feature: s3 bucket path checking

Author: sage-maker

.githooks/pre-commit

@@ -0,0 +1,31 @@
+#!/bin/sh
+# This pre-commit hook validates S3 bucket references in modified files
+
+set -e
+
+# Get list of staged files
+staged_files=$(git diff --cached --name-only --diff-filter=ACMR | grep -E '\.(py|ipynb|md|rst|yaml|yml|json)$' || true)
+
+if [ -z "$staged_files" ]; then
+    echo "No relevant files to check for S3 bucket references."
+    exit 0
+fi
+
+echo "Checking S3 bucket references in staged files..."
+has_invalid_buckets=0
+
+for file in $staged_files; do
+    echo "Validating S3 references in $file"
+    python s3_bucket_validator.py "$file"
+    if [ $? -ne 0 ]; then
+        has_invalid_buckets=1
+    fi
+done
+
+if [ $has_invalid_buckets -ne 0 ]; then
+    echo "ERROR: Invalid S3 bucket references found. Please fix them before committing."
+    exit 1
+fi
+
+echo "S3 bucket validation passed."
+exit 0

s3_bucket_validator.py

@@ -0,0 +1,49 @@
+import re
+
+import boto3
+from botocore.exceptions import ClientError
+
+
+def is_bucket_accessible(bucket_name):
+    s3 = boto3.client('s3')
+    try:
+        s3.head_bucket(Bucket=bucket_name)
+        return True
+    except ClientError as e:
+        error_code = int(e.response['Error']['Code'])
+        if error_code == 403:
+            print(f"Bucket {bucket_name} exists, but you don't have permission to access it.")
+        elif error_code == 404:
+            print(f"Bucket {bucket_name} does not exist.")
+        else:
+            print(f"Error checking bucket {bucket_name}: {e}")
+        return False
+
+def validate_s3_references(file_path):
+    with open(file_path, 'r') as file:
+        content = file.read()
+
+    s3_pattern = re.compile(r's3:\/\/([a-zA-Z0-9._-]+)')
+    matches = s3_pattern.findall(content)
+
+    invalid_buckets = []
+    for bucket in matches:
+        if not is_bucket_accessible(bucket):
+            invalid_buckets.append(bucket)
+
+    return invalid_buckets
+
+if __name__ == "__main__":
+    import sys
+    if len(sys.argv) < 2:
+        print("Usage: python s3_bucket_validator.py <file_path>")
+        sys.exit(1)
+
+    file_path = sys.argv[1]
+    invalid_buckets = validate_s3_references(file_path)
+
+    if invalid_buckets:
+        print(f"Invalid or inaccessible S3 buckets found: {', '.join(invalid_buckets)}")
+        sys.exit(1)
+    else:
+        print("All referenced S3 buckets are valid and accessible.")