Language

graphql.language.is_executable_definition_node(node: Node) → TypeGuard[ExecutableDefinitionNode]: Check whether the given node represents an executable definition.

graphql.language.is_selection_node(node: Node) → TypeGuard[SelectionNode]: Check whether the given node represents a selection.

graphql.language.is_value_node(node: Node) → TypeGuard[ValueNode]: Check whether the given node represents a value.

graphql.language.is_const_value_node(node: Node) → TypeGuard[ValueNode]: Check whether the given node represents a constant value.

graphql.language.is_type_node(node: Node) → TypeGuard[TypeNode]: Check whether the given node represents a type.

graphql.language.is_type_system_definition_node(node: Node) → TypeGuard[TypeSystemDefinitionNode]: Check whether the given node represents a type system definition.

graphql.language.is_type_definition_node(node: Node) → TypeGuard[TypeDefinitionNode]: Check whether the given node represents a type definition.

graphql.language.is_type_system_extension_node(node: Node) → TypeGuard[SchemaExtensionNode | TypeExtensionNode]: Check whether the given node represents a type system extension.

graphql.language.is_type_extension_node(node: Node) → TypeGuard[TypeExtensionNode]: Check whether the given node represents a type extension.

Lexer

class graphql.language.Lexer(source: Source)

Bases: object

GraphQL Lexer

A Lexer is a stateful stream generator in that every time it is advanced, it returns the next token in the Source. Assuming the source lexes, the final Token emitted by the lexer will be of kind EOF, after which the lexer will repeatedly return the same EOF token whenever called.

__init__(source: Source) → None: Given a Source object, initialize a Lexer for that source.

advance() → Token: Advance the token stream to the next non-ignored token.

create_token(kind: TokenKind, start: int, end: int, value: str | None = None) → Token: Create a token with line and column location information.

lookahead() → Token: Look ahead and return the next non-ignored token, but do not change state.

print_code_point_at(location: int) → str

Print the code point at the given location.

Prints the code point (or end of file reference) at a given location in a source for use in error messages.

Printable ASCII is printed quoted, while other points are printed in Unicode code point form (ie. U+1234).

read_block_string(start: int) → Token: Read a block string token from the source file.

read_comment(start: int) → Token: Read a comment token from the source file.

read_digits(start: int, first_char: str) → int: Return the new position in the source after reading one or more digits.

read_escaped_character(position: int) → EscapeSequence: Read escaped character sequence

read_escaped_unicode_fixed_width(position: int) → EscapeSequence: Read escaped unicode with fixed width

read_escaped_unicode_variable_width(position: int) → EscapeSequence: Read escaped unicode with variable width

read_name(start: int) → Token: Read an alphanumeric + underscore name from the source.

read_next_token(start: int) → Token

Get the next token from the source starting at the given position.

This skips over whitespace until it finds the next lexable token, then lexes punctuators immediately or calls the appropriate helper function for more complicated tokens.

read_number(start: int, first_char: str) → Token

Reads a number token from the source file.

This can be either a FloatValue or an IntValue, depending on whether a FractionalPart or ExponentPart is encountered.

read_string(start: int) → Token: Read a single-quote string token from the source file.

class graphql.language.TokenKind(value)

Bases: Enum

The different kinds of tokens that the lexer emits

AMP = '&'

AT = '@'

BANG = '!'

BLOCK_STRING = 'BlockString'

BRACE_L = '{'

BRACE_R = '}'

BRACKET_L = '['

BRACKET_R = ']'

COLON = ':'

COMMENT = 'Comment'

DOLLAR = '$'

EOF = '<EOF>'

EQUALS = '='

FLOAT = 'Float'

INT = 'Int'

NAME = 'Name'

PAREN_L = '('

PAREN_R = ')'

PIPE = '|'

QUESTION_MARK = '?'

SOF = '<SOF>'

SPREAD = '...'

STRING = 'String'

class graphql.language.Token(kind: TokenKind, start: int, end: int, line: int, column: int, value: str | None = None)

Bases: object

AST Token

Represents a range of characters represented by a lexical token within a Source.

