#!/usr/bin/env python3 """ Test product code extraction with enhanced logic """ import sys import os import re sys.path.append('backend') from app.utils.text_parser import ProductCodeExtractor def test_extraction(): """Test the enhanced product code extraction logic""" extractor = ProductCodeExtractor() # Test cases from user examples test_cases = [ # User's new examples { "name": "SKIMS Boyfriend 女款宽松圆领T恤 AP-TSH-0462代购6.6 MC", "expected": "AP-TSH-0462", "description": "Should extract AP-TSH-0462 and ignore Boyfriend" }, { "name": "Moose Knuckles女款灰色YUKON短款羽绒服M33LB041O 1003代购8.14M", "expected": "M33LB041O 1003", "description": "Should extract M33LB041O 1003 and ignore YUKON" }, # Previous problem cases { "name": "Alexander Wang 女款 黑色蕾丝细节黑色牛仔裤4DC1254613 8.8AT", "expected": "4DC1254613", "description": "Should extract 4DC1254613 and ignore 8.8AT" }, { "name": "Alexander Wang Bodywear白色闪闪短款上衣 8C2238047T代购6.18MC", "expected": "8C2238047T", "description": "Should extract 8C2238047T and ignore 6.18MC" }, { "name": "UMA WANG 女款绿色拼接短袖连衣裙 UP5036 UW400 DUDI 8.8AT", "expected": "UP5036 UW400 DUDI", "description": "Should extract multi-part UMA WANG code" }, { "name": "Tods女款黑色雕花细带布洛克皮鞋XXW60C0DE20EB5B999代购8.15LA", "expected": "XXW60C0DE20EB5B999", "description": "Should extract long alphanumeric code" }, # Multi-part codes with different separators { "name": "Anine Bing 女款Karter Jogger米灰色小logo 卫裤A-03-10598-GRY1", "expected": "A-03-10598-GRY1", "description": "Should extract dash-separated product code A-03-10598-GRY1" }, { "name": "小王国 Acne Studios 中性款 魔术贴小白鞋 Steffey Friend 8.18", "expected": "Steffey Friend", "description": "Should extract space-separated product code Steffey Friend and ignore date 8.18" }, { "name": "小王国 ST JOHN 紫色V领上衣 K91KQ21 8.17 LA", "expected": "K91KQ21", "description": "Should extract K91KQ21 and ignore date+procurement format 8.17 LA" }, { "name": "小王国Rag & Bone条纹短袖T恤 MAI STRIPED BOXY TEE 8.13", "expected": "MAI STRIPED BOXY TEE", "description": "Should extract multi-word product code MAI STRIPED BOXY TEE and ignore date 8.13" }, { "name": "小王国 Moose Knuckles男款蓝色狐狸毛短款羽绒服MK2000MB-402 MC", "expected": "MK2000MB-402", "description": "Should extract MK2000MB-402 and ignore procurement keyword MC" }, { "name": "商品 ABC.DEF.123 产品描述", "expected": "ABC.DEF.123", "description": "Should extract dot-separated product code ABC.DEF.123" }, # Date format exclusions { "name": "商品名称 8.19MC 其他文字", "expected": None, "description": "Should NOT extract date format 8.19MC" }, { "name": "商品名称 11.27MC 其他文字", "expected": None, "description": "Should NOT extract date format 11.27MC" }, { "name": "商品名称 15LA 其他文字", "expected": None, "description": "Should NOT extract short date+procurement format" }, # Single English word product codes (Bogner cases) { "name": "Bogner博格纳 女款 绿色 连帽棉服夹克 MAGAN 代购8.16", "expected": "MAGAN", "description": "Should extract single English word MAGAN" }, { "name": "小王国Bogner博格纳女款黑色拉链夹克ELIN代购8.16", "expected": "ELIN", "description": "Should extract single English word ELIN" }, { "name": "小王国Bogner博格纳 