excel识别

yuthon_ai_agent/models/ai_import_record.py

@@ -0,0 +1,275 @@
+# -*- coding: utf-8 -*-
+"""
+AI 附件导入记录模型
+记录每次 Excel/图片 附件的识别结果，支持后台查看和重新导入
+"""
+from odoo import models, fields, api
+from odoo import tools
+import logging
+
+_logger = logging.getLogger(__name__)
+
+
+class AiImportRecord(models.Model):
+    """
+    AI 附件导入记录
+    每次用户上传附件（图片/Excel）并触发 AI 识别后，创建一条记录。
+    用于后台审计、重新导入、数据分析。
+    """
+    _name = 'ai.import.record'
+    _description = 'AI 附件导入记录'
+    _order = 'create_date desc'
+    _rec_name = 'name'
+
+    name = fields.Char(string='名称', compute='_compute_name', store=True)
+    create_date = fields.Datetime(string='创建时间', readonly=True)
+    create_uid = fields.Many2one('res.users', string='创建人', readonly=True)
+
+    # 关联信息
+    conversation_id = fields.Many2one('ai.conversation', string='对话', ondelete='set null', index=True)
+    message_id = fields.Many2one('ai.message', string='消息', ondelete='set null', index=True)
+    user_id = fields.Many2one('res.users', string='用户', index=True)
+    wecom_userid = fields.Char(string='企微UserID', index=True)
+
+    # 附件信息
+    document_id = fields.Many2one('documents.document', string='原始附件', ondelete='set null', index=True)
+    filename = fields.Char(string='文件名')
+    file_type = fields.Selection([
+        ('image', '图片'),
+        ('excel', 'Excel'),
+        ('csv', 'CSV'),
+        ('pdf', 'PDF'),
+        ('other', '其他'),
+    ], string='文件类型', default='other', index=True)
+
+    # 识别结果
+    ocr_content = fields.Text(string='OCR/解析内容', help='图片OCR文字或Excel解析后的Markdown文本')
+    ocr_status = fields.Selection([
+        ('pending', '待处理'),
+        ('success', '成功'),
+        ('no_text', '无文字'),
+        ('failed', '失败'),
+    ], string='识别状态', default='pending', index=True)
+    ocr_error = fields.Text(string='识别错误')
+
+    # 导入信息
+    import_status = fields.Selection([
+        ('none', '未导入'),
+        ('pending', '待导入'),
+        ('success', '导入成功'),
+        ('failed', '导入失败'),
+        ('skipped', '已跳过'),
+    ], string='导入状态', default='none', index=True)
+    import_rule_id = fields.Many2one('ai.import.rule', string='使用的导入规则', ondelete='set null')
+    import_result = fields.Text(string='导入结果', help='导入成功/失败的详细信息')
+    imported_record_ids = fields.Text(string='已导入记录ID', help='JSON格式，记录导入的Odoo记录ID列表')
+
+    # AI 回复
+    ai_reply = fields.Text(string='AI回复内容', help='AI对该附件的回复')
+
+    @api.depends('filename', 'file_type', 'create_date')
+    def _compute_name(self):
+        for rec in self:
+            prefix = dict(self._fields['file_type'].selection).get(rec.file_type, '文件')
+            time_str = rec.create_date.strftime('%m-%d %H:%M') if rec.create_date else ''
+            rec.name = u'{prefix} - {rec.filename or "未命名"} - {time_str}'
+
+    def action_reparse(self):
+        """重新解析附件（管理员手动触发）"""
+        self.ensure_one()
+        if not self.document_id:
+            return {'warning': u'原始附件已删除，无法重新解析'}
+
+        if self.file_type in ('excel', 'csv'):
+            from ..utils.excel_parser import parse_excel_to_markdown
+            file_bytes = self.document_id.raw
+            if not file_bytes:
+                self.write({'ocr_status': 'failed', 'ocr_error': u'无法读取文件内容'})
+                return True
+
+            markdown_text, meta = parse_excel_to_markdown(file_bytes, self.filename or '')
+            if meta.get('error'):
+                self.write({'ocr_status': 'failed', 'ocr_error': meta['error']})
+            else:
+                self.write({
+                    'ocr_status': 'success',
+                    'ocr_content': markdown_text,
+                    'ocr_error': '',
+                })
+        elif self.file_type == 'image':
+            # 重新OCR
+            provider = self.env['ai.provider'].search([('active', '=', True)], limit=1)
