pyconverters-openai_vision 0.5.2__py3-none-any.whl

pyconverters_openai_vision/__init__.py

@@ -0,0 +1,2 @@
+"""OpenAIVision converter"""
+__version__ = "0.5.2"

pyconverters_openai_vision/openai_utils.py

@@ -0,0 +1,117 @@
+import os
+from logging import Logger
+
+import requests
+from openai import OpenAI
+from openai.lib.azure import AzureOpenAI
+from pymultirole_plugins.util import comma_separated_to_list
+from strenum import StrEnum
+
+logger = Logger("pymultirole")
+DEFAULT_CHAT_GPT_MODEL = "gpt-4o-mini"
+
+
+# Now use default retry with backoff of openai api
+def openai_chat_completion(prefix, **kwargs):
+    client = set_openai(prefix)
+    response = client.chat.completions.create(**kwargs)
+    return response
+
+
+def openai_list_models(prefix, **kwargs):
+    def sort_by_created(x):
+        if 'created' in x:
+            return x['created']
+        elif 'created_at' in x:
+            return x['created_at']
+        elif 'deprecated' in x:
+            return x['deprecated'] or 9999999999
+        else:
+            return x.id
+
+    models = []
+    client = set_openai(prefix)
+    if prefix.startswith("DEEPINFRA"):
+        deepinfra_url = client.base_url
+        deepinfra_models = {}
+        public_models_list_url = f"{deepinfra_url.scheme}://{deepinfra_url.host}/models/list"
+        response = requests.get(public_models_list_url,
+                                headers={'Accept': "application/json", 'Authorization': f"Bearer {client.api_key}"})
+        if response.ok:
+            resp = response.json()
+            mods = sorted(resp, key=sort_by_created, reverse=True)
+            mods = list(
+                {m['model_name'] for m in mods if m['type'] == 'text-generation'})
+            deepinfra_models.update({m: m for m in mods})
+
+        private_models_list_url = f"{deepinfra_url.scheme}://{deepinfra_url.host}/models/private/list"
+        response = requests.get(private_models_list_url,
+                                headers={'Accept': "application/json", 'Authorization': f"Bearer {client.api_key}"})
+        if response.ok:
+            resp = response.json()
+            mods = sorted(resp, key=sort_by_created, reverse=True)
+            mods = list(
+                {m['model_name'] for m in mods if m['type'] == 'text-generation'})
+            deepinfra_models.update({m: m for m in mods})
+
+        deployed_models_list_url = f"{deepinfra_url.scheme}://{deepinfra_url.host}/deploy/list/"
+        response = requests.get(deployed_models_list_url,
