From 94d8dcb1985adc7878264ef304e84257afb5d2be Mon Sep 17 00:00:00 2001
From: Alexandre Aubin <alexandre.aubin@cern.ch>
Date: Fri, 20 Mar 2020 18:36:53 +0100
Subject: [PATCH 1/5] Read credentials from .github_credentials instead of
 having to provide them through arguments

---
 list_builder.py | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

diff --git a/list_builder.py b/list_builder.py
index e1fc7ad..077fbe4 100755
--- a/list_builder.py
+++ b/list_builder.py
@@ -20,6 +20,10 @@ re_commit_author = re.compile(
     re.MULTILINE
 )
 
+# GitHub credentials from ./.github_credentials than should contain <user>:<token>
+# For example:      foobar:abcdef1234567890
+github_credentials = open("./.github_credentials").read().strip().split(":")
+
 
 # Helpers
 
@@ -72,7 +76,7 @@ def get_json(url, verify=True):
     try:
         # Retrieve and load manifest
         if ".github" in url:
-            r = requests.get(url, verify=verify, auth=token)
+            r = requests.get(url, verify=verify, auth=github_credentials)
         else:
             r = requests.get(url, verify=verify)
         r.raise_for_status()
@@ -106,7 +110,6 @@ parser = argparse.ArgumentParser(description='Process YunoHost application list.
 # Add arguments and options
 parser.add_argument("input", help="Path to json input file")
 parser.add_argument("-o", "--output", help="Path to result file. If not specified, '-build' suffix will be added to input filename.")
-parser.add_argument("-g", "--github", help="Github token <username>:<password>")
 
 # Parse args
 args = parser.parse_args()
@@ -133,12 +136,6 @@ if os.path.exists(args.output):
     except Exception as e:
         print("Error while trying to load already built file: %s" % e)
 
-# GitHub credentials
-if args.github:
-    token = (args.github.split(':')[0], args.github.split(':')[1])
-else:
-    token = None
-
 # Loop through every apps
 result_dict = {}
 for app, info in apps_list.items():

From 5d9d3686fa080574e583bb4bafc1dae757f4f0e1 Mon Sep 17 00:00:00 2001
From: Alexandre Aubin <alexandre.aubin@cern.ch>
Date: Fri, 20 Mar 2020 18:37:29 +0100
Subject: [PATCH 2/5] Replace should_i_rebuild.sh by a new script that
 corresponds to what we do in the current cron job nowadays

---
 rebuild.sh | 10 ++++++++++
 1 file changed, 10 insertions(+)
 create mode 100644 rebuild.sh

diff --git a/rebuild.sh b/rebuild.sh
new file mode 100644
index 0000000..dfe9bec
--- /dev/null
+++ b/rebuild.sh
@@ -0,0 +1,10 @@
+#!/bin/bash
+
+workdir=$(dirname "$0")
+log=$workdir/app_list_auto_update.log
+
+cd $workdir
+date >> $log
+git pull >/dev/null
+
+python ./list_builder.py &>> $log || sendxmpppy "[listbuilder] Rebuilding the application list failed miserably"

From 2a54e5e9b01525dc07d75866bc5e0bd7636ca3d6 Mon Sep 17 00:00:00 2001
From: Alexandre Aubin <alex.aubin@mailoo.org>
Date: Sat, 4 Apr 2020 17:03:46 +0200
Subject: [PATCH 3/5] We don't need no argument parsin'

---
 list_builder.py | 64 +++++++++++++++++++++----------------------------
 1 file changed, 27 insertions(+), 37 deletions(-)

diff --git a/list_builder.py b/list_builder.py
index 077fbe4..07ec7bf 100755
--- a/list_builder.py
+++ b/list_builder.py
@@ -5,7 +5,6 @@ import sys
 import time
 import json
 import zlib
-import argparse
 import subprocess
 import yaml
 
@@ -15,15 +14,14 @@ from dateutil.parser import parse
 
 # Regular expression patterns
 
-re_commit_author = re.compile(
-    r'^author (?P<name>.+) <(?P<email>.+)> (?P<time>\d+) (?P<tz>[+-]\d+)$',
-    re.MULTILINE
-)
+list_name = "apps.json"
+
+assert os.path.exists(".github_credentials"), "You should add github credentials in .github_credentials first. (Format <user>:<token>)"
+assert os.path.exists(list_name), "File %s doesn't exist ?" % list_name
 
