From 82b26daf37e25d3a6e3a3ebce0a0857185180823 Mon Sep 17 00:00:00 2001
From: "Paul Gauthier (aider)" <paul@aider.chat>
Date: Sun, 20 Apr 2025 11:00:49 -0700
Subject: [PATCH] feat: display matching entries side-by-side with diff
 highlighting

---
 scripts/clean_metadata.py | 39 +++++++++++++++++++++++++++++++++++----
 1 file changed, 35 insertions(+), 4 deletions(-)

diff --git a/scripts/clean_metadata.py b/scripts/clean_metadata.py
index dcdfbb5f4..defca1606 100644
--- a/scripts/clean_metadata.py
+++ b/scripts/clean_metadata.py
@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 
 import json
+import difflib
 from pathlib import Path
 
 import json5
@@ -43,12 +44,42 @@ def main():
     litellm_keys = set(litellm_data.keys())
     aider_keys = set(aider_data.keys())
 
-    common_keys = litellm_keys.intersection(aider_keys)
+    common_keys = sorted(list(litellm_keys.intersection(aider_keys)))
 
     if common_keys:
-        print("Common models found in both files:")
-        for key in sorted(list(common_keys)):
-            print(f"- {key}")
+        print("Comparing common models found in both files:\n")
+        for key in common_keys:
+            print(f"--- {key} (litellm) ---")
+            print(f"+++ {key} (aider) +++")
+
+            litellm_entry = litellm_data.get(key, {})
+            aider_entry = aider_data.get(key, {})
+
+            # Convert dicts to formatted JSON strings for comparison
+            litellm_json = json.dumps(litellm_entry, indent=4, sort_keys=True).splitlines()
+            aider_json = json.dumps(aider_entry, indent=4, sort_keys=True).splitlines()
+
+            # Generate unified diff
+            diff = difflib.unified_diff(
+                litellm_json,
+                aider_json,
+                fromfile=f"{key} (litellm)",
+                tofile=f"{key} (aider)",
+                lineterm="",
+                n=max(len(litellm_json), len(aider_json)),  # Show all lines
+            )
+
+            # Print the diff, skipping the header lines generated by unified_diff
+            diff_lines = list(diff)[2:]
+            if not diff_lines:
+                print("(No differences found)")
+            else:
+                for line in diff_lines:
+                    # Add color for better readability (optional, requires a library like 'termcolor' or manual ANSI codes)
+                    # Simple +/- indication is standard for diffs
+                    print(line)
+            print("\n" + "=" * 40 + "\n")
+
     else:
         print("No common models found between the two files.")