Promptheus

"""Essential tests for utility functions.""" import pytest from promptheus.utils import sanitize_error_message, collapse_whitespace def test_sanitize_error_message_basic(): """Test basic error message sanitization.""" message = "API call failed" result = sanitize_error_message(message) assert result == "API call failed" def test_sanitize_error_message_with_tokens(): """Test sanitization of API keys and tokens.""" message = "Request failed with API key: sk-1234567890abcdef and another key: AIza987654321" result = sanitize_error_message(message) assert "***" in result # The actual implementation replaces the entire token with *** assert "1234567890abcdef" not in result assert "987654321" not in result def test_sanitize_error_message_with_uuid(): """Test sanitization of UUID-like tokens.""" message = "Session token: 550e8400-e29b-41d4-a716-446655440000" result = sanitize_error_message(message) assert "***" in result assert "550e8400-e29b-41d4-a716-446655440000" not in result def test_sanitize_error_message_long_message(): """Test sanitization with very long message.""" long_message = "Error: " + "a" * 200 + " sk-1234567890 " + "b" * 200 result = sanitize_error_message(long_message) # Should be truncated to max length (160) assert len(result) <= 160 assert "***" in result def test_sanitize_error_message_empty(): """Test sanitization with empty message.""" result = sanitize_error_message("") assert result == "" def test_sanitize_error_message_none(): """Test sanitization with None message.""" result = sanitize_error_message(None) assert result == "" def test_sanitize_error_message_truncated(): """Test that long messages are properly truncated.""" # Create a message that is long but doesn't match token pattern long_message = "This is a long message " + "with normal words " * 20 result = sanitize_error_message(long_message, max_length=50) # Should be truncated to max_length - if adding ... would exceed max_length, # it just truncates to max_length assert len(result) <= 50 # Either exactly 50 or 53 if ... fits # Check if it was actually truncated (doesn't end with original ending) assert len(result) < len(long_message) # Definitely truncated def test_collapse_whitespace_basic(): """Test basic whitespace collapsing functionality (right strip only).""" lines = ["hello ", " world ", " test"] result = collapse_whitespace(lines) assert result == "hello\n world\n test" # Only right side is stripped def test_collapse_whitespace_mixed_whitespace(): """Test collapsing with mixed whitespace (right strip only).""" lines = [" hello\t ", " \tworld\n", "\ttest \t"] result = collapse_whitespace(lines) # Only trailing whitespace is stripped, leading whitespace remains assert result == " hello\n \tworld\n\ttest" def test_collapse_whitespace_empty_lines(): """Test handling of empty lines.""" lines = ["hello", "", "world", " ", "test"] result = collapse_whitespace(lines) # Empty lines should remain, but lines with just spaces become empty expected_lines = result.split("\n") assert expected_lines[0] == "hello" assert expected_lines[1] == "" # Empty line remains assert expected_lines[2] == "world" assert expected_lines[3] == "" # Space-only line becomes empty assert expected_lines[4] == "test" def test_collapse_whitespace_single_line(): """Test collapsing with a single line (right strip only).""" lines = [" hello world "] result = collapse_whitespace(lines) assert result == " hello world" # Only right side is stripped def test_collapse_whitespace_special_characters(): """Test collapsing with special characters.""" lines = ["hello\nworld", "test\twith\ttabs"] result = collapse_whitespace(lines) assert result == "hello\nworld\ntest\twith\ttabs" def test_sanitize_error_message_multiple_tokens(): """Test sanitization with multiple tokens in one message.""" message = "Keys: sk-1234567890 and AIza0987654321 and secret-abcdef123456" result = sanitize_error_message(message) assert result.count("***") >= 3 # Should sanitize all three tokens def test_sanitize_error_message_edge_case_short_token(): """Test sanitization doesn't affect short strings.""" message = "Short: abcd" result = sanitize_error_message(message) assert result == "Short: abcd" # Should not be sanitized def test_sanitize_error_message_edge_case_long_token(): """Test sanitization with very long token.""" long_token = "x" * 100 message = f"API key: {long_token}" result = sanitize_error_message(message) assert "***" in result assert long_token not in result def test_collapse_whitespace_unicode(): """Test collapsing with Unicode characters (right strip only).""" lines = ["héllo ", " wørld ", " tëst"] result = collapse_whitespace(lines) assert result == "héllo\n wørld\n tëst" # Only right side is stripped def test_collapse_whitespace_with_newlines(): """Test that existing newlines in content are preserved properly.""" lines = ["line1\nsubline", "line2", "line3"] result = collapse_whitespace(lines) # The function joins lines with \n, so the embedded \n becomes part of the content assert "\nsubline" in result assert "line2" in result