Better URL handling.

1 files changed, 19 insertions, 7 deletions

diff --git a/gemfeed.py b/gemfeed.py
index 5b1e68c..0cbe6a4 100644
--- a/gemfeed.py
+++ b/gemfeed.py
@@ -2,6 +2,7 @@ import argparse
 import datetime
 import glob
 import os.path
+import urllib.parse
 
 from feedgen.feed import FeedGenerator
 
@@ -15,7 +16,7 @@ def find_files():
     return files
 
 def populate_entry_from_file(filename, base_url, entry):
-    url = base_url + filename
+    url = urljoin(base_url, filename)
     entry.guid(url)
     entry.link(href=url, rel="alternate")
     updated = os.path.getctime(filename)
@@ -32,6 +33,15 @@ def populate_entry_from_file(filename, base_url, entry):
             title = filename
     entry.title(title)
 
+def urljoin(base, url):
+    base = urllib.parse.urlsplit(base)
+    base = base._replace(scheme="https")
+    base = urllib.parse.urlunsplit(base)
+    joined = urllib.parse.urljoin(base, url)
+    joined = urllib.parse.urlsplit(joined)
+    joined = joined._replace(scheme="gemini")
+    return urllib.parse.urlunsplit(joined)
+
 def main():
 
     # Parse arguments
@@ -50,11 +60,13 @@ def main():
             default="A Gemini feed", help='feed title')
     args = parser.parse_args()
 
-    # Normalise base URL, quick and dirty
-    if not args.base_url.startswith("gemini://"):
-        args.base_url = "gemini://" + args.base_url
-    if not args.base_url.endswith("/"):
-        args.base_url = args.base_url + "/"
+    # Normalise base URL
+    base_url = urllib.parse.urlsplit(args.base_url)
+    if not base_url.netloc and base_url.path:
+        # Handle a naked domain, which urlsplit will interpet at a local path
+        base_url = base_url._replace(netloc=base_url.path, path="")
+    base_url = base_url._replace(scheme="gemini")
+    args.base_url = urllib.parse.urlunsplit(base_url)
 
     # Setup feed
     feed = FeedGenerator()
@@ -70,7 +82,7 @@ def main():
     if author:
         feed.author(author)
     feed.link(href=args.base_url, rel='alternate')
-    feed.link(href=args.base_url + args.output, rel='self')
+    feed.link(href=urljoin(args.base_url, args.output), rel='self')
 
     # Add one entry per .gmi file
     files = find_files()