+            if not provider or not provider.ocr_enabled:
+                self.write({'ocr_status': 'failed', 'ocr_error': u'OCR未启用'})
+                return True
+
+            from ..models.ocr_provider import get_ocr_instance
+            ocr, init_err = get_ocr_instance(provider)
+            if ocr is None:
+                self.write({'ocr_status': 'failed', 'ocr_error': init_err or u'OCR实例创建失败'})
+                return True
+
+            file_bytes = self.document_id.raw
+            mimetype = self.document_id.mimetype or 'image/png'
+            result = ocr.recognize_general(file_bytes, mimetype)
+            if result.get('success'):
+                ocr_text = result.get('text') or ''
+                if ocr_text:
+                    self.write({
+                        'ocr_status': 'success',
+                        'ocr_content': ocr_text,
+                        'ocr_error': '',
+                    })
+                else:
+                    self.write({'ocr_status': 'no_text', 'ocr_error': u'图片中未识别到文字'})
+            else:
+                self.write({'ocr_status': 'failed', 'ocr_error': result.get('error', u'OCR调用失败')})
+
+        return True
+
+    def action_view_document(self):
+        """打开原始附件"""
+        self.ensure_one()
+        if not self.document_id:
+            return {}
+        return {
+            'type': 'ir.actions.act_window',
+            'name': u'原始附件',
+            'res_model': 'documents.document',
+            'res_id': self.document_id.id,
+            'view_mode': 'form',
+            'target': 'new',
+        }
+
+
+class AiImportRule(models.Model):
+    """
+    AI 附件导入规则
+    定义"某种类型的附件应该如何解析和导入"。
+    例如：
+    - Excel 花名册 → 导入 hr.employee
+    - Excel 供应商列表 → 导入 res.partner
+    - 图片发票 → 调用腾讯云发票OCR → 导入 account.move
+    """
+    _name = 'ai.import.rule'
+    _description = u'AI 附件导入规则'
+    _order = 'sequence, id'
+
+    name = fields.Char(string='规则名称', required=True)
+    sequence = fields.Integer(string='优先级', default=10)
+    active = fields.Boolean(string='启用', default=True)
+
+    # 匹配条件
+    file_type = fields.Selection([
+        ('image', u'图片'),
+        ('excel', u'Excel/CSV'),
+        ('any', u'任意'),
+    ], string='文件类型', default='any', required=True)
+    filename_pattern = fields.Char(
+        string='文件名模式',
+        help=u'文件名匹配正则表达式，如 "花名册.*xlsx" 表示花名册开头的xlsx文件'
+    )
+    mime_type_pattern = fields.Char(
+        string='MIME类型模式',
+        help=u'MIME类型匹配，如 "image/*" 或 "application/vnd.openxmlformats*"'
+    )
+
+    # 解析配置
+    parser_type = fields.Selection([
+        ('auto', u'自动（图片OCR，Excel解析为Markdown）'),
+        ('excel_markdown', u'Excel→Markdown'),
+        ('excel_json', u'Excel→JSON（按模板列映射）'),
+        ('image_ocr', u'图片→OCR文字'),
+        ('image_invoice_ocr', u'图片→腾讯云发票OCR'),
+        ('image_idcard_ocr', u'图片→腾讯云身份证OCR'),
+        ('custom', u'自定义解析器（Python代码）'),
+    ], string='解析方式', default='auto', required=True)
+
+    # 导入配置
+    target_model = fields.Char(
+        string='目标模型',
+        help=u'导入数据的Odoo模型，如 hr.employee、res.partner'
+    )
+    field_mapping = fields.Text(
+        string='字段映射',
+        help=u'JSON格式，定义Excel列名→Odoo字段的映射。如 {"姓名": "name", "手机号": "mobile"}'
+    )
+    import_mode = fields.Selection([
+        ('create', u'仅创建'),
+        ('update', u'仅更新（按唯一键匹配）'),
+        ('create_update', u'创建或更新'),
+    ], string='导入模式', default='create')
+
+    # 唯一键（用于 update 模式匹配已有记录）
+    unique_key = fields.Char(
+        string='唯一键字段',
+        help=u'用于匹配已有记录的字段，如 "id_card" 或 "name,mobile"'
+    )
+
+    # AI 提示词（告诉AI如何解读这个附件）
+    ai_prompt = fields.Text(
+        string='AI提示词',
+        help=u'告诉AI这个附件的用途和解读方式。如："这是员工花名册，请统计各部门人数"'
+    )
+
+    # 统计
+    match_count = fields.Integer(string='匹配次数', compute='_compute_match_count')
