Merge pull request #1519 from frappe/get-app-cache

feat: cache get-app artifacts by commit_hash

.gitignore

@@ -1,6 +1,9 @@
 # MAC OS
 .DS_Store
 
+# VS Code
+.vscode/
+
 # Vim Gitignore
 ## Swap
 [._]*.s[a-v][a-z]

bench/app.py

@@ -6,10 +6,12 @@ import re
 import shutil
 import subprocess
 import sys
+import tarfile
 import typing
 from collections import OrderedDict
 from datetime import date
 from functools import lru_cache
+from pathlib import Path
 from urllib.parse import urlparse
 
 # imports - third party imports
@@ -23,6 +25,7 @@ from bench.utils import (
 	UNSET_ARG,
 	fetch_details_from_tag,
 	get_available_folder_name,
+	get_bench_cache_path,
 	is_bench_directory,
 	is_git_url,
 	is_valid_frappe_branch,
@@ -166,6 +169,7 @@ class App(AppMeta):
 		branch: str = None,
 		bench: "Bench" = None,
 		soft_link: bool = False,
+		cache_key = None,
 		*args,
 		**kwargs,
 	):
@@ -173,6 +177,7 @@ class App(AppMeta):
 		self.soft_link = soft_link
 		self.required_by = None
 		self.local_resolution = []
+		self.cache_key =  cache_key
 		super().__init__(name, branch, *args, **kwargs)
 
 	@step(title="Fetching App {repo}", success="App {repo} Fetched")
@@ -227,6 +232,7 @@ class App(AppMeta):
 		resolved=False,
 		restart_bench=True,
 		ignore_resolution=False,
+		using_cached=False
 	):
 		import bench.cli
 		from bench.utils.app import get_app_name
@@ -247,6 +253,7 @@ class App(AppMeta):
 			skip_assets=skip_assets,
 			restart_bench=restart_bench,
 			resolution=self.local_resolution,
+			using_cached=using_cached,
 		)
 
 	@step(title="Cloning and installing {repo}", success="App {repo} Installed")
@@ -285,6 +292,134 @@ class App(AppMeta):
 		)
 		
 		
+	"""
+	Get App Cache
+
+	Since get-app affects only the `apps`, `env`, and `sites`
+	bench sub directories. If we assume deterministic builds
+	when get-app is called, the `apps/app_name` sub dir can be
+	cached.
+
+	In subsequent builds this would save time by not having to:
+	- clone repository
+	- install frontend dependencies
+	- building frontend assets
+	as all of this is contained in the `apps/app_name` sub dir.
+
+	Code that updates the `env` and `sites` subdirs still need
+	to be run.
+	"""
+		
+	def get_app_path(self) -> Path:
+		return Path(self.bench.name) / "apps" / self.app_name
+
+	def get_app_cache_path(self, is_compressed=False) -> Path:
+		assert self.cache_key is not None
+
+		cache_path = get_bench_cache_path("apps")
+		ext = "tgz" if is_compressed else "tar"
+		tarfile_name = f"{self.app_name}-{self.cache_key[:10]}.{ext}"
+		return cache_path / tarfile_name
+		
+	def get_cached(self) -> bool:
+		if not self.cache_key:
+			return False
+		
+		cache_path = self.get_app_cache_path()
+		mode = "r"
+		
+		# Check if cache exists without gzip
+		if not cache_path.is_file():
+			cache_path = self.get_app_cache_path(True)
+			mode = "r:gz"
+
+		# Check if cache exists with gzip
+		if not cache_path.is_file():
+			return False
+
+		app_path = self.get_app_path()
+		if app_path.is_dir():
+			shutil.rmtree(app_path)
+		
+		click.secho(f"Getting {self.app_name} from cache", fg="yellow")
+		with tarfile.open(cache_path, mode) as tar:
+			tar.extractall(app_path.parent)
+
+		return True
+	
+	def set_cache(self, compress_artifacts=False) -> bool:
+		if not self.cache_key:
+			return False
+
+		app_path = self.get_app_path()
+		if not app_path.is_dir():
+			return False
+
+		cwd = os.getcwd()
+		cache_path = self.get_app_cache_path(compress_artifacts)
+		mode =  "w:gz" if compress_artifacts else "w"
+		
+		message = f"Caching {self.app_name} app directory"
+		if compress_artifacts:
+			message += " (compressed)"
+		click.secho(message)
+
+		self.prune_app_directory()
+		
+		success = False
+		os.chdir(app_path.parent)
+		try:
+			with tarfile.open(cache_path, mode) as tar:
+				tar.add(app_path.name)
+			success = True
+		except Exception:
+			log(f"Failed to cache {app_path}", level=3)
+			success = False
+		finally:
+			os.chdir(cwd)
+		return success
+	
+	def prune_app_directory(self):
+		app_path = self.get_app_path()
+		remove_unused_node_modules(app_path)
+	
+
+def remove_unused_node_modules(app_path: Path) -> None:
+	"""
+	Erring a bit the side of caution; since there is no explicit way
+	to check if node_modules are utilized, this function checks if Vite
+	is being used to build the frontend code.
+	
+	Since most popular Frappe apps use Vite to build their frontends,
+	this method should suffice.
+	
+	Note: root package.json is ignored cause those usually belong to
+	apps that do not have a build step and so their node_modules are
+	utilized during runtime.
+	"""
+	
+	for p in app_path.iterdir():
+		if not p.is_dir():
+			continue
+		
+		package_json = p / "package.json"
+		if not package_json.is_file():
+			continue
+		
+		node_modules = p / "node_modules"
+		if not node_modules.is_dir():
+			continue
+		
+		can_delete = False
+		with package_json.open("r", encoding="utf-8") as f:
+			package_json = json.loads(f.read())
+			build_script = package_json.get("scripts", {}).get("build", "")
+			can_delete = "vite build" in build_script
+			
+		if can_delete:
+			shutil.rmtree(node_modules)
+
+
 def make_resolution_plan(app: App, bench: "Bench"):
 	"""
 	decide what apps and versions to install and in what order
@@ -346,6 +481,8 @@ def get_app(
 	soft_link=False,
 	init_bench=False,
 	resolve_deps=False,
+	cache_key=None,
+	compress_artifacts=False,
 ):
 	"""bench get-app clones a Frappe App from remote (GitHub or any other git server),
 	and installs it on the current bench. This also resolves dependencies based on the
@@ -360,7 +497,7 @@ def get_app(
 	from bench.utils.app import check_existing_dir
 
