commit 29500f51d3c9142f1a604fac0a103e29144e6966
from: Aleksey Ryndin <continue@to.any-key.press>
date: Fri Sep 20 10:54:12 2024 UTC

Add: proxy images from <img> HTML tags

commit - dd301323d34fe54b42e2bd9b1a70eb5b3bf27f82
commit + 29500f51d3c9142f1a604fac0a103e29144e6966
blob - fa8564a68d1bf6620e56038a86973dad30983aa0
blob + 8d02380be94eb01fdb8f37a05aacb23ef5ec581a
--- lonk.py
+++ lonk.py
@@ -1,5 +1,6 @@
 #!/usr/bin/env python3
 from json import loads as json_loads
+from mimetypes import guess_type
 from os import environ
 from pathlib import Path
 from sys import stdout
@@ -119,11 +120,12 @@ class PreformattedTag(_BaseTag):
 
 
 class HtmlToGmi(HTMLParser):
-    def __init__(self, base_url):
+    def __init__(self, base_url, fn_media_url):
         super().__init__()
         self.gmi_text = []
         self.stack = []
         self.base_url = base_url
+        self.fn_media_url = fn_media_url
 
     def feed(self, html_text):
         super().feed(html_text)
@@ -163,7 +165,10 @@ class HtmlToGmi(HTMLParser):
             else:
                 src = img.get("src")
                 if src:
-                    self.gmi_text.append(f"=> {urljoin(self.base_url, src)} {title}")
+                    img_url = urljoin(self.base_url, src)
+                    mime, _ = guess_type(img_url)
+                    img_url = self.fn_media_url(mime, img_url)
+                    self.gmi_text.append(f"=> {img_url} {title}")
         elif tag == "br":
             if self.stack:
                 self.gmi_text.append(self.stack[-1].flush())
@@ -188,29 +193,31 @@ class HtmlToGmi(HTMLParser):
         return self.stack[0]
 
 
-def _format_honk(honk, header, lonk_url):
+def _format_honk(honk, header, fn_media_url):
     assert honk["Noise"] == honk["HTML"], honk
-    lines = [header, f'=> {honk["XID"]}', HtmlToGmi(_build_url()).feed(honk["HTML"])]
+    lines = [header, f'=> {honk["XID"]}', HtmlToGmi(_build_url(), fn_media_url).feed(honk["HTML"])]
     for donk in (honk.get("Donks") or []):
-        if environ.get("LONK_PROXY_MEDIA"):
-            gemini_url = urlunsplit(
-                (
-                    "gemini",
-                    lonk_url.netloc,
-                    lonk_url.path + "/proxy",
-                    urlencode({"m": donk["Media"], "u": donk["URL"]}),
-                    ""
-                )
-            )
-            lines.append(f'=> {gemini_url}')
-        else:
-            lines.append(f'=> {donk["URL"]}')
+        lines.append(f'=> {fn_media_url(donk["Media"], donk["URL"])}')
         desc = donk.get("Desc") or donk.get("Name")
         if desc:
             lines.append(desc)
     return "\n".join(lines)
 
 
+def _proxy_url_factory(proxy_media_enabled, lonk_url):
+    def _get_proxy_url(mime, url):
+        return urlunsplit(
+            (
+                "gemini",
+                lonk_url.netloc,
+                lonk_url.path + "/proxy",
+                urlencode({"m": mime, "u": url}),
+                "",
+            )
+        ) if mime and proxy_media_enabled else url
+    return _get_proxy_url
+
+
 def demo(lonk_url):
     collected = {}
 
@@ -246,9 +253,10 @@ def demo(lonk_url):
         line += f' ({home["chatcount"]})'
     print(line)
 
+    fn_media_url = _proxy_url_factory(environ.get("LONK_PROXY_MEDIA"), lonk_url)
     for honk, header in reversed(collected.values()):
         print()
-        print(_format_honk(honk, header, lonk_url))
+        print(_format_honk(honk, header, fn_media_url))
 
 
 def proxy(mime, url):