 # GitHub credentials from ./.github_credentials than should contain <user>:<token>
 # For example:      foobar:abcdef1234567890
-github_credentials = open("./.github_credentials").read().strip().split(":")
-
+github_credentials = open(".github_credentials").read().strip().split(":")
 
 # Helpers
 
@@ -104,35 +102,15 @@ def get_zlib(url, verify=True):
 
 # Main
 
-# Create argument parser
-parser = argparse.ArgumentParser(description='Process YunoHost application list.')
-
-# Add arguments and options
-parser.add_argument("input", help="Path to json input file")
-parser.add_argument("-o", "--output", help="Path to result file. If not specified, '-build' suffix will be added to input filename.")
-
-# Parse args
-args = parser.parse_args()
-
-try:
-    # Retrieve apps list from json file
-    with open(args.input) as f:
-        apps_list = json.load(f)
-except IOError as e:
-    fail("%s file not found" % args.input)
-
 # Get list name from filename
-list_name = os.path.splitext(os.path.basename(args.input))[0]
 print(":: Building %s list..." % list_name)
 
-# Args default
-if not args.output:
-    args.output = '%s-build.json' % list_name
+output = '%s-build.json' % list_name
 
 already_built_file = {}
-if os.path.exists(args.output):
+if os.path.exists(output):
     try:
-        already_built_file = json.load(open(args.output))
+        already_built_file = json.load(open(output))
     except Exception as e:
         print("Error while trying to load already built file: %s" % e)
 
@@ -303,6 +281,11 @@ for app, info in apps_list.items():
         else:
             commit = commit[1]
 
+        re_commit_author = re.compile(
+            r'^author (?P<name>.+) <(?P<email>.+)> (?P<time>\d+) (?P<tz>[+-]\d+)$',
+            re.MULTILINE
+        )
+
         # Extract author line and commit date
         commit_author = re_commit_author.search(commit)
         if not commit_author:
@@ -355,19 +338,24 @@ for app, info in apps_list.items():
         error("Invalid app info or manifest for app %s, %s" % (app, e))
         continue
 
-## output version 2, including the categories
+#######################
+## Current version 2 ##
+#######################
 categories = yaml.load(open("categories.yml").read())
-with open(args.output.replace(".json", "-v2.json"), 'w') as f:
+with open(output.replace(".json", "-v2.json"), 'w') as f:
     f.write(json.dumps({"apps": result_dict, "categories": categories}, sort_keys=True))
 
-## output version 1
-with open(args.output, 'w') as f:
+######################
+## Legacy version 1 ##
+######################
+with open(output, 'w') as f:
     f.write(json.dumps(result_dict, sort_keys=True))
 
-print("\nDone! Written in %s" % args.output)
+print("\nDone! Written in %s" % output)
 
-
-## output version 0
+######################
+## Legacy version 0 ##
+######################
 print("\nAlso splitting the file into official and community-build.json for backward compatibility")
 
 official_apps = set(["agendav", "ampache", "baikal", "dokuwiki", "etherpad_mypads", "hextris", "jirafeau", "kanboard", "my_webapp", "nextcloud", "opensondage", "phpmyadmin", "piwigo", "rainloop", "roundcube", "searx", "shellinabox", "strut", "synapse", "transmission", "ttrss", "wallabag2", "wordpress", "zerobin"])
@@ -385,4 +373,6 @@ with open("official-build.json", 'w') as f:
 with open("community-build.json", 'w') as f:
     f.write(json.dumps(community_apps_dict, sort_keys=True))
 
+######################
+
 print("\nDone!")

From 5cbcb98354612c2131d77ec6ad00677bf6f246b2 Mon Sep 17 00:00:00 2001
From: Alexandre Aubin <alex.aubin@mailoo.org>
Date: Sun, 5 Apr 2020 02:54:24 +0200
Subject: [PATCH 4/5] Rewrite list_builder to use a local git clone cache and
 simpler logic

---
 list_builder.py | 542 +++++++++++++++++-------------------------------
 rebuild.sh      |   2 +-
 2 files changed, 187 insertions(+), 357 deletions(-)

diff --git a/list_builder.py b/list_builder.py
index 07ec7bf..9e2a071 100755
--- a/list_builder.py
+++ b/list_builder.py
@@ -1,34 +1,198 @@
-#!/usr/bin/env python2
-import re
+#!/usr/bin/python3
+
 import os
-import sys
-import time
+import re
 import json
-import zlib
 import subprocess
 import yaml
+import time
 