女款短款双面排骨马甲ALIAH代购8.16", "expected": "ALIAH", "description": "Should extract single English word ALIAH" }, { "name": "小王国Bogner博格纳女款浅色羽绒夹克KOSY 代购8.16", "expected": "KOSY", "description": "Should extract single English word KOSY" }, # Additional user test cases { "name": "ArcTeryx 始祖鸟男款极光白SIMA HOODY带帽防晒服X6592代购6.20M", "expected": "X6592", "description": "Should extract alphanumeric code X6592" }, { "name": "小王国 Bogner博格纳 女款蓝色排骨羽绒服FARAH-D 368 8.16", "expected": "FARAH-D 368", "description": "Should extract multi-part code FARAH-D 368" }, { "name": "小王国 Aritzia 蓝色腋下包 117077 8.17 LAMC", "expected": "117077", "description": "Should extract numeric code 117077" }, { "name": "小王国Stuart Weitzman女款黑色牛皮+羊毛短靴CHRLI CZY SJ787 LA", "expected": "CHRLI CZY SJ787", "description": "Should extract multi-word code CHRLI CZY SJ787" }, { "name": "Polo Ralph Lauren 大童 棉质 LOGO马标 牛津衬衫 32367 代购MC", "expected": "32367", "description": "Should extract numeric code 32367" }, { "name": "Moncler 男款 CLUNYE 羽绒服 代购 7.15MC", "expected": "CLUNYE", "description": "Should extract single English word CLUNYE" }, { "name": "小王国Bogner 男款 格纹衬衫 TIMT-2O 762 8.2", "expected": "TIMT-2O 762", "description": "Should extract multi-part code TIMT-2O 762" }, { "name": "小王国 Rag & Bone绿色短款圆领短袖T恤MOSS MAI BOXY TEE8.13", "expected": "MOSS MAI BOXY TEE", "description": "Should extract multi-word code MOSS MAI BOXY TEE" }, { "name": "Alexander McQueen 女款黑色系带马丁靴586402 1000代购7.30", "expected": "586402 1000", "description": "Should extract numeric code with space 586402 1000" }, { "name": "小王国Bogner 博格纳男款 白色带帽羽绒服TINO-D 8.2", "expected": "TINO-D", "description": "Should extract alphanumeric code TINO-D" } ] print("Testing Enhanced Product Code Extraction") print("=" * 50) passed = 0 total = len(test_cases) for i, case in enumerate(test_cases, 1): result = extractor.extract_product_code(case["name"]) expected = case["expected"] print(f"\nTest {i}: {case['description']}") print(f"Input: {case['name']}") print(f"Expected: {expected}") print(f"Got: {result}") # Debug: show what patterns matched clean_name = extractor._remove_brand(case["name"], None) print(f"Clean name: {clean_name}") procurement_keywords = {'AT', 'MC', 'GN', 'NY', 'LA', 'AP', 'SS', 'ZT', 'CN', '现货'} code_patterns = [ r'[A-Za-z0-9]+(?:[-\s/][A-Za-z0-9]+)+', r'[A-Za-z]+\d+[A-Za-z0-9]*', r'\d+[A-Za-z]+[A-Za-z0-9]*', r'\b[A-Z]{2,}[0-9]{2,}\b', r'\b\d{6,}\b', r'[A-Za-z0-9]{6,}(?=\s|[\u4e00-\u9fff]|$)', ] all_matches = [] for j, pattern in enumerate(code_patterns): matches = re.findall(pattern, clean_name) if matches: print(f" Pattern {j+1} matched: {matches}") all_matches.extend(matches) valid_codes = [] for code in all_matches: if extractor._is_valid_product_code(code, procurement_keywords): valid_codes.append(code.upper()) print(f" Valid code: {code}") else: print(f" Invalid code: {code}") if result == expected: print("✅ PASS") passed += 1 else: print("❌ FAIL") print(f"\n{'='*50}") print(f"Results: {passed}/{total} tests passed ({passed/total*100:.1f}%)") return passed == total if __name__ == "__main__": success = test_extraction() sys.exit(0 if success else 1)