chore: Bump version to 3.2.0

Continuation of the broken #21

Co-authored-by: dettlaff <dettlaff@riseup.net>
Co-authored-by: def <dettlaff@riseup.net>
Co-authored-by: Houkime <>
Reviewed-on: #112
Reviewed-by: houkime <houkime@protonmail.com>

README.md

@@ -1,6 +1,8 @@
 # SelfPrivacy GraphQL API which allows app to control your server
 
-## build
+![CI status](https://ci.selfprivacy.org/api/badges/SelfPrivacy/selfprivacy-rest-api/status.svg)
+
+## Build
 
 ```console
 $ nix build
@@ -8,7 +10,7 @@ $ nix build
 
 In case of successful build, you should get the `./result` symlink to a folder (in `/nix/store`) with build contents.
 
-## develop
+## Develop
 
 ```console
 $ nix develop
@@ -21,10 +23,10 @@ Type "help", "copyright", "credits" or "license" for more information.
 If you don't have experimental flakes enabled, you can use the following command:
 
 ```console
-nix --extra-experimental-features nix-command --extra-experimental-features flakes develop
+$ nix --extra-experimental-features nix-command --extra-experimental-features flakes develop
 ```
 
-## testing
+## Testing
 
 Run the test suite by running coverage with pytest inside an ephemeral NixOS VM with redis service enabled:
 ```console
@@ -61,7 +63,7 @@ $ TMPDIR=".nixos-vm-tmp-dir" nix run .#checks.x86_64-linux.default.driverInterac
 
 Option `-L`/`--print-build-logs` is optional for all nix commands. It tells nix to print each log line one after another instead of overwriting a single one.
 
-## dependencies and dependant modules
+## Dependencies and Dependant Modules
 
 This flake depends on a single Nix flake input - nixpkgs repository. nixpkgs repository is used for all software packages used to build, run API service, tests, etc.
 
@@ -85,6 +87,6 @@ $ nix flake metadata git+https://git.selfprivacy.org/SelfPrivacy/selfprivacy-nix
 
 Nix code for NixOS service module for API is located in NixOS configuration repository.
 
-## troubleshooting
+## Troubleshooting
 
 Sometimes commands inside `nix develop` refuse to work properly if the calling shell lacks `LANG` environment variable. Try to set it before entering `nix develop`.

flake.nix

@@ -27,6 +27,7 @@
             python-lsp-server
             pyflakes
             typer # for strawberry
+            types-redis # for mypy
           ] ++ strawberry-graphql.optional-dependencies.cli));
 
       vmtest-src-dir = "/root/source";

selfprivacy_api/dependencies.py

@@ -27,4 +27,4 @@ async def get_token_header(
 
 def get_api_version() -> str:
     """Get API version"""
-    return "3.1.0"
+    return "3.2.0"

selfprivacy_api/graphql/mutations/system_mutations.py

@@ -8,9 +8,12 @@ from selfprivacy_api.graphql.mutations.mutation_interface import (
     GenericJobMutationReturn,
     GenericMutationReturn,
     MutationReturnInterface,
+    GenericJobMutationReturn,
 )
 
 import selfprivacy_api.actions.system as system_actions
+from selfprivacy_api.graphql.common_types.jobs import job_to_api_job
+from selfprivacy_api.jobs.nix_collect_garbage import start_nix_collect_garbage
 import selfprivacy_api.actions.ssh as ssh_actions
 
 
@@ -195,3 +198,14 @@ class SystemMutations:
             message=f"Failed to pull repository changes:\n{result.data}",
             code=500,
         )
+
+    @strawberry.mutation(permission_classes=[IsAuthenticated])
+    def nix_collect_garbage(self) -> GenericJobMutationReturn:
+        job = start_nix_collect_garbage()
+
+        return GenericJobMutationReturn(
+            success=True,
+            code=200,
+            message="Garbage collector started...",
+            job=job_to_api_job(job),
+        )

selfprivacy_api/jobs/migrate_to_binds.py

@@ -67,8 +67,8 @@ def move_folder(
 
     try:
         data_path.mkdir(mode=0o750, parents=True, exist_ok=True)
-    except Exception as e:
-        print(f"Error creating data path: {e}")
+    except Exception as error:
+        print(f"Error creating data path: {error}")
         return
 
     try:

selfprivacy_api/jobs/nix_collect_garbage.py

@@ -0,0 +1,147 @@
+import re
+import subprocess
+from typing import Tuple, Iterable
+
+from selfprivacy_api.utils.huey import huey
+
+from selfprivacy_api.jobs import JobStatus, Jobs, Job
+
+
+class ShellException(Exception):
+    """Shell-related errors"""
+
+
+COMPLETED_WITH_ERROR = "Error occurred, please report this to the support chat."
