bella-openapi 1.0.2.3__py3-none-any.whl → 1.0.2.5__py3-none-any.whl

bella_openapi/__init__.py

@@ -3,7 +3,19 @@ from .log import operation_log, submit_log
 from .openapi_contexvar import trace_id_context, caller_id_context, request_url_context
 from .auth_billing import ErrorInfo, async_authenticate_decorator_args, authenticate_user, print_context, \
 get_context, set_context, clean_context, report
-from .domtree import StandardDomTree, StandardNode
+from .entity import (
+    StandardDomTree,
+    StandardNode,
+    SourceFile,
+    StandardPosition,
+    StandardImage,
+    Cell,
+    StandardRow,
+    StandardBaseElement,
+    StandardElement,
+    StandardTableElement,
+    StandardImageElement
+)
 
 __all__ = ["validate_token", "operation_log",
            "support_model",
@@ -22,5 +34,14 @@ __all__ = ["validate_token", "operation_log",
            "clean_context",
            "report",
            "StandardDomTree",
-           "StandardNode"
+           "StandardNode",
+           "SourceFile",
+           "StandardPosition",
+           "StandardImage",
+           "Cell",
+           "StandardRow",
+           "StandardBaseElement",
+           "StandardElement",
+           "StandardTableElement",
+           "StandardImageElement"
            ]

bella_openapi/entity/__init__.py

@@ -0,0 +1,27 @@
+from .standard_domtree import (
+    StandardDomTree,
+    StandardNode,
+    SourceFile,
+    StandardPosition,
+    StandardImage,
+    Cell,
+    StandardRow,
+    StandardBaseElement,
+    StandardElement,
+    StandardTableElement,
+    StandardImageElement
+)
+
+__all__ = [
+    "StandardDomTree",
+    "StandardNode", 
+    "SourceFile",
+    "StandardPosition",
+    "StandardImage",
+    "Cell",
+    "StandardRow",
+    "StandardBaseElement",
+    "StandardElement",
+    "StandardTableElement",
+    "StandardImageElement"
+]

bella_openapi/{domtree → entity}/standard_domtree.py

@@ -10,8 +10,7 @@ StandardDomTree - 新的DOM树协议定义
 
 from typing import List, Optional, Literal, Any, Union
 from pydantic import BaseModel, Field
-
-from bella_openapi.domtree.utils import count_tokens
+import tiktoken
 
 # 布局类型映射表：将旧解析器的布局类型映射到新的标准类型
 # 注意：这个映射表是临时的，后续计划统一布局类型定义
@@ -33,6 +32,7 @@ layout_type_mapping = {
 }
 
 
+
 class SourceFile(BaseModel):
     id: str  # 文件ID，唯一标识符，类型为string
     name: str  # 文件名，文档的名称，类型为string
@@ -431,7 +431,7 @@ class StandardDomTree(BaseModel):
                                       cell_data['end_col']],
                                 text=cell_text,
                                 # 目前只会有一个元素,且是Text类型，Path重新从头编号，相对cell是root
-                                nodes=[StandardNode(summary="", tokens=count_tokens(cell_text), path=[1], children=[],
+                                nodes=[StandardNode(summary="", tokens=cls.count_tokens(cell_text), path=[1], children=[],
                                                     element=StandardElement(
                                                         type='Text',
                                                         positions=[],
@@ -482,7 +482,7 @@ class StandardDomTree(BaseModel):
                     standard_node.children.append(standard_child)
 
         # 计算 token 数量：自身 text 的 token 数量 + 子节点 token 数量
-        tokens = count_tokens(text)
+        tokens = cls.count_tokens(text)
         for child in standard_node.children:
             tokens += child.tokens
 
@@ -490,3 +490,24 @@ class StandardDomTree(BaseModel):
         standard_node.tokens = tokens
 
         return standard_node
+
+    @classmethod
+    def count_tokens(cls, text: str, model: str = "gpt-4") -> int:
+        """
+        计算文本的token数量
+
+        Args:
+            text: 要计算的文本
+            model: 使用的模型名称，默认为gpt-4
+
+        Returns:
+            int: token数量
+        """
+        if not text:
+            return 0
+        encoding = tiktoken.encoding_for_model(model)
+        tokens = encoding.encode(text)
+        # 计算标记列表的长度，即标记的数量
+        token_count = len(tokens)
+        # 返回标记的数量
+        return token_count

{bella_openapi-1.0.2.3.dist-info → bella_openapi-1.0.2.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bella-openapi
-Version: 1.0.2.3
+Version: 1.0.2.5
 Summary: client for openapi service.
 Home-page: 
 Author: ['tangxiaolong', 'fanqiangwei', 'zhangxiaojia', 'liumin', 'wangyukun']