+                                headers={'Accept': "application/json", 'Authorization': f"Bearer {client.api_key}"})
+        if response.ok:
+            resp = response.json()
+            mods = sorted(resp, key=sort_by_created, reverse=True)
+            mods = list(
+                {m['model_name'] for m in mods if m['task'] == 'text-generation' and m['status'] == 'running'})
+            deepinfra_models.update({m: m for m in mods})
+        models = [m for m in deepinfra_models.keys() if 'vision' in m.lower()]
+    elif prefix.startswith("AZURE"):
+        models = comma_separated_to_list(os.getenv(prefix + "OPENAI_DEPLOYMENT_ID", None))
+    else:
+        response = client.models.list(**kwargs)
+        models = sorted(response.data, key=sort_by_created, reverse=True)
+        models = [m.id for m in models]
+    return models
+
+
+def set_openai(prefix):
+    if prefix.startswith("AZURE"):
+        client = AzureOpenAI(
+            # This is the default and can be omitted
+            api_key=os.getenv(prefix + "OPENAI_API_KEY"),
+            azure_endpoint=os.getenv(prefix + "OPENAI_API_BASE", None),
+            api_version=os.getenv(prefix + "OPENAI_API_VERSION", None),
+            # azure_deployment=os.getenv(prefix + "OPENAI_DEPLOYMENT_ID", None)
+        )
+    else:
+        client = OpenAI(
+            # This is the default and can be omitted
+            api_key=os.getenv(prefix + "OPENAI_API_KEY"),
+            base_url=os.getenv(prefix + "OPENAI_API_BASE", None)
+        )
+    return client
+
+
+def gpt_filter(m: str):
+    return m.startswith('gpt') and not m.startswith('gpt-3.5-turbo-instruct') and 'vision' not in m
+
+
+NO_DEPLOYED_MODELS = 'no deployed models - check API key'
+
+
+def create_openai_model_enum(name, prefix="", key=lambda m: m):
+    chat_gpt_models = []
+    default_chat_gpt_model = None
+    try:
+        chat_gpt_models = [m for m in openai_list_models(prefix) if key(m)]
+        if chat_gpt_models:
+            default_chat_gpt_model = DEFAULT_CHAT_GPT_MODEL if DEFAULT_CHAT_GPT_MODEL in chat_gpt_models else \
+                chat_gpt_models[0]
+    except BaseException:
+        logger.warning("Can't list models from endpoint", exc_info=True)
+
+    if len(chat_gpt_models) == 0:
+        chat_gpt_models = [NO_DEPLOYED_MODELS]
+    models = [("".join([c if c.isalnum() else "_" for c in m]), m) for m in chat_gpt_models]
+    model_enum = StrEnum(name, dict(models))
+    default_chat_gpt_model = model_enum(default_chat_gpt_model) if default_chat_gpt_model is not None else None
+    return model_enum, default_chat_gpt_model

