Add script detection, env shebang parser, and opt-in content guessing

tests/test_detect.py

@@ -70,7 +70,13 @@ class TestExtensionMap:
         assert detect_type(path="archive.tar.gz") == "text"
 
     def test_hidden_file(self) -> None:
-        assert detect_type(path=".bashrc") == "text"
+        assert detect_type(path=".bashrc") == "bash"
+
+    def test_zshrc(self) -> None:
+        assert detect_type(path=".zshrc") == "bash"
+
+    def test_profile(self) -> None:
+        assert detect_type(path=".profile") == "bash"
 
     @pytest.mark.parametrize(
         ("suffix", "expected"),
@@ -80,8 +86,73 @@ class TestExtensionMap:
         assert detect_type(path=f"example{suffix}") == expected
 
 
+class TestShebangDetection:
+    """Priority 4: detect common scripts from shebangs."""
+
+    def test_env_python3_shebang(self) -> None:
+        assert detect_type(content="#!/usr/bin/env python3\nprint('hi')\n") == "python"
+
+    def test_env_python3_with_assignment(self) -> None:
+        assert (
+            detect_type(content="#!/usr/bin/env FOO=1 python3\nprint('hi')\n")
+            == "python"
+        )
+
+    def test_env_ignore_environment_before_program(self) -> None:
+        assert detect_type(content="#!/usr/bin/env -i bash\necho hi\n") == "bash"
+
+    def test_env_split_string_shebang(self) -> None:
+        assert (
+            detect_type(content="#!/usr/bin/env -S python3 -u\nprint('hi')\n")
+            == "python"
+        )
+
+    def test_env_compact_split_string_shebang(self) -> None:
+        assert (
+            detect_type(content="#!/usr/bin/env -Spython3 -u\nprint('hi')\n")
+            == "python"
+        )
+
+    def test_env_quoted_split_string_shebang(self) -> None:
+        assert (
+            detect_type(
+                content='#!/usr/bin/env --split-string="python3 -u"\nprint("hi")\n'
+            )
+            == "python"
+        )
+
+    def test_env_long_option_with_equals(self) -> None:
+        assert (
+            detect_type(content="#!/usr/bin/env --ignore-signal=TERM bash\necho hi\n")
+            == "bash"
+        )
+
+    def test_direct_python_shebang(self) -> None:
+        assert detect_type(content="#!/usr/bin/python3\nprint('hi')\n") == "python"
+
+    def test_bash_shebang(self) -> None:
+        assert detect_type(content="#!/bin/bash\necho hi\n") == "bash"
+
+    def test_sh_shebang(self) -> None:
+        assert detect_type(content="#!/bin/sh\necho hi\n") == "bash"
+
+    def test_zsh_shebang(self) -> None:
+        assert detect_type(content="#!/bin/zsh\necho hi\n") == "bash"
+
+    def test_extension_wins_over_shebang(self) -> None:
+        assert (
+            detect_type(path="script.py", content="#!/bin/bash\necho hi\n") == "python"
+        )
+
+    def test_filename_wins_over_shebang(self) -> None:
+        assert detect_type(path=".bashrc", content="#!/usr/bin/env python3\n") == "bash"
+
+    def test_unrecognized_shebang_falls_through(self) -> None:
+        assert detect_type(content="#!/usr/bin/env ruby\nputs 'hi'\n") == "text"
+
+
 class TestContentJsonDetection:
-    """Priority 4: detect JSON from content when path gives no clue."""
+    """Priority 5: detect JSON from content when path gives no clue."""
 
     def test_json_object(self, sample_json: str) -> None:
         assert detect_type(content=sample_json) == "json"
@@ -104,9 +175,31 @@ class TestContentJsonDetection:
     def test_content_with_valid_path_takes_priority(self) -> None:
         assert detect_type(path="script.py", content='{"a": 1}') == "python"
 
+    def test_shebang_takes_priority_over_json(self) -> None:
+        assert detect_type(content='#!/usr/bin/env python3\n{"a": 1}\n') == "python"
+
+
+class TestGuessContent:
+    """Priority 6: optional heuristic content guessing."""
+
+    def test_guess_content_disabled_by_default(self) -> None:
+        diff = "--- a/file\n+++ b/file\n@@ -1 +1 @@\n-old\n+new\n"
+        assert detect_type(content=diff) == "text"
+
+    def test_guess_content_detects_diff(self) -> None:
+        diff = "--- a/file\n+++ b/file\n@@ -1 +1 @@\n-old\n+new\n"
+        assert detect_type(content=diff, guess_content=True) == "diff"
+
+    def test_guess_content_rejects_low_confidence_false_positive(self) -> None:
+        python_snippet = "def f():\n    return 1\n"
+        assert detect_type(content=python_snippet, guess_content=True) == "text"
+
+    def test_json_still_wins_over_guess_content(self, sample_json: str) -> None:
+        assert detect_type(content=sample_json, guess_content=True) == "json"
+
 
 class TestFallback:
-    """Priority 5: fallback to 'text'."""
+    """Priority 7: fallback to 'text'."""
 
     def test_no_args(self) -> None:
         assert detect_type() == "text"