Strip links from meta fields

[paperdoorknob] / glowfic.py

diff --git a/glowfic.py b/glowfic.py
index 573050872d3d3d3ef6e5647c02dcb4f238e65e21..8561cd756236bc1c5bf37c2b913ac91fad0b6a77 100644 (file)
--- a/glowfic.py
+++ b/glowfic.py
@@ -39,9 +39,9 @@ def flatURL(url: str) -> str:
 @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
@@ -74,34 +74,48 @@ def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]:
 def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
 
     def getIcon() -> str | None:
-        icon_div = chunk_dom.find_next('div', class_='post-icon')
+        icon_div = chunk_dom.findChild('div', class_='post-icon')
         if icon_div is None:
             return None
-        icon_img = icon_div.find_next('img')
+        assert isinstance(icon_div, Tag)
+        icon_img = icon_div.findChild('img')
         if icon_img is None:
             return None
         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:
+    def getByClass(css_class: str) -> Tag | None:
+        tag = chunk_dom.findChild('div', class_=css_class)
+        assert tag is None or isinstance(tag, Tag)
+        return tag
+
+    def stripHREF(tag: Tag) -> None:
+        for c in tag.findChildren("a"):
+            if "href" in c.attrs:
+                del c.attrs["href"]
+
+    def getMeta(css_class: str) -> Tag | None:
+        tag = getByClass(css_class)
+        if tag is None:
             return None
-        return div.text.strip()
+        stripHREF(tag)
+        return tag
 
-    content = chunk_dom.find_next('div', class_='post-content')
+    content = chunk_dom.findChild('div', class_='post-content')
     assert isinstance(content, Tag)
 
     return Chunk(getIcon(),
-                 getTextByClass('post-character'),
-                 getTextByClass('post-screenname'),
-                 getTextByClass('post-author'),
+                 getMeta('post-character'),
+                 getMeta('post-screenname'),
+                 getMeta('post-author'),
                  content)
 
 
-def renderIcon(icon_path: str | None, image_size: float) -> str | None:
-    params = f'width={image_size}mm,height={image_size}mm,keepaspectratio'
-    return f'\\includegraphics[{params}]{{{icon_path}}}' if icon_path else None
+def renderIcon(icon_path: str | None, image_size: float) -> bytes | None:
+    if icon_path is None:
+        return None
+    return b'\\includegraphics[width=%fmm,height=%fmm,keepaspectratio]{%s}' % (
+        image_size, image_size, icon_path.encode('UTF-8'))
 
 
 class Layout(ABC):
@@ -127,12 +141,11 @@ 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] 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]
@@ -156,7 +169,7 @@ 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))
 
 
@@ -190,6 +203,6 @@ class BesideIconLayout(Layout):
 \\strut
 
 ''' % (
-            icon.encode('UTF-8') if icon else b'',
-            b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None),
+            icon if icon else b'',
+            b'\\\\*'.join(self._texifier.texify(x) for x in meta if x is not None),
             self._texifier.texify(chunk.content))