sunholo 0.119.18__tar.gz → 0.120.1__tar.gz

{sunholo-0.119.18/src/sunholo.egg-info → sunholo-0.120.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.18
+Version: 0.120.1
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0
@@ -28,6 +28,7 @@ Provides-Extra: test
 Requires-Dist: pytest; extra == "test"
 Requires-Dist: pytest-cov; extra == "test"
 Provides-Extra: all
+Requires-Dist: aiofiles; extra == "all"
 Requires-Dist: aiohttp; extra == "all"
 Requires-Dist: anthropic[vertex]; extra == "all"
 Requires-Dist: asyncpg; extra == "all"
@@ -50,7 +51,7 @@ Requires-Dist: google-cloud-pubsub; extra == "all"
 Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "all"
 Requires-Dist: google-cloud-texttospeech; extra == "all"
 Requires-Dist: google-generativeai>=0.7.1; extra == "all"
-Requires-Dist: google-genai; extra == "all"
+Requires-Dist: google-genai>=0.2.2; extra == "all"
 Requires-Dist: gunicorn; extra == "all"
 Requires-Dist: httpcore; extra == "all"
 Requires-Dist: httpx; extra == "all"
@@ -122,6 +123,7 @@ Requires-Dist: pytesseract; extra == "pipeline"
 Requires-Dist: tabulate; extra == "pipeline"
 Requires-Dist: unstructured[all-docs,local-inference]; extra == "pipeline"
 Provides-Extra: gcp
+Requires-Dist: aiofiles; extra == "gcp"
 Requires-Dist: anthropic[vertex]; extra == "gcp"
 Requires-Dist: google-api-python-client; extra == "gcp"
 Requires-Dist: google-auth-httplib2; extra == "gcp"
@@ -136,7 +138,7 @@ Requires-Dist: google-cloud-logging; extra == "gcp"
 Requires-Dist: google-cloud-pubsub; extra == "gcp"
 Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "gcp"
 Requires-Dist: google-cloud-texttospeech; extra == "gcp"
-Requires-Dist: google-genai; extra == "gcp"
+Requires-Dist: google-genai>=0.2.2; extra == "gcp"
 Requires-Dist: google-generativeai>=0.8.3; extra == "gcp"
 Requires-Dist: langchain-google-genai>=2.0.0; extra == "gcp"
 Requires-Dist: langchain_google_alloydb_pg>=0.2.2; extra == "gcp"

{sunholo-0.119.18 → sunholo-0.120.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sunholo"
-version = "0.119.18"
+version = "0.120.1"
 description = "Large Language Model DevOps - a package to help deploy LLMs to the Cloud."
 readme = "README.md"
 requires-python = ">=3.10"
@@ -59,6 +59,7 @@ test = [
     # any other test dependencies
 ]
 all = [
+    "aiofiles",
     "aiohttp",
     "anthropic[vertex]",
     "asyncpg",
@@ -81,7 +82,7 @@ all = [
     "google-cloud-discoveryengine>=0.13.4",
     "google-cloud-texttospeech",
     "google-generativeai>=0.7.1",
-    "google-genai",
+    "google-genai>=0.2.2",
     "gunicorn",
     "httpcore",
     "httpx",
@@ -159,6 +160,7 @@ pipeline = [
     "unstructured[local-inference,all-docs]"
 ]
 gcp = [
+    "aiofiles",
     "anthropic[vertex]",
     "google-api-python-client",
     "google-auth-httplib2",
@@ -173,7 +175,7 @@ gcp = [
     "google-cloud-pubsub",
     "google-cloud-discoveryengine>=0.13.4",
     "google-cloud-texttospeech",
-    "google-genai",
+    "google-genai>=0.2.2",
     "google-generativeai>=0.8.3",
     "langchain-google-genai>=2.0.0",
     "langchain_google_alloydb_pg>=0.2.2",

{sunholo-0.119.18 → sunholo-0.120.1}/src/sunholo/auth/refresh.py

@@ -9,8 +9,9 @@ from ..custom_logging import log
 def get_default_email():
 
     # https://stackoverflow.com/questions/64234214/how-to-generate-a-blob-signed-url-in-google-cloud-run
-
-    gcs_credentials, project_id = refresh_credentials()
+    gcs_credentials, project_id = get_default_creds()
+    if gcs_credentials is None:
+        gcs_credentials, project_id = refresh_credentials()
 
     if gcs_credentials is None:
         log.error("Could not refresh the credentials properly.")

