Move get_title() to Thread

[paperdoorknob] / glowfic.py

diff --git a/glowfic.py b/glowfic.py
index 573050872d3d3d3ef6e5647c02dcb4f238e65e21..92ae4ab81a195fceeda74869fad340d9718c4095 100644 (file)
--- a/glowfic.py
+++ b/glowfic.py
@@ -5,7 +5,6 @@
 # Free Software Foundation, version 3.
 
 
 # Free Software Foundation, version 3.
 
 

-from abc import ABC, abstractmethod

 from dataclasses import dataclass
 import itertools
 from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse
 from dataclasses import dataclass
 import itertools
 from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse


@@ -39,9 +38,9 @@ def flatURL(url: str) -> str:
 @dataclass(frozen=True)
 class Chunk:
     icon: str | None
 @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
     content: Tag
 
 # We avoid the name "post" because the Glowfic community uses the term


@@ -55,141 +54,135 @@ class Chunk:
 #  * Humans in the community tend to use "posts" to mean chunks.
 
 
 #  * Humans in the community tend to use "posts" to mean chunks.
 
 

-def chunkDOMs(html: BeautifulSoup) -> Iterable[Tag]:
-    def text() -> Tag:
-        body = html.body
-        assert body
-        text = body.find_next("div", class_="post-post")
-        assert isinstance(text, Tag)
-        return text
+class Thread:

 
 

-    def the_replies() -> Iterable[Tag]:
-        rs = html.find_all("div", class_="post-reply")
-        assert all(isinstance(r, Tag) for r in rs)
-        return rs
+    def __init__(self, dom: BeautifulSoup) -> None:
+        self._dom = dom

 
 

-    return itertools.chain([text()], the_replies())
+    def title(self) -> str | None:
+        span = self._dom.findChild("span", id="post-title")
+        if not isinstance(span, Tag):
+            return None
+        return span.text.strip()
+
+    def chunkDOMs(self) -> Iterable[Tag]:
+        def text() -> Tag:
+            body = self._dom.body
+            assert body
+            text = body.find_next("div", class_="post-post")
+            assert isinstance(text, Tag)
+            return text
+
+        def the_replies() -> Iterable[Tag]:
+            rs = self._dom.find_all("div", class_="post-reply")
+            assert all(isinstance(r, Tag) for r in rs)
+            return rs
+
+        return itertools.chain([text()], the_replies())

 
 
 def makeChunk(chunk_dom: Tag, image_store: ImageStore) -> Chunk:
 
     def getIcon() -> str | None:
 
 
 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
         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'])
 
         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 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(),
     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)
 
 
                  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
-
-
-class Layout(ABC):
-
-    @abstractmethod
-    def renderChunk(self, chunk: Chunk) -> bytes:
-        raise NotImplementedError()
-
-
-class ContentOnlyLayout(Layout):
-
-    def __init__(self, texifier: Texifier) -> None:
-        self._texifier = texifier
+def renderChunk(texifier: Texifier, chunk: Chunk) -> bytes:
+    return b''.join([
+        br'\glowhead{',
+        br'\glowicon{%s}' % chunk.icon.encode('UTF-8') if chunk.icon else b'',
+        b'}{',
+        texifier.texify(chunk.character) if chunk.character else b'',
+        b'}{',
+        texifier.texify(chunk.screen_name) if chunk.screen_name else b'',
+        b'}{',
+        texifier.texify(chunk.author) if chunk.author else b'',
+        b'}',
+        texifier.texify(chunk.content)])
+
+
+ContentOnlyLayout = br'''
+\newcommand{\glowhead}[4]{}
+'''
+
+
+BelowIconLayout = br'''
+\newcommand{\glowhead}[4]{\wrapstuffclear
+\begin{wrapstuff}[l]
+\fbox{
+\begin{varwidth}{0.5\textwidth}
+  \smash{\parbox[t][0pt]{0pt}{
+    \setlength{\fboxrule}{0.2pt}
+    \setlength{\fboxsep}{0pt}
+    \vspace{-3.4pt}
+    \fbox{\hspace{107mm}}
+  }\\*}
+  \vspace{-1em}
+\begin{center}
+#1\ifnotempty
+{#1}{\\*}#2\ifnotempty
+{#2}{\\*}#3\ifnotempty
+{#3}{\\*}#4
+\end{center}
+\end{varwidth}
+}
+\end{wrapstuff}

 
 

-    def renderChunk(self, chunk: Chunk) -> bytes:
-        return self._texifier.texify(chunk.content) + b'\n'
+\strut

 
 

+\noindent}'''

 
 

-class BelowIconLayout(Layout):

 
 

-    def __init__(self, texifier: Texifier, image_size: float) -> None:
-        self._texifier = texifier
-        self._image_size = image_size
+# Why is \textwidth not the width of the text?
+# Why is the width of the text .765\textwidth?
+BesideIconLayout = br'''
+\newcommand{\glowhead}[4]{

 
 

-    def renderChunk(self, chunk: Chunk) -> bytes:
-        meta = [
-            renderIcon(chunk.icon, self._image_size),
-            chunk.character,
-            chunk.screen_name,
-            chunk.author,
-        ]
+\strut

 
 

-        return b'''\\wrapstuffclear
-\\begin{wrapstuff}[l]
-\\fbox{
-\\begin{varwidth}{0.5\\textwidth}
-  \\smash{\\parbox[t][0pt]{0pt}{
-    \\setlength{\\fboxrule}{0.2pt}
-    \\setlength{\\fboxsep}{0pt}
-    \\vspace{-3.4pt}
-    \\fbox{\\hspace{107mm}}
-  }\\\\*}
-  \\vspace{-1em}
-\\begin{center}
-%s
-\\end{center}
-\\end{varwidth}
-}
-\\end{wrapstuff}
-
-\\strut
-
-\\noindent %s
-''' % (
-            b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None),
-            self._texifier.texify(chunk.content))
-
-
-class BesideIconLayout(Layout):
-
-    def __init__(self, texifier: Texifier, image_size: float) -> None:
-        self._texifier = texifier
-        self._image_size = image_size
-
-    def renderChunk(self, chunk: Chunk) -> bytes:
-        icon = renderIcon(chunk.icon, self._image_size)
-        meta = [
-            chunk.character,
-            chunk.screen_name,
-            chunk.author,
-        ]
-
-        # Why is \textwidth not the width of the text?
-        # Why is the width of the text .765\textwidth?
-        return b'''\\noindent\\fbox{
-%s
-\\parbox[b]{.765\\textwidth}{
-\\begin{center}
-%s
-\\end{center}
+\noindent\fbox{
+#1
+\parbox[b]{.765\textwidth}{
+\begin{center}
+#2\ifnotempty
+{#2}{\\*}#3\ifnotempty
+{#3}{\\*}#4
+\end{center}

 }
 }

-}\\\\*
-\\vspace{-0.75em}\\\\*
-\\noindent %s
-
-\\strut
-
-''' % (
-            icon.encode('UTF-8') if icon else b'',
-            b'\\\\*'.join(x.encode('UTF-8') for x in meta if x is not None),
-            self._texifier.texify(chunk.content))
+}\\*
+\vspace{-0.75em}\\*
+\noindent}'''