 	bench = Bench(bench_path)
-	app = App(git_url, branch=branch, bench=bench, soft_link=soft_link)
+	app = App(git_url, branch=branch, bench=bench, soft_link=soft_link, cache_key=cache_key)
 	git_url = app.url
 	repo_name = app.repo
 	branch = app.tag
@@ -418,6 +555,10 @@ def get_app(
 		)
 		return
 	
+	if app.get_cached():
+		app.install(verbose=verbose, skip_assets=skip_assets, restart_bench=restart_bench, using_cached=True)
+		return
+
 	dir_already_exists, cloned_path = check_existing_dir(bench_path, repo_name)
 	to_clone = not dir_already_exists
 
@@ -443,6 +584,9 @@ def get_app(
 	):
 		app.install(verbose=verbose, skip_assets=skip_assets, restart_bench=restart_bench)
 		
+	app.set_cache(compress_artifacts)
+
+
 
 def install_resolved_deps(
 	bench,
@@ -452,7 +596,6 @@ def install_resolved_deps(
 	verbose=False,
 ):
 	from bench.utils.app import check_existing_dir
-
 	if "frappe" in resolution:
 		# Terminal dependency
 		del resolution["frappe"]
@@ -550,6 +693,7 @@ def install_app(
 	restart_bench=True,
 	skip_assets=False,
 	resolution=UNSET_ARG,
+	using_cached=False,
 ):
 	import bench.cli as bench_cli
 	from bench.bench import Bench
@@ -577,14 +721,14 @@ def install_app(
 	if conf.get("developer_mode"):
 		install_python_dev_dependencies(apps=app, bench_path=bench_path, verbose=verbose)
 
