davidkhala.ai 0.1.7__tar.gz → 0.2.0__tar.gz

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/PKG-INFO

@@ -1,13 +1,16 @@
 Metadata-Version: 2.4
 Name: davidkhala.ai
-Version: 0.1.7
+Version: 0.2.0
 Summary: misc AI modules
-Requires-Python: >=3.13
+Requires-Python: >=3.12
 Provides-Extra: ali
 Requires-Dist: dashscope; extra == 'ali'
+Requires-Dist: davidkhala-utils; extra == 'ali'
+Requires-Dist: wuying-agentbay-sdk; extra == 'ali'
 Provides-Extra: api
 Requires-Dist: davidkhala-utils[http-request]; extra == 'api'
 Provides-Extra: azure
+Requires-Dist: davidkhala-utils; extra == 'azure'
 Requires-Dist: openai; extra == 'azure'
 Provides-Extra: dify
 Requires-Dist: davidkhala-databases[pg]; extra == 'dify'

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/agent/dify/api/knowledge.py

@@ -8,7 +8,8 @@ from urllib.parse import urlparse
 import requests
 
 from davidkhala.ai.agent.dify.api import API, Iterator
-from davidkhala.ai.agent.dify.common import Document as DocumentBase
+from davidkhala.ai.agent.dify.model import Document as DocumentBase
+
 
 class DatasetDict(TypedDict):
     id: str

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/const.py

@@ -0,0 +1,10 @@
+from enum import Enum
+
+
+class IndexingStatus(str, Enum):
+    WAITING = "waiting"
+    PARSING = "parsing"
+    SPLITTING = 'splitting'
+    INDEXING = "indexing"
+    COMPLETED = "completed"
+    FAILED = "error"

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/interface.py

@@ -0,0 +1,3 @@
+class IndexingError(Exception):
+    """Raised when document indexing fails (indexing_status = 'error')"""
+    pass

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/model.py

@@ -0,0 +1,31 @@
+from pydantic import BaseModel, Field
+
+from davidkhala.ai.agent.dify.const import IndexingStatus
+
+
+class Document(BaseModel):
+    id: str
+    position: int
+    data_source_type: str
+    data_source_info: dict[str, str]
+    name: str
+    indexing_status: IndexingStatus
+    error: str | None
+    enabled: bool
+
+
+class Dataset(BaseModel):
+    id: str
+    name: str
+    description: str
+
+
+class JsonData(BaseModel):
+    data: list
+
+
+class NodeOutput(BaseModel):
+    """Schema for Output of a Dify node"""
+    text: str
+    files: list
+    json_: list[JsonData] = Field(alias="json") # avoid conflict with .json()

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/agent/dify/ops/console/knowledge.py

@@ -3,7 +3,9 @@ from time import sleep
 from davidkhala.utils.http_request.stream import as_sse, Request as StreamRequest
 from pydantic import BaseModel
 
-from davidkhala.ai.agent.dify.common import IndexingStatus, IndexingError, Dataset, Document
+from davidkhala.ai.agent.dify.interface import IndexingError
+from davidkhala.ai.agent.dify.model import Document, Dataset
+from davidkhala.ai.agent.dify.const import IndexingStatus
 from davidkhala.ai.agent.dify.ops.console import API
 from davidkhala.ai.agent.dify.ops.console.session import ConsoleUser
 from davidkhala.ai.agent.dify.ops.db.orm import Node

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/ops/console/plugin.py

@@ -0,0 +1,68 @@
+from time import sleep
+
+from davidkhala.ai.agent.dify.ops.console import API
+from davidkhala.ai.agent.dify.ops.console.session import ConsoleUser
+
+
+class ConsolePlugin(API):
+    def __init__(self, context: ConsoleUser):
+        super().__init__()
+        self.base_url = f"{context.base_url}/workspaces/current/plugin"
+        self.session.cookies = context.session.cookies
+        self.options = context.options
+
+    def upgrade(self, *plugin_names: str) -> list[dict]:
+        versions = self.latest_version(*plugin_names)
+        self.async_install(*versions)
+
+        current = []
+        while len(current) < len(versions):
+            current = self.get(*plugin_names)
+            sleep(1)
+        return current
+
+    def async_install(self, *plugin_versioned_names: str) -> str | None:
+        url = f"{self.base_url}/install/marketplace"
+        r = self.request(url, method="POST", json={
+            'plugin_unique_identifiers': plugin_versioned_names,
+        })
+        if r['all_installed']:
+            # plugins exist, no need to install
+            return None
+
+        return r['task_id']
+
+    def plugins(self, *, page=1, size=100):
+        url = f"{self.base_url}/list?page={page}&page_size={size}"
+        r = self.request(url, method="GET")
+        _ = r['plugins']
+        assert r['total'] == len(_)
+        return _
+
+    def get(self, *plugin_names: str) -> list[dict]:
+        "inspect installed plugins"
+        url = f"{self.base_url}/list/installations/ids"
+        r = self.request(url, method="POST", json={
+            'plugin_ids': plugin_names,
+        })
+        return r['plugins']
+
+    def latest_version(self, *plugin_names: str) -> dict:
+        url = f"{self.base_url}/list/latest-versions"
+        r = self.request(url, method="POST", json={
+            'plugin_ids': plugin_names,
+        })
+        return [r['versions'][name]['unique_identifier'] for name in plugin_names]
+
+    def uninstall(self, id: str):
+        url = f"{self.base_url}/uninstall"
+
+        r = self.request(url, method="POST", json={
+            'plugin_installation_id': id
+        })
+        assert r['success'] is True
+
+    def uninstall_by(self, *plugin_names: str):
+        for name in plugin_names:
+            r = self.get(name)
+            self.uninstall(r[0]['id'])

