_v1 = copy.deepcopy(base_catalog)
-    #############################
+    result_dict_with_manifest_v1 = {
+        name: infos
+        for name, infos in result_dict_with_manifest_v1.items()
+        if float(str(infos["manifest"].get("packaging_format", "")).strip() or "0") < 2
+    }
+    full_catalog = {
+        "apps": result_dict_with_manifest_v1,
+        "categories": categories,
+        "antifeatures": antifeatures,
+    }
 
-    result_dict_with_manifest_v1 = copy.deepcopy(result_dict)
+    target_file = target_dir / "apps.json"
-    result_dict_with_manifest_v1 = {name: infos for name, infos in result_dict_with_manifest_v1.items() if float(str(infos["manifest"].get("packaging_format", "")).strip() or "0") < 2}
+    target_file.parent.mkdir(parents=True, exist_ok=True)
+    target_file.open("w", encoding="utf-8").write(json.dumps(full_catalog, sort_keys=True))
 
-    os.system("mkdir -p ./builds/default/v2/")
-    with open("builds/default/v2/apps.json", "w") as f:
-        f.write(
-            json.dumps(
-                {
-                    "apps": result_dict_with_manifest_v1,
-                    "categories": categories,
-                    "antifeatures": antifeatures,
-                },
-                sort_keys=True,
-            )
-        )
 
-    #############################################
+def write_catalog_v3(base_catalog, target_dir: Path) -> None:
-    # Catalog catalog API v3 (with manifest v2) #
+    result_dict_with_manifest_v2 = copy.deepcopy(base_catalog)
-    #############################################
-
-    result_dict_with_manifest_v2 = copy.deepcopy(result_dict)
     for app in result_dict_with_manifest_v2.values():
         packaging_format = float(str(app["manifest"].get("packaging_format", "")).strip() or "0")
         if packaging_format < 2:
@@ -218,34 +115,31 @@ def build_catalog():
         if "manifest" in app and "resources" in app["manifest"]:
             del app["manifest"]["resources"]
 
+    logos_dir = target_dir / "logos"
+    logos_dir.mkdir(parents=True, exist_ok=True)
     for appid, app in result_dict_with_manifest_v2.items():
         appid = appid.lower()
-        if (REPO_APPS_PATH / "logos" / f"{appid}.png").exists():
+        logo_source = REPO_APPS_ROOT / "logos" / f"{appid}.png"
-            logo_hash = subprocess.check_output(["sha256sum", f"logos/{appid}.png"]).strip().decode("utf-8").split()[0]
+        if logo_source.exists():
-            os.system(f"cp logos/{appid}.png builds/default/v3/logos/{logo_hash}.png")
+            logo_hash = subprocess.check_output(["sha256sum", logo_source]).strip().decode("utf-8").split()[0]
+            shutil.copyfile(logo_source, logos_dir / f"{logo_hash}.png")
             # FIXME: implement something to cleanup old logo stuf in the builds/.../logos/ folder somehow
         else:
             logo_hash = None
         app["logo_hash"] = logo_hash
 
-    os.system("mkdir -p ./builds/default/v3/")
+    full_catalog = {
-    with open("builds/default/v3/apps.json", "w") as f:
+        "apps": result_dict_with_manifest_v2,
-        f.write(
+        "categories": categories,
-            json.dumps(
+        "antifeatures": antifeatures,
-                {
+    }
-                    "apps": result_dict_with_manifest_v2,
-                    "categories": categories,
-                    "antifeatures": antifeatures,
-                },
-                sort_keys=True,
-            )
-        )
 
-    ##############################
+    target_file = target_dir / "apps.json"
-    # Version for catalog in doc #
+    target_file.parent.mkdir(parents=True, exist_ok=True)
-    ##############################
+    target_file.open("w", encoding="utf-8").write(json.dumps(full_catalog, sort_keys=True))
-    os.system("mkdir -p ./builds/default/doc_catalog")
 
+
+def write_catalog_doc(base_catalog, target_dir: Path) -> None:
     def infos_for_doc_catalog(infos):
         level = infos.get("level")
         if not isinstance(level, int):
@@ -267,31 +161,40 @@ def build_catalog():
 
     result_dict_doc = {
         k: infos_for_doc_catalog(v)
-        for k, v in result_dict.items()
+        for k, v in base_catalog.items()
         if v["state"] == "working"
     }
-    with open("builds/default/doc_catalog/apps.json", "w") as f:
+    full_catalog = {
-        f.write(
+        "apps": result_dict_doc,
-            json.dumps(
+        "categories": categories
-                {"apps": result_dict_doc, "categories": categories}, sort_keys=True
+    }
-            )
+
-        )
+    target_file = target_dir / "apps.json"
+    target_file.parent.mkdir(parents=True, exist_ok=True)
+    target_file.open("w", encoding="utf-8").write(json.dumps(full_catalog, sort_keys=True))
 
 
 def build_app_dict(app, infos):
-
     # Make sure we have some cache
     this_app_cache = app_cache_folder(app)
-    assert this_app_cache.exists(), "No cache yet for %s" % app
+    assert this_app_cache.exists(), f"No cache yet for {app}"
 
     repo = Repo(this_app_cache)
 
-    commit_timestamps_for_this_app_in_catalog = \
+    commits_in_apps_json = Repo(REPO_APPS_ROOT).git.log(
-        repo.git.log("-G", f"cinny", "--first-parent", "--reverse", "--date=unix",
+            "-S", f"\"{app}\"", "--first-parent", "--reverse", "--date=unix",
-                     "--format=%cd", "--", "apps.json", "apps.toml")
+            "--format=%cd", "--", "apps.json").split("\n")
+    if len(commits_in_apps_json) > 1:
+        first_commit = commits_in_apps_json[0]
+    else:
+        commits_in_apps_toml = Repo(REPO_APPS_ROOT).git.log(
+                "-S", f"[{app}]", "--first-parent", "--reverse", "--date=unix",
+                "--format=%cd", "--", "apps.json", "apps.toml").split("\n")
+        first_commit = commits_in_apps_toml[0]
 
     # Assume the first entry we get (= the oldest) is the time the app was added
-    infos["added_in_catalog"] = int(commit_timestamps_for_this_app_in_catalog.split("\n")[0])
+    infos["added_in_catalog"] = int(first_commit)
+    # int(commit_timestamps_for_this_app_in_catalog.split("\n")[0])
 
     infos["branch"] = infos.get("branch", "master")
     infos["revision"] = infos.get("revision", "HEAD")
@@ -338,7 +241,7 @@ def build_app_dict(app, infos):
         "manifest": manifest,
         "state": infos["state"],
         "level": infos.get("level", "?"),
-        "maintained": not 'package-not-maintained' in infos.get('antifeatures', []),
+        "maintained": 'package-not-maintained' not in infos.get('antifeatures', []),
         "high_quality": infos.get("high_quality", False),
         "featured": infos.get("featured", False),
         "category": infos.get("category", None),
@@ -351,5 +254,9 @@ def build_app_dict(app, infos):
 
 
 if __name__ == "__main__":
-    refresh_all_caches()
+    apps_cache_update_all(get_catalog(), parallel=50)
-    build_catalog()
+
+    catalog = build_base_catalog()
+    write_catalog_v2(catalog, REPO_APPS_ROOT / "builds" / "default" / "v2")
+    write_catalog_v3(catalog, REPO_APPS_ROOT / "builds" / "default" / "v3")
+    write_catalog_doc(catalog, REPO_APPS_ROOT / "builds" / "default" / "doc_catalog")