import time from config import simplemodel from menchuangfallback import menchuangfallback example1=''' 1、包水管 做法详见图纸设计及相关图集规范 2、5#镀锌角钢基架 3、水管外包吸音材料 4、水泥纤维板 ''' answer_1=''' {'面层': '水泥纤维板'} ''' example2=''' 1、一层大厅消火栓处，做法详见图纸设计及相关图集规范 2、石膏板、阻燃基层板、40*20*3镀锌方通龙骨、金属转轴等 ''' answer_2=''' {'面层': '石膏板'} ''' example3=''' 1、不锈钢饰面 做法详见图纸设计及相关图集规范 2、1.5mm厚不锈钢面层 3、阻燃板基层，木龙骨找平阻燃处理 ''' answer_3=''' {'面层': '1.5mm厚不锈钢面层'} ''' def extra( data, #data aiclient, qwclient, sfclient, menchuang_collection, model, qita_collection, ): if '高强螺栓' in data['mc']: return '高强螺栓' if data['bianma'].startswith("0108"): sentence=["特征描述：" + data['mc'] + "\n" + data['tz']] embeddings = model.encode(sentence) result = menchuang_collection.query(query_embeddings=embeddings, n_results=10) print(result['documents'][0]) l = len([x for x in result['distances'][0] if x < 0.5]) if l < 2: l = 2 completion = aiclient.chat.completions.create( model="glm-4.5-flash", messages=[ {"role": "system", "content": "You are a helpful assistant.请将最终答案以JSON格式输出"}, {"role": "user", "content": "特征描述往往比较具体，工作内容是对特征描述的主要关键的总结提炼。以下是一些特征描述以及对应的提炼的工作内容的例子。" + '\n\n'.join(result['documents'][0][:l]) + "给定一段特征描述，内容为" + data['mc'] +data['tz'] + "。请参照示例，给出提炼的工作内容. 注意，不需要输出特征描述，仅输出工作内容"}, ], extra_body={"thinking": {"type": "disabled"}}, ) json_string = completion.choices[0].message.content print(json_string) answers = json_string.split("\n") answers = [x for x in answers if ':' in x ] answer2 = answers[0].split(":")[1].replace(" ", "") return answer2 if data['bianma'].startswith("0115"): sentence=["特征描述：" + data['mc'] + "\n" + data['tz']] embeddings = model.encode(sentence) result = qita_collection.query(query_embeddings=embeddings, n_results=10) print(result['documents'][0]) l = len([x for x in result['distances'][0] if x < 0.5]) if l < 2: l = 2 completion = aiclient.chat.completions.create( model="glm-4.5-flash", messages=[ {"role": "system", "content": "You are a helpful assistant.请将最终答案以JSON格式输出"}, {"role": "user", "content": "特征描述往往比较具体，工作内容是对特征描述的主要关键的总结提炼。以下是一些特征描述以及对应的提炼的工作内容的例子。" + '\n\n'.join(result['documents'][0][:l]) + "给定一段特征描述，内容为" + data['mc'] +data['tz'] + "。请参照示例，给出提炼的工作内容(提炼的工作内容中不得出现类似详见图纸、图集的表述). 注意，不需要输出特征描述，仅输出工作内容"}, ], extra_body={"thinking": {"type": "disabled"}}, ) json_string = completion.choices[0].message.content print(json_string) answers = json_string.split("\n") answers = [x for x in answers if ':' in x ] answer2 = answers[0].split(":")[1].replace(" ", "") return answer2 if data['bianma'].startswith("011207") or data['bianma'].startswith('011208'): completion = aiclient.chat.completions.create( model="glm-4.5-air", messages=[ {"role": "system", "content": "You are a helpful assistant.请将最终答案以JSON格式输出"}, {"role": "user", "content": "墙柱面装饰板工程往往由多道工序组成，包括底层龙骨，附着于龙骨上的基层板、吸音棉、阻燃板等，以及作为饰面的面层板。现在要求你从给定的工作内容描述中抽取出面层的描述。举个例子，给定工作内容："+example1+"\n你应该返回："+answer_1+"\n再举个例子，给定工作内容："+example2+"\n你应该返回:"+answer_2+"\n再举个例子，给定工作内容："+example3+"\n你应该返回："+answer_3+"\n现在给定工作内容："+data['mc']+" "+data['tz']+"\n请给出你的答案"}, ], extra_body={"thinking": {"type": "disabled"}}, ) json_string = completion.choices[0].message.content print(json_string) answers = json_string.split("\n") answers = [x for x in answers if ':' in x ] answer2 = answers[0].split(":")[1].replace(" ", "") return answer2 completion = aiclient.chat.completions.create( model="glm-4.5-air", messages=[ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": " 背景知识：已知预应力高强混凝土管桩（PHC）代号定义为PHC-AAA(BB)CC-DDD-E1,E2,E3,E4,其中AAA代表管桩外径，BB代表管桩壁厚，CC表示型号，DDD表示混凝土强度等级，E1/E2/E3/E4表示分段桩长。例如，PHC-500(125)-AB-C80-9,7 表示外径500mm,壁厚125mm,型号AB，混凝土强度C80, 分段桩长分别为9米和7米，总桩长16米,施工时需要将两个分段接桩"}, {"role": "user", "content": "问题描述： 给定一段工作内容描述，内容为" + data['mc'] +data['tz'] + "。请判断内容中是否包含桩的代号，如果没有，请输出“无”,如果有，请输出代号"}, ], extra_body={"thinking": {"type": "disabled"}}, ) json_string = completion.choices[0].message.content completion = aiclient.chat.completions.create( model="glm-4.5-flash", messages=[ {"role": "system", "content": "You are a helpful assistant.请将最终答案以JSON格式输出"}, {"role": "user", "content": " 给你一段文字如下， " + json_string + ",其中给出了一个代号作为答案，请将该最终答案输出"}, ], extra_body={"thinking": {"type": "disabled"}}, ) json_string = completion.choices[0].message.content answers = json_string.split("\n") answers = [x for x in answers if ':' in x ] answers = [x for x in answers if not 'true' in x] answers = [x for x in answers if not '是' in x] print(answers) if len(answers) == 0: return "无" answer2 = answers[0].split(":")[1].replace(" ", "") return answer2 def need_extra( data, #data aiclient, qwclient, sfclient, result, name_label): if data['bianma'].startswith('011207') or data['bianma'].startswith('011208'): result = [x[0].replace('换', '') for x in result] result = [name_label[x] for x in result] if len([x for x in result if '铝板' in x or '铝单板' in x]) > 0: return False left = [x.replace('在木龙骨', '') for x in result ] left = [x.replace('在龙骨', '') for x in left ] left = [x for x in left if '龙骨' not in x] left = [x for x in left if '细木工板基层' not in x] left = [x for x in left if '吸音' not in x] if len(left) == 0: return True return False if data['bianma'].startswith("0108") and len(result) == 0: return True if data['bianma'].startswith("0115") and len(result) == 0: return True if '高强螺栓' in data['mc']: return True time.sleep(1) completion = qwclient.chat.completions.create( model="Qwen/Qwen3-32B", #model="THUDM/GLM-4-9B-0414", messages=[ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "问题描述： 给定一段工作内容描述，内容为" + data['mc'] +data['tz'] + "。请判断内容是否属于打桩、压桩。请回答是或者否"}, ], #extra_body={"thinking": {"type": "disabled"}}, extra_body={"enable_thinking": False}, ) json_string = completion.choices[0].message.content print(json_string) if "是" in json_string: return True else: return False