From 7b80bfe54783bd662d0e71b3eee09c5ecf304355 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 15 Sep 2020 17:40:13 -0700
Subject: web robots.txt

Two variants depending on environment.
---
 fatcat_scholar/static/robots.allow.txt    |  5 +++++
 fatcat_scholar/static/robots.disallow.txt |  4 ++++
 fatcat_scholar/web.py                     | 13 +++++++++++++
 3 files changed, 22 insertions(+)
 create mode 100644 fatcat_scholar/static/robots.allow.txt
 create mode 100644 fatcat_scholar/static/robots.disallow.txt

diff --git a/fatcat_scholar/static/robots.allow.txt b/fatcat_scholar/static/robots.allow.txt
new file mode 100644
index 0000000..3c40454
--- /dev/null
+++ b/fatcat_scholar/static/robots.allow.txt
@@ -0,0 +1,5 @@
+# Allow most queries, but please don't crawl search results (or at least do so very politely, eg no parallel requests)
+
+User-Agent: *
+Disallow: /search
+Allow: /
diff --git a/fatcat_scholar/static/robots.disallow.txt b/fatcat_scholar/static/robots.disallow.txt
new file mode 100644
index 0000000..de20247
--- /dev/null
+++ b/fatcat_scholar/static/robots.disallow.txt
@@ -0,0 +1,4 @@
+# By default (eg, in QA/development) set robots to disallow all
+
+User-Agent: *
+Disallow: /
diff --git a/fatcat_scholar/web.py b/fatcat_scholar/web.py
index 866214e..5ad301d 100644
--- a/fatcat_scholar/web.py
+++ b/fatcat_scholar/web.py
@@ -7,6 +7,7 @@ So far there are few endpoints, so we just put them all here!
 import babel.support
 from fastapi import FastAPI, APIRouter, Request, Depends, Response
 from fastapi.staticfiles import StaticFiles
+from fastapi.responses import PlainTextResponse
 from dynaconf import settings
 from typing import Optional, Any
 
@@ -218,3 +219,15 @@ for lang_option in I18N_LANG_OPTIONS:
 app.include_router(api)
 
 app.mount("/static", StaticFiles(directory="fatcat_scholar/static"), name="static")
+
+ROBOTS_ALLOW = open("fatcat_scholar/static/robots.allow.txt", "r").read()
+ROBOTS_DISALLOW = open("fatcat_scholar/static/robots.disallow.txt", "r").read()
+
+
+@app.get("/robots.txt", include_in_schema=False)
+async def robots_txt(response_class=PlainTextResponse) -> Any:
+    print(ROBOTS_ALLOW)
+    if settings.SCHOLAR_ENV == "prod":
+        return PlainTextResponse(ROBOTS_ALLOW)
+    else:
+        return PlainTextResponse(ROBOTS_DISALLOW)
-- 
cgit v1.2.3