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/agent/dify/ops/console/session.py

@@ -1,3 +1,5 @@
+from base64 import b64encode
+
 from davidkhala.ai.agent.dify.ops.console import API
 
 
@@ -11,7 +13,7 @@ class ConsoleUser(API):
 
         r = self.request(url, "POST", json={
             'email': email,
-            'password': password,
+            'password': b64encode(password.encode()).decode(), # use base64 from dify 1.11
             'remember_me': remember_me,
             'language': language,
         })

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/agent/dify/ops/db/app.py

@@ -4,16 +4,13 @@ from sqlalchemy.orm import Session
 from sqlalchemy import desc
 
 class Studio(DB):
-    @property
-    def user_feedbacks(self):
-        sql = """SELECT mf.conversation_id,
+    user_feedbacks_sql = """SELECT mf.conversation_id,
                         mf.content,
                         m.query,
                         m.answer
                  FROM message_feedbacks mf
                           LEFT JOIN messages m ON mf.message_id = m.id
                  WHERE mf.from_source = 'user'"""
-        return self.get_dict(sql)
 
     @property
     def apps(self): return self.get_dict("select id, name, mode from apps where status = 'normal'")

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/agent/dify/ops/db/knowledge.py

@@ -26,16 +26,22 @@ class Dataset(DB):
         template = "select id from datasource_providers where name = :name and provider = :provider"
         return self.query(template, {'name': name, 'provider': provider}).scalars().all()
 
+    def documents(self, dataset_id: str):
+        template = "select id, name,created_from, created_at from documents where dataset_id = :dataset_id"
+        return self.query(template, {'dataset_id': dataset_id})
+
 
 class Document(DB):
     def hit_documents(self, top_k: int = 3):
         template = "SELECT dataset_id, document_id, content FROM document_segments ORDER BY hit_count DESC LIMIT :top_k"
         return self.get_dict(template, {'top_k': top_k})
 
-    def id_by(self, name) -> list[str]:
+    def id_by(self, name: str, dataset_id: str = None) -> list[str]:
         """multiple ids can be found"""
         template = "select id from documents where name = :name"
-        return [str(uuid) for uuid in self.query(template, {'name': name}).scalars().all()]
+        if dataset_id:
+            template = "select id from documents where name = :name and dataset_id = :dataset_id"
+        return [str(uuid) for uuid in self.query(template, {'name': name, 'dataset_id': dataset_id}).scalars().all()]
 
 
 class Pipeline(DB):

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/plugins/__init__.py

@@ -0,0 +1,7 @@
+from typing import Literal
+
+from pydantic import BaseModel
+
+
+class DataSourceTypeAware(BaseModel):
+    datasource_type: Literal["local_file", "online_document", "website_crawl"]

davidkhala_ai-0.2.0/davidkhala/ai/agent/dify/plugins/popular.py

@@ -0,0 +1,36 @@
+model_provider = [
+    'langgenius/siliconflow',
+    'langgenius/azure_openai',
+    'langgenius/tongyi',
+    'langgenius/jina',
+    'langgenius/openrouter',
+    'langgenius/deepseek',
+]
+
+class Knowledge:
+    data_source = [
+        'langgenius/firecrawl_datasource',
+    ]
+    chunk = [
+        'langgenius/parentchild_chunker',
+        'langgenius/general_chunker',
+    ]
+    api = [
+        'abesticode/knowledge_pro',
+    ]
+
+class Node:
+    format = [
+        'langgenius/json_process',
+        'langgenius/dify_extractor',
+    ]
+    agent = [
+        'langgenius/agent',
+    ]
+    data = [
+        'langgenius/chart',
+        'junjiem/db_query',
+        'junjiem/db_query_pre_auth',
+    ]
+
+