+RESULT_WAS_NOT_FOUND_ERROR = (
+    "We are sorry, garbage collection result was not found. "
+    "Something went wrong, please report this to the support chat."
+)
+CLEAR_COMPLETED = "Garbage collection completed."
+
+
+def delete_old_gens_and_return_dead_report() -> str:
+    subprocess.run(
+        ["nix-env", "-p", "/nix/var/nix/profiles/system", "--delete-generations old"],
+        check=False,
+    )
+
+    result = subprocess.check_output(["nix-store", "--gc", "--print-dead"]).decode(
+        "utf-8"
+    )
+
+    return " " if result is None else result
+
+
+def run_nix_collect_garbage() -> Iterable[bytes]:
+    process = subprocess.Popen(
+        ["nix-store", "--gc"], stdout=subprocess.PIPE, stderr=subprocess.STDOUT
+    )
+    return process.stdout if process.stdout else iter([])
+
+
+def parse_line(job: Job, line: str) -> Job:
+    """
+    We parse the string for the presence of a final line,
+    with the final amount of space cleared.
+    Simply put, we're just looking for a similar string:
+    "1537 store paths deleted, 339.84 MiB freed".
+    """
+    pattern = re.compile(r"[+-]?\d+\.\d+ \w+(?= freed)")
+    match = re.search(pattern, line)
+
+    if match is None:
+        raise ShellException("nix returned gibberish output")
+
+    else:
+        Jobs.update(
+            job=job,
+            status=JobStatus.FINISHED,
+            status_text=CLEAR_COMPLETED,
+            result=f"{match.group(0)} have been cleared",
+        )
+    return job
+
+
+def process_stream(job: Job, stream: Iterable[bytes], total_dead_packages: int) -> None:
+    completed_packages = 0
+    prev_progress = 0
+
+    for line in stream:
+        line = line.decode("utf-8")
+
+        if "deleting '/nix/store/" in line:
+            completed_packages += 1
+            percent = int((completed_packages / total_dead_packages) * 100)
+
+            if percent - prev_progress >= 5:
+                Jobs.update(
+                    job=job,
+                    status=JobStatus.RUNNING,
+                    progress=percent,
+                    status_text="Cleaning...",
+                )
+                prev_progress = percent
+
+        elif "store paths deleted," in line:
+            parse_line(job, line)
+
+
+def get_dead_packages(output) -> Tuple[int, float]:
+    dead = len(re.findall("/nix/store/", output))
+    percent = 0
+    if dead != 0:
+        percent = 100 / dead
+    return dead, percent
+
+
+@huey.task()
+def calculate_and_clear_dead_paths(job: Job):
+    Jobs.update(
+        job=job,
+        status=JobStatus.RUNNING,
+        progress=0,
+        status_text="Calculate the number of dead packages...",
+    )
+
+    dead_packages, package_equal_to_percent = get_dead_packages(
+        delete_old_gens_and_return_dead_report()
+    )
+
+    if dead_packages == 0:
+        Jobs.update(
+            job=job,
+            status=JobStatus.FINISHED,
+            status_text="Nothing to clear",
+            result="System is clear",
+        )
+        return True
+
+    Jobs.update(
+        job=job,
+        status=JobStatus.RUNNING,
+        progress=0,
+        status_text=f"Found {dead_packages} packages to remove!",
+    )
+
+    stream = run_nix_collect_garbage()
+    try:
+        process_stream(job, stream, dead_packages)
+    except ShellException as error:
+        Jobs.update(
+            job=job,
+            status=JobStatus.ERROR,
+            status_text=COMPLETED_WITH_ERROR,
+            error=RESULT_WAS_NOT_FOUND_ERROR,
+        )
+
+
+def start_nix_collect_garbage() -> Job:
+    job = Jobs.add(
+        type_id="maintenance.collect_nix_garbage",
+        name="Collect garbage",
+        description="Cleaning up unused packages",
+    )
+
+    calculate_and_clear_dead_paths(job=job)
+
+    return job

selfprivacy_api/task_registry.py

@@ -1,5 +1,14 @@
+from os import environ
+
 from selfprivacy_api.utils.huey import huey
-from selfprivacy_api.jobs.test import test_job
+
 from selfprivacy_api.backup.tasks import *
 from selfprivacy_api.services.tasks import move_service
 from selfprivacy_api.jobs.upgrade_system import rebuild_system_task
+
+from selfprivacy_api.jobs.test import test_job
+from selfprivacy_api.jobs.nix_collect_garbage import calculate_and_clear_dead_paths
+
+
+if environ.get("TEST_MODE"):
+    from tests.test_huey import sum

selfprivacy_api/utils/huey.py

@@ -1,16 +1,24 @@
 """MiniHuey singleton."""
