123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223 |
- # -*- coding: utf-8 -*-
- from textwrap import dedent
- import pytest
- from parso._compatibility import u
- from parso import parse
- from parso.python import tree
- from parso.utils import split_lines
- def test_basic_parsing(each_version):
- def compare(string):
- """Generates the AST object and then regenerates the code."""
- assert parse(string, version=each_version).get_code() == string
- compare('\na #pass\n')
- compare('wblabla* 1\t\n')
- compare('def x(a, b:3): pass\n')
- compare('assert foo\n')
- def test_subscope_names(each_version):
- def get_sub(source):
- return parse(source, version=each_version).children[0]
- name = get_sub('class Foo: pass').name
- assert name.start_pos == (1, len('class '))
- assert name.end_pos == (1, len('class Foo'))
- assert name.value == 'Foo'
- name = get_sub('def foo(): pass').name
- assert name.start_pos == (1, len('def '))
- assert name.end_pos == (1, len('def foo'))
- assert name.value == 'foo'
- def test_import_names(each_version):
- def get_import(source):
- return next(parse(source, version=each_version).iter_imports())
- imp = get_import('import math\n')
- names = imp.get_defined_names()
- assert len(names) == 1
- assert names[0].value == 'math'
- assert names[0].start_pos == (1, len('import '))
- assert names[0].end_pos == (1, len('import math'))
- assert imp.start_pos == (1, 0)
- assert imp.end_pos == (1, len('import math'))
- def test_end_pos(each_version):
- s = dedent('''
- x = ['a', 'b', 'c']
- def func():
- y = None
- ''')
- parser = parse(s, version=each_version)
- scope = next(parser.iter_funcdefs())
- assert scope.start_pos == (3, 0)
- assert scope.end_pos == (5, 0)
- def test_carriage_return_statements(each_version):
- source = dedent('''
- foo = 'ns1!'
- # this is a namespace package
- ''')
- source = source.replace('\n', '\r\n')
- stmt = parse(source, version=each_version).children[0]
- assert '#' not in stmt.get_code()
- def test_incomplete_list_comprehension(each_version):
- """ Shouldn't raise an error, same bug as #418. """
- # With the old parser this actually returned a statement. With the new
- # parser only valid statements generate one.
- children = parse('(1 for def', version=each_version).children
- assert [c.type for c in children] == \
- ['error_node', 'error_node', 'endmarker']
- def test_newline_positions(each_version):
- endmarker = parse('a\n', version=each_version).children[-1]
- assert endmarker.end_pos == (2, 0)
- new_line = endmarker.get_previous_leaf()
- assert new_line.start_pos == (1, 1)
- assert new_line.end_pos == (2, 0)
- def test_end_pos_error_correction(each_version):
- """
- Source code without ending newline are given one, because the Python
- grammar needs it. However, they are removed again. We still want the right
- end_pos, even if something breaks in the parser (error correction).
- """
- s = 'def x():\n .'
- m = parse(s, version=each_version)
- func = m.children[0]
- assert func.type == 'funcdef'
- assert func.end_pos == (2, 2)
- assert m.end_pos == (2, 2)
- def test_param_splitting(each_version):
- """
- Jedi splits parameters into params, this is not what the grammar does,
- but Jedi does this to simplify argument parsing.
- """
- def check(src, result):
- # Python 2 tuple params should be ignored for now.
- m = parse(src, version=each_version)
- if each_version.startswith('2'):
- # We don't want b and c to be a part of the param enumeration. Just
- # ignore them, because it's not what we want to support in the
- # future.
- func = next(m.iter_funcdefs())
- assert [param.name.value for param in func.get_params()] == result
- else:
- assert not list(m.iter_funcdefs())
- check('def x(a, (b, c)):\n pass', ['a'])
- check('def x((b, c)):\n pass', [])
- def test_unicode_string():
- s = tree.String(None, u('bö'), (0, 0))
- assert repr(s) # Should not raise an Error!
- def test_backslash_dos_style(each_version):
- assert parse('\\\r\n', version=each_version)
- def test_started_lambda_stmt(each_version):
- m = parse(u'lambda a, b: a i', version=each_version)
- assert m.children[0].type == 'error_node'
- def test_python2_octal(each_version):
- module = parse('0660', version=each_version)
- first = module.children[0]
- if each_version.startswith('2'):
- assert first.type == 'number'
- else:
- assert first.type == 'error_node'
- @pytest.mark.parametrize('code', ['foo "', 'foo """\n', 'foo """\nbar'])
- def test_open_string_literal(each_version, code):
- """
- Testing mostly if removing the last newline works.
- """
- lines = split_lines(code, keepends=True)
- end_pos = (len(lines), len(lines[-1]))
- module = parse(code, version=each_version)
- assert module.get_code() == code
- assert module.end_pos == end_pos == module.children[1].end_pos
- def test_too_many_params():
- with pytest.raises(TypeError):
- parse('asdf', hello=3)
- def test_dedent_at_end(each_version):
- code = dedent('''
- for foobar in [1]:
- foobar''')
- module = parse(code, version=each_version)
- assert module.get_code() == code
- suite = module.children[0].children[-1]
- foobar = suite.children[-1]
- assert foobar.type == 'name'
- def test_no_error_nodes(each_version):
- def check(node):
- assert node.type not in ('error_leaf', 'error_node')
- try:
- children = node.children
- except AttributeError:
- pass
- else:
- for child in children:
- check(child)
- check(parse("if foo:\n bar", version=each_version))
- def test_named_expression(works_ge_py38):
- works_ge_py38.parse("(a := 1, a + 1)")
- def test_extended_rhs_annassign(works_ge_py38):
- works_ge_py38.parse("x: y = z,")
- works_ge_py38.parse("x: Tuple[int, ...] = z, *q, w")
- @pytest.mark.parametrize(
- 'param_code', [
- 'a=1, /',
- 'a, /',
- 'a=1, /, b=3',
- 'a, /, b',
- 'a, /, b',
- 'a, /, *, b',
- 'a, /, **kwargs',
- ]
- )
- def test_positional_only_arguments(works_ge_py38, param_code):
- works_ge_py38.parse("def x(%s): pass" % param_code)
- @pytest.mark.parametrize(
- 'expression', [
- 'a + a',
- 'lambda x: x',
- 'a := lambda x: x'
- ]
- )
- def test_decorator_expression(works_ge_py39, expression):
- works_ge_py39.parse("@%s\ndef x(): pass" % expression)
|