{bella_openapi-1.0.2.3.dist-info → bella_openapi-1.0.2.5.dist-info}/RECORD

@@ -1,4 +1,4 @@
-bella_openapi/__init__.py,sha256=NPsPxysa625xRIKgL6je4q1ZSJfnVar4Mb-2RljNwfs,1006
+bella_openapi/__init__.py,sha256=Iq8YjMYBNWJvQvlu9gFy8uxG8Xf06jPaCyZvXtAW2jA,1462
 bella_openapi/auth_billing.py,sha256=Hn0KS8GuG48etnvnd1Faej4IfFXD3tjzalUzDnpZh7Q,3520
 bella_openapi/authorize.py,sha256=cO6J-wx9dmmkDAeqpXT7QlyCr13hO-HSC5SWQSw2gZw,2150
 bella_openapi/config.py,sha256=Dn8vnToDaOesPGboauxCCwNrW5awQLeSkmDjNjXS4bQ,319
@@ -13,13 +13,12 @@ bella_openapi/bella_trace/record_log.py,sha256=qSZXp_VTzIzMVlQNnZKLIbmyGvggFSQL5
 bella_openapi/bella_trace/trace_requests.py,sha256=ADA8J_gbC3TwUo5LWQ3c_yTmCSZRaWzq1FC0iUOnst0,1370
 bella_openapi/console/__init__.py,sha256=uSfr5v6JLRSqTlftjK_ZU1pnbkEyxAPbuQbMyYX_phk,64
 bella_openapi/console/models.py,sha256=Hh1UuYHIxFtF9r5QK-pSJPFrSqbZUHv6spLvPbCeX08,1274
-bella_openapi/domtree/__init__.py,sha256=qxFd6d9WZ4ThWQAFtHAhSseo2h2FB2-5KZWyGOooAbo,107
-bella_openapi/domtree/standard_domtree.py,sha256=lIJP2gReyF2RgAxtyncSB6xl_XAzKCpOb-FTNaf-MIw,19952
-bella_openapi/domtree/utils.py,sha256=-ItZYh9Gj8QyOkZzjCC5xWPYU-FkzJllGC0oUO21Kp4,394
+bella_openapi/entity/__init__.py,sha256=zzsYYg859pzPSgx1Py2kxB2ozQ0tt4OtTatBtpm2bAw,512
+bella_openapi/entity/standard_domtree.py,sha256=EzeYH_UH8MfcVucOcviof8qAIZ5BULVu2DO2EBN7F-c,20510
 bella_openapi/middleware/__init__.py,sha256=XWvZG1xO30ZXIn10YVYthmT1BV-9fonMEP_jVRZbAlQ,157
 bella_openapi/middleware/context_middleware.py,sha256=YawQyKAxMzvlDs_MxcuQKh90pP6VoMKzCBDS94qmlzQ,3870
-bella_openapi-1.0.2.3.dist-info/licenses/LICENSE,sha256=O-0zMbcEi6wXz1DiSdVgzMlQjJcNqNe5KDv08uYzqR0,1055
-bella_openapi-1.0.2.3.dist-info/METADATA,sha256=PbC6CexkWl5ZOR1_0dXPtD6HpydmJ9H_TsOMRH0Kp_E,9377
-bella_openapi-1.0.2.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-bella_openapi-1.0.2.3.dist-info/top_level.txt,sha256=EZuq3F6tKeF-vmZQi6_S2XzmES7SPW7HAbGN1Uv9vN8,14
-bella_openapi-1.0.2.3.dist-info/RECORD,,
+bella_openapi-1.0.2.5.dist-info/licenses/LICENSE,sha256=O-0zMbcEi6wXz1DiSdVgzMlQjJcNqNe5KDv08uYzqR0,1055
+bella_openapi-1.0.2.5.dist-info/METADATA,sha256=ohCbeW5mv4uFMn3NgoscDQxEBKW7zi168nRfIbgltc0,9377
+bella_openapi-1.0.2.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+bella_openapi-1.0.2.5.dist-info/top_level.txt,sha256=EZuq3F6tKeF-vmZQi6_S2XzmES7SPW7HAbGN1Uv9vN8,14
+bella_openapi-1.0.2.5.dist-info/RECORD,,

bella_openapi/domtree/__init__.py

@@ -1,3 +0,0 @@
-from .standard_domtree import StandardDomTree, StandardNode
-
-__all__ = ["StandardDomTree", "StandardNode"]

bella_openapi/domtree/utils.py

@@ -1,12 +0,0 @@
-import tiktoken
-
-# 自研模型均用gpt-4计算（可能有误差，可忽略）
-def count_tokens(text: str, model: str = "gpt-4") -> int:
-    if not text:
-        return 0
-    encoding = tiktoken.encoding_for_model(model)
-    tokens = encoding.encode(text)
-    # 计算标记列表的长度，即标记的数量
-    token_count = len(tokens)
-    # 返回标记的数量
-    return token_count