-import requests
-from dateutil.parser import parse
+now = time.time()
+
+catalog = json.load(open("apps.json"))
+
+my_env = os.environ.copy()
+my_env["GIT_TERMINAL_PROMPT"] = "0"
+
+os.makedirs(".apps_cache", exist_ok=True)
+os.makedirs("builds/", exist_ok=True)
+
+def app_cache_folder(app):
+    return os.path.join(".apps_cache", app)
 
 
-# Regular expression patterns
+def refresh_all_caches():
 
-list_name = "apps.json"
+    for app, infos in catalog.items():
+        app = app.lower()
+        print(app)
+        if not os.path.exists(app_cache_folder(app)):
+            try:
+                init_cache(app, infos)
+            except Exception as e:
+                error("Could not init cache for %s: %s" % (app, e))
+        else:
+            try:
+                refresh_cache(app, infos)
+            except Exception as e:
+                error("Could not refresh cache for %s: %s" % (app, e))
 
-assert os.path.exists(".github_credentials"), "You should add github credentials in .github_credentials first. (Format <user>:<token>)"
-assert os.path.exists(list_name), "File %s doesn't exist ?" % list_name
 
-# GitHub credentials from ./.github_credentials than should contain <user>:<token>
-# For example:      foobar:abcdef1234567890
-github_credentials = open(".github_credentials").read().strip().split(":")
+def init_cache(app, infos):
 
