sunholo 0.70.5__py3-none-any.whl → 0.71.0__py3-none-any.whl

sunholo/discovery_engine/discovery_engine_client.py

@@ -2,7 +2,7 @@ try:
     from google.api_core.client_options import ClientOptions
     from google.cloud import discoveryengine_v1alpha as discoveryengine
     from google.api_core.retry import Retry, if_exception_type
-    from google.api_core.exceptions import ResourceExhausted
+    from google.api_core.exceptions import ResourceExhausted, AlreadyExists
     from google.cloud.discoveryengine_v1alpha import SearchResponse, Chunk
 except ImportError:
     ClientOptions = None
@@ -313,6 +313,9 @@ class DiscoveryEngineClient:
             log.error(f"DiscoveryEngine Operation failed after retries due to quota exceeded: {e}")
 
             raise e
+        except AlreadyExists as e:
+            log.warning(f"DiscoveryEngine - Already exists: {e} - {gcs_uri=} {bigquery_table=}")
+
         except Exception as e:
             log.error(f"An unexpected DiscoveryEngine error occurred: {e}")
 

sunholo/vertex/extensions.py

@@ -6,6 +6,16 @@ from ..logging import log
 from ..utils.gcp_project import get_gcp_project
 from ..utils.parsers import validate_extension_id
 
+# https://github.com/GoogleCloudPlatform/applied-ai-engineering-samples/blob/main/genai-on-vertex-ai/vertex_ai_extensions/notebooks/pandas_code_interpreter.ipynb
+import base64
+import json
+import pprint
+import pandas
+from io import StringIO
+
+global CODE_INTERPRETER_WRITTEN_FILES
+CODE_INTERPRETER_WRITTEN_FILES = []
+
 def get_extension_import_config(
         display_name: str, 
         description: str,
@@ -102,7 +112,7 @@ def create_extension_instance(
             },
         },
     )
-    log.info(f"Creating Vertex Extension: {extension=}")
+    log.info(f"Created Vertex Extension: {extension_name}")
 
     return extension
 
@@ -125,14 +135,192 @@ def create_extension_code_interpreter(
                             }
                         }
 
+    llm_description="""
+Tool to generate and execute valid Python code from a natural
+  language description, or to execute custom Python code.
+  Use this tool to:
+  - generate and/or execute code for various tasks:
+    - perform a wide variety of mathematical calculations, for example, add,
+      subtract, multiply, divide, average, power, factorial, quotient,
+      formulae, logarithms, random numbers, trigonometric functions, and
+      equations;
+    - sort, filter, select top results, and otherwise analyze data (including
+      data acquired from other tools and Extensions);
+    - create visualizations, plot charts, draw graphs, shapes, print results,
+      etc.
+  - execute custom code and get results and output files.
+"""
+
     code_extension = create_extension_instance(
         display_name="Code Interpreter",
         description="This extension generates and executes code in the specified language",
         open_api_gcs_uri="gs://vertex-extension-public/code_interpreter.yaml",
         llm_name="code_interpreter_tool",
-        llm_description="Google Code Interpreter Extension",
+        llm_description=llm_description,
         runtime_config=runtime_config
         )
     log.info(f"Created code extension: {code_extension=}")
 
