davidkhala.ai 0.0.9__tar.gz → 0.1.1__tar.gz

{davidkhala_ai-0.0.9 → davidkhala_ai-0.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: davidkhala.ai
-Version: 0.0.9
+Version: 0.1.1
 Summary: misc AI modules
 Requires-Python: >=3.13
 Provides-Extra: ali
@@ -21,5 +21,12 @@ Provides-Extra: langchain
 Requires-Dist: langchain; extra == 'langchain'
 Requires-Dist: langchain-openai; extra == 'langchain'
 Requires-Dist: langgraph; extra == 'langchain'
+Provides-Extra: ragflow
+Requires-Dist: ragflow-sdk; extra == 'ragflow'
 Provides-Extra: telemetry
 Requires-Dist: opik; (python_version < '3.14') and extra == 'telemetry'
+Description-Content-Type: text/markdown
+
+# davidkhala.ai
+
+For usage of `azure.ai.agents`, goto https://github.com/davidkhala/azure-utils/tree/main/py 

davidkhala_ai-0.1.1/README.md

@@ -0,0 +1,3 @@
+# davidkhala.ai
+
+For usage of `azure.ai.agents`, goto https://github.com/davidkhala/azure-utils/tree/main/py 

davidkhala_ai-0.1.1/davidkhala/ai/agent/README.md

@@ -0,0 +1,7 @@
+
+# ragflow python sdk
+
+beyond using official package `ragflow-sdk`, these are highlighted community client
+- [ragflow-client](https://pypi.org/project/ragflow-client/)
+ 
+

davidkhala_ai-0.1.1/davidkhala/ai/agent/dify/__init__.py

@@ -0,0 +1,9 @@
+from datetime import datetime
+
+
+def with_datetime(filename:str):
+    if ' ' in filename:
+        raise ValueError(f"Filename '{filename}' should not contain spaces.")
+    return f"{datetime.now().strftime("%Y%m%d_%H%M%S")}.{filename}"
+
+

davidkhala_ai-0.1.1/davidkhala/ai/agent/dify/base.py

@@ -0,0 +1,7 @@
+from davidkhala.utils.http_request import Request
+
+class API(Request):
+    def __init__(self, api_key: str, base_url="https://api.dify.ai/v1"):
+        super().__init__({'bearer': api_key})
+        self.base_url = base_url
+        self.api_key = api_key