-# Helpers
+    if infos["state"] == "notworking":
+        depth = 5
+    if infos["state"] == "inprogress":
+        depth = 20
+    else:
+        depth = 40
+
+    git("clone --depth {depth} --single-branch --branch master {url} {folder}".format(depth=depth, url=infos["url"], folder=app_cache_folder(app)))
+
+
+def refresh_cache(app, infos):
+
+    # Don't refresh if already refreshed during last hour
+    fetch_head = app_cache_folder(app) + "/.git/FETCH_HEAD"
+    if os.path.exists(fetch_head) and now - os.path.getmtime(fetch_head) < 3600:
+        return
+
+    git("remote set-url origin " + infos["url"], in_folder=app_cache_folder(app))
+    git("fetch origin master --force", in_folder=app_cache_folder(app))
+    git("reset origin/master --hard", in_folder=app_cache_folder(app))
+
+
+def git(cmd, in_folder=None):
+
+    if in_folder:
+        cmd = "-C " + in_folder + " " + cmd
+    cmd = "git " + cmd
+    return subprocess.check_output(cmd.split()).strip().decode("utf-8")
+
+
+def build_catalog():
+
+    result_dict = {}
+
+    for app, infos in catalog.items():
+        print("Processing '%s'..." % app)
+
+        app = app.lower()
+
+        try:
+            app_dict = build_app_dict(app, infos)
+        except Exception as e:
+            error("Adding %s failed: %s" % (app, str(e)))
+            continue
+
+        result_dict[app_dict["id"]] = app_dict
+
+    #####################
+    # Current version 2 #
+    #####################
+    categories = yaml.load(open("categories.yml").read())
+    with open("builds/v2.json", 'w') as f:
+        f.write(json.dumps({"apps": result_dict, "categories": categories}, sort_keys=True))
+
+    ####################
+    # Legacy version 1 #
+    ####################
+    with open("builds/v1.json", 'w') as f:
+        f.write(json.dumps(result_dict, sort_keys=True))
+
+    ####################
+    # Legacy version 0 #
+    ####################
+    official_apps = set(["agendav", "ampache", "baikal", "dokuwiki", "etherpad_mypads", "hextris", "jirafeau", "kanboard", "my_webapp", "nextcloud", "opensondage", "phpmyadmin", "piwigo", "rainloop", "roundcube", "searx", "shellinabox", "strut", "synapse", "transmission", "ttrss", "wallabag2", "wordpress", "zerobin"])
+
+    official_apps_dict = {k: v for k, v in result_dict.items() if k in official_apps}
+    community_apps_dict = {k: v for k, v in result_dict.items() if k not in official_apps}
+
+    # We need the official apps to have "validated" as state to be recognized as official
+    for app, infos in official_apps_dict.items():
+        infos["state"] = "validated"
+
+    with open("builds/v0-official.json", 'w') as f:
+        f.write(json.dumps(official_apps_dict, sort_keys=True))
+
+    with open("builds/v0-community.json", 'w') as f:
+        f.write(json.dumps(community_apps_dict, sort_keys=True))
+
+
+def build_app_dict(app, infos):
+
+    assert infos["branch"] == "master"
+
+    this_app_cache = app_cache_folder(app)
+
+    assert os.path.exists(this_app_cache), "No cache yet for %s" % app
+
+    manifest = json.load(open(this_app_cache + "/manifest.json"))
+
+    if infos["revision"] == "HEAD":
+        relevant_files = ["manifest.json", "actions.json", "hooks/", "scripts/", "conf/", "sources/"]
+        most_recent_relevant_commit = "rev-list --full-history --all -n 1 -- " + " ".join(relevant_files)
+        infos["revision"] = git(most_recent_relevant_commit, in_folder=this_app_cache)
+        assert re.match(r"^[0-9a-f]+$", infos["revision"]), "Output was not a commit? '%s'" % infos["revision"]
+    else:
+        assert infos["revision"] in git("rev-list --all", in_folder=this_app_cache).split("\n"), "Revision ain't in history ? %s" % infos["revision"]
+
+    timestamp = git("show -s --format=%ct " + infos["revision"], in_folder=this_app_cache)
+    assert re.match(r"^[0-9]+$", timestamp), "Failed to get timestamp for revision ? '%s'" % timestamp
+    timestamp = int(timestamp)
+
+    return {'id':manifest["id"],
+            'git': {
+                'branch': infos['branch'],
+                'revision': infos["revision"],
+                'url': infos["url"]
+            },
+            'lastUpdate': timestamp,
+            'manifest': include_translations_in_manifest(manifest),
+            'state': infos['state'],
+            'level': infos.get('level', '?'),
+            'maintained': infos.get("maintained", True),
+            'high_quality': infos.get("high_quality", False),
+            'featured': infos.get("featured", False),
+            'category': infos.get('category', None),
+            'subtags': infos.get('subtags', []),
+            }
+
+
+def include_translations_in_manifest(manifest):
+
+    app_name = manifest["id"]
+
+    for locale in os.listdir("locales"):
+        if not locale.endswith("json"):
+            continue
+
+        if locale == "en.json":
+            continue
+
+        current_lang = locale.split(".")[0]
+        translations = json.load(open(os.path.join("locales", locale), "r"))
+
+        key = "%s_manifest_description" % app_name
+        if translations.get(key, None):
+            manifest["description"][current_lang] = translations[key]
+
+        for category, questions in manifest["arguments"].items():
+            for question in questions:
+                key = "%s_manifest_arguments_%s_%s" % (app_name, category, question["name"])
+                # don't overwrite already existing translation in manifests for now
+                if translations.get(key) and not current_lang not in question["ask"]:
+                    #print("[ask]", current_lang, key)
+                    question["ask"][current_lang] = translations[key]
+
+                key = "%s_manifest_arguments_%s_help_%s" % (app_name, category, question["name"])
+                # don't overwrite already existing translation in manifests for now
+                if translations.get(key) and not current_lang not in question.get("help", []):
+                    #print("[help]", current_lang, key)
+                    question["help"][current_lang] = translations[key]
+
+    return manifest
 
-def fail(msg, retcode=1):
-    """Show failure message and exit."""
-    print("Error: {0:s}".format(msg))
-    sys.exit(retcode)
 
 def error(msg):
     msg = "[Applist builder error] " + msg
@@ -37,342 +201,8 @@ def error(msg):
     print(msg)
 