-	if os.path.exists(os.path.join(app_path, "package.json")):
+	if not using_cached and os.path.exists(os.path.join(app_path, "package.json")):
 		yarn_install = "yarn install --verbose" if verbose else "yarn install"
 		bench.run(yarn_install, cwd=app_path)
 
 	bench.apps.sync(app_name=app, required=resolution, branch=tag, app_dir=app_path)
 
 	if not skip_assets:
-		build_assets(bench_path=bench_path, app=app)
+		build_assets(bench_path=bench_path, app=app, using_cached=using_cached)
 
 	if restart_bench:
 		# Avoiding exceptions here as production might not be set-up

bench/commands/__init__.py

@@ -72,6 +72,7 @@ bench_command.add_command(switch_to_develop)
 
 
 from bench.commands.utils import (
+	app_cache_helper,
 	backup_all_sites,
 	bench_src,
 	disable_production,
@@ -108,6 +109,7 @@ bench_command.add_command(disable_production)
 bench_command.add_command(bench_src)
 bench_command.add_command(find_benches)
 bench_command.add_command(migrate_env)
+bench_command.add_command(app_cache_helper)
 
 from bench.commands.setup import setup
 

bench/commands/make.py

@@ -151,6 +151,18 @@ def drop(path):
 	default=False,
 	help="Resolve dependencies before installing app",
 )
+@click.option(
+	"--cache-key",
+	type=str,
+	default=None,
+	help="Caches get-app artifacts if provided (only first 10 chars is used)",
+)
+@click.option(
+	"--compress-artifacts",
+	is_flag=True,
+	default=False,
+	help="Whether to gzip get-app artifacts that are to be cached",
+)
 def get_app(
 	git_url,
 	branch,
@@ -160,6 +172,8 @@ def get_app(
 	soft_link=False,
 	init_bench=False,
 	resolve_deps=False,
+	cache_key=None,
+	compress_artifacts=False,
 ):
 	"clone an app from the internet and set it up in your bench"
 	from bench.app import get_app
@@ -172,6 +186,8 @@ def get_app(
 		soft_link=soft_link,
 		init_bench=init_bench,
 		resolve_deps=resolve_deps,
+		cache_key=cache_key,
+		compress_artifacts=compress_artifacts,
 	)
 
 

bench/commands/utils.py

@@ -176,3 +176,21 @@ def migrate_env(python, backup=True):
 	from bench.utils.bench import migrate_env
 
 	migrate_env(python=python, backup=backup)
+
+
+@click.command("app-cache", help="View or remove items belonging to bench get-app cache")
+@click.option("--clear", is_flag=True, default=False, help="Remove all items")
+@click.option(
+	"--remove-app",
+	default="",
+	help="Removes all items that match provided app name",
+)
+@click.option(
+	"--remove-key",
+	default="",
+	help="Removes all items that matches provided cache key",
+)
+def app_cache_helper(clear=False, remove_app="", remove_key=""):
+	from bench.utils.bench import cache_helper
+
+	cache_helper(clear, remove_app, remove_key)

bench/utils/__init__.py

@@ -7,8 +7,9 @@ import subprocess
 import sys
 from functools import lru_cache
 from glob import glob
+from pathlib import Path
 from shlex import split
-from typing import List, Tuple
+from typing import List, Optional, Tuple
 
 # imports - third party imports
 import click
@@ -50,6 +51,15 @@ def is_frappe_app(directory: str) -> bool:
 
 	return bool(is_frappe_app)
 
+def get_bench_cache_path(sub_dir: Optional[str]) -> Path:
+	relative_path = "~/.cache/bench"
+	if sub_dir and not sub_dir.startswith("/"):
+		relative_path += f"/{sub_dir}"
+
+	cache_path = os.path.expanduser(relative_path)
+	cache_path = Path(cache_path)
+	cache_path.mkdir(parents=True, exist_ok=True)
+	return cache_path
 
 @lru_cache(maxsize=None)
 def is_valid_frappe_branch(frappe_path: str, frappe_branch: str):

bench/utils/bench.py

@@ -4,11 +4,13 @@ import json
 import logging
 import os
 import re
+import shutil
 import subprocess
 import sys
 from functools import lru_cache
 from glob import glob
 from json.decoder import JSONDecodeError
+from pathlib import Path
 
 # imports - third party imports
 import click
@@ -16,7 +18,8 @@ import click
 # imports - module imports
 import bench
 from bench.exceptions import PatchError, ValidationError
-from bench.utils import exec_cmd, get_bench_name, get_cmd_output, log, which
+from bench.utils import (exec_cmd, get_bench_cache_path, get_bench_name,
+                         get_cmd_output, log, which)
 
 logger = logging.getLogger(bench.PROJECT_NAME)
 
@@ -350,11 +353,16 @@ def restart_process_manager(bench_path=".", web_workers=False):
 		exec_cmd(f"overmind restart {worker}", cwd=bench_path)
 