     return code_extension
+
+def execute_extension(operation_id: str,
+                      operation_params: dict,
+                      extension_id: str):
+
+    # only us-central for now
+    location = "us-central1"
+    init_vertex(location=location)
+
+    if not extension_id.startswith("projects/"):
+        project_id=get_gcp_project()
+        extension_name = f"projects/{project_id}/locations/{location}/extensions/{extension_id}"
+    else:
+        extension_name=extension_id
+    
+    extension = extensions.Extension(extension_name)
+
+    response = extension.execute(
+        operation_id=operation_id,
+        # {"query": "find the max value in the list: [1,2,3,4,-5]"}
+        operation_params=operation_params,
+    )
+
+    return response
+
+def execute_code_extension(query:str, filenames: list[str]=None, gcs_files: list[str]=None):
+
+    if filenames and gcs_files:
+        raise ValueError("Can't specify both filenames and gcs_files")
+    
+    extension_code_interpreter = extensions.Extension.from_hub("code_interpreter")
+
+    file_arr=None
+    if filenames:
+        file_arr = [
+            {
+                "name": filename,
+                "contents":  base64.b64encode(open(filename, "rb").read()).decode()
+            }
+            for filename in filenames
+        ]
+
+    response = extension_code_interpreter.execute(
+        operation_id = "generate_and_execute",
+        operation_params={
+            "query": query,
+            "files": file_arr,
+            "file_gcs_uris": gcs_files
+        })
+    
+    CODE_INTERPRETER_WRITTEN_FILES.extend(
+        [item['name'] for item in response['output_files']])
+
+    if response.get('execution_error'):
+        log.error(f"Code Execution Response failed with: {response.get('execution_error')} - maybe retry?")
+    
+    return response
+
+css_styles = """
+<style>
+.main_summary {
+  font-weight: bold;
+  font-size: 14px; color: #4285F4;
+  background-color:rgba(221, 221, 221, 0.5); padding:8px;}
+</style>
+        """
+
+# Parser to visualise the content of returned files as HTML.
+def parse_files_to_html(outputFiles, save_files_locally = True):
+    IMAGE_FILE_EXTENSIONS = set(["jpg", "jpeg", "png"])
+    file_list = []
+    details_tml = """<details><summary>{name}</summary><div>{html_content}</div></details>"""
+
+    if not outputFiles:
+      return "No Files generated from the code"
+    # Sort output_files so images are displayed before other files such as JSON.
+    for output_file in sorted(
+        outputFiles,
+        key=lambda x: x["name"].split(".")[-1] not in IMAGE_FILE_EXTENSIONS,
+    ):
+        file_name = output_file.get("name")
+        file_contents = base64.b64decode(output_file.get("contents"))
+        if save_files_locally:
+          open(file_name,"wb").write(file_contents)
+
+        if file_name.split(".")[-1] in IMAGE_FILE_EXTENSIONS:
+            # Render Image
+            file_html_content = ('<img src="data:image/png;base64, '
+                                f'{output_file.get("contents")}" />')
+        elif file_name.endswith(".json"):
+            # Pretty print JSON
+            json_pp = pprint.pformat(
+                        json.loads(file_contents.decode()),
+                        compact=False,
+                        width=160)
+            file_html_content =  (f'<span>{json_pp}</span>')
+        elif file_name.endswith(".csv"):
+            # CSV
+            csv_md = pandas.read_csv(
+                  StringIO(file_contents.decode())).to_markdown(index=False)
+            file_html_content = f'<span>{csv_md}</span>'
+        elif file_name.endswith(".pkl"):
+            # PKL
+            file_html_content = f'<span>Preview N/A</span>'
+        else:
+            file_html_content = f"<span>{file_contents.decode()}</span>"
+
+        file_list.append({'name': file_name, "html_content": file_html_content})
+
+    buffer_html = [ details_tml.format(**_file) for _file in file_list ]
+    return "".join(buffer_html)
+
+# Processing code interpreter response to html visualization.
+def process_response(response: dict, save_files_locally = None) -> None:
+
+    result_template = """
+    <details open>
+        <summary class='main_summary'>{summary}:</summary>
+        <div><pre>{content}</pre></div>
+    </details>
+    """
+
+    result = ""
+    code = response.get('generated_code')
+    if 'execution_result' in response and response['execution_result']!="":
+        result = result_template.format(
+            summary="Executed Code Output",
+            content=response.get('execution_result'))
+    else:
+        result = result_template.format(
+        summary="Executed Code Output",
+        content="Code does not produce printable output.")
+
+    if response.get('execution_error', None):
+        result += result_template.format(
+            summary="Generated Code Raised a (Possibly Non-Fatal) Exception",
+            content=response.get('execution_error', None))
+
+    result += result_template.format(
+        summary="Files Created <u>(Click on filename to view content)</u>",
+        content=parse_files_to_html(
+            response.get('output_files', []),
+            save_files_locally = True))
+
+    html_content = f"""
+{css_styles}
+<div id='main'>
+    <div id="right">
+      <h3>Generated Code by Code Interpreter</h3>
+      <pre><code>{code}</code></pre>
+    </div>
+    <div id="left">
+      <h3>Code Execution Results</h3>
+      {result}
+    </div>
+</div>
+"""
+    if save_files_locally:
+        # write to local file
+        pass
+  
+    return html_content

