R中ask_chatgpt结合dplyr::mutate报错：预期对象却获字符串

问题场景 #

单独调用ask_chatgpt函数查询单个问题时运行正常，但通过dplyr::mutate批量处理数据框中的问题列表时，出现以下错误：

Error in mutate(): ! Problem while computing y = ask_chatgpt(x). Caused by error in gpt_get_completions(): ! list(message = "Invalid type for 'messages[13].content[0]': expected an object, but got a string instead.", type = "invalid_request_error", param = "messages[13].content[0]", code = "invalid_type")

错误原因 #

1. ask_chatgpt默认仅支持单个字符串输入，直接在mutate中传递向量x会导致函数接收整个向量而非逐个处理元素，触发API请求格式错误。
2. cat()函数的作用是将内容打印到控制台，其返回值为NULL，无法作为有效结果存入数据框列中。

解决方法 #

方法1：使用dplyr的rowwise()逐行处理 #

通过rowwise()让mutate按行逐个调用ask_chatgpt，并去掉多余的cat()：

# 创建测试数据框
test <- data.frame(x = c("Is Arkansas in the United States?", "Is Arkansas in India?"))

# 逐行调用ask_chatgpt并存储结果
test <- test %>%
  rowwise() %>%
  mutate(y = ask_chatgpt(x)) %>%
  ungroup() # 处理完成后取消按行分组

方法2：使用purrr包批量映射处理 #

借助purrr::map_chr逐个遍历问题向量，调用ask_chatgpt并返回字符串结果：

library(purrr)

# 创建测试数据框
test <- data.frame(x = c("Is Arkansas in the United States?", "Is Arkansas in India?"))

# 批量处理问题并添加结果列
test$y <- map_chr(test$x, ask_chatgpt)

验证结果 #

处理完成后，查看数据框test即可得到每个问题对应的ChatGPT回答：

print(test)

输出示例：

x                                                                 y
1 Is Arkansas in the United States? Yes, Arkansas is a state in the United States. It is located in the southern region of the country.
2        Is Arkansas in India? No, Arkansas is not in India. Arkansas is a state in the United States, while India is a country located in South Asia. They are two separate and distinct geographical locations.

内容的提问来源于stack exchange，提问作者computer_guy2000