Merge pull request #4 from jsickcodes/edition-url-rewrites

Add logic for mapping paths to S3 keys

.github/workflows/ci.yaml

@@ -20,6 +20,9 @@ jobs:
         with:
           python-version: ${{ matrix.python }}
 
+      - name: Install stable virtualenv # https://github.com/pre-commit/action/issues/135
+        run: pip install -U virtualenv==20.10.0
+
       - name: Run pre-commit
         uses: pre-commit/[email protected]
 

manifests/base/auth-configmap.yaml

@@ -1,7 +1,7 @@
 apiVersion: v1
 kind: ConfigMap
 metadata:
-  name: "ltdproxy-auth"
+  name: "ltd-proxy-auth"
   labels:
     app.kubernetes.io/name: "ltd-proxy"
 data:

manifests/base/configmap.yaml

@@ -1,17 +1,17 @@
 apiVersion: v1
 kind: ConfigMap
 metadata:
-  name: "ltdproxy"
+  name: "ltd-proxy"
   labels:
     app.kubernetes.io/name: "ltd-proxy"
 data:
   # These configurations are injected as environment variables into the
   # app container.
-  SAFIR_NAME: "ltdproxy"
-  SAFIR_LOGGER: "ltdproxy"
+  SAFIR_NAME: "ltd-proxy"
+  SAFIR_LOGGER: "ltd-proxy"
   SAFIR_LOG_LEVEL: "INFO"
   SAFIR_PROFILE: "production"
-  LTDPROXY_AUTH_CONFIG: "/opt/ltdproxy/auth/authrules.yaml"
+  LTDPROXY_AUTH_CONFIG: "/opt/ltd-proxy/auth/authrules.yaml"
   LTDPROXY_PATH_PREFIX: "/"
   LTDPROXY_S3_BUCKET: ""
   LTDPROXY_S3_PREFIX: ""

manifests/base/deployment.yaml

@@ -1,7 +1,7 @@
 apiVersion: apps/v1
 kind: Deployment
 metadata:
-  name: "ltdproxy"
+  name: "ltd-proxy"
   labels:
     app.kubernetes.io/name: "ltd-proxy"
 spec:
@@ -25,7 +25,7 @@ spec:
               name: "app"
           envFrom:
             - configMapRef:
-                name: "ltdproxy"
+                name: "ltd-proxy"
           securityContext:
             allowPrivilegeEscalation: false
             capabilities:
@@ -34,12 +34,12 @@ spec:
             readOnlyRootFilesystem: true
           volumeMounts:
              - name: "auth-config"
-               mountPath: "/opt/ltdproxy/auth/"
+               mountPath: "/opt/ltd-proxy/auth/"
                readOnly: true
       volumes:
         - name: "auth-config"
           configMap:
-            name: "ltdproxy-auth"
+            name: "ltd-proxy-auth"
       securityContext:
         runAsNonRoot: true
         runAsUser: 1000

manifests/base/kustomization.yaml

@@ -3,7 +3,7 @@ kind: Kustomization
 
 images:
   - name: "ghcr.io/jsickcodes/ltd-proxy"
-    newTag: 0.0.0
+    newTag: 0.1.0
 
 resources:
   - configmap.yaml

manifests/base/service.yaml

@@ -1,14 +1,14 @@
 apiVersion: v1
 kind: Service
 metadata:
-  name: "ltdproxy"
+  name: "ltd-proxy"
   labels:
     app.kubernetes.io/name: "ltd-proxy"
 spec:
   ports:
-    - name: "ltdproxy-http"
+    - name: "ltd-proxy-http"
       protocol: "TCP"
       port: 8080
       targetPort: "app"
   selector:
-    app.kubernetes.io/name: "ltdproxy"
+    app.kubernetes.io/name: "ltd-proxy"

src/ltdproxy/githubauth.py

@@ -21,6 +21,7 @@
 import yaml
 from authlib.integrations.starlette_client import OAuth
 from pydantic import BaseModel
+from structlog import get_logger
 
 from ltdproxy.config import config
 
@@ -42,6 +43,8 @@
 )
 """Type alias from the authlib GitHub OAuth client."""
 