 
-def include_translations_in_manifest(app_name, manifest):
-    for i in os.listdir("locales"):
-        if not i.endswith("json"):
-            continue
-
-        if i == "en.json":
-            continue
-
-        current_lang = i.split(".")[0]
-        translations = json.load(open(os.path.join("locales", i), "r"))
-
-        key = "%s_manifest_description" % app_name
-        if key in translations and translations[key]:
-            manifest["description"][current_lang] = translations[key]
-
-        for category, questions in manifest["arguments"].items():
-            for question in questions:
-                key = "%s_manifest_arguments_%s_%s" % (app_name, category, question["name"])
-                # don't overwrite already existing translation in manifests for now
-                if key in translations and translations[key] and not current_lang not in question["ask"]:
-                    print "[ask]", current_lang, key
-                    question["ask"][current_lang] = translations[key]
-
-                key = "%s_manifest_arguments_%s_help_%s" % (app_name, category, question["name"])
-                # don't overwrite already existing translation in manifests for now
-                if key in translations and translations[key] and not current_lang not in question.get("help", []):
-                    print "[help]", current_lang, key
-                    question["help"][current_lang] = translations[key]
-
-    return manifest
-
-
-def get_json(url, verify=True):
-
-    try:
-        # Retrieve and load manifest
-        if ".github" in url:
-            r = requests.get(url, verify=verify, auth=github_credentials)
-        else:
-            r = requests.get(url, verify=verify)
-        r.raise_for_status()
-        return r.json()
-    except requests.exceptions.RequestException as e:
-        print("-> Error: unable to request %s, %s" % (url, e))
-        return None
-    except ValueError as e:
-        print("-> Error: unable to decode json from %s : %s" % (url, e))
-        return None
-
-def get_zlib(url, verify=True):
-
-    try:
-        # Retrieve last commit information
-        r = requests.get(obj_url, verify=verify)
-        r.raise_for_status()
-        return zlib.decompress(r.content).decode('utf-8').split('\x00')
-    except requests.exceptions.RequestException as e:
-        print("-> Error: unable to request %s, %s" % (obj_url, e))
-        return None
-    except zlib.error as e:
-        print("-> Error: unable to decompress object from %s : %s" % (url, e))
-        return None
-
-# Main
-
-# Get list name from filename
-print(":: Building %s list..." % list_name)
-
-output = '%s-build.json' % list_name
-
-already_built_file = {}
-if os.path.exists(output):
-    try:
-        already_built_file = json.load(open(output))
-    except Exception as e:
-        print("Error while trying to load already built file: %s" % e)
-
-# Loop through every apps
-result_dict = {}
-for app, info in apps_list.items():
-    print("---")
-    print("Processing '%s'..." % app)
-    app = app.lower()
-
-    # Store usefull values
-    app_branch = info['branch']
-    app_url = info['url']
-    app_rev = info['revision']
-    app_state = info["state"]
-    app_level = info.get("level")
-    app_maintained = info.get("maintained", True)
-    app_featured = info.get("featured", False)
-    app_high_quality = info.get("high_quality", False)
-
-    forge_site = app_url.split('/')[2]
-    owner = app_url.split('/')[3]
-    repo = app_url.split('/')[4]
-    if forge_site == "github.com":
-        forge_type = "github"
-    elif forge_site == "framagit.org":
-        forge_type = "gitlab"
-    elif forge_site == "code.ffdn.org":
-        forge_type = "gitlab"
-    elif forge_site == "code.antopie.org":
-        forge_type = "gitea"
-    else:
-        forge_type = "unknown"
-
-    previous_state = already_built_file.get(app, {}).get("state", {})
-
-    manifest = {}
-    timestamp = None
-
-    previous_rev = already_built_file.get(app, {}).get("git", {}).get("revision", None)
-    previous_url = already_built_file.get(app, {}).get("git", {}).get("url")
-    previous_level = already_built_file.get(app, {}).get("level")
-    previous_maintained = already_built_file.get(app, {}).get("maintained")
-    previous_featured = already_built_file.get(app, {}).get("featured")
-    previous_high_quality = already_built_file.get(app, {}).get("high_quality")
-
-    if app_rev == "HEAD":
-        app_rev = subprocess.check_output(["git", "ls-remote", app_url, "refs/heads/"+app_branch]).split()[0]
-        if not re.match(r"^[0-9a-f]+$", app_rev):
