Added local-runner requirements validation step (#712)

* added local-runner requirements validation step

* added post check for ollama toolkit

* fixed tests

Author: mogith-pn

clarifai/cli/model.py

@@ -5,7 +5,12 @@
 import click
 
 from clarifai.cli.base import cli, pat_display
-from clarifai.utils.cli import validate_context
+from clarifai.utils.cli import (
+    check_ollama_installed,
+    check_requirements_installed,
+    customize_ollama_model,
+    validate_context,
+)
 from clarifai.utils.constants import (
     DEFAULT_LOCAL_RUNNER_APP_ID,
     DEFAULT_LOCAL_RUNNER_COMPUTE_CLUSTER_CONFIG,
@@ -22,49 +27,6 @@
 from clarifai.utils.misc import GitHubDownloader, clone_github_repo, format_github_repo_url
 
 
-def customize_ollama_model(model_path, model_name, port, context_length):
-    """Customize the Ollama model name in the cloned template files.
-    Args:
-     model_path: Path to the cloned model directory
-     model_name: The model name to set (e.g., 'llama3.1', 'mistral')
-
-    """
-    model_py_path = os.path.join(model_path, "1", "model.py")
-
-    if not os.path.exists(model_py_path):
-        logger.warning(f"Model file {model_py_path} not found, skipping model name customization")
-        return
-
-    try:
-        # Read the model.py file
-        with open(model_py_path, 'r') as file:
-            content = file.read()
-        if model_name:
-            # Replace the default model name in the load_model method
-            content = content.replace(
-                'self.model = os.environ.get("OLLAMA_MODEL_NAME", \'llama3.2\')',
-                f'self.model = os.environ.get("OLLAMA_MODEL_NAME", \'{model_name}\')',
-            )
-
-        if port:
-            # Replace the default port variable in the model.py file
-            content = content.replace("PORT = '23333'", f"PORT = '{port}'")
-
-        if context_length:
-            # Replace the default context length variable in the model.py file
-            content = content.replace(
-                "context_length = '8192'", f"context_length = '{context_length}'"
-            )
-
-        # Write the modified content back to model.py
-        with open(model_py_path, 'w') as file:
-            file.write(content)
-
-    except Exception as e:
-        logger.error(f"Failed to customize Ollama model name in {model_py_path}: {e}")
-        raise
-
-
 @cli.group(
     ['model'], context_settings={'max_content_width': shutil.get_terminal_size().columns - 10}
 )
@@ -164,6 +126,11 @@ def init(
 
     # --toolkit option
     if toolkit == 'ollama':
+        if not check_ollama_installed():
+            logger.error(
+                "Ollama is not installed. Please install it from `https://ollama.com/` to use the Ollama toolkit."
+            )
+            raise click.Abort()
         github_url = DEFAULT_OLLAMA_MODEL_REPO
         branch = DEFAULT_OLLAMA_MODEL_REPO_BRANCH
 
@@ -858,6 +825,18 @@ def local_runner(ctx, model_path, pool_size):
         ModelBuilder._save_config(config_file, config)
 
     builder = ModelBuilder(model_path, download_validation_only=True)
+    if not check_requirements_installed(model_path):
+        logger.error(f"Requirements not installed for model at {model_path}.")
+        raise click.Abort()
+
+    # Post check while running `clarifai model local-runner` we check if the toolkit is ollama
+    if builder.config.get('toolkit', {}).get('provider') == 'ollama':
+        if not check_ollama_installed():
+            logger.error(
+                "Ollama is not installed. Please install it from `https://ollama.com/` to use the Ollama toolkit."
+            )
+            raise click.Abort()
+
     # don't mock for local runner since you need the dependencies to run the code anyways.
     method_signatures = builder.get_method_signatures(mocking=False)
 

