test: snapshot deeply-nested alternation, combined lookaround, named-backref, and ReDoS edge fixtures

10 files changed, 135 insertions, 3 deletions

diff --git a/tests/docs/snapshots.test.py b/tests/docs/snapshots.test.py
index 031a382..f734452 100644
--- a/tests/docs/snapshots.test.py
+++ b/tests/docs/snapshots.test.py
@@ -8,12 +8,11 @@ change surfaces every doc example the change touches so their prose stays
 truthful.
 """
 
+import re
 from pathlib import Path
 
 import pytest
 
-import re
-
 import edify
 import edify.library as edify_library
 from edify import Pattern
@@ -103,7 +102,9 @@ _BLOCKS_DEFERRED_TO_DOCS_REWRITE = frozenset(
 @pytest.mark.parametrize(
     ("rst_path", "block_start", "block_source", "relative_stem"),
     DISCOVERED_BLOCKS,
-    ids=[f"{relative_stem}:{block_start}" for _, block_start, _, relative_stem in DISCOVERED_BLOCKS],
+    ids=[
+        f"{relative_stem}:{block_start}" for _, block_start, _, relative_stem in DISCOVERED_BLOCKS
+    ],
 )
 def test_doc_code_block_produces_the_snapshotted_regex(
     rst_path, block_start, block_source, relative_stem
diff --git a/tests/fixtures/edge_cases.test.py b/tests/fixtures/edge_cases.test.py
new file mode 100644
index 0000000..4e74983
--- /dev/null
+++ b/tests/fixtures/edge_cases.test.py
@@ -0,0 +1,123 @@
+"""Snapshot suite for the compile-path edge-case fixtures.
+
+Each fixture builds a small, intentionally-tricky pattern and pins the
+emitted regex string. The fixtures are the ones most likely to silently
+drift when the compile path is refactored: deeply nested alternation,
+combined lookaround, item-4 named backreferences, and constructs that
+:mod:`edify` flags for ReDoS.
+"""
+
+from pathlib import Path
+
+import pytest
+
+from edify import Pattern, RegexBuilder
+from edify.testing import assert_snapshot
+
+_SNAPSHOT_ROOT = Path(__file__).parent.parent / "snapshots" / "fixtures"
+
+
+def _deeply_nested_alternation():
+    inner_alt = RegexBuilder().any_of("abc", "def", "ghi")
+    outer_alt = RegexBuilder().any_of("first", "second", "third")
+    return (
+        RegexBuilder()
+        .start_of_input()
+        .subexpression(inner_alt)
+        .string("-")
+        .subexpression(outer_alt)
+        .end_of_input()
+    )
+
+
+def _combined_lookaround_positive_and_negative():
+    return (
+        RegexBuilder()
+        .assert_ahead()
+        .digit()
+        .end()
+        .assert_not_ahead()
+        .string("0")
+        .end()
+        .assert_behind()
+        .string("$")
+        .end()
+        .digit()
+        .digit()
+        .digit()
+    )
+
+
+def _named_backreference_pair():
+    return (
+        RegexBuilder()
+        .named_capture("open")
+        .any_of("[", "(", "{")
+        .end()
+        .zero_or_more()
+        .word()
+        .named_back_reference("open")
+    )
+
+
+def _nested_unbounded_quantifier_redos():
+    return RegexBuilder().one_or_more().group().one_or_more().digit().end()
+
+
+def _alternation_with_overlapping_prefixes_redos():
+    return RegexBuilder().one_or_more().any_of("aa", "aaa", "aaaa")
+
+
+def _hex_color_pattern():
+    return (
+        RegexBuilder()
+        .start_of_input()
+        .string("#")
+        .any_of_chars("0123456789abcdefABCDEF")
+        .exactly(6)
+        .word()
+        .end_of_input()
+    )
+
+
+def _email_shape_pattern():
+    return (
+        RegexBuilder()
+        .start_of_input()
+        .one_or_more()
+        .any_of_chars("a-zA-Z0-9._%+-")
+        .string("@")
+        .one_or_more()
+        .any_of_chars("a-zA-Z0-9.-")
+        .string(".")
+        .at_least(2)
+        .letter()
+        .end_of_input()