davidkhala_ai-0.1.1/davidkhala/ai/agent/dify/knowledge.py

@@ -0,0 +1,156 @@
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from typing import Iterable, TypedDict, Callable
+from urllib.parse import urlparse
+
+import requests
+
+from davidkhala.ai.agent.dify.base import API
+
+
+class DatasetDict(TypedDict):
+    id: str
+    name: str
+    description: str
+    provider: str
+    permission: str
+    data_source_type: str
+    indexing_technique: str
+    doc_form: str
+    runtime_mode: str
+    is_published: bool
+    enable_api: bool
+    # stats
+    app_count: int
+    document_count: int
+    word_count: int
+    total_documents: int
+    total_available_documents: int
+    # embedding
+    embedding_available: bool
+    embedding_model: str
+    embedding_model_provider: str
+    retrieval_model_dict: dict
+    external_retrieval_model: dict
+    external_knowledge_info: dict
+
+
+class Iterator(Iterable):
+    def __iter__(self):
+        return self
+
+    def __init__(self, get_fn: Callable, r):
+        self.response = r
+        self.fn = get_fn
+
+    def __next__(self):
+        if self.response and not self.response['has_more']:
+            raise StopIteration
+        self.response = self.fn()
+        return self.response['data']
+
+
+class DocumentDict(TypedDict):
+    id: str
+    position: int
+    data_source_type: str
+    data_source_info: dict[str, str]
+    data_source_detail_dict: dict[str, dict]
+    dataset_process_rule_id: str
+    name: str
+    created_from: str
+    created_by: str
+    created_at: int
+    tokens: int
+    indexing_status: str
+    error: str
+    enabled: bool
+    archived: bool
+    display_status: str
+    word_count: int
+    hit_count: int
+    doc_form: str
+    doc_metadata: dict
+    disabled_at: int
+    disabled_by: str
+
+
+class Dataset(API):
+    def __init__(self, api_key: str, base_url="https://api.dify.ai/v1"):
+        super().__init__(api_key, f"{base_url}/datasets")
+
+    def list(self):
+        return self.request(self.base_url, "GET")
+
+    def list_all(self) -> Iterable[DatasetDict]:
+        return Iterator(self.list, None)
+
+    @property
+    def ids(self):
+        for sub_list in self.list_all():
+            for dataset in sub_list:
+                yield dataset['id']
+
+    class Instance(API):
+        def __init__(self, d: Dataset, dataset_id: str):
+            super().__init__(d.api_key, f"{d.base_url}/{dataset_id}")
+
+        def get(self):
+            return self.request(self.base_url, "GET")
+
+        def upload(self, filename, *, path=None, url=None, document_id=None):
+            """
+            don't work for html
+            work for markdown
+            TODO how to simulate console
+            """
+            files = {}
+            if path:
+                with open(path, 'rb') as f:
+                    content = f.read()
+                if not filename:
+                    filename = os.path.basename(path)
+            elif url:
+                r = requests.get(url)
+                r.raise_for_status()
+                if not filename:
+                    parsed_url = urlparse(url)
+                    filename = Path(parsed_url.path).name
+                content = r.content
+            files['file'] = (filename, content)
+            if document_id:
+                # don't work for html
+                r = requests.post(f"{self.base_url}/documents/{document_id}/update-by-file", files=files,
+                                  **self.options)
+            else:
+                r = requests.post(f"{self.base_url}/document/create-by-file", files=files, **self.options)
+            r = self.on_response(r)
+            return r['document']
+
+        def list(self):
+            return self.request(f"{self.base_url}/documents", "GET")
+
+        def list_documents(self) -> Iterable[DocumentDict]:
+            return Iterator(self.list, None)
+
+        def has_document(self, name) -> bool:
+            return any(name == item['name'] for row in self.list_documents() for item in row)
+
+
+class Document(API):
+    def __init__(self, d: Dataset.Instance, document_id: str):
+        super().__init__(d.api_key, f"{d.base_url}/documents/{document_id}")
+
+    def get(self):
+        return self.request(self.base_url, "GET")
+
+    def exist(self):
+        try:
+            self.get()
+            return True
+        except requests.exceptions.HTTPError as e:
+            if e.response.status_code == 404:
+                return False
+            raise e

davidkhala_ai-0.1.1/davidkhala/ai/agent/ragflow.py

@@ -0,0 +1,8 @@
+from ragflow_sdk import RAGFlow
+class Client:
+    def __init__(self, api_key:str, base_url ="http://localhost:9380/"):
+        self.client = RAGFlow(api_key=api_key, base_url=base_url)
+
+    @property
+    def datasets(self):
+        return self.client.list_datasets()

{davidkhala_ai-0.0.9 → davidkhala_ai-0.1.1}/davidkhala/ai/api/siliconflow.py

@@ -48,7 +48,7 @@ class SiliconFlow(API):
         response = self.request(f"{self.base_url}/embeddings", "POST", json=json)
         return [_['embedding'] for _ in response['data']]
 
-    def which(self, query: str, documents: list[str], **kwargs):
+    def which(self, query: str, documents: list[str], **kwargs)->tuple[str,int]:
         json = {
             'model': self.model,
             'query': query,
@@ -56,6 +56,7 @@ class SiliconFlow(API):
             **kwargs
         }
         response = self.request(f"{self.base_url}/rerank", "POST", json=json)
-        most_relevant = max(response['results'], key=lambda x: x['relevance_score'])
-        return documents[most_relevant['index']]
+        most_relevant_index = max(response['results'], key=lambda x: x['relevance_score'])['index']
+
+        return documents[most_relevant_index], most_relevant_index
 

davidkhala_ai-0.1.1/davidkhala/ai/huggingface/inference.py

@@ -0,0 +1,13 @@
+from huggingface_hub import InferenceApi
+
+
+class API:
+    def __init__(self, token):
+        self.inference = None
+        self.token = token
+
+    def as_model(self, repo_id):
+        self.inference = InferenceApi(repo_id=repo_id, token=self.token)
+
+    def call(self, **kwargs):
+        return self.inference(**kwargs)