-import os
-from huey import SqliteHuey
+from os import environ
+from huey import RedisHuey
+
+from selfprivacy_api.utils.redis_pool import RedisPool
+
+HUEY_DATABASE_NUMBER = 10
+
+
+def immediate() -> bool:
+    if environ.get("HUEY_QUEUES_FOR_TESTS"):
+        return False
+    if environ.get("TEST_MODE"):
+        return True
+    return False
 
-HUEY_DATABASE = "/etc/selfprivacy/tasks.db"
 
 # Singleton instance containing the huey database.
-
-test_mode = os.environ.get("TEST_MODE")
-
-huey = SqliteHuey(
+huey = RedisHuey(
     "selfprivacy-api",
-    filename=HUEY_DATABASE if not test_mode else None,
-    immediate=test_mode == "true",
+    url=RedisPool.connection_url(dbnumber=HUEY_DATABASE_NUMBER),
+    immediate=immediate(),
     utc=True,
 )

selfprivacy_api/utils/redis_pool.py

@@ -1,8 +1,8 @@
 """
 Redis pool module for selfprivacy_api
 """
-from os import environ
 import redis
+
 from selfprivacy_api.utils.singleton_metaclass import SingletonMetaclass
 
 REDIS_SOCKET = "/run/redis-sp-api/redis.sock"
@@ -14,20 +14,20 @@ class RedisPool(metaclass=SingletonMetaclass):
     """
 
     def __init__(self):
-        if "USE_REDIS_PORT" in environ:
-            self._pool = redis.ConnectionPool(
-                host="127.0.0.1",
-                port=int(environ["USE_REDIS_PORT"]),
-                decode_responses=True,
-            )
-
-        else:
-            self._pool = redis.ConnectionPool.from_url(
-                f"unix://{REDIS_SOCKET}",
-                decode_responses=True,
-            )
+        self._pool = redis.ConnectionPool.from_url(
+            RedisPool.connection_url(dbnumber=0),
+            decode_responses=True,
+        )
         self._pubsub_connection = self.get_connection()
 
+    @staticmethod
+    def connection_url(dbnumber: int) -> str:
+        """
+        redis://[[username]:[password]]@localhost:6379/0
+        unix://[username@]/path/to/socket.sock?db=0[&password=password]
+        """
+        return f"unix://{REDIS_SOCKET}?db={dbnumber}"
+
     def get_connection(self):
         """
         Get a connection from the pool.

setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="selfprivacy_api",
-    version="3.1.0",
+    version="3.2.0",
     packages=find_packages(),
     scripts=[
         "selfprivacy_api/app.py",

tests/conftest.py

@@ -99,23 +99,14 @@ def generic_userdata(mocker, tmpdir):
 
 
 @pytest.fixture
-def huey_database(mocker, shared_datadir):
-    """Mock huey database."""
