truncate articles from the rss feed

Gradio_UI.py

@@ -290,8 +290,7 @@ class GradioUI:
                 [stored_messages, text_input],
             ).then(self.interact_with_agent, [stored_messages, chatbot], [chatbot])
 
-        share = not os.getenv("GRADIO_SHARE") == "false"
-        print("GRADIO_SHARE is set to", os.getenv("GRADIO_SHARE"))
-        demo.launch(debug=True, share=share, **kwargs)
+        gradio_share = not os.getenv("GRADIO_SHARE") == "false"
+        demo.launch(debug=True, share=gradio_share, **kwargs)
 
 __all__ = ["stream_to_gradio", "GradioUI"]

app.py

@@ -32,7 +32,8 @@ def get_current_time_in_timezone(timezone: str) -> str:
 
 def choose_model():
     if os.getenv("OLLAMA_MODEL"):
-        print("Using an Ollama model")
+        print("Using an Ollama model: ", os.getenv("OLLAMA_MODEL"))
+
         return LiteLLMModel(
             model_id=os.getenv("OLLAMA_MODEL"),
             api_base=os.getenv("OLLAMA_ENDPOINT"),

ollama.env

@@ -1,5 +1,5 @@
 OLLAMA_ENDPOINT="http://localhost:11434"
-OLLAMA_MODEL="ollama_chat/llama3-groq-tool-use"
+OLLAMA_MODEL="ollama_chat/mistral:7b"
 OLLAMA_KEY=""
 
 GRADIO_SHARE=false

tools/read_rss_feed.py

@@ -1,13 +1,35 @@
 from smolagents.tools import Tool
+from markdownify import MarkdownConverter
+
+class CustomMarkdownConverter(MarkdownConverter):
+    """
+    Create a custom Markdown converter that ignores images and
+    ignores links, keeping only their text.
+    """
+    def convert_img(self, el, text, parent_tags):
+        return ""
+    def convert_a(self, el, text, parent_tags):
+        return text
+
+def markdown_summary(html, article_limit):
+    rendered = CustomMarkdownConverter().convert(html).strip()
+    if len(rendered) <= article_limit:
+        return rendered
+    else:
+        return rendered[:article_limit] + '...'
 
 class ReadRssFeedTool(Tool):
     name = "read_rss_feed"
-    description = "Read the articles from an RSS feed."
-    inputs = {'url': {'type': 'string', 'description': 'The url of the RSS feed (example: https://www.eldiario.es/rss).'}}
+    description = "Read the articles from an RSS feed and return them in a single text with markdown format."
+    inputs = {
+        'url': {'type': 'string', 'description': 'The url of the RSS feed (example: https://www.eldiario.es/rss).'},
+        'article_limit': {'type': 'integer', 'nullable': True, 'default': 512, 'description': 'Number of characters that will be considered for each article'},
+        'feed_limit': {'type': 'integer', 'nullable': True, 'default': 4096, 'description': 'Number of characters that will be considered for the whole article'},
+    }
     output_type = "string"
 
-    def forward(self, url: str) -> str:
-        """Read articles from an RSS feed."""
+    def forward(self, url: str, article_limit: int=512, feed_limit: int=4096) -> str:
+        """Read all the articles from a given RSS feed and return them as a single markdown formatted text."""
         import re
         from requests import get
         from rss_parser import RSSParser
@@ -18,6 +40,6 @@ class ReadRssFeedTool(Tool):
         for item in rss.channel.items:
             articles += "## " + item.title.content + "\n"
             articles += item.pub_date.content + "\n\n"
-            content = markdownify(item.description.content).strip()
+            content = markdown_summary(item.description.content, article_limit)
             articles += content + "\n\n"
-        return articles
+        return articles[:feed_limit]