修改简历上传为多路上传

2026-04-03 11:20:17 +08:00
parent e82159278a
commit 7ac1e723a4
6 changed files with 164 additions and 122 deletions
@@ -0,0 +1,61 @@
+"""简历并行提取：将完整简历文本拆分为5个AI任务并行提取"""
+
+import asyncio
+
+from langchain_core.output_parsers import JsonOutputParser
+from langchain_core.prompts import ChatPromptTemplate
+
+from app.ai.models import LLM
+from app.ai.resume_extractor.prompts import (
+    PROFILE_PROMPT, EDUCATION_PROMPT, WORK_PROMPT,
+    PROJECT_PROMPT, COMPETITION_PROMPT,
+)
+from app.core.logger import log
+
+
+def _build_chain(prompt: str):
+    """构建单个提取链：prompt → LLM → JSON解析"""
+    return (
+        ChatPromptTemplate.from_messages([("system", prompt), ("human", "{text}")])
+        | LLM.DOUBAO_PRO_32K.create(temperature=0)
+        | JsonOutputParser()
+    )
+
+
+# 5 条独立的提取链
+_profile_chain = _build_chain(PROFILE_PROMPT)
+_education_chain = _build_chain(EDUCATION_PROMPT)
+_work_chain = _build_chain(WORK_PROMPT)
+_project_chain = _build_chain(PROJECT_PROMPT)
+_competition_chain = _build_chain(COMPETITION_PROMPT)
+
+
+async def extract_all(text: str) -> dict:
+    """asyncio.gather 并行提取简历所有模块，返回合并后的结构化数据"""
+    log.info("开始5路并行AI提取")
+    inp = {"text": text}
+
+    profile, education, work_intern, project, competition = await asyncio.gather(
+        _safe_invoke(_profile_chain, inp, "个人信息"),
+        _safe_invoke(_education_chain, inp, "教育经历"),
+        _safe_invoke(_work_chain, inp, "工作+实习经历"),
+        _safe_invoke(_project_chain, inp, "项目经历"),
+        _safe_invoke(_competition_chain, inp, "竞赛经历"),
+    )
+
+    result = profile if isinstance(profile, dict) else {}
+    result["education"] = education if isinstance(education, list) else []
+    result["work"] = work_intern.get("work", []) if isinstance(work_intern, dict) else []
+    result["internship"] = work_intern.get("internship", []) if isinstance(work_intern, dict) else []
+    result["project"] = project if isinstance(project, list) else []
+    result["competition"] = competition if isinstance(competition, list) else []
+    return result
+
+
+async def _safe_invoke(chain, inp: dict, label: str):
+    """单个链调用，失败返回空"""
+    try:
+        return await chain.ainvoke(inp)
+    except Exception as e:
+        log.warning(f"AI提取[{label}]失败: {e}")
+        return {} if "个人信息" in label else []