 
-def build_assets(bench_path=".", app=None):
+def build_assets(bench_path=".", app=None, using_cached=False):
 	command = "bench build"
 	if app:
 		command += f" --app {app}"
-	exec_cmd(command, cwd=bench_path, env={"BENCH_DEVELOPER": "1"})
+
+	env = {"BENCH_DEVELOPER": "1"}
+	if using_cached:
+		env["USING_CACHED"] = "1"
+
+	exec_cmd(command, cwd=bench_path, env=env)
 
 
 def handle_version_upgrade(version_upgrade, bench_path, force, reset, conf):
@@ -635,3 +643,115 @@ To switch to your required branch, run the following commands: bench switch-to-b
 			)
 
 			sys.exit(1)
+
+
+def cache_helper(clear=False, remove_app="", remove_key="") -> None:
+	can_remove = bool(remove_key or remove_app)
+	if not clear and not can_remove:
+		cache_list()
+	elif can_remove:
+		cache_remove(remove_app, remove_key)
+	elif clear:
+		cache_clear()
+	else:
+		pass  # unreachable
+
+
+def cache_list() -> None:
+	from datetime import datetime
+
+	tot_size = 0
+	tot_items = 0
+
+	printed_header = False
+	for item in get_bench_cache_path("apps").iterdir():
+		if item.suffix not in [".tar", ".tgz"]:
+			continue
+
+		stat = item.stat()
+		size_mb = stat.st_size / 1_000_000
+		created = datetime.fromtimestamp(stat.st_ctime)
+		accessed = datetime.fromtimestamp(stat.st_atime)
+
+		app = item.name.split("-")[0]
+		tot_items += 1
+		tot_size += stat.st_size
+		compressed = item.suffix == ".tgz"
+
+		if not printed_header:
+			click.echo(
+				f"{'APP':15}  "
+				f"{'FILE':25}  "
+				f"{'SIZE':>13}  "
+				f"{'COMPRESSED'}  "
+				f"{'CREATED':19}  "
+				f"{'ACCESSED':19}  "
+			)
+			printed_header = True
+
+		click.echo(
+			f"{app:15}  "
+			f"{item.name:25}  "
+			f"{size_mb:10.3f} MB  "
+			f"{str(compressed):10}  "
+			f"{created:%Y-%m-%d %H:%M:%S}  "
+			f"{accessed:%Y-%m-%d %H:%M:%S}  "
+		)
+
+	if tot_items:
+		click.echo(f"Total size {tot_size / 1_000_000:.3f} MB belonging to {tot_items} items")
+	else:
+		click.echo("No cached items")
+
+
+def cache_remove(app: str = "", key: str = "") -> None:
+	rem_items = 0
+	rem_size = 0
+	for item in get_bench_cache_path("apps").iterdir():
+		if not should_remove_item(item, app, key):
+			continue
+
+		rem_items += 1
+		rem_size += item.stat().st_size
+		item.unlink(True)
+		click.echo(f"Removed {item.name}")
+
+	if rem_items:
+		click.echo(f"Cleared {rem_size / 1_000_000:.3f} MB belonging to {rem_items} items")
+	else:
+		click.echo("No items removed")
+
+
+def should_remove_item(item: Path, app: str, key: str) -> bool:
+	if item.suffix not in [".tar", ".tgz"]:
+		return False
+
+	name = item.name
+	if app and key and name.startswith(f"{app}-{key[:10]}."):
+		return True
+
+	if app and name.startswith(f"{app}-"):
+		return True
+
+	if key and f"-{key[:10]}." in name:
+		return True
+
+	return False
+
+
+def cache_clear() -> None:
+	cache_path = get_bench_cache_path("apps")
+	tot_items = len(os.listdir(cache_path))
+	if not tot_items:
+		click.echo("No cached items")
+		return
+
+	tot_size = get_dir_size(cache_path)
+	shutil.rmtree(cache_path)
+
+	if tot_items:
+		click.echo(f"Cleared {tot_size / 1_000_000:.3f} MB belonging to {tot_items} items")
+
+
+def get_dir_size(p: Path) -> int:
+	return sum(i.stat(follow_symlinks=False).st_size for i in p.iterdir())

pyproject.toml

@@ -22,7 +22,7 @@ dependencies = [
     "Click>=7.0",
     "GitPython~=3.1.30",
     "honcho",
-    "Jinja2~=3.0.3",
+    "Jinja2~=3.1.3",
     "python-crontab~=2.6.0",
     "requests",
     "semantic-version~=2.8.2",