davidkhala_ai-0.2.0/davidkhala/ai/ali/agentbay.py

@@ -0,0 +1,39 @@
+from agentbay import AgentBay, Session, Config, AgentBayLogger, BrowserOption
+from davidkhala.utils.syntax.interface import ContextAware
+
+AgentBayLogger.setup(level='WARNING')  # Default to INFO
+
+
+class Client(ContextAware):
+    def __init__(self, api_key, *, timeout_ms=10000):
+        self.agent = AgentBay(
+            api_key=api_key,
+            cfg=Config(endpoint="wuyingai.ap-southeast-1.aliyuncs.com", timeout_ms=timeout_ms)
+        )
+        self.session: Session | None = None
+
+    def open(self):
+        r = self.agent.create()
+        if not r.success:
+            return False
+        self.session = r.session
+        return True
+
+    def close(self):
+        self.agent.delete(self.session)
+        del self.session
+
+
+class Browser(ContextAware):
+    def __init__(self, session: Session):
+        self.session = session
+        self.option = BrowserOption()
+        self.endpoint_url: str | None = None
+
+    def open(self) -> bool:
+        success = self.session.browser.initialize(self.option)
+        self.endpoint_url = self.session.browser.get_endpoint_url()
+        return success
+
+    def close(self):
+        self.session.browser.destroy()

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/davidkhala/ai/openai/azure.py

@@ -1,5 +1,4 @@
-import warnings
-
+from davidkhala.utils.syntax.compat import deprecated
 from openai import AzureOpenAI, OpenAI
 
 from davidkhala.ai.openai import Client
@@ -20,7 +19,7 @@ class ModelDeploymentClient(AzureHosted):
         )
 
 
-@warnings.deprecated("Azure Open AI is deprecated. Please migrate to Azure AI Foundry")
+@deprecated("Azure Open AI is deprecated. Please migrate to Microsoft Foundry")
 class OpenAIClient(AzureHosted):
 
     def __init__(self, api_key, project):

{davidkhala_ai-0.1.7 → davidkhala_ai-0.2.0}/pyproject.toml

@@ -1,9 +1,9 @@
 [project]
 name = "davidkhala.ai"
-version = "0.1.7"
+version = "0.2.0"
 description = "misc AI modules"
 readme = "README.md"
-requires-python = ">=3.13"
+requires-python = ">=3.12"
 
 [project.optional-dependencies]
 langchain = [
@@ -18,8 +18,8 @@ hf = [
     'onnxruntime', 'onnx' # for test only
 ]
 openrouter = ["openrouter"]
-ali = ["dashscope"]
-azure = ["openai"]
+ali = ["dashscope", "wuying-agentbay-sdk", "davidkhala.utils"]
+azure = ["openai", "davidkhala.utils"]
 telemetry = [
     "opik ; python_version < '3.14'" # limit py version:　PyO3 v0.22.6 currently supports up to Python 3.13.
 ]

davidkhala_ai-0.1.7/davidkhala/ai/agent/dify/common.py

@@ -1,36 +0,0 @@
-from enum import Enum
-
-from pydantic import BaseModel
-
-from davidkhala.ai.agent.dify.plugins.firecrawl import DataSourceInfo
-
-
-class IndexingStatus(str, Enum):
-    WAITING = "waiting"
-    PARSING = "parsing"
-    SPLITTING = 'splitting'
-    INDEXING = "indexing"
-    COMPLETED = "completed"
-    FAILED = "error"
-
-
-class Document(BaseModel):
-    id: str
-    position: int
-    data_source_type: str
-    data_source_info: dict[str, str]
-    name: str
-    indexing_status: IndexingStatus
-    error: str | None
-    enabled: bool
-
-
-class Dataset(BaseModel):
-    id: str
-    name: str
-    description: str
-
-
-class IndexingError(Exception):
-    """Raised when document indexing fails (indexing_status = 'error')"""
-    pass

davidkhala_ai-0.1.7/davidkhala/ai/agent/dify/plugins/__init__.py

@@ -1,14 +0,0 @@
-from typing import Literal
-
-from pydantic import BaseModel
-
-class JsonEntry(BaseModel):
-    data: list
-
-class Output(BaseModel):
-    """Class for result of a Dify node"""
-    text: str
-    files: list
-    json: list[JsonEntry]
-class DataSourceTypeAware(BaseModel):
-    datasource_type: Literal["local_file", "online_document", "website_crawl"]