sunholo/vertex/extensions_class.py

@@ -0,0 +1,241 @@
+from vertexai.preview import extensions
+from .init import init_vertex
+from ..logging import log
+from ..utils.gcp_project import get_gcp_project
+from ..utils.parsers import validate_extension_id
+import base64
+import json
+from io import StringIO
+
+class VertexAIExtensions:
+    def __init__(self):
+        self.CODE_INTERPRETER_WRITTEN_FILES = []
+        self.css_styles = """
+        <style>
+        .main_summary {
+          font-weight: bold;
+          font-size: 14px; color: #4285F4;
+          background-color:rgba(221, 221, 221, 0.5); padding:8px;}
+        </style>
+        """
+        self.IMAGE_FILE_EXTENSIONS = set(["jpg", "jpeg", "png"])
+        self.location = "us-central1"
+
+    def get_extension_import_config(self, display_name: str, description: str,
+                                    api_spec_gcs: dict, service_account_name: dict, tool_use_examples: list):
+        tool_use_examples = [
+            {
+                "extensionOperation": {
+                    "operationId": "say_hello",
+                },
+                "displayName": "Say hello in the requested language",
+                "query": "Say hello in French",
+                "requestParams": {
+                    "fields": [
+                        {
+                            "key": "apiServicePrompt",
+                            "value": {
+                                "string_value": "French",
+                            }
+                        }
+                    ]
+                },
+                "responseParams": {
+                    "fields": [
+                        {
+                            "key": "apiServiceOutput",
+                            "value": {
+                                "string_value": "bonjour",
+                            },
+                        }
+                    ],
+                },
+                "responseSummary": "Bonjour"
+            }
+        ]
+
+        return {
+            "displayName": display_name,
+            "description": description,
+            "manifest": {
+                "name": "EXTENSION_NAME_LLM",
+                "description": "DESCRIPTION_LLM",
+                "apiSpec": {
+                    "openApiGcsUri": api_spec_gcs,
+                },
+                "authConfig": {
+                    "authType": "OAUTH",
+                    "oauthConfig": {"service_account": service_account_name}
+                }
+            },
+            "toolUseExamples": tool_use_examples,
+        }
+
+    def create_extension_instance(self, display_name: str, description: str, open_api_gcs_uri: str,
+                                  llm_name: str = None, llm_description: str = None, runtime_config: dict = None, service_account: str = None):
+        project_id = get_gcp_project()
+        extension_name = f"projects/{project_id}/locations/us-central1/extensions/{validate_extension_id(display_name)}"
+
+        extension = extensions.Extension.create(
+            extension_name=extension_name,
+            display_name=display_name,
+            description=description,
+            runtime_config=runtime_config or None,
+            manifest={
+                "name": llm_name or display_name,
+                "description": llm_description or description,
+                "api_spec": {
+                    "open_api_gcs_uri": open_api_gcs_uri
+                },
+                "auth_config": {
+                    "auth_type": "GOOGLE_SERVICE_ACCOUNT_AUTH",
+                    "google_service_account_config": service_account or {},
+                },
+            },
+        )
+        log.info(f"Created Vertex Extension: {extension_name}")
+
+        return extension
+
+    def execute_extension(self, operation_id: str, operation_params: dict, extension_id: str):
+        init_vertex(location=self.location)
+
+        if not extension_id.startswith("projects/"):
+            project_id = get_gcp_project()
+            extension_name = f"projects/{project_id}/locations/{self.location}/extensions/{extension_id}"
+        else:
+            extension_name = extension_id
+
+        extension = extensions.Extension(extension_name)
+
+        response = extension.execute(
+            operation_id=operation_id,
+            operation_params=operation_params,
+        )
+
+        return response
+
+    def execute_code_extension(self, query: str, filenames: list[str] = None, gcs_files: list[str] = None):
+        if filenames and gcs_files:
+            raise ValueError("Can't specify both filenames and gcs_files")
+
+        extension_code_interpreter = extensions.Extension.from_hub("code_interpreter")
+
+        file_arr = None
+        if filenames:
+            file_arr = [
+                {
+                    "name": filename,
+                    "contents": base64.b64encode(open(filename, "rb").read()).decode()
+                }
+                for filename in filenames
+            ]
+
+        response = extension_code_interpreter.execute(
+            operation_id="generate_and_execute",
+            operation_params={
+                "query": query,
+                "files": file_arr,
+                "file_gcs_uris": gcs_files
+            })
+
+        self.CODE_INTERPRETER_WRITTEN_FILES.extend(
+            [item['name'] for item in response['output_files']])
+
+        if response.get('execution_error'):
+            log.error(f"Code Execution Response failed with: {response.get('execution_error')} - maybe retry?")
+
+        return response
+
+    def parse_files_to_html(self, outputFiles, save_files_locally=True):
+        file_list = []
+        details_tml = """<details><summary>{name}</summary><div>{html_content}</div></details>"""
+
+        if not outputFiles:
+            return "No Files generated from the code"
+        # Sort output_files so images are displayed before other files such as JSON.
+        for output_file in sorted(
+                outputFiles,
+                key=lambda x: x["name"].split(".")[-1] not in self.IMAGE_FILE_EXTENSIONS,
+        ):
+            file_name = output_file.get("name")
+            file_contents = base64.b64decode(output_file.get("contents"))
+            if save_files_locally:
+                open(file_name, "wb").write(file_contents)
+
+            if file_name.split(".")[-1] in self.IMAGE_FILE_EXTENSIONS:
+                # Render Image
+                file_html_content = ('<img src="data:image/png;base64, '
+                                     f'{output_file.get("contents")}" />')
+            elif file_name.endswith(".json"):
+                import pprint
+                # Pretty print JSON
+                json_pp = pprint.pformat(
+                    json.loads(file_contents.decode()),
+                    compact=False,
+                    width=160)
+                file_html_content = (f'<span>{json_pp}</span>')
+            elif file_name.endswith(".csv"):
+                # CSV
+                try:
+                    import pandas
+                except ImportError:
+                    log.error("Need pandas for csv processing")
+                csv_md = pandas.read_csv(
+                    StringIO(file_contents.decode())).to_markdown(index=False)
+                file_html_content = f'<span>{csv_md}</span>'
+            elif file_name.endswith(".pkl"):
+                # PKL
+                file_html_content = f'<span>Preview N/A</span>'
+            else:
+                file_html_content = f"<span>{file_contents.decode()}</span>"
+
+            file_list.append({'name': file_name, "html_content": file_html_content})
+
+        buffer_html = [details_tml.format(**_file) for _file in file_list]
+        return "".join(buffer_html)
+    
+    def process_response(self, response: dict, save_files_locally=None) -> str:
+        result_template = """
+        <details open>
+            <summary class='main_summary'>{summary}:</summary>
+            <div><pre>{content}</pre></div>
+        </details>
+        """
+
+        result = ""
+        code = response.get('generated_code')
+        if 'execution_result' in response and response['execution_result'] != "":
+            result = result_template.format(
+                summary="Executed Code Output",
+                content=response.get('execution_result'))
+        else:
+            result = result_template.format(
+                summary="Executed Code Output",
+                content="Code does not produce printable output.")
+
+        if response.get('execution_error', None):
+            result += result_template.format(
+                summary="Generated Code Raised a (Possibly Non-Fatal) Exception",
+                content=response.get('execution_error', None))
+
+        result += result_template.format(
+            summary="Files Created <u>(Click on filename to view content)</u>",
+            content=self.parse_files_to_html(
+                response.get('output_files', []),
+                save_files_locally=True))
+
+        html_content = f"""
+{self.css_styles}
+<div id='main'>
+    <h3>Generated Code by Code Interpreter</h3>
+    <pre><code>{code}</code></pre>
+    <h3>Code Execution Results</h3>
+    {result}
+</div>
+"""
+        if save_files_locally:
+            with open('code_execution_results.html', 'w') as file:
+                file.write(html_content)
+
+        return html_content

