]> git.scottworley.com Git - paperdoorknob/commitdiff
Move get_title() to Thread
authorScott Worley <scottworley@scottworley.com>
Fri, 12 Jan 2024 03:00:56 +0000 (19:00 -0800)
committerScott Worley <scottworley@scottworley.com>
Fri, 12 Jan 2024 03:00:56 +0000 (19:00 -0800)
glowfic.py
glowfic_test.py
paperdoorknob.py

index 746a7a2b7165cb6df2d9f6bb691d5e68ce20be24..92ae4ab81a195fceeda74869fad340d9718c4095 100644 (file)
@@ -59,6 +59,12 @@ class Thread:
     def __init__(self, dom: BeautifulSoup) -> None:
         self._dom = dom
 
     def __init__(self, dom: BeautifulSoup) -> None:
         self._dom = dom
 
+    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
     def chunkDOMs(self) -> Iterable[Tag]:
         def text() -> Tag:
             body = self._dom.body
index ed32945d50f1bcce42f507346146761c9a029468..64da1649b1209b2bb5b37f914e6f40f65aae89b7 100644 (file)
@@ -72,5 +72,20 @@ class TestMakeChunk(unittest.TestCase):
             PandocTexifier("pandoc").texify(c.character), b'{Keltham}\n')
 
 
             PandocTexifier("pandoc").texify(c.character), b'{Keltham}\n')
 
 
+class TestThread(unittest.TestCase):
+
+    def testTitle(self) -> None:
+        t = Thread(BeautifulSoup(b'''
+            <html><body>
+              <div class="content-header">
+                <span id="post-title">
+                  <a href="/posts/1234">Teh Story!</a>
+                </span>
+              </div>
+              <div class="post-container post-post">The "post"</div>
+            </body></html>''', 'html.parser'))
+        self.assertEqual(t.title(), 'Teh Story!')
+
+
 if __name__ == '__main__':
     unittest.main()
 if __name__ == '__main__':
     unittest.main()
index ade0b885d42addb973f415599798bd2e153ac961..cf500571604c9590920d36090e8484f4ba8ad498 100644 (file)
@@ -7,7 +7,6 @@
 from typing import Any, Iterable
 
 from bs4 import BeautifulSoup
 from typing import Any, Iterable
 
 from bs4 import BeautifulSoup
-from bs4.element import Tag
 
 from args import spec_from_commandline_args
 from glowfic import flatURL, makeChunk, renderChunk, Thread
 
 from args import spec_from_commandline_args
 from glowfic import flatURL, makeChunk, renderChunk, Thread
@@ -22,13 +21,6 @@ def ilen(it: Iterable[Any]) -> int:
     return sum(1 for _ in it)
 
 
     return sum(1 for _ in it)
 
 
-def get_title(dom: BeautifulSoup) -> str | None:
-    span = dom.findChild("span", id="post-title")
-    if not isinstance(span, Tag):
-        return None
-    return span.text
-
-
 def process(spec: Spec) -> None:
     spec.texout.write(br'''\documentclass{article}
 \usepackage{booktabs}
 def process(spec: Spec) -> None:
     spec.texout.write(br'''\documentclass{article}
 \usepackage{booktabs}
@@ -60,7 +52,7 @@ def process(spec: Spec) -> None:
     thread = Thread(dom)
     spec.log('Counting chunks...\r')
     num_chunks = ilen(thread.chunkDOMs())
     thread = Thread(dom)
     spec.log('Counting chunks...\r')
     num_chunks = ilen(thread.chunkDOMs())
-    title = get_title(dom) or "chunk"
+    title = thread.title() or "chunk"
     for i, r in enumerate(thread.chunkDOMs()):
         percent = 100.0 * i / num_chunks
         spec.log(f'Processing {title} {i} of {num_chunks} ({percent:.1f}%)\r')
     for i, r in enumerate(thread.chunkDOMs()):
         percent = 100.0 * i / num_chunks
         spec.log(f'Processing {title} {i} of {num_chunks} ({percent:.1f}%)\r')