Dhall.CsvToDhall

Convert CSV data to Dhall providing an expected Dhall type necessary to know which type will be interpreted.

The translation process will produce a Dhall expression where its type is a List of records and the type of each field of the records is one of the following:

• Bools
• Naturals
• Integers
• Doubles
• Texts
• Optionals (of valid field types)
• unions (of empty alternatives or valid record field types)

It is exactly the same as dhall-to-csv supported input types.

You can use this code as a library (this module) or as an executable named csv-to-dhall, which is used in the examples below.

For now, csv-to-dhall does not support type inference so you must always specify the Dhall type you expect.

$ cat example.csv
example
1
$ csv-to-dhall 'List { example : Integer }' < example.csv
[{ example = +1 }]

When using the csv-to-dhall executable you can specify that the CSV you want to translate does not have a header with the flag `--no-header`. In this case the resulting record fields will be named _1, _2, ... in the same order they were in the input CSV. You must still provide the expected Dhall type taking this into consideration.

$ cat no-header-example.csv
1,3.14,Hello
-1,2.68,Goodbye
$ csv-to-dhall --no-header 'List { _1 : Integer, _2 : Double, _3 : Text } < no-header-example.csv
[ { _1 = +1, _2 = 3.14, _3 = "Hello" }
, { _1 = -1, _2 = 2.68, _3 = "Goodbye" }
]

Primitive types

Strings true and false can translate to Dhall Bools

$ cat example.csv
exampleBool
true
false
$ csv-to-dhall 'List { exampleBool : Bool }' < example.csv
[ { exampleBool = True }, { exampleBool = False } ]

Numeric strings can translate to Dhall numbers:

$ cat example.csv
exampleNatural,exampleInt,exampleDouble
1,2,3
0,-2,3.14
0,+2,-3.14
$ csv-to-dhall 'List { exampleNatural : Natural, exampleInt : Integer, exampleDouble : Double }' < example.csv
[ { exampleNatural = 1, exampleInt = +2, exampleDouble = 3.0 }
, { exampleNatural = 0, exampleInt = -2, exampleDouble = 3.14 }
, { exampleNatural = 0, exampleInt = +2, exampleDouble = -3.14 }
]

Every CSV Field can translate directly to Dhall Text:

$ cat example.csv
exampleText
Hello
false

","
$ csv-to-dhall 'List { exampleText : Text }' < example.csv
[ { exampleText = "Hello" }
, { exampleText = "false" }
, { exampleText = "" }
, { exampleText = "," }
]

Unions and Optionals

By default, when a union is expected, the first alternative that matches the CSV field is chosen. With the `--unions-strict` flag one can make sure that only one alternative matches. With the `--unions-none` unions are not allowed.

An union alternative matches a CSV field if

• It's an empty alternative and the name is the same as the text in the CSV field.
• It's a non-empty alternative and the CSV field can be converted to the underlying type.

$ cat example.csv
exampleUnion
Hello
1
1.11
$ csv-to-dhall 'List { exampleUnion : <Hello | Nat : Natural | Dob : Double> }' < example.csv
[ { exampleUnion = <Hello | Nat : Natural | Dob : Double>.Hello }
, { exampleUnion = <Hello | Nat : Natural | Dob : Double>.Nat 1 }
, { exampleUnion = <Hello | Nat : Natural | Dob : Double>.Dob 1.11 }
]

Optional values can be either missing or have the expected value. The missing value is represented by the empty string. If a field's expected value is an Optional and the field is not in the CSV, then all the values will be None.

$ cat example.csv
exampleOptional
1

3
$ csv-to-dhall 'List { exampleOptional : Optional Natural, exampleMissing : Optional Natural }' < example.csv
[ { exampleOptional = Some 1, exampleMissing = None Natural }
, { exampleOptional = None Natural, exampleMissing = None Natural }
, { exampleOptional = Some 3, exampleMissing = None Natural }
]

CSV to Dhall

dhallFromCsv :: Conversion -> ExprX -> [NamedRecord] -> Either CompileError ExprX

Convert a list of CSV NameRecord to a Dhall expression given the expected Dhall Type of the output.

parseConversion :: Parser Conversion

Standard parser for options related to the conversion method

defaultConversion :: Conversion

Default conversion options

resolveSchemaExpr

Arguments

:: Text

type code (schema)

-> IO ExprX

Parse schema code and resolve imports

typeCheckSchemaExpr :: (Exception e, MonadCatch m) => (CompileError -> e) -> ExprX -> m ExprX

Check that the Dhall type expression actually has type Type

data Conversion

CSV-to-dhall translation options

Constructors

Conversion

Fields

• strictRecs :: Bool
• unions :: UnionConv

Instances

Show Conversion

Defined in Dhall.CsvToDhall

Methods

showsPrec :: Int -> Conversion -> ShowS

show :: Conversion -> String

showList :: [Conversion] -> ShowS

Exceptions

data CompileError

This is the exception type for errors that can arise when converting from CSV to Dhall.

It contains information on the specific cases that might fail to give a better insight.

Constructors

Unsupported ExprX

NotAList ExprX

NotARecord ExprX

TypeError (TypeError Src Void)

BadSchemaType ExprX ExprX

MissingField Text

UnhandledFields [Text]

Mismatch ExprX Text Text

ContainsUnion ExprX

UndecidableUnion ExprX Text Text [ExprX]

UndecidableMissingUnion ExprX Text [ExprX]

UnicodeError UnicodeException

Instances

Show CompileError

Defined in Dhall.CsvToDhall

Methods

showsPrec :: Int -> CompileError -> ShowS

show :: CompileError -> String

showList :: [CompileError] -> ShowS

Exception CompileError

Defined in Dhall.CsvToDhall

Methods

toException :: CompileError -> SomeException

fromException :: SomeException -> Maybe CompileError

displayException :: CompileError -> String