panda
bd5bfbac2d
Root cause: LLM receiving full 34k-char JRXML would regenerate from scratch
instead of modifying coordinates in-place, shrinking output to ~3k chars.
Solution (programmatic node control, not prompt engineering):
- New agent/jrxml_windower.py: decompose JRXML into header (never sent to
LLM) + individual bands. Split bands >4000 chars at element boundaries.
Reassemble with element count validation (>10% change = rollback).
- Rewrite refine_layout: per-band windowed LLM processing (~2-4k chars
each). LLM cannot "reimagine" the entire report.
- Rewrite map_fields: 100% programmatic regex $F{field_N} -> real name
replacement. Zero LLM calls, zero content loss.
- _sanitize_field_name: non-ASCII chars escaped to _uXXXX_ format for
valid JRXML identifiers.
- Tests: 48 new unit tests (windower 28 + map_fields 20). All passing.
Full suite 385 tests, zero regressions.
65 lines
2.2 KiB
Python
65 lines
2.2 KiB
Python
"""LangGraph JRXML 生成代理工作流的状态定义。"""
|
|
|
|
from typing import TypedDict, List
|
|
|
|
|
|
class AgentState(TypedDict, total=False):
|
|
# 核心工作流字段
|
|
conversation_history: List[dict]
|
|
current_jrxml: str
|
|
user_input: str
|
|
status: str
|
|
error_msg: str
|
|
natural_explanation: str
|
|
retry_count: int
|
|
user_modification_request: str
|
|
final_jrxml: str
|
|
stage: str
|
|
retrieved_context: str
|
|
|
|
# 需求1：智能上下文压缩
|
|
full_conversation_history: List[dict]
|
|
compressed_history: str
|
|
current_token_count: int
|
|
|
|
# 需求2：多会话持久化
|
|
session_id: str
|
|
session_name: str
|
|
created_at: str
|
|
updated_at: str
|
|
|
|
# 需求3：意图识别
|
|
intent: str
|
|
history_states: List[dict]
|
|
|
|
# 需求4：JRXML 版本历史（用于下载历史版本）
|
|
jrxml_versions: List[dict]
|
|
|
|
# 需求5：错误自增长（记录修正前的状态，供 validate 节点判断是否入知识库）
|
|
last_error_case: dict
|
|
|
|
# 需求6：失败上下文传递 — 重试耗尽后暂存失败信息，下次用户输入时自动注入
|
|
pending_failure_context: dict
|
|
|
|
# 需求7：OCR 单据字段精确提取结果
|
|
ocr_extraction_result: dict
|
|
uploaded_file_path: str
|
|
|
|
# 需求8：图片批注检测（圈选/箭头标记）
|
|
annotation_result: dict
|
|
|
|
# 需求9：分层精确生成
|
|
layout_schema: dict # extract_layout_schema() 输出，列+区域结构
|
|
ocr_elements: list # OCR 原始行数据（用于阶段二坐标采样）
|
|
|
|
# 需求10：多租户知识库
|
|
kb_id: str # 当前会话绑定的知识库 ID
|
|
kb_fields: list # KB 提取的字段定义 [{name, description, type, required}]
|
|
kb_field_mapping: dict # OCR 字段 → KB 字段映射 {"工单号": "billNo", ...}
|
|
uploaded_template_jrxml: str # 对话中上传的 JRXML 模板原文
|
|
uploaded_template_params: list # 解析出的参数 [{name, type}]
|
|
kb_template_jrxml: str # 从 KB 检索到的模板 JRXML
|
|
kb_template_name: str # 检索到的模板名称
|
|
datasource_mode: str # "parameter" 或 "jdbc"
|
|
db_config: dict # JDBC 连接配置
|