Show git head in health infos

README.md

@@ -9,4 +9,7 @@ Query and manipulate the Netz39 entities database.
 The service is configured via the following environment variables:
 * `PORT`: Service port. defaults to 8080
 * `AUTH`: Authentication tokens, defaults to None. Example Configuration : `AUTH={"token_1": "user_1", "token_2": "user_2"}`
-
+* `GIT_ORIGIN`: URL for the origin Git repository, including the user name
+* `GIT_PASSWORD`: The git password for the user encoded in the origin URL
+* `GIT_PULL_INTV`: Time interval between automated pull operations (default: 30s)
+* `GIT_WC_PATH`: Set a path for the working copy. Will create a temporary checkout if not provided.

app.py

@@ -12,6 +12,7 @@ import json
 
 import util
 from auth import AuthProvider
+from gitmgr import GitManagerConfiguration, GitManager
 
 
 startup_timestamp = datetime.now()
@@ -19,8 +20,9 @@ startup_timestamp = datetime.now()
 
 class HealthHandler(tornado.web.RequestHandler, metaclass=ABCMeta):
     # noinspection PyAttributeOutsideInit
-    def initialize(self):
+    def initialize(self, sources=None):
         self.git_version = self._load_git_version()
+        self.sources = sources
 
     @staticmethod
     def _load_git_version():
@@ -52,6 +54,12 @@ class HealthHandler(tornado.web.RequestHandler, metaclass=ABCMeta):
         health['timestamp'] = isodate.datetime_isoformat(datetime.now())
         health['uptime'] = isodate.duration_isoformat(datetime.now() - startup_timestamp)
 
+        if self.sources:
+            for s in self.sources:
+                h = s()
+                if h is not None:
+                    health = {**health, **h}
+
         self.set_header("Content-Type", "application/json")
         self.write(json.dumps(health, indent=4))
         self.set_status(200)
@@ -69,10 +77,11 @@ class Oas3Handler(tornado.web.RequestHandler, metaclass=ABCMeta):
         self.finish()
 
 
-def make_app(_auth_provider=None):
+def make_app(_auth_provider=None, gitmgr=None):
     version_path = r"/v[0-9]"
     return tornado.web.Application([
-        (version_path + r"/health", HealthHandler),
+        (version_path + r"/health", HealthHandler,
+         {"sources": [lambda: {"git-head":  gitmgr.head_sha}] if gitmgr else None}),
         (version_path + r"/oas3", Oas3Handler),
     ])
 
@@ -83,10 +92,16 @@ def main():
     # Setup
     auth_provider = AuthProvider.from_environment()
 
-    util.run_tornado_server(make_app(auth_provider),
+    gitcfg = GitManagerConfiguration.from_environment()
+    gitmgr = GitManager(configuration=gitcfg)
+    gitmgr.setup()
+    gitmgr.printout()
+
+    util.run_tornado_server(make_app(auth_provider, gitmgr),
                             server_port=port)
 
     # Teardown
+    gitmgr.teardown()
 
     print("Server stopped")
 