+
+    def _compute_match_count(self):
+        for rec in self:
+            rec.match_count = self.env['ai.import.record'].search_count([
+                ('import_rule_id', '=', rec.id)
+            ])
+
+    def test_match(self, filename, file_type, mime_type=''):
+        """
+        测试文件是否匹配此规则
+        :return: True / False
+        """
+        self.ensure_one()
+        # 文件类型匹配
+        if self.file_type != 'any' and self.file_type != file_type:
+            return False
+
+        # 文件名模式匹配
+        if self.filename_pattern:
+            import re
+            try:
+                if not re.search(self.filename_pattern, filename):
+                    return False
+            except re.error:
+                _logger.warning('[ImportRule] 文件名正则无效: %s', self.filename_pattern)
+                return False
+
+        # MIME类型匹配
+        if self.mime_type_pattern:
+            import fnmatch
+            if not fnmatch.fnmatch(mime_type, self.mime_type_pattern):
+                return False
+
+        return True
+
+    @api.model
+    def find_matching_rule(self, filename, file_type, mime_type=''):
+        """
+        根据文件名、类型找到最适合的导入规则（按优先级）
+        :return: ai.import.rule 记录或 False
+        """
+        rules = self.search([('active', '=', True)], order='sequence, id')
+        for rule in rules:
+            if rule.test_match(filename, file_type, mime_type):
+                _logger.info('[ImportRule] 匹配规则: %s (file=%s, type=%s)', rule.name, filename, file_type)
+                return rule
+        return False
+
+    def action_view_matched_records(self):
+        """查看使用该规则的历史记录"""
+        self.ensure_one()
+        return {
+            'type': 'ir.actions.act_window',
+            'name': u'%s - 历史记录' % self.name,
+            'res_model': 'ai.import.record',
+            'view_mode': 'tree,form',
+            'domain': [('import_rule_id', '=', self.id)],
+            'context': {'default_import_rule_id': self.id},
+        }

yuthon_ai_agent/utils/__init__.py

@@ -0,0 +1,2 @@
+# -*- coding: utf-8 -*-
+from . import excel_parser

yuthon_ai_agent/utils/excel_parser.py

@@ -0,0 +1,234 @@
+# -*- coding: utf-8 -*-
+"""
+Excel 解析工具：将 Excel 文件转为 AI 可读的 Markdown 文本
+支持 .xlsx / .xls / .csv
+"""
+import io
+import logging
+from typing import Tuple, List, Dict, Any
+
+_logger = logging.getLogger(__name__)
+
+# 默认限制
+MAX_ROWS = 500          # 单 Sheet 最大行数（超出的给统计摘要）
+MAX_TOTAL_ROWS = 2000   # 所有 Sheet 总行数上限
+MAX_FILE_SIZE_MB = 20    # 文件大小上限（MB）
+
+
+def parse_excel_to_markdown(
+    file_bytes: bytes,
+    filename: str,
+    max_rows: int = MAX_ROWS,
+    max_total_rows: int = MAX_TOTAL_ROWS,
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    将 Excel/CSV 解析为 Markdown 文本，供 AI 阅读。
+    
+    :param file_bytes: 文件二进制内容
+    :param filename: 文件名（用于判断类型）
+    :param max_rows: 单 Sheet 最大行数
+    :param max_total_rows: 所有 Sheet 总行数上限
+    :return: (markdown_text, meta_dict)
+            meta_dict: {sheets, total_rows, truncated, file_type, error}
+    """
+    meta = {
+        'sheets': [],
+        'total_rows': 0,
+        'truncated': False,
+        'file_type': '',
+        'error': '',
+    }
+    
+    try:
+        import pandas as pd
+    except ImportError:
+        meta['error'] = 'pandas 未安装，无法解析 Excel 文件。请运行：pip install pandas openpyxl'
+        _logger.error('[ExcelParser] pandas 未安装')
+        return '', meta
+    
+    file_type = _detect_file_type(file_bytes, filename)
+    meta['file_type'] = file_type
+    
+    try:
+        if file_type == 'csv':
+            return _parse_csv(file_bytes, max_rows, max_total_rows, meta)
+        elif file_type in ('xlsx', 'xls'):
+            return _parse_excel(file_bytes, max_rows, max_total_rows, meta)
+        else:
+            meta['error'] = f'不支持的文件类型：{file_type}'
