bella-openapi 1.0.3.1__py3-none-any.whl → 1.0.4.1__py3-none-any.whl

bella_openapi/entity/standard_domtree.py

@@ -9,7 +9,7 @@ StandardDomTree - 新的DOM树协议定义
 """
 
 from typing import List, Optional, Literal, Any, Union
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, root_validator
 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
@@ -77,12 +77,26 @@ class StandardTableElement(StandardBaseElement):
     name: Optional[str] = None  # 如果类型是Table、Figure为其名字
     description: Optional[str] = None  # 如果类型是Table、Figure为其描述
     rows: List[StandardRow] = Field(default_factory=list)  # 表格行
+    
+    @root_validator(pre=True)
+    def validate_table_type(cls, values):
+        """确保只有type='Table'的数据才能创建StandardTableElement"""
+        if values.get('type') != 'Table':
+            raise ValueError(f"StandardTableElement只能用于type='Table'的数据，当前type='{values.get('type')}'")
+        return values
 
 
 class StandardImageElement(StandardElement):
     name: Optional[str] = None  # 如果类型是Table、Figure为其名字
     description: Optional[str] = None  # 如果类型是Table、Figure为其描述
     image: Optional[StandardImage] = None  # 图片信息
+    
+    @root_validator(pre=True)
+    def validate_image_type(cls, values):
+        """确保只有type='Figure'的数据才能创建StandardImageElement"""
+        if values.get('type') != 'Figure':
+            raise ValueError(f"StandardImageElement只能用于type='Figure'的数据，当前type='{values.get('type')}'")
+        return values
 
 
 class StandardNode(BaseModel):
@@ -90,8 +104,9 @@ class StandardNode(BaseModel):
     summary: Optional[str] = None  # 摘要，文档的简要概述
     tokens: Optional[int] = None  # token预估数量，文档中的token数量估计
     path: Optional[List[int]] = Field(default_factory=list)  # 编号的层级信息，例如：1.2.1
-    element: Optional[Union[StandardElement, StandardImageElement, StandardTableElement]] = None  # 元素信息，当前节点的元素详情
+    element: Optional[Union[StandardTableElement, StandardImageElement, StandardElement]] = None  # 元素信息，当前节点的元素详情
     children: Optional[List["StandardNode"]] = Field(default_factory=list)  # 子节点信息，当前节点的所有子节点
+    
 
 
 class StandardDomTree(BaseModel):
@@ -129,7 +144,7 @@ class StandardDomTree(BaseModel):
                     # 添加表格名称
                     if node.element.name:
                         markdown_res += f"**{node.element.name}**\n\n"
-                    table_md = self._list_to_markdown_table(node.element.rows)
+                    table_md = self._list_to_html_table(node.element.rows)
                     markdown_res += f"{table_md}\n\n"
                     # 添加表格描述
                     if node.element.description:
@@ -168,25 +183,28 @@ class StandardDomTree(BaseModel):
         quoted_lines = ['> ' + line for line in lines]
         return '\n'.join(quoted_lines)
 
-    def _list_to_markdown_table(self, rows: List[StandardRow]) -> str:
-        """将表格行转换为 markdown 表格格式"""
+    def _list_to_html_table(self, rows: List[StandardRow]) -> str:
+        """将表格行转换为HTML表格"""
         if not rows:
             return ""
 
-        markdown_table = ""
-
-        # 添加表头
-        if rows:
-            header_cells = [cell.text or "" for cell in rows[0].cells]
-            markdown_table += "| " + " | ".join(header_cells) + " |\n"
-            markdown_table += "| " + " | ".join(["---"] * len(header_cells)) + " |\n"
-
-        # 添加数据行
+        html_text = "<table>"
         for row in rows:
-            cells = [cell.text or "" for cell in row.cells]
-            markdown_table += "| " + " | ".join(cells) + " |\n"
+            html_text += "<tr>"
+            for cell in row.cells:
+                # 从path中提取rowspan和colspan信息
+                if len(cell.path) >= 4:
+                    start_row, end_row, start_col, end_col = cell.path[:4]
+                    rowspan = end_row - start_row + 1
+                    colspan = end_col - start_col + 1
+                else:
+                    rowspan = colspan = 1
 
-        return markdown_table
+                cell_text = cell.text or ""
+                html_text += f"<td rowspan='{rowspan}' colspan='{colspan}'>{cell_text}</td>"
+            html_text += "</tr>"
+        html_text += "</table>"
+        return html_text
 
     @classmethod
     def from_domtree_dict(cls, domtree: dict, file_info):
@@ -397,19 +415,22 @@ class StandardDomTree(BaseModel):
                     url=element['image_link']
                 )
 
-            # 创建 StandardNode
-            standard_node = StandardNode(
+            # 创建StandardImageElement实例
+            image_element = StandardImageElement(
+                type=element_type,
+                positions=positions,
+                name="",
+                description="",
+                text=text,
+                image=image,
+            )
+            
+            # 使用construct方法跳过validator，保持正确的element类型
+            standard_node = StandardNode.construct(
                 summary="",
                 tokens=0,  # 先设置为 0，后面再计算
                 path=[],  # 初始化为空列表，后续再计算
-                element=StandardImageElement(
-                    type=element_type,
-                    positions=positions,
-                    name="",
-                    description="",
-                    text=text,
-                    image=image,
-                ),
+                element=image_element,
                 children=[]
             )
         elif element_type == "Table":
@@ -444,17 +465,21 @@ class StandardDomTree(BaseModel):
             # 将所有单元格的文本合并，用于计算 token 数量
             text = " ".join(cell_texts)
 
