sunholo 0.139.0__tar.gz → 0.139.1__tar.gz

{sunholo-0.139.0/src/sunholo.egg-info → sunholo-0.139.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sunholo
-Version: 0.139.0
+Version: 0.139.1
 Summary: AI DevOps - a package to help deploy GenAI to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0

{sunholo-0.139.0 → sunholo-0.139.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sunholo"
-version = "0.139.0"
+version = "0.139.1"
 description = "AI DevOps - a package to help deploy GenAI to the Cloud."
 readme = "README.md"
 requires-python = ">=3.10"

{sunholo-0.139.0 → sunholo-0.139.1}/src/sunholo/gcs/download_gcs_text.py

@@ -3,12 +3,21 @@ import json
 
 from ..custom_logging import log
 from ..utils.mime import get_mime_type_gemini
+from .metadata import check_gcs_file_size
 from .download_url import get_bytes_from_gcs
 
-def download_gcs_source_to_string(source:str) -> str:
+def download_gcs_source_to_string(source:str, max_size_bytes: int = 1024*1024) -> str:
     """
-    source: str The Google Cloud Storage URI of the file to download (e.g., 'gs://bucket_name/file_name').
+    Download a file from Google Cloud Storage and convert it to a string.
+    
+    Args:
+        source: str The Google Cloud Storage URI of the file to download (e.g., 'gs://bucket_name/file_name').
+        max_size_bytes: int Maximum file size to download, defaults to 1MB (1024*1024 bytes)
+        
+    Returns:
+        str: The contents of the file as a string, or an empty string if the file could not be downloaded.
     """
+
     mime_type = get_mime_type_gemini(source)
     if mime_type == "":
         log.warning(f"Can not download to string file source {source}")
@@ -52,6 +61,15 @@ def download_gcs_source_to_string(source:str) -> str:
 
     try:
         log.info(f"Extracting text for {source}")
+        # Check file size before downloading
+        file_size = check_gcs_file_size(source)
+        if file_size == -1:
+            log.warning(f"Could not determine file size for {source}")
+            return ""
+        elif file_size > max_size_bytes:
+            log.warning(f"File size {file_size} bytes exceeds maximum size limit of {max_size_bytes} bytes for {source}")
+            return ""
+        
         bytes = get_bytes_from_gcs(source)
         string = bytes.decode('utf-8', errors='replace')
         log.info(f"Extracted {len(string)} characters from {source}: {string[:100]}")

sunholo-0.139.1/src/sunholo/gcs/metadata.py

@@ -0,0 +1,66 @@
+try:
+    from google.cloud import storage
+except ImportError:
+    storage = None
+
+from ..custom_logging import log
+
+
+def get_object_metadata(bucket_name, object_name):
+
+    if not storage:
+        return None
+
+    if bucket_name is None or object_name is None:
+        log.warning("Got invalid bucket name and object name")
+        return None
+    try:
+        storage_client = storage.Client()
+    except Exception as e:
+        log.warning(f"Could not connect to Google Cloud Storage for metadata: {str(e)}")
+        return None
+
+    bucket = storage_client.bucket(bucket_name)
+    blob = bucket.blob(object_name)
+
+    # Fetch the blob's metadata
+    blob.reload()  # Make sure to reload the blob to get the most up-to-date metadata
+
+    # Access custom metadata
+    custom_metadata = blob.metadata
+
+    log.info(f"Custom Metadata for {object_name}: {custom_metadata}")
+    return custom_metadata
+
+def check_gcs_file_size(source: str) -> int:
+    """
+    Check the size of a file in Google Cloud Storage without downloading the entire file.
+    
+    Args:
+        source: str The Google Cloud Storage URI of the file to check (e.g., 'gs://bucket_name/file_name').
+        
+    Returns:
+        int: The size of the file in bytes, or -1 if the size cannot be determined.
+    """
+    from google.cloud import storage
+    
+    try:
+        # Parse the GCS URI
+        if not source.startswith('gs://'):
+            log.warning(f"Invalid GCS URI format: {source}")
+            return -1
+            
+        bucket_name, blob_path = source[5:].split('/', 1)
+        
+        # Create a client and get the bucket
+        storage_client = storage.Client()
+        bucket = storage_client.bucket(bucket_name)
+        
+        # Get the blob (file) and retrieve its metadata
+        blob = bucket.blob(blob_path)
+        blob.reload()  # Fetch the latest metadata
+        
+        return blob.size
+    except Exception as err:
+        log.error(f"Error checking file size for {source}: {str(err)}")
+        return -1

{sunholo-0.139.0 → sunholo-0.139.1/src/sunholo.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sunholo
-Version: 0.139.0
+Version: 0.139.1
 Summary: AI DevOps - a package to help deploy GenAI to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0

sunholo-0.139.0/src/sunholo/gcs/metadata.py

@@ -1,33 +0,0 @@
-try:
-    from google.cloud import storage
-except ImportError:
-    storage = None
-
-from ..custom_logging import log
-
-
-def get_object_metadata(bucket_name, object_name):
-
-    if not storage:
-        return None
-
-    if bucket_name is None or object_name is None:
-        log.warning("Got invalid bucket name and object name")
-        return None
-    try:
-        storage_client = storage.Client()
-    except Exception as e:
-        log.warning(f"Could not connect to Google Cloud Storage for metadata: {str(e)}")
-        return None
-
-    bucket = storage_client.bucket(bucket_name)
-    blob = bucket.blob(object_name)
-
-    # Fetch the blob's metadata
-    blob.reload()  # Make sure to reload the blob to get the most up-to-date metadata
-
-    # Access custom metadata
-    custom_metadata = blob.metadata
-
-    log.info(f"Custom Metadata for {object_name}: {custom_metadata}")
-    return custom_metadata