{davidkhala_ai-0.0.9 → davidkhala_ai-0.1.1}/davidkhala/ai/openai/__init__.py

@@ -22,7 +22,7 @@ class Client(AbstractClient):
         )
         return [item.embedding for item in response.data]
 
-    def chat(self, *user_prompt):
+    def chat(self, *user_prompt, **kwargs):
 
         messages = [
             *self.messages,
@@ -47,7 +47,8 @@ class Client(AbstractClient):
         response = self.client.chat.completions.create(
             model=self.model,
             messages=messages,
-            n=self.n
+            n=self.n,
+            **kwargs
         )
         contents = [choice.message.content for choice in response.choices]
         assert len(contents) == self.n

{davidkhala_ai-0.0.9 → davidkhala_ai-0.1.1}/davidkhala/ai/openai/azure.py

@@ -5,7 +5,13 @@ from openai import AzureOpenAI, OpenAI
 from davidkhala.ai.openai import Client
 
 
-class ModelDeploymentClient(Client):
+class AzureHosted(Client):
+    def chat(self, *user_prompt, **kwargs):
+        if 'web_search_options' in kwargs:
+            raise ValueError('Web search options not supported in any models of Azure AI Foundry')
+        return super().chat(*user_prompt, **kwargs)
+
+class ModelDeploymentClient(AzureHosted):
     def __init__(self, key, deployment):
         self.client = AzureOpenAI(
             api_version="2024-12-01-preview",  # mandatory
@@ -15,7 +21,7 @@ class ModelDeploymentClient(Client):
 
 
 @warnings.deprecated("Azure Open AI is deprecated. Please migrate to Azure AI Foundry")
-class OpenAIClient(Client):
+class OpenAIClient(AzureHosted):
 
     def __init__(self, api_key, project):
         self.client = OpenAI(

davidkhala_ai-0.1.1/davidkhala/ai/openai/native.py

@@ -0,0 +1,22 @@
+from typing import Optional, Literal
+
+from openai import OpenAI
+
+from davidkhala.ai.openai import Client
+
+
+class NativeClient(Client):
+    def __init__(self, api_key, base_url=None):
+        self.client = OpenAI(
+            api_key=api_key,
+            base_url=base_url
+        )
+
+    def chat(self, *user_prompt, web_search:Optional[Literal["low", "medium", "high"]]=None, **kwargs):
+        opts = {
+            **kwargs
+        }
+        if web_search:
+            from openai.types.chat.completion_create_params import WebSearchOptions
+            opts['web_search_options'] = WebSearchOptions(search_context_size=web_search)
+        return super().chat(*user_prompt, **opts)

{davidkhala_ai-0.0.9 → davidkhala_ai-0.1.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "davidkhala.ai"
-version = "0.0.9"
+version = "0.1.1"
 description = "misc AI modules"
 readme = "README.md"
 requires-python = ">=3.13"
@@ -10,6 +10,7 @@ langchain = [
     "langgraph", "langchain",
     "langchain-openai" # required by openRouter
 ]
+ragflow = ["ragflow-sdk"]
 google = ["google-genai", "google-adk"]
 api = ["davidkhala.utils[http_request]"]
 hf = [

davidkhala_ai-0.0.9/davidkhala/ai/openai/native.py

@@ -1,12 +0,0 @@
-from openai import OpenAI
-
-from davidkhala.ai.openai import Client
-
-
-class NativeClient(Client):
-    def __init__(self, api_key, base_url=None):
-        self.client = OpenAI(
-            api_key=api_key,
-            base_url=base_url
-        )
-