-    mock = mocker.patch(
-        "selfprivacy_api.utils.huey.HUEY_DATABASE", shared_datadir / "huey.db"
-    )
-    return mock
-
-
-@pytest.fixture
-def client(huey_database, redis_repo_with_tokens):
+def client(redis_repo_with_tokens):
     from selfprivacy_api.app import app
 
     return TestClient(app)
 
 
 @pytest.fixture
-def authorized_client(huey_database, redis_repo_with_tokens):
+def authorized_client(redis_repo_with_tokens):
     """Authorized test client fixture."""
     from selfprivacy_api.app import app
 
@@ -127,7 +118,7 @@ def authorized_client(huey_database, redis_repo_with_tokens):
 
 
 @pytest.fixture
-def wrong_auth_client(huey_database, redis_repo_with_tokens):
+def wrong_auth_client(redis_repo_with_tokens):
     """Wrong token test client fixture."""
     from selfprivacy_api.app import app
 

tests/test_graphql/test_nix_collect_garbage.py

@@ -0,0 +1,229 @@
+# pylint: disable=redefined-outer-name
+# pylint: disable=unused-argument
+# pylint: disable=missing-function-docstring
+
+import pytest
+from selfprivacy_api.utils.huey import huey
+
+from selfprivacy_api.jobs import JobStatus, Jobs
+from tests.test_graphql.common import (
+    get_data,
+    assert_ok,
+    assert_empty,
+)
+
+from selfprivacy_api.jobs.nix_collect_garbage import (
+    get_dead_packages,
+    parse_line,
+    ShellException,
+)
+
+OUTPUT_PRINT_DEAD = """
+finding garbage collector roots...
+determining live/dead paths...
+/nix/store/02k8pmw00p7p7mf2dg3n057771w7liia-python3.10-cchardet-2.1.7
+/nix/store/03vc6dznx8njbvyd3gfhfa4n5j4lvhbl-python3.10-async-timeout-4.0.2
+/nix/store/03ybv2dvfk7c3cpb527y5kzf6i35ch41-python3.10-pycparser-2.21
+/nix/store/04dn9slfqwhqisn1j3jv531lms9w5wlj-python3.10-hypothesis-6.50.1.drv
+/nix/store/04hhx2z1iyi3b48hxykiw1g03lp46jk7-python-remove-bin-bytecode-hook
+"""
+
+
+OUTPUT_COLLECT_GARBAGE = """
+removing old generations of profile /nix/var/nix/profiles/per-user/def/channels
+finding garbage collector roots...
+deleting garbage...
+deleting '/nix/store/02k8pmw00p7p7mf2dg3n057771w7liia-python3.10-cchardet-2.1.7'
+deleting '/nix/store/03vc6dznx8njbvyd3gfhfa4n5j4lvhbl-python3.10-async-timeout-4.0.2'
+deleting '/nix/store/03ybv2dvfk7c3cpb527y5kzf6i35ch41-python3.10-pycparser-2.21'
+deleting '/nix/store/04dn9slfqwhqisn1j3jv531lms9w5wlj-python3.10-hypothesis-6.50.1.drv'
+deleting '/nix/store/04hhx2z1iyi3b48hxykiw1g03lp46jk7-python-remove-bin-bytecode-hook'
+deleting unused links...
+note: currently hard linking saves -0.00 MiB
+190 store paths deleted, 425.51 MiB freed
+"""
+
+OUTPUT_COLLECT_GARBAGE_ZERO_TRASH = """
+removing old generations of profile /nix/var/nix/profiles/per-user/def/profile
+removing old generations of profile /nix/var/nix/profiles/per-user/def/channels
+finding garbage collector roots...
+deleting garbage...
+deleting unused links...