+logger = get_logger(config.logger_name)
+
 
 class GitHubOAuth:
     """This class maintains an OAuth instance that is registered for GitHub
@@ -90,7 +93,8 @@ async def set_serialized_github_memberships(
     # These orgs and teams are mentioned in the GitHub Auth configuration,
     # and therefore are ones to pay attention to in the cookie.
     relevant_orgs = github_auth.relevant_orgs
-    relevant_teams = github_auth.relevant_teams
+
+    logger.debug("Relevant orgs", orgs=relevant_orgs)
 
     github_client = gidgethub.httpx.GitHubAPI(
         http_client, "ltd-proxy", oauth_token=github_token
@@ -106,11 +110,13 @@ async def set_serialized_github_memberships(
     user_teams: List[Tuple[str, str]] = []
     async for team in github_client.getiter("/user/teams"):
         team_id = (team["organization"]["login"], team["name"])
-        if team_id in relevant_teams:
+        if team_id[0] in relevant_orgs:
+            logger.debug("Found relevant team", team=team)
             user_teams.append(team_id)
 
     # Serialize memberships to JSON to pack inside the session cookie
     memberships = json.dumps({"orgs": user_orgs, "teams": user_teams})
+    logger.debug("GitHub user memberships", orgs=user_orgs, teams=user_teams)
     session["github_memberships"] = memberships
 
 
@@ -147,6 +153,11 @@ class PathRule(BaseModel):
     def path_matches(self, url_path: str) -> bool:
         """Test if a URL path matches the rule's patten."""
         if self.pattern.match(url_path):
+            logger.debug(
+                "Path matches PathRule",
+                pattern=self.pattern,
+                url_path=url_path,
+            )
             return True
         else:
             return False
@@ -172,6 +183,12 @@ def is_user_authorized(
                     return True
 
         # no matches
+        logger.debug(
+            "No authorization match",
+            pattern=self.pattern,
+            user_orgs=user_orgs,
+            user_teams=user_teams,
+        )
         return False
 
 
@@ -257,13 +274,11 @@ def relevant_orgs(self) -> Set[str]:
         all_orgs: Set[str] = set()
 
         for github_group in self.default:
-            if not github_group.is_team:
-                all_orgs.add(github_group.org)
+            all_orgs.add(github_group.org)
 
         for path_rule in self.paths:
             for github_group in path_rule.authorized:
-                if not github_group.is_team:
-                    all_orgs.add(github_group.org)
+                all_orgs.add(github_group.org)
 
         return all_orgs
 

src/ltdproxy/handlers/external.py

@@ -1,5 +1,6 @@
 """Handlers for the app's external root, ``/ltdproxy/``."""
 
+import posixpath
 from typing import Optional, Union
 from urllib.parse import urlencode, urlparse
 
@@ -12,6 +13,7 @@
 from starlette.requests import Request
 from starlette.responses import (
     HTMLResponse,
+    PlainTextResponse,
     RedirectResponse,
     StreamingResponse,
 )
@@ -27,13 +29,19 @@
     set_serialized_github_memberships,
 )
 from ltdproxy.s3 import Bucket, bucket_dependency
+from ltdproxy.urlmap import map_s3_path
 
 __all__ = ["get_s3", "external_router"]
 
 external_router = APIRouter()
 """FastAPI router for all external handlers."""
 
 
+@external_router.get("/", name="homepage")
+async def get_homepage() -> PlainTextResponse:
+    return PlainTextResponse("OK", status_code=200)
+
+
 @external_router.get("/auth", name="get_oauth_callback")
 async def get_oauth_callback(
     ref: Optional[str],
@@ -147,20 +155,48 @@ async def get_s3(
 
