Serve archived pages

awd

@@ -0,0 +1,58 @@
+#!/usr/bin/python3
+
+import argparse
+import http
+import http.server as hs
+import ssl
+
+import psycopg2
+import psycopg2.extras
+
+class WSqlShRH(hs.BaseHTTPRequestHandler):
+    def do_GET(self):
+        self.log_message(f"checking {self.path}")
+        db.rollback()
+        csr = db.cursor(cursor_factory=psycopg2.extras.DictCursor)
+        csr.execute("select * from documents where url = %s",
+                    (args.baseurl + self.path,))
+        dr = csr.fetchone()
+        if dr:
+            self.send_response(dr["code"])
+            self.send_header("Content-Type", dr["content_type"])
+            self.send_header("Last-Modified", dr["last_modified"])
+            self.end_headers()
+            csr.execute("select * from contents where sha512 = %s",
+                        (dr["content_sha512"],))
+            cr = csr.fetchone()
+            if cr:
+                self.wfile.write(cr["content"])
+            else:
+                self.log_message(f"content missing for {dr['sha512']}")
+        else:
+            self.error404()
+
+
+    def error404(self):
+        self.send_response(http.HTTPStatus.NOT_FOUND)
+        self.send_header("Content-Type", "text/html; charset=utf-8")
+        self.end_headers()
+        self.wfile.write(b"<h1>404</h1><p>Nothing to see here</p>")
+
+
+
+ap = argparse.ArgumentParser()
+ap.add_argument("--baseurl", required=True)
+args = ap.parse_args()
+
+db = psycopg2.connect("")
+
+srv = hs.HTTPServer(("127.0.0.1", 8888), WSqlShRH)
+srv.socket = ssl.wrap_socket(
+                    srv.socket,
+                    server_side=True,
+                    certfile="localhost.crt",
+                    keyfile="localhost.key",
+                    ssl_version=ssl.PROTOCOL_TLSv1_2
+                 )
+srv.serve_forever()
+