sunholo/vertex/memory_tools.py

@@ -90,13 +90,13 @@ def get_vertex_memories(vector_name):
                     de = DiscoveryEngineClient(vector_name, project_id=get_gcp_project())
                     log.info(f"Found vectorstore {vectorstore}")
 
-                    data_store_path = de.data_store_path()
+                    data_store_path = f"{de.data_store_path()}/dataStores/{vector_name}"
                     corpus_tool = Tool.from_retrieval(
                         grounding.Retrieval(grounding.VertexAISearch(datastore=data_store_path))
                     )
                     tools.append(corpus_tool)
                 except Exception as err:
-                    log.error(f"Failed to fetch DiscoveryEngine groudning - {str(err)} - skipping")
+                    log.error(f"Failed to fetch DiscoveryEngine grounding - {str(err)} - skipping")
                     continue
                 
 

{sunholo-0.70.5.dist-info → sunholo-0.71.0.dist-info}/METADATA

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.70.5
+Version: 0.71.0
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Home-page: https://github.com/sunholo-data/sunholo-py
-Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.70.5.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.71.0.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0
@@ -56,6 +56,7 @@ Requires-Dist: langchain-anthropic >=0.1.13 ; extra == 'all'
 Requires-Dist: langfuse ; extra == 'all'
 Requires-Dist: pg8000 ; extra == 'all'
 Requires-Dist: pgvector ; extra == 'all'
+Requires-Dist: playwright ; extra == 'all'
 Requires-Dist: psycopg2-binary ; extra == 'all'
 Requires-Dist: pypdf ; extra == 'all'
 Requires-Dist: python-socketio ; extra == 'all'
@@ -110,6 +111,8 @@ Requires-Dist: GitPython ; extra == 'pipeline'
 Requires-Dist: lark ; extra == 'pipeline'
 Requires-Dist: pypdf ; extra == 'pipeline'
 Requires-Dist: tabulate ; extra == 'pipeline'
+Provides-Extra: tools
+Requires-Dist: playwright ; extra == 'tools'
 
 ## Introduction
 This is the Sunholo Python project, a comprehensive toolkit for working with language models and vector stores on Google Cloud Platform. It provides a wide range of functionalities and utilities to facilitate the development and deployment of language model applications.