pyconverters_openai_vision/openai_vision.py

@@ -0,0 +1,230 @@
+import base64
+import os
+from enum import Enum
+from logging import Logger
+from typing import List, cast, Type, Dict, Any
+
+import filetype as filetype
+from pydantic import Field, BaseModel
+from pymultirole_plugins.v1.converter import ConverterParameters, ConverterBase
+from pymultirole_plugins.v1.schema import Document
+from starlette.datastructures import UploadFile
+
+from .openai_utils import NO_DEPLOYED_MODELS, \
+    openai_chat_completion, create_openai_model_enum
+
+logger = Logger("pymultirole")
+
+
+class OpenAIVisionBaseParameters(ConverterParameters):
+    model_str: str = Field(
+        None, extra="internal"
+    )
+    model: str = Field(
+        None, extra="internal"
+    )
+    prompt: str = Field(
+        "Describe the image with a lot of details",
+        description="""Contains the prompt as a string""",
+        extra="multiline",
+    )
+    max_tokens: int = Field(
+        256,
+        description="""The maximum number of tokens to generate in the completion.
+    The token count of your prompt plus max_tokens cannot exceed the model's context length.
+    Most models have a context length of 2048 tokens (except for the newest models, which support 4096).""",
+    )
+    system_prompt: str = Field(
+        None,
+        description="""Contains the system prompt""",
+        extra="multiline,advanced",
+    )
+    temperature: float = Field(
+        1.0,
+        description="""What sampling temperature to use, between 0 and 2.
+    Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.
+    We generally recommend altering this or `top_p` but not both.""",
+        extra="advanced",
+    )
+    top_p: int = Field(
+        1,
+        description="""An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass.
+    So 0.1 means only the tokens comprising the top 10% probability mass are considered.
+    We generally recommend altering this or `temperature` but not both.""",
+        extra="advanced",
+    )
+    n: int = Field(
+        1,
+        description="""How many completions to generate for each prompt.
+    Note: Because this parameter generates many completions, it can quickly consume your token quota.
+    Use carefully and ensure that you have reasonable settings for `max_tokens`.""",
+        extra="advanced",
+    )
+    best_of: int = Field(
+        1,
+        description="""Generates best_of completions server-side and returns the "best" (the one with the highest log probability per token).
+    Results cannot be streamed.
+    When used with `n`, `best_of` controls the number of candidate completions and `n` specifies how many to return – `best_of` must be greater than `n`.
+    Use carefully and ensure that you have reasonable settings for `max_tokens`.""",
+        extra="advanced",
+    )
+    presence_penalty: float = Field(
+        0.0,
+        description="""Number between -2.0 and 2.0.
+    Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.""",
+        extra="advanced",
+    )
+    frequency_penalty: float = Field(
+        0.0,
+        description="""Number between -2.0 and 2.0.
+    Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.""",
+        extra="advanced",
+    )
+
+
+class OpenAIVisionModel(str, Enum):
+    gpt_4o_mini = "gpt-4o-mini"
+    gpt_4o = "gpt-4o"
+
+
+class OpenAIVisionParameters(OpenAIVisionBaseParameters):
+    model: OpenAIVisionModel = Field(
+        OpenAIVisionModel.gpt_4o_mini,
+        description="""The [OpenAI model](https://platform.openai.com/docs/models) used for speech to text transcription. Options currently available:</br>
+                        <li>`whisper-1` - state-of-the-art open source large-v2 Whisper model.
+                        """, extra="pipeline-naming-hint"
+    )
+
+
+DEEPINFRA_PREFIX = "DEEPINFRA_"
+DEEPINFRA_VISION_MODEL_ENUM, DEEPINFRA_DEFAULT_VISION_MODEL = create_openai_model_enum('DeepInfraVisionModel',
+                                                                                       prefix=DEEPINFRA_PREFIX)
+
+
+class DeepInfraOpenAIVisionParameters(OpenAIVisionBaseParameters):
+    model: DEEPINFRA_VISION_MODEL_ENUM = Field(
+        None,
+        description="""The [DeepInfra 'OpenAI compatible' model](https://deepinfra.com/models?type=automatic-speech-recognition) used for speech to text transcription. It must be deployed on your [DeepInfra dashboard](https://deepinfra.com/dash).
+                         """, extra="pipeline-naming-hint"
+    )
+
+
+# AZURE_PREFIX = "AZURE_"
+#
+#
+# class AzureOpenAIVisionParameters(OpenAIVisionBaseParameters):
+#     model: OpenAIVisionModel = Field(
+#         OpenAIVisionModel.whisper_1,
