Split integration tests to separate files and add test for salam pax's blog

Author: bohdanbobrowski

blog2epub/crawlers/article_factory/blogspot.py

@@ -9,8 +9,9 @@
 class BlogspotArticleFactory(DefaultArticleFactory):
     def get_title(self) -> Optional[str]:
         title = super().get_title()
-        if self.blog_title is not None and title == self.blog_title:
-            title = "Pomidor"
+        date = self.get_date()
+        if self.blog_title is not None and title == self.blog_title and date is not None:
+            title = date.strftime("%A, %d %B %Y, %H:%M")
         return title
 
     def process(self) -> ArticleModel:

blog2epub/crawlers/article_factory/default.py

@@ -22,9 +22,13 @@ def get_title(self) -> Optional[str]:
                 if title_pattern.xpath:
                     title = self.tree.xpath(title_pattern.xpath)
                     if len(title) > 1:
-                        title = title[0]
-                        title = html.unescape(title.strip())
+                        title = html.unescape(title[0].strip())
                         break
+        while isinstance(title, list):
+            try:
+                title = title[0]
+            except IndexError:
+                title = None
         return title
 
     def get_date(self) -> Optional[datetime]:

poetry.lock

@@ -31,6 +31,7 @@ click = "^8.1.7"
 webencodings = "^0.5.1"
 soupsieve = "^2.6"
 pytz = "^2024.2"
+ftfy = "^6.3.1"
 
 [tool.poetry.group.docs.dependencies]
 mkdocs = "^1.6.0"