gitmgr.py

@@ -0,0 +1,188 @@
+import git
+import os
+import shutil
+import tempfile
+import time
+
+from util import load_env
+
+
+class GitManagerConfiguration:
+    @staticmethod
+    def from_environment():
+        origin = load_env("GIT_ORIGIN", None)
+        wc_path = load_env("GIT_WC_PATH", None)
+        git_pw = load_env("GIT_PASSWORD", None)
+        pull_intv = load_env("GIT_PULL_INTV", None)
+
+        return GitManagerConfiguration(origin=origin,
+                                       git_pw=git_pw,
+                                       wc_path=wc_path,
+                                       pull_intv=pull_intv)
+
+    def __init__(self, origin, git_pw=None, wc_path=None, pull_intv=None):
+        if not origin:
+            raise ValueError("Git origin cannot be empty!")
+
+        self._origin = origin
+        self._git_pw = git_pw
+        self._wc_path = wc_path
+        self._pull_intv = 30 if pull_intv is None else int(pull_intv)
+
+    @property
+    def origin(self):
+        return self._origin
+
+    @property
+    def git_pw(self):
+        return self._git_pw
+
+    @property
+    def wc_path(self):
+        return self._wc_path
+
+    @property
+    def pull_intv(self):
+        return self._pull_intv
+
+
+class GitManager:
+    def __init__(self, configuration):
+        if configuration is None:
+            raise ValueError("GitManager must be initialized with a configuration!")
+
+        self._configuration = configuration
+        self._wc = None
+        self._last_pull = 0
+
+    @property
+    def configuration(self):
+        return self._configuration
+
+    def _setup_wc(self):
+        if self._wc is not None:
+            return
+
+        _wc = self.configuration.wc_path
+
+        if _wc is None:
+            _wc = tempfile.mkdtemp(prefix='entities_git_')
+
+        if not os.path.isdir(_wc):
+            raise ValueError("Configured directory for the working copy does not exist!")
+
+        self._wc = _wc
+
+    def _teardown_wc(self):
+        if self._wc is None:
+            return
+
+        if self.configuration.wc_path is not None:
+            print("NOTE: Not tearing down externally configured working copy.")
+            return
+
+        shutil.rmtree(self._wc)
+
+        self._wc = None
+
+    def _assert_wc(self):
+        """Assert working copy matches origin and is a valid repository.
+
+           A failed assertion will throw exceptions and lead to service abort,
+           as this error is not recoverable.
+
+           Returns False if the WC path is an empty directory"""
+
+        # Check if WC is empty
+        if not os.listdir(self._wc):
+            return False
+
+        # Create a repository object
+        # This fails if there is no valid repository
+        repo = git.Repo(self._wc)
+
+        # Assert that this is not a bare repo
+        if repo.bare:
+            raise ValueError("WC path points to a bare git repository!")
+
+        origin = repo.remote('origin')
+        if self.configuration.origin not in origin.urls:
+            raise ValueError("Origin URL does not match!")
+
+        # We're good here.
+        return True
+
+    def _askpass_script(self):
+        # Passwords are impossible to store in scripts, as they may contain any character ...
+        # We convert the password into a list of integers and create a little script
+        # that reconstructs the password and writes it to the console.
+        # Python will be installed anyways.
+
+        pw_chars = [ord(c) for c in self.configuration.git_pw]
+
+        script = "#!/usr/bin/env python3\n"
+        script += "l = %s\n" % str(list(pw_chars))
+        script += "p = [chr(c) for c in l]\n"
+        script += f"print(\"\".join(p))\n"
+        return script
+
+    def _init_repo(self):
+        # Assert working copy is valid,
+        # return false if cloning is necessary
+        if not self._assert_wc():
+            print("Cloning new git working copy ...")
+
+            # Create a temporary script file for GIT_ASKPASS
+            with tempfile.NamedTemporaryFile(mode='w+t') as askpass:
+                askpass.write(self._askpass_script())
+                askpass.file.close()
+                os.chmod(path=askpass.name, mode=0o700)
+                self.repo = git.Repo.clone_from(url=self.configuration.origin,
+                                                to_path=self._wc,
+                                                env={'GIT_ASKPASS': askpass.name})
+        else:
+            print("Reusing existing git working copy ...")
+            self.repo = git.Repo(self._wc)
+
+    def setup(self):
+        self._setup_wc()
+        self._init_repo()
+        self.pull(force=True)
+
+    def teardown(self):
+        self._teardown_wc()
+
+    def printout(self):
+        print("Git Manager:")
+        print(f"\tGit origin is %s" % self.configuration.origin)
+        print(f"\tUsing working copy path %s" % self._wc)
+        if not self._wc == self.configuration.wc_path:
+            print("\tUsing a temporary working copy.")
+
+    @property
+    def head_sha(self):
+        return None if self.repo is None else self.repo.head.object.hexsha
+
+    def pull(self, force=False):
+        """Pull from origin.
+
+        Arguments:
+            `force` -- Do a pull even though the pull interval has not elapsed
+
+        Returns: True if pull was executed
+        """
+
+        if not force and (time.time() - self._last_pull < self.configuration.pull_intv):
+            return False
+
+        self._last_pull = time.time()
+
+        old_head = self.head_sha
+
+        # get the origin
+        # (We verified during initialization that this origin exists.)
+        origin = self.repo.remote('origin')
+
+        origin.pull(rebase=True)
+
+        return self.head_sha != old_head

requirements.txt

@@ -1,3 +1,4 @@
 tornado==6.0.4
 isodate==0.6.0
 pytest==5.4.1
+GitPython==3.1.12