+#         description="""The [Azure OpenAI model](https://platform.openai.com/docs/models) used for speech to text transcription. Options currently available:</br>
+#                         <li>`whisper-1` - state-of-the-art open source large-v2 Whisper model.
+#                         """, extra="pipeline-naming-hint"
+#     )
+
+
+class OpenAIVisionConverterBase(ConverterBase):
+    __doc__ = """Generate text using [OpenAI Text Completion](https://platform.openai.com/docs/guides/completion) API
+    You input some text as a prompt, and the model will generate a text completion that attempts to match whatever context or pattern you gave it."""
+    PREFIX: str = ""
+
+    def compute_args(self, params: OpenAIVisionBaseParameters, source: UploadFile
+                     ) -> Dict[str, Any]:
+        data = source.file.read()
+        rv = base64.b64encode(data)
+        messages = [{"role": "system", "content": params.system_prompt}] if params.system_prompt is not None else []
+        messages.append({"role": "user",
+                         "content": [
+                             {
+                                 "type": "text",
+                                 "text": params.prompt
+                             },
+                             {
+                                 "type": "image_url",
+                                 "image_url": {
+                                     "url": f"data:image/jpeg;base64,{rv.decode('utf-8')}"
+                                 }
+                             }]})
+        kwargs = {
+            'model': params.model_str,
+            'messages': messages,
+            'max_tokens': params.max_tokens,
+            'temperature': params.temperature,
+            'top_p': params.top_p,
+            'n': params.n,
+            'frequency_penalty': params.frequency_penalty,
+            'presence_penalty': params.presence_penalty,
+        }
+        return kwargs
+
+    def compute_result(self, **kwargs):
+        response = openai_chat_completion(self.PREFIX, **kwargs)
+        contents = []
+        for choice in response.choices:
+            if choice.message.content:
+                contents.append(choice.message.content)
+        if contents:
+            result = "\n".join(contents)
+        return result
+
+    def convert(self, source: UploadFile, parameters: ConverterParameters) \
+            -> List[Document]:
+
+        params: OpenAIVisionBaseParameters = cast(
+            OpenAIVisionBaseParameters, parameters
+        )
+        OPENAI_MODEL = os.getenv(self.PREFIX + "OPENAI_MODEL", None)
+        if OPENAI_MODEL:
+            params.model_str = OPENAI_MODEL
+        doc = None
+        try:
+            kind = filetype.guess(source.file)
+            source.file.seek(0)
+            if kind.mime.startswith("image"):
+                result = None
+                kwargs = self.compute_args(params, source)
+                if kwargs['model'] != NO_DEPLOYED_MODELS:
+                    result = self.compute_result(**kwargs)
+                if result:
+                    doc = Document(identifier=source.filename, text=result)
+                    doc.properties = {"fileName": source.filename}
+        except BaseException as err:
+            raise err
+        if doc is None:
+            raise TypeError(f"Conversion of audio file {source.filename} failed")
+        return [doc]
+
+    @classmethod
+    def get_model(cls) -> Type[BaseModel]:
+        return OpenAIVisionBaseParameters
+
+
+class OpenAIVisionConverter(OpenAIVisionConverterBase):
+    __doc__ = """Convert audio using [OpenAI Audio](https://platform.openai.com/docs/guides/speech-to-text) API"""
+
+    def convert(self, source: UploadFile, parameters: ConverterParameters) \
+            -> List[Document]:
+        params: OpenAIVisionParameters = cast(
+            OpenAIVisionParameters, parameters
+        )
+        params.model_str = params.model.value
+        return super().convert(source, params)
+
+    @classmethod
+    def get_model(cls) -> Type[BaseModel]:
+        return OpenAIVisionParameters
+
+
+class DeepInfraOpenAIVisionConverter(OpenAIVisionConverterBase):
+    __doc__ = """Convert images using [DeepInfra Vision](https://deepinfra.com/docs/tutorials/whisper) API"""
+    PREFIX = DEEPINFRA_PREFIX
+
+    def convert(self, source: UploadFile, parameters: ConverterParameters) \
+            -> List[Document]:
+        params: DeepInfraOpenAIVisionParameters = cast(
+            DeepInfraOpenAIVisionParameters, parameters
+        )
+        params.model_str = params.model.value
+        return super().convert(source, params)
+
+    @classmethod
+    def get_model(cls) -> Type[BaseModel]:
+        return DeepInfraOpenAIVisionParameters