{sunholo-0.119.18 → sunholo-0.120.1}/src/sunholo/genai/file_handling.py

@@ -3,6 +3,7 @@ from ..gcs import get_bytes_from_gcs
 
 from functools import partial
 import mimetypes
+import uuid
 import asyncio
 import tempfile
 import re
@@ -11,6 +12,7 @@ import traceback
 try:
     import google.generativeai as genai
     from google import genai as genaiv2
+
 except ImportError:
     genai = None
     genaiv2 = None
@@ -77,32 +79,44 @@ ALLOWED_MIME_TYPES = set(AUDIO_MIMES + VIDEO_MIMES + IMAGE_MIMES + DOCUMENT_MIME
 # ]
 
 def sanitize_file(filename):
+    """
+    Sanitize filename to conform to Gemini API requirements:
+    - Only lowercase alphanumeric characters and dashes
+    - Cannot begin or end with a dash
+    """
     # Split the filename into name and extension
     name, extension = os.path.splitext(filename)
     
-    # Sanitize the name by removing invalid characters and converting to lowercase
-    sanitized_name = re.sub(r'[^a-z0-9-]', '', name.lower())
-    sanitized_name = re.sub(r'^-+|-+$', '', sanitized_name)  # Remove leading or trailing dashes
+    # Convert to lowercase
+    name = name.lower()
+    
+    # Replace any non-alphanumeric characters (including underscores) with dashes
+    sanitized_name = re.sub(r'[^a-z0-9]', '-', name)
+    
+    # Remove consecutive dashes
+    sanitized_name = re.sub(r'-+', '-', sanitized_name)
     
-    # Reattach the original extension
+    # Remove leading or trailing dashes
+    sanitized_name = sanitized_name.strip('-')
+    
+    # If the name is empty after sanitization, use a default
+    if not sanitized_name:
+        sanitized_name = 'file'
+    
+    # Limit the length
     return sanitized_name[:40]
 
-async def construct_file_content(gs_list, bucket:str, genai_lib=False):
+async def construct_file_content(gs_list, bucket:str, genai_lib=False, timeout=60):
     """
     Args:
     - gs_list: a list of dicts representing files in a bucket
-       - contentType: The content type of the file on GCS
-       - storagePath: The path in the bucket
-       - name: The name of the file
-       - url: The URL of the file that can be used to display the contents
     - bucket: The bucket the files are in
-    - genai: whether its using the genai SDK
-
+    - genai_lib: whether its using the genai SDK
+    - timeout: timeout in seconds for the gather operation
     """
     
     file_list = []
     for element in gs_list:
-
         the_mime_type = element.get('contentType')
         if the_mime_type is None:
             continue
@@ -114,7 +128,7 @@ async def construct_file_content(gs_list, bucket:str, genai_lib=False):
             log.warning(f'{the_mime_type} is not in allowed MIME types for {element.get("name")}')
     
     if not file_list:
-        return {"role": "user", "parts": [{"text": "No eligible contentTypes were found"}]}
+        return [{"role": "user", "parts": [{"text": "No eligible contentTypes were found"}]}]
 
     content = []
     