+    )
+
+
+def _pattern_composition_via_use():
+    reusable = Pattern().at_least(2).letter()
+    return RegexBuilder().start_of_input().subexpression(reusable).end_of_input()
+
+
+_FIXTURES = [
+    ("deeply_nested_alternation", _deeply_nested_alternation),
+    ("combined_lookaround", _combined_lookaround_positive_and_negative),
+    ("named_backreference_pair", _named_backreference_pair),
+    ("nested_unbounded_quantifier_redos", _nested_unbounded_quantifier_redos),
+    ("alternation_with_overlapping_prefixes_redos", _alternation_with_overlapping_prefixes_redos),
+    ("hex_color_pattern", _hex_color_pattern),
+    ("email_shape_pattern", _email_shape_pattern),
+    ("pattern_composition_via_use", _pattern_composition_via_use),
+]
+
+
[email protected](
+    ("fixture_name", "builder_factory"), _FIXTURES, ids=[name for name, _ in _FIXTURES]
+)
+def test_edge_case_fixture_emits_the_snapshotted_regex(fixture_name, builder_factory):
+    builder = builder_factory()
+    emitted = builder.to_regex_string()
+    snapshot_path = _SNAPSHOT_ROOT / f"{fixture_name}.regex"
+    assert_snapshot(emitted, snapshot_path)
diff --git a/tests/snapshots/fixtures/alternation_with_overlapping_prefixes_redos.regex b/tests/snapshots/fixtures/alternation_with_overlapping_prefixes_redos.regex
new file mode 100644
index 0000000..8d3f7b5
--- /dev/null
+++ b/tests/snapshots/fixtures/alternation_with_overlapping_prefixes_redos.regex
@@ -0,0 +1 @@
+(?:aa|aaa|aaaa)+
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/combined_lookaround.regex b/tests/snapshots/fixtures/combined_lookaround.regex
new file mode 100644
index 0000000..baa7e8f
--- /dev/null
+++ b/tests/snapshots/fixtures/combined_lookaround.regex
@@ -0,0 +1 @@
+(?=\d)(?!0)(?<=\$)\d\d\d
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/deeply_nested_alternation.regex b/tests/snapshots/fixtures/deeply_nested_alternation.regex
new file mode 100644
index 0000000..0f2c26b
--- /dev/null
+++ b/tests/snapshots/fixtures/deeply_nested_alternation.regex
@@ -0,0 +1 @@
+^(?:abc|def|ghi)\-(?:first|second|third)$
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/email_shape_pattern.regex b/tests/snapshots/fixtures/email_shape_pattern.regex
new file mode 100644
index 0000000..bd17d4d
--- /dev/null
+++ b/tests/snapshots/fixtures/email_shape_pattern.regex
@@ -0,0 +1 @@
+^[a\-zA\-Z0\-9._%+-]+@[a\-zA\-Z0\-9.-]+\.[a-zA-Z]{2,}$
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/hex_color_pattern.regex b/tests/snapshots/fixtures/hex_color_pattern.regex
new file mode 100644
index 0000000..c2d63c0
--- /dev/null
+++ b/tests/snapshots/fixtures/hex_color_pattern.regex
@@ -0,0 +1 @@
+^\#[0123456789abcdefABCDEF]\w{6}$
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/named_backreference_pair.regex b/tests/snapshots/fixtures/named_backreference_pair.regex
new file mode 100644
index 0000000..1e3fe4a
--- /dev/null
+++ b/tests/snapshots/fixtures/named_backreference_pair.regex
@@ -0,0 +1 @@
+(?P<open>[\[\(\{])\w*(?P=open)
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/nested_unbounded_quantifier_redos.regex b/tests/snapshots/fixtures/nested_unbounded_quantifier_redos.regex
new file mode 100644
index 0000000..e5aaf8b
--- /dev/null
+++ b/tests/snapshots/fixtures/nested_unbounded_quantifier_redos.regex
@@ -0,0 +1 @@
+(?:\d+)+
\ No newline at end of file
diff --git a/tests/snapshots/fixtures/pattern_composition_via_use.regex b/tests/snapshots/fixtures/pattern_composition_via_use.regex
new file mode 100644
index 0000000..70b3c7c
--- /dev/null
+++ b/tests/snapshots/fixtures/pattern_composition_via_use.regex
@@ -0,0 +1 @@
+^[a-zA-Z]{2,}$
\ No newline at end of file