pyconverters_openai_vision-0.5.2.dist-info/METADATA

@@ -0,0 +1,81 @@
+Metadata-Version: 2.1
+Name: pyconverters-openai_vision
+Version: 0.5.2
+Summary: OpenAIVision converter
+Home-page: https://kairntech.com/
+Author: Olivier Terrier
+Author-email: olivier.terrier@kairntech.com
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Classifier: Intended Audience :: Information Technology
+Classifier: Intended Audience :: System Administrators
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python
+Classifier: Topic :: Internet
+Classifier: Topic :: Software Development :: Libraries :: Application Frameworks
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Classifier: Topic :: Software Development :: Libraries
+Classifier: Topic :: Software Development
+Classifier: Typing :: Typed
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Web Environment
+Classifier: Framework :: AsyncIO
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Topic :: Internet :: WWW/HTTP :: HTTP Servers
+Classifier: Topic :: Internet :: WWW/HTTP
+Requires-Dist: pymultirole-plugins>=0.5.0,<0.6.0
+Requires-Dist: openai==1.9.0
+Requires-Dist: Jinja2
+Requires-Dist: tenacity
+Requires-Dist: log-with-context
+Requires-Dist: StrEnum
+Requires-Dist: filetype==1.0.13
+Requires-Dist: requests
+Requires-Dist: flit ; extra == "dev"
+Requires-Dist: pre-commit ; extra == "dev"
+Requires-Dist: bump2version ; extra == "dev"
+Requires-Dist: sphinx ; extra == "docs"
+Requires-Dist: sphinx-rtd-theme ; extra == "docs"
+Requires-Dist: m2r2 ; extra == "docs"
+Requires-Dist: sphinxcontrib.apidoc ; extra == "docs"
+Requires-Dist: jupyter_sphinx ; extra == "docs"
+Requires-Dist: pytest>=7.0 ; extra == "test"
+Requires-Dist: pytest-cov ; extra == "test"
+Requires-Dist: pytest-flake8 ; extra == "test"
+Requires-Dist: pytest-black ; extra == "test"
+Requires-Dist: flake8==3.9.2 ; extra == "test"
+Requires-Dist: tox ; extra == "test"
+Requires-Dist: dirty-equals ; extra == "test"
+Requires-Dist: werkzeug==2.0.0 ; extra == "test"
+Requires-Dist: flask==2.1.3 ; extra == "test"
+Provides-Extra: dev
+Provides-Extra: docs
+Provides-Extra: test
+
+## Requirements
+
+- Python 3.8+
+- Flit to put Python packages and modules on PyPI
+- Pydantic for the data parts.
+
+## Installation
+```
+pip install flit
+pip install pymultirole-plugins
+```
+
+## Publish the Python Package to PyPI
+- Increment the version of your package in the `__init__.py` file:
+```
+"""An amazing package!"""
+
+__version__ = 'x.y.z'
+```
+- Publish
+```
+flit publish
+```
+

pyconverters_openai_vision-0.5.2.dist-info/RECORD

@@ -0,0 +1,7 @@
+pyconverters_openai_vision/__init__.py,sha256=59Qc-bsfVK9kVVX4vnt2PQF_YysgYAQ-T0KEABdUD64,51
+pyconverters_openai_vision/openai_utils.py,sha256=YNOcbeh1sJKeRvkaMh6FnixrUcOfF4rJIGKG5fmnUDo,4806
+pyconverters_openai_vision/openai_vision.py,sha256=MoEyjYCZ75jWiMonaQdR1erpFBpWgYEiqabp9D8102Y,9562
+pyconverters_openai_vision-0.5.2.dist-info/entry_points.txt,sha256=BeU_sWrRTfIdBnkWCBALpYNOrYOYIlRjQdpR2XQhkSU,285
+pyconverters_openai_vision-0.5.2.dist-info/WHEEL,sha256=EZbGkh7Ie4PoZfRQ8I0ZuP9VklN_TvcZ6DSE5Uar4z4,81
+pyconverters_openai_vision-0.5.2.dist-info/METADATA,sha256=d2SvC6t_svBjtYRBo-JLShoD_9fVg5ODMvlI9pDULRo,2635
+pyconverters_openai_vision-0.5.2.dist-info/RECORD,,

pyconverters_openai_vision-0.5.2.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: flit 3.9.0
+Root-Is-Purelib: true
+Tag: py3-none-any

pyconverters_openai_vision-0.5.2.dist-info/entry_points.txt

@@ -0,0 +1,5 @@
+[pyconverters.plugins]
+azure_openai_vision=pyconverters_openai_vision.openai_vision:AzureOpenAIVisionConverter
+deepinfra_openai_vision=pyconverters_openai_vision.openai_vision:DeepInfraOpenAIVisionConverter
+openai_vision=pyconverters_openai_vision.openai_vision:OpenAIVisionConverter
+