feat: Enhances wiki article info fetching

Refactors to integrate fetching article info, badges, and interwiki links in a single request. Adds support to display category members and improves interwiki link processing by translating links where feasible. Adds error handling for interwiki processing. Updates HTML template to display category members.

Improves performance and reduces external requests, enhancing user navigation and data retrieval efficiency.

Fixes #19.

src/wikimore/app.py

@@ -392,25 +392,55 @@ def wiki_article(
 
     logger.debug(f"Fetching {title} from {base_url}")
 
-    api_request = urllib.request.Request(
+    # Check if the article is something we need to handle differently
-        f"{base_url}/api/rest_v1/page/html/{escape(quote(title.replace(' ', '_')), True).replace('/', '%2F')}",
+    info_api_request = urllib.request.Request(
-        headers=HEADERS,
+        f"{base_url}/w/api.php?action=query&format=json&titles={escape(quote(title.replace(' ', '_')), True)}&prop=info|pageprops|categoryinfo|langlinks&lllimit=500",
-    )
-
-    logger.debug(f"Request URL: {api_request.full_url}")
-
-    # Use the MediaWiki API to fetch any badges for the article
-    api_request_badges = urllib.request.Request(
-        f"{base_url}/w/api.php?action=query&format=json&titles={escape(quote(title.replace(' ', '_')), True)}&prop=pageprops",
         headers=HEADERS,
     )
 
+    category_members = []
+    interwiki = []
     badges = []
 
-    with urllib.request.urlopen(api_request_badges) as response:
+    with urllib.request.urlopen(info_api_request) as response:
-        logger.debug(f"Tried to fetch badges from {api_request_badges.full_url}")
+        logger.debug(
+            f"Tried to fetch info for {title} from {info_api_request.full_url}"
+        )
         data = json.loads(response.read().decode())
         page = data["query"]["pages"].popitem()[1]
+
+        langlinks = page.get("langlinks", [])
+
+        logger.debug(f"Original Interwiki links for {title}: {langlinks}")
+
+        # Get interwiki links and translate them to internal links where possible
+        for link in langlinks:
+            try:
+                interwiki_lang = link["lang"]
+                interwiki_title = link["*"]
+
+                logger.debug(
+                    f"Generating interwiki link for: {interwiki_lang}.{project}/{interwiki_title}"
+                )
+
+                interwiki_url = url_for(
+                    "wiki_article",
+                    project=project,
+                    lang=interwiki_lang,
+                    title=interwiki_title,
+                )
+                link["url"] = interwiki_url
+
+                link["langname"] = app.languages[interwiki_lang]["name"]
+
+                interwiki.append(link)
+
+            except KeyError as e:
+                logger.error(
+                    f"Error processing interwiki link for title {title} in language {lang}: {e}"
+                )
+
+        # Get badges (e.g. "Good Article", "Featured Article")
         props = page.get("pageprops", {})
 
         for prop in props:
@@ -418,6 +448,7 @@ def wiki_article(
                 try:
                     badge_id = prop.replace("wikibase-badge-", "")
 
+                    # Fetch the badge data from Wikidata
                     badge_request = urllib.request.Request(
                         f"https://www.wikidata.org/w/api.php?action=wbgetentities&format=json&ids={badge_id}&languages={lang}",
                         headers=HEADERS,
@@ -449,52 +480,58 @@ def wiki_article(
                 except Exception as e:
                     logger.error(f"Error fetching badge {prop}: {e}")
 
-    # Use the MediaWiki API to fetch interwiki links
+        # If the article is a category, fetch the category members
-    api_request_interwiki = urllib.request.Request(
+        if "categoryinfo" in page:
-        f"{base_url}/w/api.php?action=query&format=json&titles={escape(quote(title.replace(' ', '_')), True)}&prop=langlinks&lllimit=500",
+            category_api_url = f"{base_url}/w/api.php?action=query&format=json&list=categorymembers&cmtitle={escape(quote(title.replace(' ', '_')), True)}&cmlimit=500"
+
+            category_api_request = urllib.request.Request(
+                category_api_url,
                 headers=HEADERS,
             )
 
-    with urllib.request.urlopen(api_request_interwiki) as response:
+            all_members = []
+
+            with urllib.request.urlopen(category_api_request) as category_api_response:
                 logger.debug(
-            f"Tried to fetch interwiki links from {api_request_interwiki.full_url}"
+                    f"Tried to fetch category members for {title} from {category_api_request.full_url}"
                 )
-        data = json.loads(response.read().decode())
+                data = json.loads(category_api_response.read().decode())
-        langlinks = data["query"]["pages"].popitem()[1].get("langlinks", [])
+                category_members = data["query"]["categorymembers"]
+                all_members += category_members
 
-    logger.debug(f"Original Interwiki links: {langlinks}")
+                if "continue" in data:
-
+                    continue_params = f"&cmcontinue={data['continue']['cmcontinue']}"
-    interwiki = []
+                    category_api_request = urllib.request.Request(
-
+                        category_api_url + continue_params,
-    # Translate the interwiki links to internal links where possible
+                        headers=HEADERS,
-    for link in langlinks:
-        try:
-            interwiki_lang = link["lang"]
-            interwiki_title = link["*"]
-
-            logger.debug(
-                f"Generating interwiki link for: {interwiki_lang}.{project}/{interwiki_title}"
                     )
 
-            interwiki_url = url_for(
+                    with urllib.request.urlopen(
+                        category_api_request
+                    ) as category_api_response:
+                        data = json.loads(category_api_response.read().decode())
+                        all_members += data["query"]["categorymembers"]
+
+            category_members = all_members
+
+            for member in category_members:
+                member["url"] = url_for(
                     "wiki_article",
                     project=project,
-                lang=interwiki_lang,
+                    lang=lang,
-                title=interwiki_title,
+                    title=member["title"],
-            )
-            link["url"] = interwiki_url
-
-            link["langname"] = app.languages[interwiki_lang]["name"]
-
-            interwiki.append(link)
-
-        except KeyError as e:
-            logger.error(
-                f"Error processing interwiki link for title {title} in language {lang}: {e}"
                 )
 
     interwiki = langsort(interwiki)
 
+    # Prepare the API request to fetch the article content
+    api_request = urllib.request.Request(
+        f"{base_url}/api/rest_v1/page/html/{escape(quote(title.replace(' ', '_')), True).replace('/', '%2F')}",
+        headers=HEADERS,
+    )
+
+    logger.debug(f"Article content URL: {api_request.full_url}")
+
     # Add the `variant` header if the `variant` query parameter is present
     # This is used to fetch articles in a specific script variant (https://www.mediawiki.org/wiki/Writing_systems/LanguageConverter)
     if request.args.get("variant", None):
@@ -679,6 +716,7 @@ def wiki_article(
         license=license,
         interwiki=interwiki,
         badges=badges,
+        category_members=category_members,
     )
 
 

src/wikimore/templates/article.html

@@ -32,6 +32,14 @@
         </div>
     {% endif %}
     {{ content|safe }}
+    {% if category_members %}
+        <h2>Pages in category "{{ title }}"</h2>
+        <ul>
+            {% for member in category_members %}
+                <li><a href="{{ member.url }}">{{ member.title }}</a></li>
+            {% endfor %}
+        </ul>
+    {% endif %}
 {% endblock %}
 
 {% block license %}