fix: Use raw string values instead of repr() for str types in display_string

[kgilpin]

• Replace isinstance(val, str) with issubclass(type(val), str) to avoid __class__ lookups on proxy/lazy objects
• Update the display_string docstring/comment to reflect that str types return raw values (not repr())
• Verify all tests pass

[Copilot]

Pull request overview

Adjusts AppMap parameter/return-value rendering so str values are recorded as their raw string content (not repr()-quoted), improving consistency with log output and enabling reliable secret-in-log substring matching.

Changes:

• Update display_string to return raw str values instead of repr(val) for strings.
• Update unit tests to assert raw string values for parameters/return values.
• Update JSON “expected” AppMap fixtures to match the new string formatting.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
_appmap/event.py	Changes display_string to emit raw str values instead of repr() output.
_appmap/test/web_framework.py	Updates web framework capture assertions to expect raw string values.
_appmap/test/test_params.py	Updates parameter-capture assertions/parameterized expectations for str values.
_appmap/test/test_http.py	Updates HTTP client capture assertions for raw string query param values.
_appmap/test/test_events.py	Updates labeled params/return value assertions and comment to reflect raw strings.
_appmap/test/data/unittest/expected/unittest.appmap.json	Updates expected recorded parameter/return str values (and formatting) to raw strings.
_appmap/test/data/unittest/expected/unittest-no-test-cases.appmap.json	Updates expected recorded str values to raw strings.
_appmap/test/data/unittest/expected/pytest.appmap.json	Updates expected recorded str values to raw strings.
_appmap/test/data/pytest/expected/pytest-numpy2.appmap.json	Updates expected recorded str return values to raw strings.
_appmap/test/data/pytest/expected/pytest-numpy2-no-test-cases.appmap.json	Updates expected recorded str return values to raw strings.
_appmap/test/data/pytest/expected/pytest-numpy1.appmap.json	Updates expected recorded str return values to raw strings.
_appmap/test/data/pytest/expected/pytest-numpy1-no-test-cases.appmap.json	Updates expected recorded str return values to raw strings.
_appmap/test/data/expected.appmap.json	Updates expected recorded str parameter/return values (including newline escaping) to raw strings.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Using isinstance(val, str) here can trigger class lookups on proxy/lazy objects (the file already avoids isinstance for this reason in _is_list_or_dict), potentially causing side effects or premature evaluation when params display is enabled. Consider avoiding isinstance and instead check the real type (e.g., type(val) is str or issubclass(type(val), str)) before returning the raw value.

Suggested change

	value = val if isinstance(val, str) else repr(val)
	val_type = type(val)
	value = val if (val_type is str or issubclass(val_type, str)) else repr(val)

[kgilpin]

What about:

issubclass(type(val), str)

[Copilot]

Changed to issubclass(type(val), str) in commit 622ec72.

[Copilot]

The comment above still states that parameter display uses repr() to compute a string value, but the implementation now returns the raw value for str types. Please update the comment to reflect the new behavior so it doesn't mislead future readers.

[kgilpin]

@copilot apply changes based on this feedback

[Copilot]

Updated the comment in commit 622ec72 to clarify that str types are returned as-is and other types use repr().

[dividedmind]

Note many tests require APPMAP_DISPLAY_PARAMS=true (which tox sets across the board).

Also note uv run is an alternative to using venv directly.

[dividedmind]

Looks good to me. We went back and forth on whether strings should be represented in maps verbatim or as a representation; the scanner usage does finally provide a good argument on one over the other.

The test is also a good addition.

Before merging, please:

• fixup the issubclass and comment fix into the main feature commit,
• amend the main feature commit message to include BREAKING CHANGE: note; this changes the format of the stringified parameters.

One thing I wonder: since we're special casing strings, does it make sense to always capture them, regardless of appmap_display_params value? If it's a string, there is should be no risk of extra serialization overhead. Two potential pitfalls: 1) making sure the class is also always captured in this case, 2) do we want to truncate to avoid putting huge strings in the appmap?