davidkhala.ai 0.1.0__py3-none-any.whl → 0.1.2__py3-none-any.whl

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/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/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/agent/dify/knowledge.py

@@ -0,0 +1,163 @@
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from typing import Iterable, TypedDict, Callable, Any, Optional
+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[[int, int], Any], r: Optional[dict]):
+        self.response = r
+        self.fn = get_fn
+
+    def __next__(self):
+        if self.response and not self.response['has_more']:
+            raise StopIteration
+        page = 1 if not self.response else self.response['page'] + 1
+        limit = None if not self.response else self.response['limit']
+        self.response = self.fn(page, limit)
+        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 paginate_datasets(self, page=1, size=20):
+        r = self.request(self.base_url, "GET", params={
+            'page': page,
+            'limit': size,
+        })
+        return r
+
+    def list_datasets(self) -> Iterable[list[DatasetDict]]:
+        return Iterator(self.paginate_datasets, None)
+
+    @property
+    def ids(self):
+        for sub_list in self.list_datasets():
+            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 paginate_documents(self, page=1, size=20):
+            return self.request(f"{self.base_url}/documents", "GET", params={
+                'page': page,
+                'limit': size
+            })
+
+        def list_documents(self) -> Iterable[list[DocumentDict]]:
+            return Iterator(self.paginate_documents, 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}")
+        try:
+            self.get()
+            self.exist = True
+        except requests.exceptions.HTTPError as e:
+            if e.response.status_code == 404:
+                self.exist = False
+            raise e
+
+    def get(self):
+        return self.request(self.base_url, "GET")

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/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/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/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/openai/native.py

@@ -1,3 +1,5 @@
+from typing import Optional, Literal
+
 from openai import OpenAI
 
 from davidkhala.ai.openai import Client
@@ -10,3 +12,11 @@ class NativeClient(Client):
             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-0.1.0.dist-info → davidkhala_ai-0.1.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: davidkhala.ai
-Version: 0.1.0
+Version: 0.1.2
 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-0.1.0.dist-info → davidkhala_ai-0.1.2.dist-info}/RECORD

@@ -1,8 +1,13 @@
 davidkhala/ai/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 davidkhala/ai/model.py,sha256=1wcXC8X8oqerMatlcPbZmuxZ-nJWdJKmaDSDgiGlUGw,647
 davidkhala/ai/opik.py,sha256=YU1XuweMUAzUkhpjxhltt-SBBDBkR3z-PCNo0DqzBRs,39
+davidkhala/ai/agent/README.md,sha256=kIPsx3gOjrpOw7w2qhNEALuCEQkuh4nYp6uBnijdvHE,178
 davidkhala/ai/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 davidkhala/ai/agent/langgraph.py,sha256=jrc_Yvgo7eJjd3y5UJn0t1FzpnObDGYscwgsuVl2O_I,1052
+davidkhala/ai/agent/ragflow.py,sha256=UaK31us6V0NhAPCthGo07rQsm72vlR-McmihC_NDe1g,273
+davidkhala/ai/agent/dify/__init__.py,sha256=GNgNwwX75NGBNsyKnA_Qz63QnbwOwF4nKd5fhHwdcKM,246
+davidkhala/ai/agent/dify/base.py,sha256=V4BIhG9oTg9ayPZnYVxxA1Yobl2C4ukSWEOu2U_l0Tw,259
+davidkhala/ai/agent/dify/knowledge.py,sha256=FH-aSMx3EG7uXYBPL0olbOKfgeYEwkgL5SHYFsPSsKU,4980
 davidkhala/ai/ali/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 davidkhala/ai/ali/dashscope.py,sha256=SZIzRhVHlLx3s5I2RNUh2-u8OoSdrbvoN5e1k8Mh8N0,1943
 davidkhala/ai/api/__init__.py,sha256=q2Ro5nhW5kJx2CYR1MRVamjTT5tTexPZwhrS2hwAvFM,1319
@@ -13,9 +18,10 @@ davidkhala/ai/google/adk.py,sha256=QwxYoOzT2Hol03V4NM0PF_HAzUGb4fB18VUAYacYbAY,6
 davidkhala/ai/google/gemini.py,sha256=Xf4HDOOcK4-jEBERzuLnQNFsU61P2fFx4K0z-ijvNHE,214
 davidkhala/ai/huggingface/BAAI.py,sha256=LZ9kp5Gfql4UzuTn4osyekI6VV1H3RIfED2IolXFj5c,341
 davidkhala/ai/huggingface/__init__.py,sha256=FJyU8eOfWQWKAvkIa5qwubF9ghsSQ8C0e6p6DKyomgs,521
-davidkhala/ai/openai/__init__.py,sha256=UalrIvniAFSbCAQeCrDgBVEWCFTCPxFnoOAGVnVWkkI,1864
-davidkhala/ai/openai/azure.py,sha256=jAtZNW1d8ub3odhB_f1MXxM7MImg64MXDnqLm1-Idkk,828
-davidkhala/ai/openai/native.py,sha256=-XuHn0KSQnuFdJ4jTwZla1ts5iOW2L2Hosfo8o8iip8,257
-davidkhala_ai-0.1.0.dist-info/METADATA,sha256=MnG5FylnaOFamw3sjPcMMVCUoNEtMWOfxnoyqP1A40Q,877
-davidkhala_ai-0.1.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-davidkhala_ai-0.1.0.dist-info/RECORD,,
+davidkhala/ai/huggingface/inference.py,sha256=bYN0PtLF2CaIHzdTP4LaTALJhcawvuLnLR7rhMVqwDE,333
+davidkhala/ai/openai/__init__.py,sha256=GXzWaw2ER3YFGHG6TPD9SmAHV6Tpsnqxj6tXlaWsrko,1897
+davidkhala/ai/openai/azure.py,sha256=QR1uZj8qAyhpCjo3Ks5zNV8GfOp3-enyZs6fBvV-MkA,1110
+davidkhala/ai/openai/native.py,sha256=MB0nDnzCOj_M42RMhdK3HTMVnxGnwpLT2GeLwSrepwI,704
+davidkhala_ai-0.1.2.dist-info/METADATA,sha256=QHtyUo2dYlHTg2hpLKgVLGbDNg_z4fTpLesFKUCCxCY,1098
+davidkhala_ai-0.1.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+davidkhala_ai-0.1.2.dist-info/RECORD,,