+            return '', meta
+    except Exception as e:
+        _logger.exception('[ExcelParser] 解析失败: %s', e)
+        meta['error'] = f'解析失败：{e}'
+        return '', meta
+
+
+def _detect_file_type(file_bytes: bytes, filename: str) -> str:
+    """根据文件内容和扩展名判断类型"""
+    # 1. 优先看扩展名
+    lower_name = filename.lower()
+    if lower_name.endswith('.csv'):
+        return 'csv'
+    if lower_name.endswith('.xlsx'):
+        return 'xlsx'
+    if lower_name.endswith('.xls'):
+        return 'xls'
+    
+    # 2. 看魔数（文件头）
+    if file_bytes[:2] == b'PK':  # ZIP (xlsx)
+        return 'xlsx'
+    if file_bytes[:2] == b'\xd0\xcf':  # Old Excel (xls)
+        return 'xls'
+    # 尝试当作 CSV
+    try:
+        file_bytes.decode('utf-8-sig')
+        return 'csv'
+    except:
+        pass
+    return 'unknown'
+
+
+def _parse_csv(bytes_data: bytes, max_rows: int, max_total: int, meta: dict) -> Tuple[str, dict]:
+    """解析 CSV 文件"""
+    import pandas as pd
+    
+    # 尝试多种编码
+    encoding = _detect_encoding(bytes_data)
+    df = pd.read_csv(io.BytesIO(bytes_data), encoding=encoding, nrows=max_total + 1)
+    
+    return _df_to_markdown(df, 'CSV', max_rows, max_total, meta)
+
+
+def _parse_excel(bytes_data: bytes, max_rows: int, max_total: int, meta: dict) -> Tuple[str, dict]:
+    """解析 Excel 文件（支持多 Sheet）"""
+    import pandas as pd
+    
+    xl = pd.ExcelFile(io.BytesIO(bytes_data))
+    sheet_names = xl.sheet_names
+    
+    meta['sheets'] = sheet_names
+    
+    parts = []
+    total_rows = 0
+    
+    for sheet_name in sheet_names:
+        df = pd.read_excel(xl, sheet_name=sheet_name, nrows=max_total + 1 - total_rows)
+        
+        # 检查是否超出总行数
+        if total_rows + len(df) > max_total:
+            df = df.head(max_total - total_rows)
+            meta['truncated'] = True
+            _logger.warning('[ExcelParser] 总行数超过 %s，已截断', max_total)
+        
+        sheet_md, sheet_meta = _df_to_markdown(df, sheet_name, max_rows, max_total, meta, is_single_sheet=(len(sheet_names) == 1))
+        parts.append(sheet_md)
+        total_rows += len(df)
+        
+        if meta.get('truncated'):
+            break
+    
+    meta['total_rows'] = total_rows
+    return '\n\n'.join(parts), meta
+
+
+def _df_to_markdown(
+    df,
+    sheet_name: str,
+    max_rows: int,
+    max_total: int,
+    meta: dict,
+    is_single_sheet: bool = False,
+) -> Tuple[str, dict]:
+    """
+    将 DataFrame 转为 Markdown。
+    策略：
+    - 行数 <= max_rows：完整输出
+    - 行数 > max_rows：输出统计摘要 + 前 max_rows 行
+    """
+    row_count = len(df)
+    col_count = len(df.columns)
+    
+    if not is_single_sheet:
+        parts = [f'## Sheet: {sheet_name}']
+    else:
+        parts = []
+    
+    parts.append(f'（{row_count} 行 × {col_count} 列）\n')
+    
+    if row_count == 0:
+        parts.append('（空表）')
+        return '\n'.join(parts), meta
+    
+    # 统计摘要（数值列）
+    numeric_cols = df.select_dtypes(include=['number']).columns.tolist()
+    if numeric_cols:
+        parts.append('【数值列统计】')
+        for col in numeric_cols[:10]:  # 最多10个数值列
+            series = df[col].dropna()
+            if len(series) > 0:
+                parts.append(f'  - {col}：最小={series.min():.2f}，最大={series.max():.2f}，平均={series.mean():.2f}，合计={series.sum():.2f}')
+        parts.append('')
+    
+    # 分类列（唯一值数量）
+    cat_cols = [c for c in df.columns if c not in numeric_cols]
+    if cat_cols:
+        parts.append('【分类列概况】')
+        for col in cat_cols[:5]:  # 最多5个分类列
+            unique_count = df[col].nunique()
+            top_val = df[col].mode().iloc[0] if len(df[col].mode()) > 0 else '（无）'
+            parts.append(f'  - {col}：共 {unique_count} 种值，最常见「{top_val}」')
+        parts.append('')
+    
+    # 数据内容
+    if row_count <= max_rows:
+        parts.append('【完整数据】')
