Fix extension checks

Merged latest version from main branch

Author: lassoan

.github/workflows/extension-ai-analysis.yml

@@ -70,7 +70,9 @@ jobs:
 
       - name: Run AI analysis
         env:
-          NEBULA_API_KEY: ${{ secrets.NEBULA_API_KEY }}
+          INFERENCE_URL: ${{ vars.INFERENCE_URL }}
+          INFERENCE_MODEL: ${{ vars.INFERENCE_MODEL }}
+          INFERENCE_API_KEY: ${{ secrets.INFERENCE_API_KEY }}
         id: ai-analysis
         run: |
           RESULT=$(python scripts/extension_ai_analysis.py \

scripts/check_description_files.py

@@ -105,9 +105,10 @@ def check_json_file_format(extension_name, metadata, extension_file_path):
             extension_name, check_name,
             f"Invalid JSON format: {str(e)}")
     # Force using LF-only line endings
-    with open(extension_file_path, 'r', encoding='utf-8') as f:
+    # Must open in binary mode to detect line endings
+    with open(extension_file_path, 'rb') as f:
         content = f.read()
-    if '\r\n' in content or '\r' in content:
+    if b'\r\n' in content or b'\r' in content:
         raise ExtensionCheckError(
             extension_name, check_name,
             "File contains non-LF line endings (CR or CRLF). Please convert to LF-only line endings.")

scripts/extension_ai_analysis.py

@@ -15,13 +15,20 @@
 import subprocess
 import shutil
 
-# Use Nebula Block API endpoint for chat completions.
-# It offers capable models for free with an OpenAI-compatible API.
-INFERENCE_URL = "https://inference.nebulablock.com/v1/chat/completions"
-INFERENCE_MODEL = "mistralai/Mistral-Small-3.2-24B-Instruct-2506"
-INFERENCE_RESPONSE_PER_MINUTE_LIMIT = 4 #  slow down to not exceed token per minute (tpm) limit of 60k
-INFERENCE_API_KEY = os.getenv("NEBULA_API_KEY")
-INFERENCE_MAX_CHARACTERS = 100000  # max characters in all files provided to the model, approximately 25k tokens (limit is 32k)
+
+# Get inference server configuration from environment variables
+INFERENCE_URL = os.getenv("INFERENCE_URL")
+if not INFERENCE_URL:
+    raise ValueError("INFERENCE_URL environment variable is not set. Please set it before running the script.")
+INFERENCE_MODEL = os.getenv("INFERENCE_MODEL")
+if not INFERENCE_MODEL:
+    raise ValueError("INFERENCE_MODEL environment variable is not set. Please set it before running the script.")
+INFERENCE_API_KEY = os.getenv("INFERENCE_API_KEY")
+if not INFERENCE_API_KEY:
+    raise ValueError("INFERENCE_API_KEY environment variable is not set. Please set it before running the script.")
+
+INFERENCE_RESPONSE_PER_MINUTE_LIMIT = 10 #  slow down to not exceed token per minute (tpm) limit
+INFERENCE_MAX_CHARACTERS = 400000  # max characters in all files provided to the model, approximately 100k tokens
 
 QUESTIONS = [
     ["Is there a EXTENSION_DESCRIPTION variable in the CMakeLists.txt file that describes what the extension does in a few sentences that can be understood by a person knowledgeable in medical image computing?", ["cmake"]],
@@ -130,7 +137,9 @@ def collect_analyzed_files(folder):
 def ask_question(system_msg, question):
     headers = {
         "Content-Type": "application/json",
-        "Authorization": f"Bearer {INFERENCE_API_KEY}"
+        "Authorization": f"Bearer {INFERENCE_API_KEY}",
+        "HTTP-Referer": "slicer.org", # Optional. Site URL for rankings on openrouter.ai.
+        "X-Title": "3D Slicer", # Optional. Site title for rankings on openrouter.ai.
     }
 
     messages = [
@@ -157,6 +166,11 @@ def ask_question(system_msg, question):
     try:
         answer = response.json()["choices"][0]["message"]["content"]
     except Exception as e:
+        import traceback
+        traceback.print_exc()
+        print(f"Request data: {data}", file=sys.stderr)
+        print(f"Response status code: {response.status_code}", file=sys.stderr)
+        print(f"Response content: {response.text}", file=sys.stderr)
         raise RuntimeError(f"Error or unexpected response: {response.json()["error"]["message"]}")
 
     return answer