feat: allow running server on custom port

fixes #75

.github/workflows/test.yml

@@ -55,4 +55,4 @@ jobs:
 
             - name: Run pytest
               run: |
-                  poetry run pytest . -vvs
+                  poetry run pytest . -vvs --timeout=60

docs/server.md

@@ -0,0 +1,23 @@
+<!-- markdownlint-disable MD046 -->
+# SeaGOAT-server
+
+The seagoat-server is an integral component of the Seagoat command-line tool
+designed to analyze your codebase and create vector embeddings for it.
+
+While it serves as a backend for the command-line tool, also allows you to
+use it through HTTP to build your own SeaGOAT-based applications.
+
+## Starting the server
+
+To boot up the server for a specific repository, use:
+
+```bash
+seagoat-server start <repo_path> [--port=<custom_port>]
+```
+
+* `repo_path`: Path to your Git repository
+* `--port`: (Optional) Run the server on a specific port
+
+If you don't specify a custom port, a random port will be assigned to your
+server. Don't worry, SeaGOAT will still be able to automatically find
+the server corresponding to a specific repository.

mkdocs.yml

@@ -29,3 +29,9 @@ theme:
     features:
         - content.action.edit
         - content.code.copy
+        - navigation.instant
+        - navigation.tracking
+nav:
+    - Getting Started: index.md
+    - Usage Reference: usage.md
+    - SeaGOAT Server: server.md

pyproject.toml

@@ -55,6 +55,7 @@ python-semantic-release = "^8.0.4"
 pytest-sugar = "^0.9.7"
 pytest-profiling = "^1.7.0"
 pytest-timeout = "^2.1.0"
+psutil = "^5.9.5"
 
 [build-system]
 requires = ["poetry-core"]
@@ -134,4 +135,4 @@ dist_glob_patterns = ["dist/*"]
 upload_to_vcs_release = true
 
 [tool.pytest.ini_options]
-timeout = 30
+timeout = 15

seagoat/server.py

@@ -85,9 +85,10 @@ def get_free_port():
     return port
 
 
-def start_server(repo_path, port=None):
+def start_server(repo_path, custom_port=None):
     os.environ["TOKENIZERS_PARALLELISM"] = "false"
     app = create_app(repo_path)
+    port = custom_port
 
     if port is None:
         port = get_free_port()
@@ -125,8 +126,9 @@ def wait_for(condition_function, timeout, period=0.05):
         time.sleep(period)
 
 
-def get_server(repo_path, port=None):
+def get_server(repo_path, custom_port=None):
     server_info_file = get_server_info_file(repo_path)
+    port = None
 
     if os.path.exists(server_info_file):
         host, port, _, server_address = load_server_info(server_info_file)
@@ -136,11 +138,14 @@ def get_server(repo_path, port=None):
             return server_address
         os.remove(server_info_file)
 
+    if custom_port is not None:
+        port = custom_port
+
     os.environ["TOKENIZERS_PARALLELISM"] = "true"
     temp_app = create_app(repo_path)
     del temp_app
 
-    start_server(str(repo_path), port)
+    start_server(str(repo_path), custom_port=port)
 
     wait_for(lambda: os.path.exists(server_info_file), timeout=60)
 
@@ -161,9 +166,10 @@ def server():
 
 @server.command()
 @click.argument("repo_path")
-def start(repo_path):
+@click.option("--port", type=int, help="The port to start the server on", default=None)
+def start(repo_path, port):
     """Starts the server."""
-    get_server(repo_path, port=None)
+    get_server(repo_path, custom_port=port)
     click.echo("Server running.")
 
 

tests/conftest.py

@@ -3,8 +3,10 @@
 import multiprocessing
 import os
 import shutil
+import subprocess
 import tempfile
 from collections import defaultdict
+from contextlib import contextmanager
 from datetime import datetime
 from datetime import timedelta
 from datetime import timezone
@@ -390,3 +392,25 @@ def client(repo):
 def mock_queue(client):
     # pylint: disable=protected-access
     yield client._mock_queue
+
+
+@pytest.fixture
+def managed_process():
+    processes = []
+
+    @contextmanager
+    def _process(*args, **kwargs):
+        proc = subprocess.Popen(*args, **kwargs)
+        processes.append(proc)
+        try:
+            yield proc
+        finally:
+            proc.terminate()
+            proc.wait()
+            processes.remove(proc)
+
+    yield _process
+
+    for proc in processes:
+        proc.terminate()
+        proc.wait()

tests/test_server.py

@@ -1,6 +1,7 @@
 # pylint: disable=no-member
 import copy
 import json
+import os
 import re
 import subprocess
 from unittest.mock import ANY
@@ -10,8 +11,11 @@
 import requests
 
 from seagoat import __version__
+from seagoat.server import get_server_info_file
 from seagoat.server import get_status_data
+from seagoat.server import load_server_info
 from seagoat.server import server as seagoat_server
+from seagoat.server import wait_for
 
 
 def normalize_full_paths(data, repo):
@@ -126,11 +130,14 @@ def assert_server_status(repo, running):
 
 
 def simulate_server_dying(repo):
-    pid = get_status_data(repo.working_dir)["pid"]
+    try:
+        pid = get_status_data(repo.working_dir)["pid"]
 
-    process = psutil.Process(pid)
-    process.terminate()
-    process.wait(timeout=10)
+        process = psutil.Process(pid)
+        process.terminate()
+        process.wait(timeout=10)
+    except psutil.TimeoutExpired:
+        pass
 
 
 @pytest.mark.usefixtures("server")
@@ -186,29 +193,23 @@ def test_version_option(runner):
     assert result.output.strip() == f"seagoat, version {__version__}"
 
 
-@pytest.mark.parametrize("port", [8080, 8081])
-def test_start_server_on_specific_port(port, repo):
-    subprocess.run(
-        [
-            "python",
-            "-m",
-            "seagoat.server",
-            "start",
-            "--port",
-            str(port),
-            repo.working_dir,
-        ],
-        capture_output=True,
-        text=True,
-        check=False,
-    )
+@pytest.mark.parametrize("custom_port", [7483, 99081])
+def test_start_server_on_specific_port(custom_port, repo, mocker, managed_process):
+    mocker.patch("seagoat.server.TaskQueue")
 
-    response = requests.get(f"http://localhost:{port}/query/hello")
-    assert response.status_code == 200
+    server_cmd = [
+        "python",
+        "-m",
+        "seagoat.server",
+        "start",
+        "--port",
+        str(custom_port),
+        repo.working_dir,
+    ]
+
+    with managed_process(server_cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE):
+        wait_for(lambda: os.path.exists(get_server_info_file(repo.working_dir)), 8)
 
-    # subprocess.run(
-    #     ["python", "-m", "seagoat.server", "stop", repo.working_dir],
-    #     capture_output=True,
-    #     text=True,
-    #     check=False,
-    # )
+        server_info_file = get_server_info_file(repo.working_dir)
+        _, _, _, server_address = load_server_info(server_info_file)
+        assert str(custom_port) in server_address