@@ -124,34 +138,59 @@ async def construct_file_content(gs_list, bucket:str, genai_lib=False):
         img_url = f"gs://{bucket}/{file_info['storagePath']}"
         display_url = file_info.get('url')
         mime_type = file_info['contentType']
-        name = sanitize_file(file_info['name'])
+        # Generate a unique name for each file
+        import uuid
+        original_name = sanitize_file(file_info['name'])
+        unique_id = str(uuid.uuid4())[:8]
+        unique_name = sanitize_file(f"{original_name}-{unique_id}")
+
         display_name = file_info['name']
-        log.info(f"Processing {name=} {display_name=}")
+        log.info(f"Processing {unique_name=} {display_name=}")
+        
         try:
             if not genai_lib:
-                myfile = genai.get_file(name)
+                myfile = genai.get_file(unique_name)
             else:
                 client = genaiv2.Client()
-                myfile = client.files.get(name=name)
+                myfile = client.files.get(name=unique_name)
             content.append(myfile)
             content.append(f"You have been given the ability to work with file {display_name=} with {mime_type=} {display_url=}")
-            log.info(f"Found existing genai.get_file {name=}")
+            log.info(f"Found existing genai.get_file {unique_name=}")
         except Exception as e:
-            log.info(f"Not found checking genai.get_file: '{name}' {str(e)}")
+            log.info(f"Not found checking genai.get_file: '{unique_name}' {str(e)}")
             tasks.append(
                 download_gcs_upload_genai(img_url, 
-                                          mime_type=mime_type, 
-                                          name=name, 
-                                          display_url=display_url, 
-                                          display_name=display_name,
-                                          genai_lib=genai_lib)
-                                          )
+                                         mime_type=mime_type, 
+                                         name=unique_name, 
+                                         display_url=display_url, 
+                                         display_name=display_name,
+                                         genai_lib=genai_lib)
+                                        )
 
-    # Run all tasks in parallel
+    # Process files with timeout
     if tasks:
-        task_content = await asyncio.gather(*tasks)
-        content.extend(task_content)
-
+        try:
+            # Add timeout to prevent hanging
+            task_results = await asyncio.wait_for(asyncio.gather(*tasks, return_exceptions=True), timeout=timeout)
+            
+            # Process results, handle any exceptions
+            for result in task_results:
+                if isinstance(result, Exception):
+                    log.error(f"Task failed with error: {str(result)}")
+                    # Add error message to content
+                    content.append({"role": "user", "parts": [{"text": f"Error processing file: {str(result)}"}]})
+                else:
+                    # Normalize the result structure based on genai_lib
+                    if genai_lib and isinstance(result, list):
+                        content.append(result[0])  # The file object
+                        content.append({"role": "user", "parts": [{"text": result[1]}]})  # The message
+                    else:
+                        content.append(result)
+        except asyncio.TimeoutError:
+            log.error(f"Timeout occurred after {timeout} seconds while processing files")
+            content.append({"role": "user", "parts": [{"text": f"Some files could not be processed within the time limit ({timeout} seconds)"}]})
+    
+    log.info(f"construct_file_content completed with {len(content)} items")
     return content
 
 # Helper function to handle each file download with error handling
@@ -169,10 +208,9 @@ async def download_gcs_upload_genai(img_url,
                                     display_url=None, 
                                     display_name=None, 
                                     retries=3, delay=2, genai_lib=False):
-    import aiofiles
-    from google.generativeai.types import file_types
     """
     Downloads and uploads a file with retries in case of failure.
+    Thread-safe implementation using unique file paths.
     
     Args:
       - img_url: str The URL of the file to download.
@@ -184,9 +222,10 @@ async def download_gcs_upload_genai(img_url,
     Returns:
       - downloaded_content: The result of the file upload if successful.
     """
+    import aiofiles
     for attempt in range(retries):
         try:
-            log.info(f"Upload {attempt} for {img_url=}")
+            log.info(f"Upload attempt [{attempt}] for {img_url=}")
             # Download the file bytes asynchronously
             file_bytes = await asyncio.to_thread(get_bytes_from_gcs, img_url)
             if not file_bytes:
@@ -200,50 +239,72 @@ async def download_gcs_upload_genai(img_url,
 
             if file_size > 19434343:
                 log.warning(f"File size for {img_url}: {file_size} is too big.")
-                msg =  f"The file for {img_url} is too large ({file_size} bytes) to be used directly.  Use RAG instead or {display_url=}"
+                msg = f"The file for {img_url} is too large ({file_size} bytes) to be used directly. Use RAG instead or {display_url=}"
                 return {"role": "user", "parts": [{"text": msg}]}
             
             extension = mimetypes.guess_extension(mime_type)
             
