我希望能够通过指定一个JSON查找来从巨大的（100 GB以上）JSON文件中提取特定的JSON对象。（虽然我的最终目标是有一个用于查找的路径类型字符串，但我非常乐意将此字符串处理为jq过滤器-用于流媒体的jq过滤器是我遇到的问题）。理想情况下，我希望只使用jq。
我的文件格式如下：
文件1：

{
   "data": {
      "<userId1>": {
         "things": {
            "<thingId1>": {
               "subfield1": "blah1",
               "subfield3": "foobar"
            },
            "<thingId2>": {
               "subfield2": "blah2"
            }
         }
      },
      "<userId2>": {
         "things": {
            "<thingId1>": {
               "subfield4": "blah3"
            },
            "<thingId2>": {
               "subfield3": "blah4"
            }
         }
      }
   },
   "users": {
      "<userId1>": {
         "name": "user1",
         "email": "[email protected]"
      },
      "<userId2>": {
         "name": "user2"
      },
      "<userId3>": {
         "email": "[email protected]"
      }
   }
}

字符串
文件2（一个更简单的对象键列表）：

{
   "<key1>": {
      "subfield1": "blah1",
      "subfield3": "foobar"
   },
   "<key2>": {
      "subfield2": "blah2"
   }
}

型
文件3（一个深度键控的对象列表）：

{
   "<key1>": {
      "<subKey1A>": {
         "<deeperKey1A>": {
            "subfield1": "blah1",
            "subfield3": "foobar"
         },
         "<deeperKey2A>": {
            "subfield1": "blah1",
            "subfield4": "foobar4"
         }
      },
      "<subKey2A>": {
         "<deeperKey2A>": {
            "subfield4": "foobar4"
         }
      }
   },
   "<key2>": {
      "<subKey1B>": {
         "<deeperKey2B>": {
            "subfield1": "blah1",
            "subfield4": "foobar4"
         }
      },
      "<subKey2B>": {
         "<deeperKey1B>": {
            "subfield1": "blah1",
            "subfield3": "foobar"
         },
         "<deeperKey3B>": {
            "subfield4": "foobar6"
         }
      }
   }
}

型
在上述两种情况下，任何<blah>键都是任意ID。
我想输出以下文件-搜索是我想实现的搜索类型查找-认为$<>路径是“这匹配任何东西”（也许使用*更清楚，但我喜欢它有助于识别什么是阅读它时的jq组件.）.我还包括了特定的jq过滤器为每一个情况下，我已经得到的工作。
文件1 -用于data/$userId/things/$thingId搜索：

{ "path": ["data", "<userId1>", "things", "<thingId1>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["data", "<userId1>", "things", "<thingId2>"], "value": { "subfield2": "blah2" } }
{ "path": ["data", "<userId2>", "things", "<thingId1>"], "value": { "subfield4": "blah3" } }
{ "path": ["data", "<userId2>", "things", "<thingId1>"], "value": { "subfield3": "blah4" }

型
jq过滤器：

foreach (

  ( inputs | select(has(1) and ((.[0] | length) >= 4) and .[0][0] == "data" and .[0][2] == "things")
  | (first | [rindex("things")] | max | values) as $p
  | [.[0][$p+2:], .[0][:$p+2], .[1]]), [[]]

) as [$vpath, $path, $value] ([];

  if [first.path, $path] | unique[1]
  then [{$path}, first] else .[:1] end
  | first.value |= setpath($vpath; $value);

  .[1] | values
)

型
文件1 -用于users/$userId搜索：

{ "path": ["users", "<userId1>"], "value": { "name": "user1", "email": "[email protected]" } }
{ "path": ["users", "<userId2>"], "value": { "name": "user2" } }
{ "path": ["users", "<userId3>"], "value": { "email": "[email protected]" } }

型
jq过滤器：

foreach (

  ( inputs | select(has(1) and ((.[0] | length) >= 1) and .[0][0] == "users")
  | (first | [rindex("users")] | max | values) as $p
  | [.[0][$p+2:], .[0][:$p+2], .[1]]), [[]]

) as [$vpath, $path, $value] ([];

  if [first.path, $path] | unique[1]
  then [{$path}, first] else .[:1] end
  | first.value |= setpath($vpath; $value);

  .[1] | values
)

型
文件2 -对于$key搜索：

{ "path": ["<key1>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["<key2>"], "value": { "subfield2": "blah2" } }

型
我偶然发现了一个部分解决方案，同时胡乱调整上述过滤器：

foreach (

  ( inputs | select(has(1) and ((.[0] | length) >= 1))
  | (first | [-1] | max | values) as $p
  | [.[0][$p+2:], .[0][:$p+2], .[1]]), [[]]

) as [$vpath, $path, $value] ([];

  if [first.path, $path] | unique[1]
  then [{$path}, first] else .[:1] end
  | first.value |= setpath($vpath; $value);

  .[1] | values
)

型
这几乎可以工作，但输出重复的条目（与子对象中的键数相同）：

{ "path": ["<key1>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["<key1>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["<key2>"], "value": { "subfield2": "blah2" } }

型
我真的不明白为什么用[-1]替换[rindex("some-key")]会产生这样的输出！
文件3 -对于$key/$subKey/$deeperKey搜索：

{ "path": ["<key1>", "<subKey1A>", "<deeperKey1A>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["<key1>", "<subKey1A>", "<deeperKey2A>"], "value": { "subfield1": "blah1", "subfield4": "foobar4" } }
{ "path": ["<key1>", "<subKey2A>", "<deeperKey2A>"], "value": { "subfield4": "foobar4" } }
{ "path": ["<key2>", "<subKey1B>", "<deeperKey2B>"], "value": { "subfield1": "blah1", "subfield4": "foobar4" } }
{ "path": ["<key2>", "<subKey2B>", "<deeperKey1B>"], "value": { "subfield1": "blah1", "subfield3": "foobar" } }
{ "path": ["<key2>", "<subKey2B>", "<deeperKey1C>"], "value": { "subfield4": "foobar6" } }

型
我已经设法让它再次与上面的过滤器一起工作，但我所能打印出来的只是第一个键（<key1>），value是嵌套在其父路径（{ "<subKey1A>": { "<deeperKey1A>": { "subfield1": "blah1", "subfield3": "foobar" } } }）下的对象。

注意事项

这可以被归类为my original question的副本，虽然这是一个很好的答案，但我仍然需要更多的灵活性，我不想改变这个问题（我已经有过一次），从优秀的答案中删除意义。