在Databricks PySpark中使用Azure认知文本分析时遭遇「不支持混合字符串与字典/对象文档输入」类型错误的解决求助

Let's break down what's causing this error and how to fix it quickly.

Your Scenario Recap #

You're trying to extract key phrases from a text file stored in DBFS using Azure Text Analytics, but hitting a type error: Mixing string and dictionary/object document input unsupported. Here's your original code flow:

1. Reading the file with binary mode:

with open("/dbfs/mnt/lake/RAW/export/dummy.txt", "rb") as fd:
    documents = fd.read()

2. Calling the Text Analytics API:

response = text_analytics_client.extract_key_phrases(documents, language="en")

Root Cause #

The extract_key_phrases method expects an iterable of strings (like a list of documents), not a single raw bytes object or a single string. When you pass the raw documents variable (which is bytes from rb mode) or even a single string, the SDK misinterprets the input format, leading to the type mismatch error.

Fixed Solution #

Here's the corrected code with explanations:

1. Read the file as text (not binary)
   Use text mode (r) instead of binary (rb) to get a proper string, and strip any extra whitespace to avoid unintended issues:

   with open("/dbfs/mnt/lake/RAW/export/dummy.txt", "r", encoding="utf-8") as fd:
       document_text = fd.read().strip()
   

2. Wrap the single document in a list
   The API is built to handle multiple documents at once, so even for one file, you need to pass it as a list to match the expected input format:

   from azure.core.credentials import AzureKeyCredential
   from azure.ai.textanalytics import TextAnalyticsClient
   
   credential = AzureKeyCredential("xxxxxxxxxxxxxxxxxxxx")
   endpoint= "https://xxxxxxxx.cognitiveservices.azure.com/"
   text_analytics_client = TextAnalyticsClient(endpoint, credential)
   
   # Pass the text as a list (even for a single document)
   response = text_analytics_client.extract_key_phrases([document_text], language="en")
   result = [doc for doc in response if not doc.is_error]
   for doc in result:
       print(doc.key_phrases)
   

Expected Output #

Running this corrected code will give you the desired result:

['King County', 'United States', 'Redmond', 'city', 'Washington', 'Seattle']

内容的提问来源于stack exchange，提问作者Patterson