change migrate-config to use yaml parse tree instead

tests/commands/migrate_config_test.py

@@ -134,6 +134,27 @@ def test_migrate_config_sha_to_rev(tmpdir):
     )
 
 
+def test_migrate_config_sha_to_rev_json(tmp_path):
+    contents = """\
+{"repos": [{
+    "repo": "https://github.com/pre-commit/pre-commit-hooks",
+    "sha": "v1.2.0",
+    "hooks": []
+}]}
+"""
+    expected = """\
+{"repos": [{
+    "repo": "https://github.com/pre-commit/pre-commit-hooks",
+    "rev": "v1.2.0",
+    "hooks": []
+}]}
+"""
+    cfg = tmp_path.joinpath('cfg.yaml')
+    cfg.write_text(contents)
+    assert not migrate_config(str(cfg))
+    assert cfg.read_text() == expected
+
+
 def test_migrate_config_language_python_venv(tmp_path):
     src = '''\
 repos:
@@ -167,6 +188,31 @@ repos:
     assert cfg.read_text() == expected
 
 
+def test_migrate_config_quoted_python_venv(tmp_path):
+    src = '''\
+repos:
+-   repo: local
+    hooks:
+    -   id: example
+        name: example
+        entry: example
+        language: "python_venv"
+'''
+    expected = '''\
+repos:
+-   repo: local
+    hooks:
+    -   id: example
+        name: example
+        entry: example
+        language: "python"
+'''
+    cfg = tmp_path.joinpath('cfg.yaml')
+    cfg.write_text(src)
+    assert migrate_config(str(cfg)) == 0
+    assert cfg.read_text() == expected
+
+
 def test_migrate_config_invalid_yaml(tmpdir):
     contents = '['
     cfg = tmpdir.join(C.CONFIG_FILE)

tests/yaml_rewrite_test.py

@@ -0,0 +1,47 @@
+from __future__ import annotations
+
+import pytest
+
+from pre_commit.yaml import yaml_compose
+from pre_commit.yaml_rewrite import MappingKey
+from pre_commit.yaml_rewrite import MappingValue
+from pre_commit.yaml_rewrite import match
+from pre_commit.yaml_rewrite import SequenceItem
+
+
+def test_match_produces_scalar_values_only():
+    src = '''\
+-   name: foo
+-   name: [not, foo]  # not a scalar: should be skipped!
+-   name: bar
+'''
+    matcher = (SequenceItem(), MappingValue('name'))
+    ret = [n.value for n in match(yaml_compose(src), matcher)]
+    assert ret == ['foo', 'bar']
+
+
+@pytest.mark.parametrize('cls', (MappingKey, MappingValue))
+def test_mapping_not_a_map(cls):
+    m = cls('s')
+    assert list(m.match(yaml_compose('[foo]'))) == []
+
+
+def test_sequence_item_not_a_sequence():
+    assert list(SequenceItem().match(yaml_compose('s: val'))) == []
+
+
+def test_mapping_key():
+    m = MappingKey('s')
+    ret = [n.value for n in m.match(yaml_compose('s: val\nt: val2'))]
+    assert ret == ['s']
+
+
+def test_mapping_value():
+    m = MappingValue('s')
+    ret = [n.value for n in m.match(yaml_compose('s: val\nt: val2'))]
+    assert ret == ['val']
+
+
+def test_sequence_item():
+    ret = [n.value for n in SequenceItem().match(yaml_compose('[a, b, c]'))]
+    assert ret == ['a', 'b', 'c']