+note: currently hard linking saves 0.00 MiB
+0 store paths deleted, 0.00 MiB freed
+"""
+
+
+# ---
+
+
+def test_parse_line():
+    txt = "note: currently hard linking saves -0.00 MiB 190 store paths deleted, 425.51 MiB freed"
+
+    job = Jobs.add(
+        name="name",
+        type_id="parse_line",
+        description="description",
+    )
+
+    output = parse_line(job, txt)
+    assert output.result == "425.51 MiB have been cleared"
+    assert output.status == JobStatus.FINISHED
+    assert output.error is None
+
+
+def test_parse_line_with_blank_line():
+    txt = ""
+    job = Jobs.add(
+        name="name",
+        type_id="parse_line",
+        description="description",
+    )
+
+    with pytest.raises(ShellException):
+        output = parse_line(job, txt)
+
+
+def test_get_dead_packages():
+    assert get_dead_packages(OUTPUT_PRINT_DEAD) == (5, 20.0)
+
+
+def test_get_dead_packages_zero():
+    assert get_dead_packages("") == (0, 0)
+
+
+RUN_NIX_COLLECT_GARBAGE_MUTATION = """
+mutation CollectGarbage {
+    system {
+        nixCollectGarbage {
+            success
+            message
+            code
+            job {
+                uid,
+                typeId,
+                name,
+                description,
+                status,
+                statusText,
+                progress,
+                createdAt,
+                updatedAt,
+                finishedAt,
+                error,
+                result,
+            }
+        }
+    }
+}
+"""
+
+
+def test_graphql_nix_collect_garbage(authorized_client, fp):
+    assert huey.immediate is True
+
+    fp.register(
+        ["nix-env", "-p", "/nix/var/nix/profiles/system", "--delete-generations old"],
+        stdout="",
+    )
+    fp.register(["nix-store", "--gc", "--print-dead"], stdout=OUTPUT_PRINT_DEAD)
+    fp.register(["nix-store", "--gc"], stdout=OUTPUT_COLLECT_GARBAGE)
+
+    response = authorized_client.post(
+        "/graphql",
+        json={
+            "query": RUN_NIX_COLLECT_GARBAGE_MUTATION,
+        },
+    )
+
+    output = get_data(response)["system"]["nixCollectGarbage"]
+    assert_ok(output)
+    assert output["job"] is not None
+    assert output["job"]["status"] == "FINISHED"
+    assert output["job"]["error"] is None
+
+    assert (
+        fp.call_count(
+            [
+                "nix-env",
+                "-p",
+                "/nix/var/nix/profiles/system",
+                "--delete-generations old",
+            ]
+        )
+        == 1
+    )
+    assert fp.call_count(["nix-store", "--gc", "--print-dead"]) == 1
+    assert fp.call_count(["nix-store", "--gc"]) == 1
+
+
+def test_graphql_nix_collect_garbage_return_zero_trash(authorized_client, fp):
+    assert huey.immediate is True
+
+    fp.register(
+        ["nix-env", "-p", "/nix/var/nix/profiles/system", "--delete-generations old"],
+        stdout="",
+    )
+    fp.register(["nix-store", "--gc", "--print-dead"], stdout=OUTPUT_PRINT_DEAD)
+    fp.register(["nix-store", "--gc"], stdout=OUTPUT_COLLECT_GARBAGE_ZERO_TRASH)
+
+    response = authorized_client.post(
+        "/graphql",
+        json={
+            "query": RUN_NIX_COLLECT_GARBAGE_MUTATION,
+        },
+    )
+
+    output = get_data(response)["system"]["nixCollectGarbage"]
+    assert_ok(output)
+    assert output["job"] is not None
+    assert output["job"]["status"] == "FINISHED"
+    assert output["job"]["error"] is None
+
+    assert (
+        fp.call_count(
+            [
+                "nix-env",
+                "-p",
+                "/nix/var/nix/profiles/system",
+                "--delete-generations old",
+            ]
+        )
+        == 1
+    )
+    assert fp.call_count(["nix-store", "--gc", "--print-dead"]) == 1
+    assert fp.call_count(["nix-store", "--gc"]) == 1
+
+
+def test_graphql_nix_collect_garbage_not_authorized_client(client, fp):
+    assert huey.immediate is True
+
+    fp.register(
+        ["nix-env", "-p", "/nix/var/nix/profiles/system", "--delete-generations old"],
+        stdout="",
+    )
+    fp.register(["nix-store", "--gc", "--print-dead"], stdout=OUTPUT_PRINT_DEAD)
+    fp.register(["nix-store", "--gc"], stdout=OUTPUT_COLLECT_GARBAGE)
+
+    response = client.post(
+        "/graphql",
+        json={
