Escape character names

[paperdoorknob] / glowfic.py

diff --git a/glowfic.py b/glowfic.py
index d1966cae628267177b08789aaf8714c81c01f6d3..a700995bf93f6950d2b63a403ee57c4f757c0d9a 100644 (file)
--- a/glowfic.py
+++ b/glowfic.py
@@ -8,6 +8,7 @@
 from abc import ABC, abstractmethod
 from dataclasses import dataclass
 import itertools
+from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse
 
 from typing import Iterable
 
@@ -18,12 +19,29 @@ from images import ImageStore
 from texify import Texifier
 
 
+def _removeViewFromURL(url: str) -> str:
+    u = urlparse(url)
+    old_qs = parse_qsl(u.query)
+    new_qs = [(k, v) for k, v in old_qs if k != 'view']
+    return urlunparse(u._replace(query=urlencode(new_qs)))
+
+
+def nonFlatURL(url: str) -> str:
+    return _removeViewFromURL(url)
+
+
+def flatURL(url: str) -> str:
+    u = urlparse(_removeViewFromURL(url))
+    qs = parse_qsl(u.query) + [('view', 'flat')]
+    return urlunparse(u._replace(query=urlencode(qs)))
+
+
 @dataclass(frozen=True)
 class Chunk:
     icon: str | None
-    character: str | None
-    screen_name: str | None
-    author: str | None
+    character: Tag | None
+    screen_name: Tag | None
+    author: Tag | None
     content: Tag
 
 # We avoid the name "post" because the Glowfic community uses the term
@@ -65,19 +83,18 @@ def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
         assert isinstance(icon_img, Tag)
         return image_store.get_image(icon_img.attrs['src'])
 
-    def getTextByClass(css_class: str) -> str | None:
-        div = chunk_dom.find_next('div', class_=css_class)
-        if div is None:
-            return None
-        return div.text.strip()
+    def getByClass(css_class: str) -> Tag | None:
+        tag = chunk_dom.find_next('div', class_=css_class)
+        assert tag is None or isinstance(tag, Tag)
+        return tag
 
     content = chunk_dom.find_next('div', class_='post-content')
     assert isinstance(content, Tag)
 
     return Chunk(getIcon(),
-                 getTextByClass('post-character'),
-                 getTextByClass('post-screenname'),
-                 getTextByClass('post-author'),
+                 getByClass('post-character'),
+                 getByClass('post-screenname'),
+                 getByClass('post-author'),
                  content)
 
 
@@ -109,12 +126,11 @@ class BelowIconLayout(Layout):
         self._image_size = image_size
 
     def renderChunk(self, chunk: Chunk) -> bytes:
-        meta = [
-            renderIcon(chunk.icon, self._image_size),
-            chunk.character,
-            chunk.screen_name,
-            chunk.author,
-        ]
+        icon = renderIcon(chunk.icon, self._image_size)
+        meta = [icon.encode('UTF-8')] if icon else []
+        meta += [self._texifier.texify(x)
+                 for x in [chunk.character, chunk.screen_name, chunk.author]
+                 if x is not None]
 
         return b'''\\wrapstuffclear
 \\begin{wrapstuff}[l]
@@ -138,7 +154,7 @@ class BelowIconLayout(Layout):
 
 \\noindent %s
 ''' % (
-            b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None),
+            b'\\\\*'.join(meta),
             self._texifier.texify(chunk.content))
 
 
@@ -173,5 +189,5 @@ class BesideIconLayout(Layout):
 
 ''' % (
             icon.encode('UTF-8') if icon else b'',
-            b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None),
+            b'\\\\*'.join(self._texifier.texify(x) for x in meta if x is not None),
             self._texifier.texify(chunk.content))