-            # Use aiofiles for asynchronous file operations
-            temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=extension)
-            downloaded_file = temp_file.name
-
-            sanitized_file = sanitize_file(downloaded_file)
-
-            log.info(f"Writing file {sanitized_file}")
-            async with aiofiles.open(sanitized_file, 'wb') as f:
+            # Create a unique directory for this upload task
+            unique_id = str(uuid.uuid4())
+            temp_dir = os.path.join(tempfile.gettempdir(), f"upload_{unique_id}")
+            os.makedirs(temp_dir, exist_ok=True)
+            
+            # Create a file with unique path
+            file_path = os.path.join(temp_dir, f"file_{unique_id}{extension}")
+            
+            log.info(f"Writing file {file_path}")
+            async with aiofiles.open(file_path, 'wb') as f:
                 await f.write(file_bytes)
 
-            # Upload the file and get its content reference
-            if not genai_lib:
-                try:
-                    downloaded_content: file_types.File = await asyncio.to_thread(
+            try:
+                if not genai_lib:
+                    downloaded_content = await asyncio.to_thread(
                         partial(genai.upload_file, name=name, mime_type=mime_type, display_name=display_name), 
-                        sanitized_file
-                        )
+                        file_path
+                    )
+                    
+                    # Clean up after successful upload
+                    try:
+                        os.remove(file_path)
+                        os.rmdir(temp_dir)
+                    except OSError as e:
+                        log.warning(f"Cleanup error (non-critical): {str(e)}")
+                        
                     return {"role": "user", "parts": [{"file_data": downloaded_content}, 
-                                                    {"text": f"You have been given the ability to read and work with filename '{display_name=}' with {mime_type=} {display_url=}"}
-                                                    ]}
-                except Exception as err:
-                    msg = f"Could not upload {sanitized_file} to genai.upload_file: {str(err)} {traceback.format_exc()} {display_url=}"
-                    log.error(msg)
-                    return {"role": "user", "parts": [{"text": msg}]}
-            else:
-                try:
+                                                    {"text": f"You have been given the ability to read and work with filename '{display_name}' with {mime_type=} {display_url=}"}
+                                                   ]}
+                else:
                     client = genaiv2.Client()
-                    downloaded_content = await asyncio.to_thread(
-                        client.files.upload, 
-                        file=sanitized_file,  
-                        config=dict(mime_type=mime_type, display_name=display_name)
-                    )
+                    
+                    # Use semaphore to limit concurrent uploads
+                    async with upload_semaphore:
+                        downloaded_content = await asyncio.to_thread(
+                            client.files.upload, 
+                            file=file_path,  
+                            config=dict(mime_type=mime_type, display_name=display_name)
+                        )
+                    
+                    # Clean up after successful upload
+                    try:
+                        os.remove(file_path)
+                        os.rmdir(temp_dir)
+                    except OSError as e:
+                        log.warning(f"Cleanup error (non-critical): {str(e)}")
+                        
                     return [downloaded_content, 
-                            f"You have been given the ability to read and work with filename '{display_name=}' with {mime_type=} {display_url=}"]
+                            f"You have been given the ability to read and work with filename '{display_name}' with {mime_type=} {display_url=}"]
 
-                except Exception as err:
-                    msg = f"Could not upload {sanitized_file} to genaiv2.client.files.upload: {str(err)} {traceback.format_exc()} {display_url=}"
-                    log.error(msg)
-                    return {"role": "user", "parts": [{"text": msg}]}
+            except Exception as err:
+                # Clean up on error
+                try:
+                    os.remove(file_path)
+                    os.rmdir(temp_dir)
+                except OSError:
+                    pass
+                
+                msg = f"Could not upload {file_path} to {'genai.upload_file' if not genai_lib else 'genaiv2.client.files.upload'}: {str(err)} {traceback.format_exc()} {display_url=}"
+                log.error(msg)
+                return {"role": "user", "parts": [{"text": msg}]}
         
         except Exception as err:
             log.error(f"Error processing file {img_url} {mime_type=} on attempt {attempt + 1}/{retries}: {str(err)}")
@@ -255,3 +316,7 @@ async def download_gcs_upload_genai(img_url,
             else:
                 raise err  # Raise the error after max retries
 
