From c89a9df6e258606c1826a4bcdfce64735dfb27d8 Mon Sep 17 00:00:00 2001
From: Shane <shanehill@mail.com>
Date: Mon, 12 May 2025 21:16:53 +1000
Subject: [PATCH] update insert link naturally

---
 foodie_utils.py | 91 ++++++++++++++++++++++++++++++++++---------------
 1 file changed, 63 insertions(+), 28 deletions(-)

diff --git a/foodie_utils.py b/foodie_utils.py
index 5afe36e..eb9396f 100644
--- a/foodie_utils.py
+++ b/foodie_utils.py
@@ -613,45 +613,80 @@ def insert_link_naturally(summary, source_name, source_url):
             logging.error("No valid paragraphs to insert link.")
             return summary
 
+        # Find paragraphs with at least two sentences
         eligible_paragraphs = [p for p in paragraphs if p.strip() and len(re.split(r'(?<=[.!?])\s+', p.strip())) >= 2]
         if not eligible_paragraphs:
-            logging.warning("No paragraph with multiple sentences found, appending to last paragraph.")
-            target_para = paragraphs[-1].strip()
-            link_pattern = f'<a href="{source_url}">{source_name}</a>'
-            new_para = f"{target_para} Source: {link_pattern}."
-            paragraphs[-1] = new_para
-            new_summary = '\n'.join(paragraphs)
-            logging.info(f"Appended link to summary: {new_summary!r}")
-            return new_summary
-
-        target_para = random.choice(eligible_paragraphs)
-        sentences = re.split(r'(?<=[.!?])\s+', target_para.strip())
-        
-        eligible_sentences = [(i, s) for i, s in enumerate(sentences) if s.strip()]
-        if not eligible_sentences:
-            logging.error("No eligible sentences found for link insertion.")
-            return summary
-        
-        sentence_idx, sentence = random.choice(eligible_sentences)
+            logging.warning("No paragraph with multiple sentences found, using fallback.")
+            return append_link_as_fallback(summary, source_name, source_url)
+
+        # Alternative phrases for variety
+        link_phrases = [
+            "according to {source}",
+            "as reported by {source}",
+            "{source} notes that"
+        ]
+
+        best_candidate = None
+        best_score = -1
+        best_paragraph = None
+
+        # Score each paragraph and sentence for suitability
+        for para in eligible_paragraphs:
+            sentences = re.split(r'(?<=[.!?])\s+', para.strip())
+            eligible_sentences = [
+                (i, s) for i, s in enumerate(sentences)
+                if s.strip() and not s.endswith('?')  # Exclude sentences ending with '?'
+                and not s.endswith('!')  # Exclude exclamations for smoother integration
+            ]
+            if not eligible_sentences:
+                continue
+
+            # Score sentences based on suitability (prefer declarative sentences)
+            for idx, sentence in eligible_sentences:
+                score = 0
+                # Favor sentences with factual content (simplified heuristic)
+                if any(word in sentence.lower() for word in ["is", "are", "has", "shows", "reveals"]):
+                    score += 2
+                # Prefer longer sentences for better context
+                score += len(sentence.split()) // 5
+                # Prefer middle sentences for natural flow
+                score += abs(idx - len(sentences) / 2) * -1  # Penalize sentences far from the middle
+
+                if score > best_score:
+                    best_score = score
+                    best_candidate = (idx, sentence)
+                    best_paragraph = para
+
+        if best_candidate is None:
+            logging.warning("No suitable sentence found, using fallback.")
+            return append_link_as_fallback(summary, source_name, source_url)
+
+        # Select a link phrase based on sentence structure
+        sentence_idx, sentence = best_candidate
+        link_phrase = random.choice(link_phrases)
         link_pattern = f'<a href="{source_url}">{source_name}</a>'
-        
-        # Insert the link at the end of the sentence
-        new_sentence = f"{sentence.rstrip('.')} according to {link_pattern}."
-        
+        formatted_link = link_phrase.format(source=link_pattern)
+
+        # Insert the link at the end of the selected sentence (no capitalization needed)
+        new_sentence = f"{sentence.rstrip('.')} {formatted_link}."
         sentences[sentence_idx] = new_sentence
         new_para = ' '.join(sentences)
-        paragraphs[paragraphs.index(target_para)] = new_para
-        
+        paragraphs[paragraphs.index(best_paragraph)] = new_para
+
         new_summary = '\n'.join(paragraphs)
         logging.info(f"Summary with naturally embedded link: {new_summary!r}")
         return new_summary
 
     except Exception as e:
         logging.error(f"Link insertion failed: {e}")
-        link_pattern = f'<a href="{source_url}">{source_name}</a>'
-        new_summary = f"{summary}\n\nSource: {link_pattern}."
-        logging.info(f"Fallback summary with link: {new_summary!r}")
-        return new_summary
+        return append_link_as_fallback(summary, source_name, source_url)
+
+def append_link_as_fallback(summary, source_name, source_url):
+    """Fallback method to append the link in a polished way."""
+    link_pattern = f'<a href="{source_url}">{source_name}</a>'
+    new_summary = f"{summary}\n\n<p>We learned about this from {link_pattern}.</p>"
+    logging.info(f"Fallback summary with link: {new_summary!r}")
+    return new_summary
 
 def generate_category_from_summary(summary):
     try: