fix: 修复简历解析换行粘连与专业误判(主修课程干扰)

backend/app/services/resume_parser_core.py

@@ -52,12 +52,13 @@ def read_text(path: Path) -> str:
 
 def normalize(text: str) -> str:
     text = text.replace("\r", "\n")
-    text = re.sub(r"(?<=[一-龥])\s+(?=[一-龥])", "", text)
+    # 只压缩汉字之间的横向空白(空格/制表符/全角空格)，保留换行，避免把相邻行粘连
+    text = re.sub(r"(?<=[一-龥])[ \t\u3000]+(?=[一-龥])", "", text)
     return re.sub(r"\n{3,}", "\n\n", text).strip()
 
 
 def compact_chinese(text: str) -> str:
-    return re.sub(r"(?<=[一-龥])\s+(?=[一-龥])", "", text or "")
+    return re.sub(r"(?<=[一-龥])[ \t\u3000]+(?=[一-龥])", "", text or "")
 
 
 def first_match(patterns, text):
@@ -324,6 +325,27 @@ def infer_school(text: str) -> str:
     return next((school for school in candidates if school.endswith("大学")), candidates[0] if candidates else "")
 
 
+def _major_from_adjacent_line(line: str) -> str:
+    """从学校紧邻的一行里提取专业名；不是专业行时返回空串。"""
+    line = line.strip()
+    if not line or re.match(r"主修|主要课程|课程|专业课|选修", line):
+        return ""
+    # 去掉行首时间范围，如 2017.09—2021.07 / 2017.09-2021.07 / 2017.09 / 2017年
+    line = re.sub(
+        r"^(?:\d{4}\s*[年./\-]\s*\d{1,2}\s*(?:—|-|~|至|到)\s*\d{4}\s*[年./\-]?\s*\d{0,2}\s*[年]?"
+        r"|\d{4}\s*[年./\-]\s*\d{1,2}\s*[年]?"
+        r"|\d{4}\s*年)[\s\u3000]*",
+        "",
+        line,
+    )
+    # 含句读/顿号的是句子或课程列表，不当作专业名
+    if re.search(r"[。，、；：]", line):
+        return ""
+    line = re.sub(r"(?:（|\(|\s)?(?:本科|硕士|博士|大专|专业)(?:）|\))?$", "", line).strip()
+    match = re.match(r"^([一-龥A-Za-z0-9（）()·\-]{2,30})$", line)
+    return match.group(1) if match else ""
+
+
 def infer_major(text: str) -> str:
     text = compact_chinese(text)
     explicit = first_match([
@@ -337,16 +359,32 @@ def infer_major(text: str) -> str:
             return cleaned
     school = infer_school(text)
     if school and school in text:
-        segment = text.split(school, 1)[1][:180]
-        dictionary_major = next((major for major in sorted(COMMON_MAJORS, key=len, reverse=True) if major in segment), "")
+        after = text.split(school, 1)[1]
+        segment = after[:180]
+        # 优先从学校紧邻行取专业（同行剩余/后一行，跳过日期行），
+        # 避免误扫进"主修课程"把课程名当成专业
+        for line in [part.strip() for part in after.splitlines() if part.strip()][:3]:
+            adjacent = _major_from_adjacent_line(line)
+            if adjacent:
+                return adjacent
+        # 排除"主修课程/课程"之后的内容再按词典匹配
+        pre_course = re.split(r"(?:主修课程|主要课程|主修|课程|专业课|选修)", segment, maxsplit=1)[0]
+        dictionary_major = next(
+            (major for major in sorted(COMMON_MAJORS, key=len, reverse=True) if major in pre_course), ""
+        )
         if dictionary_major:
             return dictionary_major
         match = re.search(r"([^\s，,；;。]{2,30}?)(?:专业)?\s*(?:博士|硕士|本科|大专)", segment.strip())
         if match:
             return re.sub(r"专业$", "", match.group(1)).strip()
-    education_block_match = re.search(r"(?:教育背景|教育经历|学习经历)[\s\S]{0,500}?(?:工作经历|项目经历|实习经历|技能|证书|自我评价|$)", text)
+    education_block_match = re.search(
+        r"(?:教育背景|教育经历|学习经历)[\s\S]{0,500}?(?:工作经历|项目经历|实习经历|技能|证书|自我评价|$)", text
+    )
     education_block = education_block_match.group(0) if education_block_match else ""
-    dictionary_major = next((major for major in sorted(COMMON_MAJORS, key=len, reverse=True) if major in education_block), "")
+    pre_course_block = re.split(r"(?:主修课程|主要课程|主修|课程|专业课|选修)", education_block, maxsplit=1)[0]
+    dictionary_major = next(
+        (major for major in sorted(COMMON_MAJORS, key=len, reverse=True) if major in pre_course_block), ""
+    )
     if dictionary_major:
         return dictionary_major
     return ""

backend/tests/test_resume_parser.py

-from backend.app.services.resume_parser_core import main, parse_resume
+from backend.app.services.resume_parser_core import infer_major, main, normalize, parse_resume
 
 
 def test_parse_text_resume(tmp_path):
@@ -8,6 +8,8 @@ def test_parse_text_resume(tmp_path):
     assert parsed["name"] == "张三"
     assert parsed["phone"] == "13812345678"
     assert parsed["email"] == "zhangsan@example.com"
+    assert parsed["school"] == "清华大学"
+    assert parsed["major"] == "计算机科学与技术"
     assert "React" in parsed["skills"]
 
 
@@ -17,3 +19,60 @@ def test_parser_help(capsys):
     except SystemExit as exc:
         assert exc.code == 0
     assert "解析 PDF/DOCX/TXT 简历" in capsys.readouterr().out
+
+
+def test_normalize_preserves_line_breaks_between_chinese_lines():
+    text = "基本信息\n姓 名： 薛庆霞\n民 族 ： 汉\n电 话： 15110312431"
+    normalized = normalize(text)
+    assert "基本信息\n姓名：" in normalized
+    assert "薛庆霞\n民族" in normalized
+
+
+def test_parse_name_not_merged_with_next_line(tmp_path):
+    resume = tmp_path / "薛庆霞.txt"
+    resume.write_text(
+        "个人简历\n细心从每一个小细节开始。\nPersonal resume\n基本信息\n"
+        "姓 名： 薛庆霞\n民 族 ： 汉\n电 话： 15110312431\n"
+        "邮 箱： 1176053788@qq.com\n住 址： 山西省太原市万柏林区\n出生年月： 1993.04.30\n"
+        "身 高： 170cm\n政治面貌 ：群众\n毕业院校： 山西警察学院\n籍 贯： 山西太原\n教育背景",
+        encoding="utf-8",
+    )
+    parsed = parse_resume(resume)
+    assert parsed["name"] == "薛庆霞"
+    assert parsed["phone"] == "15110312431"
+    assert parsed["email"] == "1176053788@qq.com"
+
+
+def test_major_prefers_adjacent_line_over_course_names(tmp_path):
+    resume = tmp_path / "张晓燕.txt"
+    resume.write_text(
+        "张晓燕\n性别：女年龄：28\n电话：18734915261 邮箱：1822742034@qq.com\n销售支持专员\n教育经历\n"
+        "2017.09—2021.07 甘肃农业大学\n农林经济管理\n"
+        "主修课程：市场营销、管理学原理、人力资源管理、企业经营战略、财务管理、会计学原理、区域经济学、发展经济\n"
+        "学等。专业基础扎实，系统掌握市场营销、企业管理、经济分析相关知识，具备销售业务逻辑、客户管理及商务协作\n"
+        "理论功底。",
+        encoding="utf-8",
+    )
+    parsed = parse_resume(resume)
+    assert parsed["school"] == "甘肃农业大学"
+    assert parsed["major"] == "农林经济管理"
+
+
+def test_major_same_line_after_school():
+    text = "教育经历\n2017.09—2021.07 甘肃农业大学 农林经济管理 本科\n主修课程：市场营销、人力资源管理"
+    assert infer_major(text) == "农林经济管理"
+
+
+def test_major_on_line_after_date_range():
+    text = "教育经历\n甘肃农业大学\n2017.09—2021.07\n农林经济管理\n主修课程：人力资源管理"
+    assert infer_major(text) == "农林经济管理"
+
+
+def test_major_explicit_field_wins():
+    text = "毕业院校：甘肃农业大学\n专业：人力资源管理\n主修课程：市场营销"
+    assert infer_major(text) == "人力资源管理"
+
+
+def test_major_course_names_not_picked_when_no_major():
+    text = "教育经历\n2017.09—2021.07 甘肃农业大学\n主修课程：市场营销、人力资源管理"
+    assert infer_major(text) == ""

shared/resume-dictionaries.json

@@ -149,7 +149,8 @@
     "工商管理",
     "市场营销",
     "财务管理",
-    "会计学"
+    "会计学",
+    "农林经济管理"
   ],
   "skills": [
     "电力交易",

vue-app/src/utils/inference.js

@@ -9,7 +9,7 @@ export function hasProfileValue(value) {
 }
 
 export function compactChineseText(text = '') {
-  return String(text || '').replace(/(?<=[\u4e00-\u9fa5])\s+(?=[\u4e00-\u9fa5])/g, '')
+  return String(text || '').replace(/(?<=[\u4e00-\u9fa5])[ \t\u3000]+(?=[\u4e00-\u9fa5])/g, '')
 }
 
 export function inferPhone(text = '') {
@@ -165,6 +165,21 @@ export function inferSchool(text = '') {
   return candidates.find((school) => school.endsWith('大学')) || candidates[0] || ''
 }
 
+function majorFromAdjacentLine(line = '') {
+  let value = line.trim()
+  if (!value || /^(主修|主要课程|课程|专业课|选修)/.test(value)) return ''
+  // 去掉行首时间范围，如 2017.09—2021.07 / 2017.09-2021.07 / 2017.09 / 2017年
+  value = value.replace(
+    /^(?:\d{4}\s*[年./\-]\s*\d{1,2}\s*(?:—|-|~|至|到)\s*\d{4}\s*[年./\-]?\s*\d{0,2}\s*[年]?|\d{4}\s*[年./\-]\s*\d{1,2}\s*[年]?|\d{4}\s*年)[\s\u3000]*/,
+    ''
+  )
+  // 含句读/顿号的是句子或课程列表，不当作专业名
+  if (/[。，、；：]/.test(value)) return ''
+  value = value.replace(/(?:（|\(|\s)?(?:本科|硕士|博士|大专|专业)(?:）|\))?$/, '').trim()
+  const match = value.match(/^([\u4e00-\u9fa5A-Za-z0-9（）()·\-]{2,30})$/)
+  return match ? match[1] : ''
+}
+
 export function inferMajor(text = '') {
   const source = compactChineseText(text)
   const explicit = source.match(/(?:专业|所学专业)[:：\s]+([\u4e00-\u9fa5A-Za-z0-9（）()·\-]{2,30})/)
@@ -179,9 +194,23 @@ export function inferMajor(text = '') {
       /(?:教育背景|教育经历|学习经历)[\s\S]{0,500}?(?:工作经历|项目经历|实习经历|技能|证书|自我评价|$)/
     )?.[0] || ''
   const educationSegment = schoolIndex >= 0 ? source.slice(schoolIndex, schoolIndex + 180) : educationBlock || source
-  const dictionaryMajor =
-    [...commonMajors].sort((a, b) => b.length - a.length).find((major) => educationSegment.includes(major)) ||
-    [...commonMajors].sort((a, b) => b.length - a.length).find((major) => educationBlock.includes(major))
+  // 优先从学校紧邻行取专业（同行剩余/后一行，跳过日期行），避免误扫进"主修课程"把课程名当成专业
+  if (schoolIndex >= 0) {
+    const after = source.slice(schoolIndex + school.length)
+    const adjacent = after
+      .split('\n')
+      .map((line) => line.trim())
+      .filter(Boolean)
+      .slice(0, 3)
+      .map(majorFromAdjacentLine)
+      .find(Boolean)
+    if (adjacent) return adjacent
+  }
+  // 排除"主修课程/课程"之后的内容再按词典匹配
+  const preCourse = educationSegment.split(/(?:主修课程|主要课程|主修|课程|专业课|选修)/)[0]
+  const dictionaryMajor = [...commonMajors]
+    .sort((a, b) => b.length - a.length)
+    .find((major) => preCourse.includes(major))
   if (dictionaryMajor) return dictionaryMajor
   const educationLine = source.match(
     /(?:大学|学院)\s+([\u4e00-\u9fa5A-Za-z0-9（）()·\-]{2,30}?)(?:专业)?\s+(?:博士|硕士|本科|大专)/
@@ -196,7 +225,11 @@ export function inferMajor(text = '') {
     const major = segment.trim().match(/([^\s，,；;。]{2,30}?)(?:专业)?\s*(?:博士|硕士|本科|大专)/)
     if (major) return major[1].replace(/专业$/g, '').trim()
   }
-  return ''
+  const preCourseBlock = educationBlock.split(/(?:主修课程|主要课程|主修|课程|专业课|选修)/)[0]
+  const fallbackMajor = [...commonMajors]
+    .sort((a, b) => b.length - a.length)
+    .find((major) => preCourseBlock.includes(major))
+  return fallbackMajor || ''
 }
 
 export function schoolTags(school = '') {