如何将yield生成器输出的JSON字符串合并为单个JSON文件？

我通过yield生成器获取到多个嵌套字典形式的JSON字符串，当前输出为独立的单个JSON对象，希望将它们合并为一个JSON文件。可接受格式为包含这些对象的JSON数组，理想格式是按域名聚合浏览器数据的结构化JSON。

当前输出示例 #

{"domain.com": {"Chrome": "19362.344607264396"}}
{"domain.com": {"ChromeMobile": "7177.498437391487"}}
{"another.com": {"MobileSafari": "6237.433155080214"}}
{"another.com": {"Safari": "5895.409403430795"}}

可接受合并格式 #

[
  {"domain.com": {"Chrome": "19362.344607264396"}},
  {"domain.com": {"ChromeMobile": "7177.498437391487"}},
  {"another.com": {"MobileSafari": "6237.433155080214"}},
  {"another.com": {"Safari": "5895.409403430795"}}
]

理想合并格式 #

{
  "browsers": [
    {
      "domain.com": {
        "Chrome": "19362.344607264396",
        "ChromeMobile": "7177.498437391487"
      },
      "another.com": {
        "MobileSafari": "6237.433155080214",
        "Safari": "5895.409403430795"
      }
    }
  ]
}

我的Python代码 #

# Cloudflare zone bandwidth total
def browser_map_page_views(domain_zone):
    cloudflare = prom.custom_query(
        query="topk(5, sum by(family) (increase(browser_map_page_views_count{job='cloudflare', zone='"f'{domain_zone}'"'}[10d])))"
    )
    for domain_z in cloudflare:
        user_agent = domain_z['metric']['family']
        value = domain_z['value'][1]
        yield {domain_zone: {user_agent: {'value': value}}}

# Get list of zones from Prometheus based on Host Tracker data
def domain_zones():
    zones_domain = prom.custom_query(
        query="host_tracker_uptime_percent{job='donodeexporter'}"
    )
    for domain_z in zones_domain:
        yield domain_z['metric']['zone']

# Get a list of domains and substitution each one into a request of Prometheus query.
for domain_list in domain_zones():
    for dict in browser_map_page_views(domain_zone=domain_list):
        dicts = dict
        print(json.dumps(dicts))

解决方案 #

注意事项 #

原代码中browser_map_page_views函数yield的结构是{domain_zone: {user_agent: {'value': value}}}，但示例输出里是直接{user_agent: value}，如果要和示例格式一致，需要修改yield语句为：

yield {domain_zone: {user_agent: value}}

1. 生成可接受的JSON数组格式 #

将所有生成的JSON对象收集到列表中，最后统一序列化为JSON数组：

import json

# 保留原有的browser_map_page_views和domain_zones函数

# 收集所有对象到列表
result_array = []
for domain in domain_zones():
    for item in browser_map_page_views(domain_zone=domain):
        result_array.append(item)

# 打印输出或写入文件
print(json.dumps(result_array, indent=2))
# 写入文件
with open('output_array.json', 'w', encoding='utf-8') as f:
    json.dump(result_array, f, indent=2, ensure_ascii=False)

2. 生成理想的聚合格式 #

按域名聚合浏览器数据，构建结构化JSON：

import json

# 保留原有的browser_map_page_views和domain_zones函数

# 初始化理想格式的结构
aggregated_result = {"browsers": [{}]}
domain_dict = aggregated_result["browsers"][0]

for domain in domain_zones():
    # 为域名初始化空字典（避免KeyError）
    if domain not in domain_dict:
        domain_dict[domain] = {}
    # 遍历当前域名的所有浏览器数据并合并
    for item in browser_map_page_views(domain_zone=domain):
        domain_dict[domain].update(item[domain])

# 打印输出或写入文件
print(json.dumps(aggregated_result, indent=2))
# 写入文件
with open('output_ideal.json', 'w', encoding='utf-8') as f:
    json.dump(aggregated_result, f, indent=2, ensure_ascii=False)

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