+# Add this at the module level
+# Create a semaphore to limit concurrent uploads
+upload_semaphore = asyncio.Semaphore(5)  # Adjust the value based on your needs
+

{sunholo-0.119.18 → sunholo-0.120.1/src/sunholo.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.18
+Version: 0.120.1
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0
@@ -28,6 +28,7 @@ Provides-Extra: test
 Requires-Dist: pytest; extra == "test"
 Requires-Dist: pytest-cov; extra == "test"
 Provides-Extra: all
+Requires-Dist: aiofiles; extra == "all"
 Requires-Dist: aiohttp; extra == "all"
 Requires-Dist: anthropic[vertex]; extra == "all"
 Requires-Dist: asyncpg; extra == "all"
@@ -50,7 +51,7 @@ Requires-Dist: google-cloud-pubsub; extra == "all"
 Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "all"
 Requires-Dist: google-cloud-texttospeech; extra == "all"
 Requires-Dist: google-generativeai>=0.7.1; extra == "all"
-Requires-Dist: google-genai; extra == "all"
+Requires-Dist: google-genai>=0.2.2; extra == "all"
 Requires-Dist: gunicorn; extra == "all"
 Requires-Dist: httpcore; extra == "all"
 Requires-Dist: httpx; extra == "all"
@@ -122,6 +123,7 @@ Requires-Dist: pytesseract; extra == "pipeline"
 Requires-Dist: tabulate; extra == "pipeline"
 Requires-Dist: unstructured[all-docs,local-inference]; extra == "pipeline"
 Provides-Extra: gcp
+Requires-Dist: aiofiles; extra == "gcp"
 Requires-Dist: anthropic[vertex]; extra == "gcp"
 Requires-Dist: google-api-python-client; extra == "gcp"
 Requires-Dist: google-auth-httplib2; extra == "gcp"
@@ -136,7 +138,7 @@ Requires-Dist: google-cloud-logging; extra == "gcp"
 Requires-Dist: google-cloud-pubsub; extra == "gcp"
 Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "gcp"
 Requires-Dist: google-cloud-texttospeech; extra == "gcp"
-Requires-Dist: google-genai; extra == "gcp"
+Requires-Dist: google-genai>=0.2.2; extra == "gcp"
 Requires-Dist: google-generativeai>=0.8.3; extra == "gcp"
 Requires-Dist: langchain-google-genai>=2.0.0; extra == "gcp"
 Requires-Dist: langchain_google_alloydb_pg>=0.2.2; extra == "gcp"

{sunholo-0.119.18 → sunholo-0.120.1}/src/sunholo.egg-info/requires.txt

@@ -6,6 +6,7 @@ ruamel.yaml
 tenacity
 
 [all]
+aiofiles
 aiohttp
 anthropic[vertex]
 asyncpg
@@ -28,7 +29,7 @@ google-cloud-pubsub
 google-cloud-discoveryengine>=0.13.4
 google-cloud-texttospeech
 google-generativeai>=0.7.1
-google-genai
+google-genai>=0.2.2
 gunicorn
 httpcore
 httpx
@@ -97,6 +98,7 @@ requests
 rich
 
 [gcp]
+aiofiles
 anthropic[vertex]
 google-api-python-client
 google-auth-httplib2
@@ -111,7 +113,7 @@ google-cloud-logging
 google-cloud-pubsub
 google-cloud-discoveryengine>=0.13.4
 google-cloud-texttospeech
-google-genai
+google-genai>=0.2.2
 google-generativeai>=0.8.3
 langchain-google-genai>=2.0.0
 langchain_google_alloydb_pg>=0.2.2