-            error("Revision for %s did not match expected regex" % app)
-            continue
-
-        if previous_rev is None:
-            previous_rev = 'HEAD'
-
-        # If this is a github repo, we are able to optimize things a bit by looking at the diff
-        # and not actually updating the app if only README or other not-so-important files were edited
-        if previous_rev != app_rev and forge_type == "github":
-
-            url = "https://api.github.com/repos/{}/{}/compare/{}...{}".format(owner, repo, previous_rev, app_branch)
-            diff = get_json(url)
-
-            if not diff or not diff["commits"]:
-                app_rev = previous_rev if previous_rev != 'HEAD' else app_rev
-            else:
-                # Only if those files got updated, do we want to update the
-                # commit (otherwise that would trigger an unecessary upgrade)
-                ignore_files = [ "README.md", "LICENSE", ".gitignore", "check_process", ".travis.yml" ]
-                diff_files = [ f for f in diff["files"] if f["filename"] not in ignore_files ]
-
-                if diff_files:
-                    print("This app points to HEAD and significant changes where found between HEAD and previous commit")
-                    app_rev = diff["commits"][-1]["sha"]
-                else:
-                    print("This app points to HEAD but no significant changes where found compared to HEAD, so keeping the previous commit")
-                    app_rev = previous_rev if previous_rev != 'HEAD' else app_rev
-
-    print("Previous commit : %s" % previous_rev)
-    print("Current commit : %s" % app_rev)
-
-    if previous_rev == app_rev and previous_url == app_url:
-        print("Already up to date, ignoring")
-        result_dict[app] = already_built_file[app]
-        if previous_state != app_state:
-            result_dict[app]["state"] = app_state
-            print("... but has changed of state, updating it from '%s' to '%s'" % (previous_state, app_state))
-        if previous_level != app_level or app_level is None:
-            result_dict[app]["level"] = app_level
-            print("... but has changed of level, updating it from '%s' to '%s'" % (previous_level, app_level))
-        if previous_maintained != app_maintained:
-            result_dict[app]["maintained"] = app_maintained
-            print("... but maintained status changed, updating it from '%s' to '%s'" % (previous_maintained, app_maintained))
-        if previous_featured != app_featured:
-            result_dict[app]["featured"] = app_featured
-            print("... but featured status changed, updating it from '%s' to '%s'" % (previous_featured, app_featured))
-        if previous_high_quality != app_high_quality:
-            result_dict[app]["high_quality"] = app_high_quality
-            print("... but high_quality status changed, updating it from '%s' to '%s'" % (previous_high_quality, app_high_quality))
-
-        print "update translations but don't download anything"
-        result_dict[app]['manifest'] = include_translations_in_manifest(app, result_dict[app]['manifest'])
-
-        continue
-
-    print("Revision changed ! Updating...")
-
-    raw_url = 'https://%(forge_site)s/%(owner)s/%(repo)s/raw/%(app_rev)s/manifest.json' % {
-        "forge_site": forge_site, "owner": owner, "repo": repo, "app_rev": app_rev
-        }
-
-    manifest = get_json(raw_url, verify=True)
-    if manifest is None:
-        error("Manifest is empty for app %s ?" % app)
-        continue
-
-    # Hosted on GitHub
-    if forge_type == "github":
-        api_url = 'https://api.github.com/repos/%(owner)s/%(repo)s/commits/%(app_rev)s' % {
-            "owner": owner, "repo": repo, "app_rev": app_rev
-        }
-
-        info2 = get_json(api_url)
-        if info2 is None:
-            error("Commit info is empty for app %s ?" % app)
-            continue
-
-        commit_date = parse(info2['commit']['author']['date'])
-        timestamp = int(time.mktime(commit_date.timetuple()))
-
-    # Gitlab-type forge
-    elif forge_type == "gitlab":
-        api_url = 'https://%(forge_site)s/api/v4/projects/%(owner)s%%2F%(repo)s/repository/commits/%(app_rev)s' % {
-            "forge_site": forge_site, "owner": owner, "repo": repo, "app_rev": app_rev
-        }
-        commit = get_json(api_url)
-        if commit is None:
-            error("Commit info is empty for app %s ?" % app)
-            continue
-
-        commit_date = parse(commit["authored_date"])
-        timestamp = int(time.mktime(commit_date.timetuple()))
-
-    elif forge_type == "gitea":
-        api_url = 'https://%(forge_site)s/api/v1/repos/%(owner)s/%(repo)s/git/commits/%(app_rev)s' % {
