我有一个看起来像这样的JSON数据源:
{ "fields": [
{ "type": "datetime",
"name": "Observation Valid",
"description": "Observation Valid Time"},
{ "type": "datetime",
"name": "Observation Valid UTC",
"description": "Observation Valid Time UTC"},
{ "type": "number",
"name": "Air Temperature[F]",
"description": "Air Temperature at 2m AGL"},
{ "type": "number",
"name": "Wind Speed[kt]",
"description": "Wind Speed"},
{ "type": "number",
"name": "Wind Gust[kt]",
"description": "Wind Gust"},
{ "type": "number", "name":
"Wind Direction[deg]",
"description": "Wind Direction"}
],
"rows": [
["2018-04-22T00:10:00", "2018-04-22T05:10:00Z", 50.0, 9.0, null, 50.0],
["2018-04-22T00:15:00", "2018-04-22T05:15:00Z", 50.0, 9.0, null, 60.0],
["2018-04-22T00:20:00", "2018-04-22T05:20:00Z", 50.0, 8.0, null, 60.0],
["2018-04-22T00:30:00", "2018-04-22T05:30:00Z", 50.0, 9.0, null, 60.0]
]
}
( https://mesonet.agron.iastate.edu/json/obhistory.py?station=TVK&network=AWOS&date=2018-04-22 )
并尝试了几个数据描述,最后是:
data Entry = -- Data entries
Entry { time :: Text -- Observation Valid Time
, timeUTC :: Text -- Observation Valid Time UTC
, airTemp :: Float -- Air Temperature[F] at 2m AGL
, wind :: Float -- Wind Speed [kt]
, gust :: Float -- Wind Gust [kt]
, direction :: Int -- Wind Direction[deg]
} deriving (Show,Generic)
data Field = -- Schema Definition
Field { ftype :: String --
, name :: String --
, description :: String --
} deriving (Show,Generic)
data Record =
Record { fields :: [Field] --
, rows :: [Entry] -- data
} deriving (Show,Generic)
-- Instances to convert our type to/from JSON.
instance FromJSON Entry
instance FromJSON Field
instance FromJSON Record
-- Get JSON data and decode it
dat <- (eitherDecode <$> getJSON) :: IO (Either String Record)
给出此错误: $ .fields [0]中的错误:键" ftype"不存在
(第一个(错误来自字段定义(我不使用(。在JSON中,条目是混合类型的数组,但是在Haskell中,它只是一个数据结构,而不是数组 - 不确定如何调和这些。
毫无疑问,初学者的错误 - 但我还没有找到任何似乎具有这种结构的例子。我需要为此写一个自定义解析器吗?
三件事阻止了此操作按预期工作:
- JSON数据包含一个名为"类型"的字段。
Field
记录类型的自定义FromJson
实例可以处理此操作。 -
Entry
类型中的数据未命名,因此它可以更好地表示为没有字段名称的data
记录或tuple
。 - 代表风阵的
Float
有时为null
,因此应该是Maybe Float
下面的代码包含所有这些修改并解析您的示例JSON数据:
{-# LANGUAGE DeriveGeneric #-}
{-# LANGUAGE OverloadedStrings #-}
import Data.ByteString.Lazy as BSL
import Data.Text (Text)
import Data.Aeson
import GHC.Generics
-- Either this tuple definition of Entry or the data definition without
-- names (commented out) will work.
type Entry = -- Data entries
( Text -- Observation Valid Time
, Text -- Observation Valid Time UTC
, Float -- Air Temperature[F] at 2m AGL
, Float -- Wind Speed [kt]
, Maybe Float -- Wind Gust [kt]
, Int -- Wind Direction[deg]
)
-- data Entry = -- Data entries
-- Entry Text -- Observation Valid Time
-- Text -- Observation Valid Time UTC
-- Float -- Air Temperature[F] at 2m AGL
-- Float -- Wind Speed [kt]
-- (Maybe Float) -- Wind Gust [kt]
-- Int -- Wind Direction[deg]
-- deriving (Show,Generic)
-- instance FromJSON Entry
data Field = -- Schema Definition
Field { ftype :: String --
, name :: String --
, description :: String --
} deriving (Show,Generic)
instance FromJSON Field where
parseJSON = withObject "Field" $ v -> Field
<$> v .: "type"
<*> v .: "name"
<*> v .: "description"
data Record =
Record { fields :: [Field] --
, rows :: [Entry] -- data
} deriving (Show,Generic)
instance FromJSON Record
getJSON :: IO ByteString
getJSON = BSL.readFile "json.txt"
main :: IO()
main = do
-- Get JSON data and decode it
dat <- (eitherDecode <$> getJSON) :: IO (Either String Record)
case dat of
Right parsed -> print parsed
Left err -> print err
字段具有ftype
字段,因此Aeson试图在JSON中找到FTYPE,但无法(因为它包含FTYPE(。我了解您不能在Haskell中命名type
字段,因此您需要找到一种方法,使Aeson使用其他名称。您需要相应地使用模板Haskell并设置FieldLabelModifier。另外,手动编写坚持可能会更简单。