如何在Python中读取单文件内的多个独立JSON数据？

Got it, let's tackle this problem step by step. You're dealing with that annoying scenario where each individual JSON object is valid, but the whole file isn't a proper JSON array—super common when working with log files or bulk-exported data! Here are two reliable approaches to get the exact result you want:

Method 1: Base Python + Pandas (Handles Nested JSON & Unformatted Files) #

This method is the most robust, especially if your JSON objects are squished together without line breaks or have nested JSON structures. We'll use Python's built-in json module to safely parse each complete JSON object one by one:

import json
import pandas as pd

# Replace with your file path
file_path = '~/Desktop/data.json'

# Read the entire file content
with open(file_path, 'r') as f:
    content = f.read()

# Use JSONDecoder to parse each valid JSON object sequentially
decoder = json.JSONDecoder()
current_pos = 0
json_objects = []

while current_pos < len(content):
    try:
        # Parse the next valid JSON object and get the end position
        obj, current_pos = decoder.raw_decode(content, current_pos)
        json_objects.append(obj)
        
        # Skip any whitespace between objects (spaces, newlines, tabs)
        while current_pos < len(content) and content[current_pos].isspace():
            current_pos += 1
    except json.JSONDecodeError:
        # If we hit invalid content, skip one character and keep going (adjust this if needed)
        current_pos += 1

# Convert the list of JSON objects (Python dicts) to a DataFrame
df = pd.DataFrame({"jsons": json_objects})

Notes: #

• This correctly handles nested JSON because raw_decode identifies the complete structure of each JSON object, so it won't get tripped up by nested {} pairs.
• If you want the cells to hold JSON strings instead of Python dictionaries, replace json_objects.append(obj) with json_objects.append(json.dumps(obj)).

Method 2: Quick Pandas Adjustment (If lines=True Already Works) #

If your file does have one JSON object per line (even if they're not comma-separated), you can use your existing pd.read_json(..., lines=True) approach and then merge the split columns back into a single JSON object per row:

import pandas as pd

# First read the file with lines=True (splits into columns)
df_split = pd.read_json('~/Desktop/data.json', lines=True)

# Convert each row back to a full JSON object/dictionary
df = pd.DataFrame({"jsons": df_split.apply(lambda row: row.to_dict(), axis=1)})

Notes: #

• If you want JSON strings instead of dictionaries, use row.to_json() instead of row.to_dict().
• This is faster and simpler, but only works if each JSON object is on its own line (which lines=True relies on).

Either method will give you a DataFrame where each row's jsons column holds the full, original JSON object—no more split columns messing up your structure!

内容的提问来源于stack exchange，提问作者Outcast