__init__(kind: TokenKind, start: int, end: int, line: int, column: int, value: str | None = None) → None

column: int

property desc: str: A helper property to describe a token as a string for debugging

end: int

kind: TokenKind

line: int

next: Token | None

prev: Token | None

start: int

value: str | None

Location

graphql.language.get_location(source: Source, position: int) → SourceLocation

Get the line and column for a character position in the source.

Takes a Source and a UTF-8 character offset, and returns the corresponding line and column as a SourceLocation.

class graphql.language.SourceLocation(line: int, column: int)

Bases: NamedTuple

Represents a location in a Source.

__init__()

column: int: Alias for field number 1

count(value, /): Return number of occurrences of value.

property formatted: graphql.language.FormattedSourceLocation: Get formatted source location.

index(value, start=0, stop=9223372036854775807, /)

Return first index of value.

Raises ValueError if the value is not present.

line: int: Alias for field number 0

graphql.language.print_location(location: Location) → str: Render a helpful description of the location in the GraphQL Source document.

class graphql.language.FormattedSourceLocation

Bases: TypedDict

Formatted source location

column: int

line: int

Parser

graphql.language.parse(source: Source | str, no_location: bool = False, max_tokens: int | None = None, allow_legacy_fragment_variables: bool = False, experimental_client_controlled_nullability: bool = False) → DocumentNode

Given a GraphQL source, parse it into a Document.

Throws GraphQLError if a syntax error is encountered.

By default, the parser creates AST nodes that know the location in the source that they correspond to. The no_location option disables that behavior for performance or testing.

Parser CPU and memory usage is linear to the number of tokens in a document, however in extreme cases it becomes quadratic due to memory exhaustion. Parsing happens before validation so even invalid queries can burn lots of CPU time and memory. To prevent this you can set a maximum number of tokens allowed within a document.

Legacy feature (will be removed in v3.3):

If allow_legacy_fragment_variables is set to True, the parser will understand and parse variable definitions contained in a fragment definition. They’ll be represented in the variable_definitions field of the FragmentDefinitionNode.

The syntax is identical to normal, query-defined variables. For example:

fragment A($var: Boolean = false) on T  {
  ...
}

EXPERIMENTAL:

If enabled, the parser will understand and parse Client Controlled Nullability Designators contained in Fields. They’ll be represented in the nullability_assertion field of the FieldNode.

The syntax looks like the following:

 {
   nullableField!
   nonNullableField?
   nonNullableSelectionSet? {
     childField!
  }
}

Note: this feature is experimental and may change or be removed in the future.

graphql.language.parse_type(source: Source | str, no_location: bool = False, max_tokens: int | None = None, allow_legacy_fragment_variables: bool = False) → TypeNode

Parse the AST for a given string containing a GraphQL Type.

Throws GraphQLError if a syntax error is encountered.

This is useful within tools that operate upon GraphQL Types directly and in isolation of complete GraphQL documents.

Consider providing the results to the utility function: value_from_ast().

graphql.language.parse_value(source: Source | str, no_location: bool = False, max_tokens: int | None = None, allow_legacy_fragment_variables: bool = False) → ValueNode

Parse the AST for a given string containing a GraphQL value.

Throws GraphQLError if a syntax error is encountered.

This is useful within tools that operate upon GraphQL Values directly and in isolation of complete GraphQL documents.

Consider providing the results to the utility function: value_from_ast().

Parse the AST for a given string containing a GraphQL constant value.

Similar to parse_value, but raises a arse error if it encounters a variable. The return type will be a constant value.

Printer

graphql.language.print_ast(ast: Node) → str

Convert an AST into a string.

The conversion is done using a set of reasonable formatting rules.

Source

class graphql.language.Source(body: str, name: str = 'GraphQL request', location_offset: SourceLocation = (1, 1))

Bases: object

A representation of source input to GraphQL.

__init__(body: str, name: str = 'GraphQL request', location_offset: SourceLocation = (1, 1)) → None

Initialize source input.

The name and location_offset parameters are optional, but they are useful for clients who store GraphQL documents in source files. For example, if the GraphQL input starts at line 40 in a file named Foo.graphql, it might be useful for name to be "Foo.graphql" and location to be (40, 0).