-            standard_node = StandardNode(
+            # 创建StandardTableElement实例
+            table_element = StandardTableElement(
+                type=element_type,
+                positions=positions,
+                name="",
+                description="",
+                rows=rows
+            )
+            
+            # 使用construct方法跳过validator，保持正确的element类型
+            standard_node = StandardNode.construct(
                 summary="",
                 tokens=0,  # 先设置为 0，后面再计算
                 path=[],  # 初始化为空列表，后续再计算
-                element=StandardTableElement(
-                    type=element_type,
-                    positions=positions,
-                    name="",
-                    description="",
-                    rows=rows
-                ),
+                element=table_element,
                 children=[]
             )
         else:
@@ -507,4 +532,9 @@ class StandardDomTree(BaseModel):
         # 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
+        return token_count
+
+
+# 更新forward references
+StandardNode.update_forward_refs()
+Cell.update_forward_refs()

{bella_openapi-1.0.3.1.dist-info → bella_openapi-1.0.4.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bella-openapi
-Version: 1.0.3.1
+Version: 1.0.4.1
 Summary: client for openapi service.
 Home-page: 
 Author: ['tangxiaolong', 'fanqiangwei', 'zhangxiaojia', 'liumin', 'wangyukun']
@@ -12,7 +12,8 @@ License-File: LICENSE
 Requires-Dist: httpx<=0.26.0,>=0.10.0
 Requires-Dist: Werkzeug==3.0.1
 Requires-Dist: tiktoken>=0.5.0
-Requires-Dist: pydantic==2.11.7
+Requires-Dist: pydantic==1.10.14
+Requires-Dist: fastapi==0.110.0
 Dynamic: author
 Dynamic: classifier
 Dynamic: description
@@ -260,3 +261,5 @@ get接口地址：/v1/openapi/log/{requestId}
     * rename 包名
 * 1.0.2
     * 修复bug
+* 1.0.3 
+    * 新增standard domtree

{bella_openapi-1.0.3.1.dist-info → bella_openapi-1.0.4.1.dist-info}/RECORD

@@ -14,11 +14,11 @@ bella_openapi/bella_trace/trace_requests.py,sha256=ADA8J_gbC3TwUo5LWQ3c_yTmCSZRa
 bella_openapi/console/__init__.py,sha256=uSfr5v6JLRSqTlftjK_ZU1pnbkEyxAPbuQbMyYX_phk,64
 bella_openapi/console/models.py,sha256=Hh1UuYHIxFtF9r5QK-pSJPFrSqbZUHv6spLvPbCeX08,1274
 bella_openapi/entity/__init__.py,sha256=zzsYYg859pzPSgx1Py2kxB2ozQ0tt4OtTatBtpm2bAw,512
-bella_openapi/entity/standard_domtree.py,sha256=yJ2FN3xwhT_SPbQSuvMlIvESmjNdQgcu2hzalDq9Gcs,20099
+bella_openapi/entity/standard_domtree.py,sha256=X6C9azEVahiz7OMFBUC3RLDZEKv1pMiQDD1DT8ILPh0,21375
 bella_openapi/middleware/__init__.py,sha256=XWvZG1xO30ZXIn10YVYthmT1BV-9fonMEP_jVRZbAlQ,157
 bella_openapi/middleware/context_middleware.py,sha256=YawQyKAxMzvlDs_MxcuQKh90pP6VoMKzCBDS94qmlzQ,3870
-bella_openapi-1.0.3.1.dist-info/licenses/LICENSE,sha256=O-0zMbcEi6wXz1DiSdVgzMlQjJcNqNe5KDv08uYzqR0,1055
-bella_openapi-1.0.3.1.dist-info/METADATA,sha256=9td_a-VIyt3wOFv03zo5b_reHr8qLleSxGM9wJflCno,9409
-bella_openapi-1.0.3.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-bella_openapi-1.0.3.1.dist-info/top_level.txt,sha256=EZuq3F6tKeF-vmZQi6_S2XzmES7SPW7HAbGN1Uv9vN8,14
-bella_openapi-1.0.3.1.dist-info/RECORD,,
+bella_openapi-1.0.4.1.dist-info/licenses/LICENSE,sha256=O-0zMbcEi6wXz1DiSdVgzMlQjJcNqNe5KDv08uYzqR0,1055
+bella_openapi-1.0.4.1.dist-info/METADATA,sha256=ly-PHCzIuwpyNxhajy5HNbBI4KGHpjnJAwB9wg6VeZQ,9480
+bella_openapi-1.0.4.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+bella_openapi-1.0.4.1.dist-info/top_level.txt,sha256=EZuq3F6tKeF-vmZQi6_S2XzmES7SPW7HAbGN1Uv9vN8,14
+bella_openapi-1.0.4.1.dist-info/RECORD,,