+            "query": RUN_NIX_COLLECT_GARBAGE_MUTATION,
+        },
+    )
+
+    assert_empty(response)
+
+    assert (
+        fp.call_count(
+            [
+                "nix-env",
+                "-p",
+                "/nix/var/nix/profiles/system",
+                "--delete-generations old",
+            ]
+        )
+        == 0
+    )
+    assert fp.call_count(["nix-store", "--gc", "--print-dead"]) == 0
+    assert fp.call_count(["nix-store", "--gc"]) == 0

tests/test_huey.py

@@ -0,0 +1,132 @@
+import pytest
+import redis
+from typing import List
+
+import subprocess
+from subprocess import Popen, check_output, TimeoutExpired
+from os import environ, path, set_blocking
+from io import BufferedReader
+from huey.exceptions import HueyException
+
+from selfprivacy_api.utils.huey import huey, immediate, HUEY_DATABASE_NUMBER
+from selfprivacy_api.utils.redis_pool import RedisPool, REDIS_SOCKET
+
+
+@huey.task()
+def sum(a: int, b: int) -> int:
+    return a + b
+
+
+def reset_huey_storage():
+    huey.storage = huey.create_storage()
+
+
+def flush_huey_redis_forcefully():
+    url = RedisPool.connection_url(HUEY_DATABASE_NUMBER)
+
+    pool = redis.ConnectionPool.from_url(url, decode_responses=True)
+    connection = redis.Redis(connection_pool=pool)
+    connection.flushdb()
+
+
+# TODO: may be useful in other places too, move to utils/ tests common if using it somewhere
+def read_all_ready_output(stream: BufferedReader) -> str:
+    set_blocking(stream.fileno(), False)
+    output: List[bytes] = []
+    while True:
+        line = stream.readline()
+        raise ValueError(line)
+        if line == b"":
+            break
+        else:
+            output.append(line)
+
+    set_blocking(stream.fileno(), True)
+
+    result = b"".join(output)
+    return result.decode("utf-8")
+
+
+@pytest.fixture()
+def not_immediate():
+    assert environ["TEST_MODE"] == "true"
+
+    old_immediate = huey.immediate
+    environ["HUEY_QUEUES_FOR_TESTS"] = "Yes"
+    huey.immediate = False
+    assert huey.immediate is False
+
+    yield
+
+    del environ["HUEY_QUEUES_FOR_TESTS"]
+    huey.immediate = old_immediate
+    assert huey.immediate == old_immediate
+
+
+@pytest.fixture()
+def huey_socket_consumer(not_immediate):
+    """
+    Same as above, but with socketed redis
+    """
+
+    flush_huey_redis_forcefully()
+    command = ["huey_consumer.py", "selfprivacy_api.task_registry.huey"]
+
+    # First assert that consumer does not fail by itself
+    # Idk yet how to do it more elegantly
+    try:
+        check_output(command, timeout=2)
+    except TimeoutExpired:
+        pass
+
+    # Then open it for real
+    consumer_handle = Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+
+    assert path.exists(REDIS_SOCKET)
+
+    yield consumer_handle
+
+    consumer_handle.kill()
+
+
+def test_huey_over_redis_socket(huey_socket_consumer):
+    assert huey.immediate is False
+    assert immediate() is False
+
+    assert "unix" in RedisPool.connection_url(HUEY_DATABASE_NUMBER)
+    try:
+        assert (
+            RedisPool.connection_url(HUEY_DATABASE_NUMBER)
+            in huey.storage_kwargs.values()
+        )
+    except AssertionError:
+        raise ValueError(
+            "our test-side huey does not connect over socket: ", huey.storage_kwargs
+        )
+
+    result = sum(2, 5)
+    try:
+        assert result(blocking=True, timeout=100) == 7
+
+    except HueyException as error:
+        if "timed out" in str(error):
+            output = read_all_ready_output(huey_socket_consumer.stdout)
+            errorstream = read_all_ready_output(huey_socket_consumer.stderr)
+            raise TimeoutError(
+                f"Huey timed out: {str(error)}",
+                f"Consumer output: {output}",
+                f"Consumer errorstream: {errorstream}",
+            )
+        else:
+            raise error
+
+
+@pytest.mark.xfail(reason="cannot yet schedule with sockets for some reason")
+def test_huey_schedule(huey_queues_socket):
+    # We do not schedule tasks anywhere, but concerning that it fails.
+    sum.schedule((2, 5), delay=10)
+
+    try:
+        assert len(huey.scheduled()) == 1
+    except AssertionError:
+        raise ValueError("have wrong amount of scheduled tasks", huey.scheduled())