{sunholo-0.70.5.dist-info → sunholo-0.71.0.dist-info}/RECORD

@@ -63,7 +63,7 @@ sunholo/database/sql/sb/setup.sql,sha256=CvoFvZQev2uWjmFa3aj3m3iuPFzAAJZ0S7Qi3L3
 sunholo/discovery_engine/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sunholo/discovery_engine/chunker_handler.py,sha256=H1HHDqWMCkchJER1_oU9TOLxqf2PygiMO6CL3uKZP64,4563
 sunholo/discovery_engine/create_new.py,sha256=7oZG78T6lW0EspRzlo7-qRyXFSuFxDn2dfSAVEaqlqY,978
-sunholo/discovery_engine/discovery_engine_client.py,sha256=lZDV-eGXTMa6e4xBKFWvudbSHVJayWXoXB17TRZVrck,14655
+sunholo/discovery_engine/discovery_engine_client.py,sha256=QiPB9DExPW-qN-dZq1llXie1hTnWydSZan7Y3mp9dOc,14803
 sunholo/embedder/__init__.py,sha256=sI4N_CqgEVcrMDxXgxKp1FsfsB4FpjoXgPGkl4N_u4I,44
 sunholo/embedder/embed_chunk.py,sha256=d_dIzeNF630Q0Ar-u1hxos60s0tLIImJccAvuo_LTIw,6814
 sunholo/gcs/__init__.py,sha256=DtVw_AZwQn-IguR5BJuIi2XJeF_FQXizhJikzRNrXiE,50
@@ -108,13 +108,14 @@ sunholo/utils/timedelta.py,sha256=BbLabEx7_rbErj_YbNM0MBcaFN76DC4PTe4zD2ucezg,49
 sunholo/utils/user_ids.py,sha256=SQd5_H7FE7vcTZp9AQuQDWBXd4FEEd7TeVMQe1H4Ny8,292
 sunholo/utils/version.py,sha256=jjU_4anXBikJxPg0Wur0X-B7-ec1tC7jToykAnAG9Dg,108
 sunholo/vertex/__init__.py,sha256=JvHcGFuv6R_nAhY2AdoqqhMpJ5ugeWPZ_svGhWrObBk,136
-sunholo/vertex/extensions.py,sha256=LDeuCp8RcHHACdlYhGyGjMoG5v70132HWA_2u-tzcAo,4533
+sunholo/vertex/extensions.py,sha256=d-Ikt9gHFf-jUMPmyU-xHwYe22QtEyr90Ua1LDKgTws,11026
+sunholo/vertex/extensions_class.py,sha256=0-XMrMvfhMN380ZdGXl11Mt7R9kCu9rB4Vduiflk8QA,9202
 sunholo/vertex/init.py,sha256=RLjQppTUwubWgwf2PoAke-EtcwlVkFPaPMYvUsMw1KQ,2029
-sunholo/vertex/memory_tools.py,sha256=l2jNRFLjYpiNJULDZQq5b8fEEaqBQFPM4ehM9NhDHss,5718
+sunholo/vertex/memory_tools.py,sha256=hy_UhKl05Bka5zK7KWoO8ZPVM-p8AQkcgFVLHUg9jZs,5748
 sunholo/vertex/safety.py,sha256=3meAX0HyGZYrH7rXPUAHxtI_3w_zoy_RX7Shtkoa660,1275
-sunholo-0.70.5.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
-sunholo-0.70.5.dist-info/METADATA,sha256=a5HWOEKwPqJ7sfghYwuPxXeGZm7Cevs7qYCeKBjNB3Y,6240
-sunholo-0.70.5.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
-sunholo-0.70.5.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
-sunholo-0.70.5.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
-sunholo-0.70.5.dist-info/RECORD,,
+sunholo-0.71.0.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
+sunholo-0.71.0.dist-info/METADATA,sha256=FBr0EOFvBeqhlFja8LOxDiHKK12wmtiAj59fJ97jKns,6350
+sunholo-0.71.0.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
+sunholo-0.71.0.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
+sunholo-0.71.0.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
+sunholo-0.71.0.dist-info/RECORD,,