+        parts.append(df.to_markdown(index=False))
+    else:
+        parts.append(f'【前 {max_rows} 行数据】（共 {row_count} 行，已截断）')
+        parts.append(df.head(max_rows).to_markdown(index=False))
+        meta['truncated'] = True
+    
+    return '\n'.join(parts), meta
+
+
+def _detect_encoding(bytes_data: bytes) -> str:
+    """检测 CSV 文件编码"""
+    for enc in ['utf-8-sig', 'utf-8', 'gbk', 'gb2312', 'iso-8859-1']:
+        try:
+            bytes_data.decode(enc)
+            return enc
+        except:
+            continue
+    return 'utf-8'
+
+
+def validate_excel_file(file_bytes: bytes, filename: str) -> Dict[str, Any]:
+    """
+    验证 Excel 文件是否合法、大小是否超限。
+    返回：{valid, error, file_type, size_mb}
+    """
+    size_mb = len(file_bytes) / 1024 / 1024
+    
+    if size_mb > MAX_FILE_SIZE_MB:
+        return {
+            'valid': False,
+            'error': f'文件过大（{size_mb:.1f}MB），上限 {MAX_FILE_SIZE_MB}MB',
+            'file_type': '',
+            'size_mb': round(size_mb, 1),
+        }
+    
+    file_type = _detect_file_type(file_bytes, filename)
+    if file_type == 'unknown':
+        return {
+            'valid': False,
+            'error': f'不支持的文件格式，仅支持 .xlsx / .xls / .csv',
+            'file_type': file_type,
+            'size_mb': round(size_mb, 1),
+        }
+    
+    return {
+        'valid': True,
+        'error': '',
+        'file_type': file_type,
+        'size_mb': round(size_mb, 1),
+    }

yuthon_ai_agent/views/ai_import_record_views.xml

@@ -0,0 +1,207 @@
+<?xml version="1.0" encoding="utf-8"?>
+<odoo>
+    <!-- ========== 附件导入记录 ========== -->
+    <record id="view_ai_import_record_tree" model="ir.ui.view">
+        <field name="name">ai.import.record.tree</field>
+        <field name="model">ai.import.record</field>
+        <field name="arch" type="xml">
+            <tree editable="bottom" decoration-warning="ocr_status == 'failed'" decoration-muted="import_status == 'skipped'">
+                <field name="create_date" string="时间" optional="hide"/>
+                <field name="filename" string="文件名"/>
+                <field name="file_type" string="类型"/>
+                <field name="ocr_status" string="识别状态" widget="badge"
+                       decoration-success="ocr_status == 'success'"
+                       decoration-warning="ocr_status == 'failed'"
+                       decoration-info="ocr_status == 'pending'"
+                       decoration-muted="ocr_status == 'no_text'"/>
+                <field name="import_status" string="导入状态" widget="badge"
+                       decoration-success="import_status == 'success'"
+                       decoration-warning="import_status == 'failed'"
+                       decoration-info="import_status == 'pending'"
+                       decoration-muted="import_status == 'skipped'"/>
+                <field name="create_uid" string="用户" optional="hide"/>
+                <field name="wecom_userid" string="企微ID" optional="hide"/>
+                <field name="import_rule_id" string="导入规则" optional="hide"/>
+            </tree>
+        </field>
+    </record>
+
+    <record id="view_ai_import_record_form" model="ir.ui.view">
+        <field name="name">ai.import.record.form</field>
+        <field name="model">ai.import.record</field>
+        <field name="arch" type="xml">
+            <form>
+                <header>
+                    <button name="action_reparse" string="重新解析" type="object"
+                            class="btn-secondary" confirm="确定重新解析此附件？"/>
+                    <button name="action_view_document" string="查看原始附件" type="object"
+                            class="btn-secondary"/>
+                </header>
+                <sheet>
+                    <div class="oe_title">
+                        <h1><field name="name"/></h1>