-            "forge_site": forge_site, "owner": owner, "repo": repo, "app_rev": app_rev
-        }
-        info2 = get_json(api_url)
-        if info2 is None:
-            error("Commit info is empty for app %s ?" % app)
-            continue
-
-        commit_date = parse(info2['commit']['author']['date'])
-        timestamp = int(time.mktime(commit_date.timetuple()))
-
-    # Gogs-type forge
-    elif forge_type == "gogs":
-        if not app_url.endswith('.git'):
-            app_url += ".git"
-
-        obj_url = '%s/objects/%s/%s' % (
-            app_url, app_rev[0:2], app_rev[2:]
-        )
-        commit = get_zlib(obj_url, verify=False)
-
-        if commit is None or len(commit) < 2:
-            error("Commit info is empty for app %s ?" % app)
-            continue
-        else:
-            commit = commit[1]
-
-        re_commit_author = re.compile(
-            r'^author (?P<name>.+) <(?P<email>.+)> (?P<time>\d+) (?P<tz>[+-]\d+)$',
-            re.MULTILINE
-        )
-
-        # Extract author line and commit date
-        commit_author = re_commit_author.search(commit)
-        if not commit_author:
-            error("Author line in commit not found for app %s" % app)
-            continue
-
-        # Construct UTC timestamp
-        timestamp = int(commit_author.group('time'))
-        tz = commit_author.group('tz')
-        if len(tz) != 5:
-            error("Unexpected timezone length in commit for app %s" % app)
-            continue
-        elif tz != '+0000':
-            tdelta = (int(tz[1:3]) * 3600) + (int(tz[3:5]) * 60)
-            if tz[0] == '+':
-                timestamp -= tdelta
-            elif tz[0] == '-':
-                timestamp += tdelta
-            else:
-                error("Unexpected timezone format in commit for app %s" % app)
-                continue
-    else:
-        error("Unsupported VCS and/or protocol for app %s" % app)
-        continue
-
-    if manifest["id"] != app or manifest["id"] != repo.replace("_ynh", ""):
-        print("Warning: IDs different between list.json, manifest and repo name")
-        print(" Manifest id       : %s" % manifest["id"])
-        print(" Name in json list : %s" % app)
-        print(" Repo name         : %s" % repo.replace("_ynh", ""))
-
-    try:
-        result_dict[manifest['id']] = {
-            'git': {
-                'branch': info['branch'],
-                'revision': app_rev,
-                'url': app_url
-            },
-            'lastUpdate': timestamp,
-            'manifest': include_translations_in_manifest(manifest['id'], manifest),
-            'state': info['state'],
-            'level': info.get('level', '?'),
-            'maintained': app_maintained,
-            'high_quality': app_high_quality,
-            'featured': app_featured,
-            'category': info.get('category', None),
-            'subtags': info.get('subtags', []),
-        }
-    except KeyError as e:
-        error("Invalid app info or manifest for app %s, %s" % (app, e))
-        continue
-
-#######################
-## Current version 2 ##
-#######################
-categories = yaml.load(open("categories.yml").read())
-with open(output.replace(".json", "-v2.json"), 'w') as f:
-    f.write(json.dumps({"apps": result_dict, "categories": categories}, sort_keys=True))
-
-######################
-## Legacy version 1 ##
-######################
-with open(output, 'w') as f:
-    f.write(json.dumps(result_dict, sort_keys=True))
-
-print("\nDone! Written in %s" % output)
-
-######################
-## Legacy version 0 ##
-######################
-print("\nAlso splitting the file into official and community-build.json for backward compatibility")
-
-official_apps = set(["agendav", "ampache", "baikal", "dokuwiki", "etherpad_mypads", "hextris", "jirafeau", "kanboard", "my_webapp", "nextcloud", "opensondage", "phpmyadmin", "piwigo", "rainloop", "roundcube", "searx", "shellinabox", "strut", "synapse", "transmission", "ttrss", "wallabag2", "wordpress", "zerobin"])
-
-official_apps_dict = {k: v for k, v in result_dict.items() if k in official_apps}
-community_apps_dict = {k: v for k, v in result_dict.items() if k not in official_apps}
-
-# We need the official apps to have "validated" as state to be recognized as official
-for app, infos in official_apps_dict.items():
-    infos["state"] = "validated"
-
-with open("official-build.json", 'w') as f:
-    f.write(json.dumps(official_apps_dict, sort_keys=True))
-
-with open("community-build.json", 'w') as f:
-    f.write(json.dumps(community_apps_dict, sort_keys=True))
-
 ######################
 
