使用jq对json去重的一个小细节

jq是一个操作json的非常好的工具，这里记录一下使用jq去重踩到的一个坑

例一

• json文件内容如下

admin@pc-1:~$ cat raw_0.json 
{
    "cid": 100,
    "info": {
        "desc": "this is 100",
        "color": "green"
    }
}
{
    "cid": 200,
    "info": {
        "desc": "this is 200",
        "color": "red"
    }
}
{
    "cid": 100,
    "info": {
        "desc": "this is 100",
        "color": "green"
    }
}
admin@pc-1:~$ 

• 注意:
  • 开头结尾没有用"[“和”]"包裹起来
  • item之间没有用逗号隔开
• 使用jq根据"cid"去重正常

admin@pc-1:~$  jq -s 'unique_by(.cid)' raw_0.json 
[
 {
   "cid": 100,
   "info": {
     "desc": "this is 100",
     "color": "green"
   }
 },
 {
   "cid": 200,
   "info": {
     "desc": "this is 200",
     "color": "red"
   }
 }
]
admin@pc-1:~$ 

例二

• json文件内容如下,这个应该是json的标准格式，有"[]"，有逗号分隔

admin@pc-1:~$ cat raw_1.json 
[
    {
        "cid": 100,
        "info": {
            "desc": "this is 100",
            "color": "green"
        }
    },
    {
        "cid": 200,
        "info": {
            "desc": "this is 200",
            "color": "red"
        }
    },
    {
        "cid": 100,
        "info": {
            "desc": "this is 100",
            "color": "green"
        }
    }
]
admin@pc-1:~$ 

• 还用原来的命令会报错

admin@pc-1:~$  jq -s 'unique_by(.cid)' raw_1.json 
jq: error (at raw_1.json:23): Cannot index array with string "cid"
admin@pc-1:~$ 

• 需要去掉"-s",就可以正常去重了

admin@pc-1:~$ jq 'unique_by(.cid)' raw_1.json 
[
  {
    "cid": 100,
    "info": {
      "desc": "this is 100",
      "color": "green"
    }
  },
  {
    "cid": 200,
    "info": {
      "desc": "this is 200",
      "color": "red"
    }
  }
]
admin@pc-1:~$ 

• 看一下命令行帮助

admin@pc-1:~$ jq -h | grep "\-s "
  -s               read (slurp) all inputs into an array; apply filter to it;
admin@pc-1:~$ 

• 有趣的是可以从raw_1.json转成raw_0.json的格式

admin@pc-1:~$ cat raw_1.json | jq '.[]'
{
  "cid": 100,
  "info": {
    "desc": "this is 100",
    "color": "green"
  }
}
{
  "cid": 200,
  "info": {
    "desc": "this is 200",
    "color": "red"
  }
}
{
  "cid": 100,
  "info": {
    "desc": "this is 100",
    "color": "green"
  }
}
admin@pc-1:~$