The line and column attributes in location_offset are 1-indexed.

body

get_location(position: int) → SourceLocation: Get source location.

location_offset

name

graphql.language.print_source_location(source: Source, source_location: SourceLocation) → str: Render a helpful description of the location in the GraphQL Source document.

Visitor

graphql.language.visit(root: Node, visitor: Visitor, visitor_keys: Dict[str, Tuple[str, ...]] | None = None) → Any

Visit each node in an AST.

visit() will walk through an AST using a depth-first traversal, calling the visitor’s enter methods at each node in the traversal, and calling the leave methods after visiting that node and all of its child nodes.

By returning different values from the enter and leave methods, the behavior of the visitor can be altered, including skipping over a sub-tree of the AST (by returning False), editing the AST by returning a value or None to remove the value, or to stop the whole traversal by returning BREAK.

When using visit() to edit an AST, the original AST will not be modified, and a new version of the AST with the changes applied will be returned from the visit function.

To customize the node attributes to be used for traversal, you can provide a dictionary visitor_keys mapping node kinds to node attributes.

class graphql.language.Visitor

Bases: object

Visitor that walks through an AST.

Visitors can define two generic methods “enter” and “leave”. The former will be called when a node is entered in the traversal, the latter is called after visiting the node and its child nodes. These methods have the following signature:

def enter(self, node, key, parent, path, ancestors):
    # The return value has the following meaning:
    # IDLE (None): no action
    # SKIP: skip visiting this node
    # BREAK: stop visiting altogether
    # REMOVE: delete this node
    # any other value: replace this node with the returned value
    return

def leave(self, node, key, parent, path, ancestors):
    # The return value has the following meaning:
    # IDLE (None) or SKIP: no action
    # BREAK: stop visiting altogether
    # REMOVE: delete this node
    # any other value: replace this node with the returned value
    return

The parameters have the following meaning:

Parameters:

node – The current node being visiting.
key – The index or key to this node from the parent node or Array.
parent – the parent immediately above this node, which may be an Array.
path – The key path to get to this node from the root node.
ancestors – All nodes and Arrays visited before reaching parent of this node. These correspond to array indices in path. Note: ancestors includes arrays which contain the parent of visited node.

You can also define node kind specific methods by suffixing them with an underscore followed by the kind of the node to be visited. For instance, to visit field nodes, you would define the methods enter_field() and/or leave_field(), with the same signature as above. If no kind specific method has been defined for a given node, the generic method is called.

BREAK = True

IDLE = None

REMOVE = Ellipsis

SKIP = False

__init__() → None

enter_leave_map: dict[str, EnterLeaveVisitor]

get_enter_leave_for_kind(kind: str) → EnterLeaveVisitor: Given a node kind, return the EnterLeaveVisitor for that kind.

class graphql.language.ParallelVisitor(visitors: Collection[Visitor])

Bases: Visitor

A Visitor which delegates to many visitors to run in parallel.

Each visitor will be visited for each node before moving on.

If a prior visitor edits a node, no following visitors will see that node.

BREAK = True

IDLE = None

REMOVE = Ellipsis

SKIP = False

__init__(visitors: Collection[Visitor]) → None: Create a new visitor from the given list of parallel visitors.

enter_leave_map: dict[str, EnterLeaveVisitor]

get_enter_leave_for_kind(kind: str) → EnterLeaveVisitor: Given a node kind, return the EnterLeaveVisitor for that kind.

The module also exports the following enumeration that can be used as the return type for Visitor methods:

class graphql.language.visitor.VisitorActionEnum(value)

Bases: Enum

Special return values for the visitor methods.

You can also use the values of this enum directly.

BREAK = True

REMOVE = Ellipsis

SKIP = False

The module also exports the values of this enumeration directly. These can be used as return values of Visitor methods to signal particular actions:

graphql.language.BREAK (same as ``True``): This return value signals that no further nodes shall be visited.

graphql.language.SKIP (same as ``False``): This return value signals that the current node shall be skipped.

graphql.language.REMOVE (same as``Ellipsis``): This return value signals that the current node shall be deleted.

graphql.language.IDLE = None: This return value signals that no additional action shall take place.