     elif github_auth_result == AuthResult.authorized:
         # User is authorized; stream from S3.
-        if path == "" or path.endswith("/"):
-            # redwrite "*/" as "*/index.html" for static sites in S3
-            bucket_path = f"{config.s3_bucket_prefix}{path}index.html"
-        else:
-            bucket_path = f"{config.s3_bucket_prefix}{path}"
+        bucket_path = map_s3_path(config.s3_bucket_prefix, path)
+        logger.debug(
+            "computed bucket path",
+            bucket_path=bucket_path,
+            request_url=str(request.url),
+        )
         stream = await bucket.stream_object(http_client, bucket_path)
         if stream.status_code == 404:
-            raise HTTPException(status_code=404, detail="Does not exist.")
+            if not path.endswith("/") and posixpath.splitext(path)[1] == "":
+                # try a redirect
+                parsed_url = urlparse(str(request.url))
+                parsed_url = parsed_url._replace(path=f"{parsed_url.path}/")
+                return RedirectResponse(url=parsed_url.geturl())
+            else:
+                raise HTTPException(status_code=404, detail="Does not exist.")
         logger.debug("stream headers", headers=stream.headers)
         response_headers = {
             "Content-type": stream.headers["Content-type"],
             "Content-length": stream.headers["Content-length"],
             "Etag": stream.headers["Etag"],
         }
+        # FIXME hack to override content-type headers
+        if bucket_path.endswith(".html"):
+            logger.debug("is html")
+            response_headers["Content-type"] = "text/html"
+        elif bucket_path.endswith(".css"):
+            logger.debug("is css")
+            response_headers["Content-type"] = "text/css"
+        elif bucket_path.endswith(".js"):
+            logger.debug("is js")
+            response_headers["Content-type"] = "application/javascript"
+        elif bucket_path.endswith(".pdf"):
+            logger.debug("is pdf")
+            response_headers["Content-type"] = "application/pdf"
+        elif bucket_path.endswith(".png"):
+            logger.debug("is png")
+            response_headers["Content-type"] = "image/png"
+        else:
+            logger.debug("did not change response content-type")
+
+        logger.debug("response headers", headers=response_headers)
+
         return StreamingResponse(
             stream.aiter_raw(),
             background=BackgroundTask(stream.aclose),

src/ltdproxy/main.py

@@ -7,9 +7,11 @@
 called.
 """
 
+import structlog
 from fastapi import FastAPI
 from safir.dependencies.http_client import http_client_dependency
 from safir.logging import configure_logging
+from safir.metadata import get_metadata
 from safir.middleware.x_forwarded import XForwardedMiddleware
 from starlette.middleware.sessions import SessionMiddleware
 
@@ -36,6 +38,12 @@
 
 @app.on_event("startup")
 async def startup_event() -> None:
+    logger = structlog.get_logger(config.logger_name)
+    metadata = get_metadata(
+        package_name="ltd-proxy",
+        application_name=config.name,
+    )
+    logger.info("Starting up", version=metadata.version)
     app.add_middleware(XForwardedMiddleware)
 
 

src/ltdproxy/urlmap.py

@@ -0,0 +1,40 @@
+"""Domain model for mapping a request URL to a resource in the S3 bucket."""
+
+from __future__ import annotations
+
+__all__ = ["map_s3_path"]
+
+
+def map_s3_path(bucket_prefix: str, request_path: str) -> str:
+    """Map a request URL to an S3 bucket key."""
+    # decompose the path into the project and whether it is a /v/ edition or
+    # not
+    parts = request_path.split("/")
+    project_name = parts[0].lower()
+
+    if (len(parts) >= 3) and parts[1].lower() == "v":
+        edition_name = parts[2]
+        edition_path = "/".join(parts[3:])
+    else:
+        edition_name = "__main"  # default edition
+        edition_path = "/".join(parts[1:])
+
+    # if edition_path == "" or edition_path.endswith("/"):
+    if request_path.endswith("/"):
+        edition_path = f"{edition_path}index.html"
+
+    if bucket_prefix == "":
+        path_parts = [project_name, "v", edition_name, edition_path]
+    else:
+        path_parts = [
+            bucket_prefix,
+            project_name,
+            "v",
+            edition_name,
+            edition_path,
+        ]
+
+    bucket_path = "/".join(path_parts)
+    bucket_path = bucket_path.rstrip("/")  # happens if edition_path is ""
+
+    return bucket_path