clean-npm-cache: Expand clean-npm-cache.

Expands `clean-npm-cache` so that it can be used in production
environment as well. Also moves it to `scripts/` from `tools/`.

.travis.yml

@@ -20,7 +20,7 @@ install:
 
   # Clean any npm cache which was generated but not required anymore to avoid
   # our cache becoming huge.
-  - mispipe "tools/clean-npm-cache --travis" ts
+  - mispipe "scripts/clean-npm-cache --threshold 0" ts
 
   # Clean any emoji cache which was generated but not required anymore to avoid
   # our cache becoming huge.

scripts/clean-npm-cache

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+import argparse
+import os
+import subprocess
+import sys
+
+if False:
+    from typing import Set, Text
+
+ZULIP_PATH = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.append(ZULIP_PATH)
+from scripts.lib.node_cache import generate_sha1sum_node_modules
+from scripts.lib.zulip_tools import GENERIC_CACHE_SCRIPT_PARSER, \
+    get_caches_to_be_purged, get_environment, get_recent_deployments, \
+    purge_unused_caches
+
+ENV = get_environment()
+NODE_MODULES_CACHE_PATH = "/srv/zulip-npm-cache"
+if ENV == "travis":
+    NODE_MODULES_CACHE_PATH = os.path.join(os.environ["HOME"], "zulip-npm-cache")
+    try:
+        subprocess.check_output(["/srv/zulip-yarn/bin/yarn", '--version'])
+    except OSError:
+        print('yarn not found. Most probably we are running static-analysis and '
+              'hence yarn is not installed. Exiting without cleaning npm cache.')
+        sys.exit(0)
+
+def parse_args():
+    # type: () -> argparse.Namespace
+    parser = argparse.ArgumentParser(description="This script cleans unused zulip npm caches.",
+                                     parents=[GENERIC_CACHE_SCRIPT_PARSER, ])
+    args = parser.parse_args()
+    return args
+
+def get_caches_in_use(threshold_days):
+    # type: (int) -> Set[Text]
+    setups_to_check = set([ZULIP_PATH, ])
+    caches_in_use = set()
+
+    if ENV == "prod":
+        setups_to_check |= get_recent_deployments(threshold_days)
+    if ENV == "dev":
+        # In dev always include the currently active cache in order
+        # not to break current installation in case dependencies
+        # are updated with bumping the provision version.
+        CURRENT_CACHE = os.path.dirname(os.path.realpath(os.path.join(ZULIP_PATH, "node_modules")))
+        caches_in_use.add(CURRENT_CACHE)
+
+    for setup_dir in setups_to_check:
+        PACKAGES_FILE = os.path.join(setup_dir, "package.json")
+        if os.path.exists(PACKAGES_FILE):
+            # If 'package.json' file doesn't exist then no node_modules
+            # cache is associated with this setup.
+            sha1sum = generate_sha1sum_node_modules(setup_dir=setup_dir)
+            caches_in_use.add(os.path.join(NODE_MODULES_CACHE_PATH, sha1sum))
+
+    return caches_in_use
+
+def main():
+    # type: () -> None
+    args = parse_args()
+    caches_in_use = get_caches_in_use(args.threshold_days)
+    purge_unused_caches(
+        NODE_MODULES_CACHE_PATH, caches_in_use, args.threshold_days, args.dry_run, "node modules")
+
+if __name__ == "__main__":
+    main()

tools/clean-npm-cache

@@ -1,33 +0,0 @@
-#!/usr/bin/env python3
-from __future__ import absolute_import
-from __future__ import print_function
-import os
-import subprocess
-import sys
-
-sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
-from scripts.lib.node_cache import generate_sha1sum_node_modules
-
-NODE_MODULES_CACHE_PATH = "/srv/zulip-npm-cache"
-if "--travis" in sys.argv:
-    NODE_MODULES_CACHE_PATH = os.path.join(os.environ["HOME"], "zulip-npm-cache")
-    try:
-        subprocess.check_output([os.path.join(NODE_MODULES_CACHE_PATH, "yarn/bin/yarn"), '--version'])
-    except OSError:
-        print('yarn not found. Most probably we are running static-analysis and '
-              'hence yarn is not installed. Exiting without cleaning npm cache.')
-        sys.exit(0)
-
-sha1sum = generate_sha1sum_node_modules()
-current_cache_dir_base = os.path.join(NODE_MODULES_CACHE_PATH, sha1sum)
-current_success_stamp = os.path.join(current_cache_dir_base, '.success-stamp')
-
-print("Current cache stamp at %s" % (current_success_stamp,))
-
-for cache_dir_base in os.listdir(NODE_MODULES_CACHE_PATH):
-    node_modules_dir = os.path.join(NODE_MODULES_CACHE_PATH, cache_dir_base)
-    if node_modules_dir == current_cache_dir_base and os.path.exists(current_success_stamp):
-        print("Keeping used node_modules cache dir %s" % (node_modules_dir,))
-    else:
-        print("Cleaning unused node_modules cache dir %s" % (node_modules_dir,))
-        subprocess.check_call(["sudo", "rm", "-rf", node_modules_dir])