-print("\nDone!")
+if __name__ == "__main__":
+    refresh_all_caches()
+    build_catalog()
diff --git a/rebuild.sh b/rebuild.sh
index dfe9bec..2d7c9c9 100644
--- a/rebuild.sh
+++ b/rebuild.sh
@@ -7,4 +7,4 @@ cd $workdir
 date >> $log
 git pull >/dev/null
 
-python ./list_builder.py &>> $log || sendxmpppy "[listbuilder] Rebuilding the application list failed miserably"
+./list_builder.py &>> $log || sendxmpppy "[listbuilder] Rebuilding the application list failed miserably"

From acbe07ba77472e461df99f3bfa91b1be0ae81e44 Mon Sep 17 00:00:00 2001
From: Alexandre Aubin <alex.aubin@mailoo.org>
Date: Sun, 5 Apr 2020 03:27:44 +0200
Subject: [PATCH 5/5] More comments

---
 list_builder.py | 33 ++++++++++++++++++++++-----------
 1 file changed, 22 insertions(+), 11 deletions(-)

diff --git a/list_builder.py b/list_builder.py
index 9e2a071..d0bcd0f 100755
--- a/list_builder.py
+++ b/list_builder.py
@@ -17,10 +17,22 @@ my_env["GIT_TERMINAL_PROMPT"] = "0"
 os.makedirs(".apps_cache", exist_ok=True)
 os.makedirs("builds/", exist_ok=True)
 
+###################################
+# App git clones cache management #
+###################################
+
 def app_cache_folder(app):
     return os.path.join(".apps_cache", app)
 
 
+def git(cmd, in_folder=None):
+
+    if in_folder:
+        cmd = "-C " + in_folder + " " + cmd
+    cmd = "git " + cmd
+    return subprocess.check_output(cmd.split(), env=my_env).strip().decode("utf-8")
+
+
 def refresh_all_caches():
 
     for app, infos in catalog.items():
@@ -62,13 +74,9 @@ def refresh_cache(app, infos):
     git("reset origin/master --hard", in_folder=app_cache_folder(app))
 
 
-def git(cmd, in_folder=None):
-
-    if in_folder:
-        cmd = "-C " + in_folder + " " + cmd
-    cmd = "git " + cmd
-    return subprocess.check_output(cmd.split()).strip().decode("utf-8")
-
+################################
+# Actual list build management #
+################################
 
 def build_catalog():
 
@@ -82,7 +90,7 @@ def build_catalog():
         try:
             app_dict = build_app_dict(app, infos)
         except Exception as e:
-            error("Adding %s failed: %s" % (app, str(e)))
+            error("Processing %s failed: %s" % (app, str(e)))
             continue
 
         result_dict[app_dict["id"]] = app_dict
@@ -123,24 +131,27 @@ def build_app_dict(app, infos):
 
     assert infos["branch"] == "master"
 
+    # Make sure we have some cache
     this_app_cache = app_cache_folder(app)
-
     assert os.path.exists(this_app_cache), "No cache yet for %s" % app
 
-    manifest = json.load(open(this_app_cache + "/manifest.json"))
-
+    # If using head, find the most recent meaningful commit in logs
     if infos["revision"] == "HEAD":
         relevant_files = ["manifest.json", "actions.json", "hooks/", "scripts/", "conf/", "sources/"]
         most_recent_relevant_commit = "rev-list --full-history --all -n 1 -- " + " ".join(relevant_files)
         infos["revision"] = git(most_recent_relevant_commit, in_folder=this_app_cache)
         assert re.match(r"^[0-9a-f]+$", infos["revision"]), "Output was not a commit? '%s'" % infos["revision"]
+    # Otherwise, validate commit exists
     else:
         assert infos["revision"] in git("rev-list --all", in_folder=this_app_cache).split("\n"), "Revision ain't in history ? %s" % infos["revision"]
 
+    # Find timestamp corresponding to that commit
     timestamp = git("show -s --format=%ct " + infos["revision"], in_folder=this_app_cache)
     assert re.match(r"^[0-9]+$", timestamp), "Failed to get timestamp for revision ? '%s'" % timestamp
     timestamp = int(timestamp)
 
+    # Build the dict with all the infos
+    manifest = json.load(open(this_app_cache + "/manifest.json"))
     return {'id':manifest["id"],
             'git': {
                 'branch': infos['branch'],