Skip to content
TheChatArchiveThe Chat Archive
o3-mini· OpenAIHugging Face LMArena Human Preference 140K

I am getting: Object of type datetime is not JSON serializable

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Geminigemini-2.5-pro-preview-05-06’s answer.
  1. User

    I am getting: Object of type datetime is not JSON serializable

  2. Assistant

    The error:

    Object of type datetime is not JSON serializable
    

    occurs because the json module doesn't know how to convert a datetime object into a JSON-compatible format (like a string) by default.

    Here are some ways to fix the error:


    1. Convert the datetime to a String

    Before adding the datetime object to your dictionary, convert it to a string. For example, use the isoformat() method:

    from datetime import datetime
    
    now_timestamp = datetime.utcnow().isoformat() + "+00:00"
    
    item = {
        "url": "https://example.com",
        "index_timestamp": now_timestamp,  # now a string, not a datetime object
        "title": "",
        "text": "",
        "date": "",
        "tags": ""
    }
    

    This way, when you call json.dump(), there is no datetime object to serialize.


    2. Use a Custom JSON Encoder

    If you want to directly pass datetime objects, you can create a custom JSON encoder that knows how to handle datetime objects:

    import json
    from datetime import datetime
    
    class DateTimeEncoder(json.JSONEncoder):
        def default(self, obj):
            if isinstance(obj, datetime):
                return obj.isoformat() + "+00:00"
            return super().default(obj)
    
    # Example usage:
    item = {
        "url": "https://example.com",
        "index_timestamp": datetime.utcnow(),  # still a datetime object
        "title": "",
        "text": "",
        "date": "",
        "tags": ""
    }
    
    with open('articles.json', 'w') as sp:
        json.dump(item, sp, cls=DateTimeEncoder, indent=4)
    

    This encoder checks if the object is a datetime instance and converts it to a string accordingly.


    Update Your Function Example

    Using the Conversion Approach (Simpler):

    import json
    import os
    from datetime import datetime
    import logging
    
    SPIS_ARTYKULOW = 'articles.json'
    logger = logging.getLogger(__name__)
    
    def dodajAdresArtykulu(urlArtykulu):
        # Initialize the list if the file doesn't exist
        if not os.path.exists(SPIS_ARTYKULOW):
            spisArtykulow = []
        else:
            with open(SPIS_ARTYKULOW, 'r') as sp:
                spisArtykulow = json.load(sp)
    
        # Check if the article already exists
        artykulIstnieje = any(artykul["url"] == urlArtykulu for artykul in spisArtykulow)
        
        if not artykulIstnieje:
            logger.debug(f"Dodać nowy adres do bazy ({urlArtykulu})")
            
            nowyArtykul = {
                "url": urlArtykulu,
                "index_timestamp": datetime.utcnow().isoformat() + "+00:00",
                "title": "",
                "text": "",
                "date": "",
                "tags": ""
            }
            
            spisArtykulow.append(nowyArtykul)
            
            with open(SPIS_ARTYKULOW, 'w') as sp:
                json.dump(spisArtykulow, sp, indent=4, ensure_ascii=False)
        else:
            logger.debug(f"Adres już w bazie ({urlArtykulu})")
    

    Using the Custom Encoder:

    import json
    import os
    from datetime import datetime
    import logging
    
    SPIS_ARTYKULOW = 'articles.json'
    logger = logging.getLogger(__name__)
    
    class DateTimeEncoder(json.JSONEncoder):
        def default(self, obj):
            if isinstance(obj, datetime):
                return obj.isoformat() + "+00:00"
            return super().default(obj)
    
    def dodajAdresArtykulu(urlArtykulu):
        if not os.path.exists(SPIS_ARTYKULOW):
            spisArtykulow = []
        else:
            with open(SPIS_ARTYKULOW, 'r') as sp:
                spisArtykulow = json.load(sp)
    
        artykulIstnieje = any(artykul["url"] == urlArtykulu for artykul in spisArtykulow)
        
        if not artykulIstnieje:
            logger.debug(f"Dodać nowy adres do bazy ({urlArtykulu})")
            
            nowyArtykul = {
                "url": urlArtykulu,
                "index_timestamp": datetime.utcnow(),  # still a datetime object
                "title": "",
                "text": "",
                "date": "",
                "tags": ""
            }
            
            spisArtykulow.append(nowyArtykul)
            
            with open(SPIS_ARTYKULOW, 'w') as sp:
                json.dump(spisArtykulow, sp, cls=DateTimeEncoder, indent=4, ensure_ascii=False)
        else:
            logger.debug(f"Adres już w bazie ({urlArtykulu})")
    

    Either method works. The first method is simpler, while the custom encoder provides more flexibility if you have multiple datetime fields or other non-serializable objects.


    Choose the method that best fits your needs, but make sure all datetime objects are converted to strings before writing the data into JSON.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.