+                    </div>
+                    <group>
+                        <group string="基本信息">
+                            <field name="filename"/>
+                            <field name="file_type"/>
+                            <field name="create_date"/>
+                            <field name="create_uid"/>
+                            <field name="wecom_userid"/>
+                        </group>
+                        <group string="识别信息">
+                            <field name="ocr_status" widget="badge"
+                                   decoration-success="ocr_status == 'success'"
+                                   decoration-warning="ocr_status == 'failed'"
+                                   decoration-info="ocr_status == 'pending'"/>
+                            <field name="ocr_error" attrs="{'invisible': [('ocr_error', '=', '')]}"/>
+                            <field name="import_status" widget="badge"/>
+                            <field name="import_rule_id"/>
+                        </group>
+                    </group>
+                    <group string="关联">
+                        <field name="user_id"/>
+                        <field name="conversation_id"/>
+                        <field name="document_id"/>
+                    </group>
+                    <notebook>
+                        <page string="识别结果">
+                            <field name="ocr_content" nolabel="1" readonly="1"
+                                   widget="html" class="oe_read_only"/>
+                        </page>
+                        <page string="AI回复">
+                            <field name="ai_reply" nolabel="1" readonly="1"
+                                   widget="html"/>
+                        </page>
+                        <page string="导入结果">
+                            <field name="import_result" nolabel="1"/>
+                            <field name="imported_record_ids" nolabel="1"/>
+                        </page>
+                    </notebook>
+                </sheet>
+            </form>
+        </field>
+    </record>
+
+    <record id="view_ai_import_record_search" model="ir.ui.view">
+        <field name="name">ai.import.record.search</field>
+        <field name="model">ai.import.record</field>
+        <field name="arch" type="xml">
+            <search>
+                <field name="filename"/>
+                <field name="ocr_content"/>
+                <filter name="status_success" string="识别成功"
+                        domain="[('ocr_status', '=', 'success')]"/>
+                <filter name="status_failed" string="识别失败"
+                        domain="[('ocr_status', 'in', ['failed', 'no_text'])]"/>
+                <separator/>
+                <filter name="file_image" string="图片"
+                        domain="[('file_type', '=', 'image')]"/>
+                <filter name="file_excel" string="Excel"
+                        domain="[('file_type', 'in', ['excel', 'csv'])]"/>
+                <separator/>
+                <filter name="my_imports" string="我的导入"
+                        domain="[('user_id', '=', uid)]"/>
+                <group expand="1" string="分组">
+                    <filter name="group_file_type" string="按文件类型"
+                            context="{'group_by': 'file_type'}"/>
+                    <filter name="group_status" string="按识别状态"
+                            context="{'group_by': 'ocr_status'}"/>
+                    <filter name="group_user" string="按用户"
+                            context="{'group_by': 'user_id'}"/>
+                </group>
+            </search>
+        </field>
+    </record>
+
+    <record id="action_ai_import_record" model="ir.actions.act_window">
+        <field name="name">附件导入记录</field>
+        <field name="res_model">ai.import.record</field>