clarifai/utils/cli.py

@@ -220,3 +220,130 @@ def validate_context_auth(pat: str, user_id: str, api_base: str = None):
             logger.error(f"❌ Validation failed: \n{error_msg}")
             logger.error("Please check your credentials and try again.")
         raise click.Abort()  # Exit without saving the configuration
+
+
+def customize_ollama_model(model_path, model_name, port, context_length):
+    """Customize the Ollama model name in the cloned template files.
+    Args:
+     model_path: Path to the cloned model directory
+     model_name: The model name to set (e.g., 'llama3.1', 'mistral')
+
+    """
+    model_py_path = os.path.join(model_path, "1", "model.py")
+
+    if not os.path.exists(model_py_path):
+        logger.warning(f"Model file {model_py_path} not found, skipping model name customization")
+        return
+
+    try:
+        # Read the model.py file
+        with open(model_py_path, 'r') as file:
+            content = file.read()
+        if model_name:
+            # Replace the default model name in the load_model method
+            content = content.replace(
+                'self.model = os.environ.get("OLLAMA_MODEL_NAME", \'llama3.2\')',
+                f'self.model = os.environ.get("OLLAMA_MODEL_NAME", \'{model_name}\')',
+            )
+
+        if port:
+            # Replace the default port variable in the model.py file
+            content = content.replace("PORT = '23333'", f"PORT = '{port}'")
+
+        if context_length:
+            # Replace the default context length variable in the model.py file
+            content = content.replace(
+                "context_length = '8192'", f"context_length = '{context_length}'"
+            )
+
+        # Write the modified content back to model.py
+        with open(model_py_path, 'w') as file:
+            file.write(content)
+
+    except Exception as e:
+        logger.error(f"Failed to customize Ollama model name in {model_py_path}: {e}")
+        raise
+
+
+def check_ollama_installed():
+    """Check if the Ollama CLI is installed."""
+    try:
+        import subprocess
+
+        result = subprocess.run(
+            ['ollama', '--version'], capture_output=True, text=True, check=False
+        )
+        if result.returncode == 0:
+            return True
+        else:
+            return False
+    except FileNotFoundError:
+        return False
+
+
+def _is_package_installed(package_name):
+    """Helper function to check if a single package in requirements.txt is installed."""
+    import importlib.metadata
+
+    try:
+        importlib.metadata.distribution(package_name)
+        logger.debug(f"✅ {package_name} - installed")
+        return True
+    except importlib.metadata.PackageNotFoundError:
+        logger.debug(f"❌ {package_name} - not installed")
+        return False
+    except Exception as e:
+        logger.warning(f"Error checking {package_name}: {e}")
+        return False
+
+
+def check_requirements_installed(model_path):
+    """Check if all dependencies in requirements.txt are installed."""
+    import re
+    from pathlib import Path
+
+    requirements_path = Path(model_path) / "requirements.txt"
+
+    if not requirements_path.exists():
+        logger.warning(f"requirements.txt not found at {requirements_path}")
+        return True
+
+    try:
+        package_pattern = re.compile(r'^([a-zA-Z0-9_-]+)')
+
+        # Getting package name and version (for logging)
+        requirements = [
+            (match.group(1), pack)
+            for line in requirements_path.read_text().splitlines()
+            if (pack := line.strip())
+            and not line.startswith('#')
+            and (match := package_pattern.match(line))
+        ]
+
+        if not requirements:
+            logger.info("No dependencies found in requirements.txt")
+            return True
+
+        logger.info(f"Checking {len(requirements)} dependencies...")
+
+        missing = [
+            full_req
+            for package_name, full_req in requirements
+            if not _is_package_installed(package_name)
+        ]
+
+        if not missing:
+            logger.info(f"✅ All {len(requirements)} dependencies are installed!")
+            return True
+
+        # Report missing packages
+        logger.error(
+            f"❌ {len(missing)} of {len(requirements)} required packages are missing in the current environment"
+        )
+        logger.error("\n".join(f"  - {pkg}" for pkg in missing))
+        logger.warning(f"To install: pip install -r {requirements_path}")
+        return False
+
+    except Exception as e:
+        logger.error(f"Failed to check requirements: {e}")
+        return False

tests/workflow/fixtures/single_branch_with_custom_cropper_model.yml

@@ -14,6 +14,6 @@ workflow:
           description: Custom crop model
           output_info:
             params:
-              margin: 1.33
+              margin: 1.3
       node_inputs:
         - node_id: detector