+        <field name="view_mode">tree,form</field>
+        <field name="search_view_id" ref="view_ai_import_record_search"/>
+    </record>
+
+    <menuitem id="menu_ai_import_record"
+              name="附件导入记录"
+              parent="yuthon_ai_agent.menu_ai_root"
+              action="action_ai_import_record"
+              sequence="50"
+              groups="yuthon_ai_agent.group_ai_agent_admin"/>
+
+
+    <!-- ========== 导入规则 ========== -->
+    <record id="view_ai_import_rule_tree" model="ir.ui.view">
+        <field name="name">ai.import.rule.tree</field>
+        <field name="model">ai.import.rule</field>
+        <field name="arch" type="xml">
+            <tree editable="bottom">
+                <field name="sequence" widget="handle"/>
+                <field name="name"/>
+                <field name="file_type"/>
+                <field name="parser_type"/>
+                <field name="target_model"/>
+                <field name="match_count" string="匹配次数"/>
+                <field name="active"/>
+            </tree>
+        </field>
+    </record>
+
+    <record id="view_ai_import_rule_form" model="ir.ui.view">
+        <field name="name">ai.import.rule.form</field>
+        <field name="model">ai.import.rule</field>
+        <field name="arch" type="xml">
+            <form>
+                <header>
+                    <button name="action_view_matched_records" string="查看匹配记录"
+                            type="object" class="btn-secondary"/>
+                </header>
+                <sheet>
+                    <div class="oe_title">
+                        <h1><field name="name"/></h1>
+                    </div>
+                    <group>
+                        <group string="基本配置">
+                            <field name="sequence"/>
+                            <field name="active"/>
+                            <field name="file_type"/>
+                        </group>
+                        <group string="匹配条件">
+                            <field name="filename_pattern" placeholder="如：花名册.*xlsx"/>
+                            <field name="mime_type_pattern" placeholder="如：application/vnd.openxmlformats*"/>
+                        </group>
+                    </group>
+                    <group string="解析与导入配置">
+                        <field name="parser_type"/>
+                        <field name="target_model" placeholder="如：hr.employee"/>
+                        <field name="import_mode"/>
+                        <field name="unique_key" placeholder="如：id_card"
+                               attrs="{'invisible': [('import_mode', '=', 'create')]}"/>
+                    </group>
+                    <group string="字段映射（JSON格式）">
+                        <field name="field_mapping" nolabel="1" widget="ace"
+                               options="{'mode': 'json'}"
+                               placeholder='{"Excel列名": "odoo字段名"}  范例：{"姓名": "name", "手机号": "mobile_phone"}'/>
+                    </group>
+                    <group string="AI提示词">
+                        <field name="ai_prompt" nolabel="1"
+                               placeholder="告诉AI这个附件的用途和解读方式"/>
+                    </group>
+                </sheet>
+            </form>
+        </field>
+    </record>
+
+    <record id="action_ai_import_rule" model="ir.actions.act_window">
+        <field name="name">导入规则配置</field>
+        <field name="res_model">ai.import.rule</field>
+        <field name="view_mode">tree,form</field>
+    </record>
+
+    <menuitem id="menu_ai_import_rule"
+              name="导入规则"
+              parent="yuthon_ai_agent.menu_ai_root"
+              action="action_ai_import_rule"
+              sequence="55"
+              groups="yuthon_ai_agent.group_ai_agent_admin"/>
+
+</odoo>