✅ Snyk checks have passed. No issues have been found so far.

Status	Scanner	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues
✅	Licenses	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.

✅ PR preview is ready!

Summary

This PR adds bind="query-params" support to st.datetime_input and st.time_input, enabling two-way sync between widget values and URL query parameters. Key changes include:

• Proto changes: Added optional string query_param_key to both DateTimeInput.proto (field 15) and TimeInput.proto (field 11).
• Wire format modernization: st.datetime_input wire format changed from YYYY/MM/DD, HH:mm to ISO 8601 YYYY-MM-DDThh:mm, with backward-compatible fallback parsing on both frontend and backend.
• Backend: DateTimeInputSerde and TimeInputSerde updated to handle invalid/out-of-bounds URL values by reverting to defaults. bind parameter wired through to register_widget.
• Frontend: DateTimeInput.tsx and TimeInput.tsx construct queryParamBinding objects from the new queryParamKey proto field.
• AppTest: element_tree.py updated to parse both ISO and legacy datetime formats.

The implementation follows the patterns established in PR #14034 for st.date_input.

Code Quality

The code is well-structured, clean, and follows existing Streamlit patterns closely. Specific observations:

• Backend (time_widgets.py): The DateTimeInputSerde and TimeInputSerde classes are well-designed. The ISO format constants and fallback parsing logic are clear.
• Frontend: Both DateTimeInput.tsx and TimeInput.tsx correctly construct queryParamBinding objects and follow React best practices (memoization, useCallback, derived state via useMemo).
• dateTimeInputUtils.ts: Good separation of concerns with the dual-format parsing via DATE_TIME_PARSE_FORMATS.
• Docstrings: The bind parameter documentation for both widgets is thorough and matches the pattern from st.date_input.

One inconsistency found — In _datetime_input (line 1331-1335), when value_needs_reset is true due to dynamic bounds changes with bind="query-params", the stale URL param is NOT cleared:

        if value_needs_reset and key is not None:
            # Update session_state so subsequent accesses in this run
            # return the corrected value. Use reset_state_value to avoid
            # the "cannot be modified after widget instantiated" error.
            get_session_state().reset_state_value(key, current_value)

Compare with _date_input (line 1790-1799) which does clear it:

        if value_needs_reset and key is not None:
            # Update session_state so subsequent accesses in this run
            # return the corrected value. Use reset_state_value to avoid
            # the "cannot be modified after widget instantiated" error.
            get_session_state().reset_state_value(key, current_value)

            # Clear stale URL param when an out-of-bounds URL value was reset.
            if bind == "query-params":
                with get_session_state().query_params() as qp:
                    qp.remove_param(str(key))

This means that if a user has bind="query-params" on a datetime_input AND dynamically changes min_value/max_value such that the current (URL-seeded) value becomes out of bounds, the widget value will be reset but the stale URL param will remain. For the initial URL seeding case (tested in E2E), this is handled correctly by _seed_widget_from_url detecting default equality. But the dynamic bounds + query-params combination would leave stale URL params.

Test Coverage

Test coverage is comprehensive across all layers:

• Python unit tests (datetime_input_test.py, time_input_test.py): Tests cover bind with key, bind without key (error), invalid bind value, bind with explicit value, bind with None value, and serde deserialization for ISO/legacy/invalid/empty/out-of-bounds formats. The TestDateTimeInputSerdeISO and TestTimeInputSerdeStepSnapping classes thoroughly test the deserialization logic.
• Frontend unit tests (DateTimeInput.test.tsx, TimeInput.test.tsx): Tests cover query param binding registration/unregistration, clearable behavior, and the absence of binding when queryParamKey is not set.
• E2E tests (st_datetime_input_test.py, st_time_input_test.py): Tests cover URL seeding, clearable empty params, invalid URL values reverting to defaults, out-of-range values, and step-unaligned values for time input.
• Typing tests: Both datetime_input_types.py and time_input_types.py verify the bind parameter types with assert_type.

The NUM_DATETIME_INPUTS and NUM_TIME_INPUTS counters are correctly updated to account for the new bound widgets in the E2E app scripts.

Backwards Compatibility

This change is fully backward-compatible:

• Proto changes: Adding optional fields (15 and 11) is additive and safe. Existing clients ignore unknown fields.
• Wire format: Both frontend (DATE_TIME_PARSE_FORMATS) and backend (DateTimeInputSerde.deserialize) accept both the new ISO format and the legacy YYYY/MM/DD, HH:mm format. The element_tree.py _parse_dt helper also tries both formats.
• API surface: The bind parameter defaults to None, so existing code is unaffected. The parameter follows the same pattern already established by st.date_input, st.radio, st.selectbox, etc.
• User-facing display: The format parameter (controlling display) is completely unaffected by the wire format change.

Security & Risk

No security concerns identified:

• Invalid URL values are silently reverted to widget defaults (no error messages that could leak information).
• Out-of-bounds URL values are rejected and reset, preventing injection of unexpected values.
• The implementation follows the same safe patterns used by all other bound widgets.

Low risk of regression: The wire format change is internal and protected by dual-format parsing. The only regression risk is in the edge case noted in Code Quality (stale URL params with dynamic bounds), which is a minor inconsistency rather than a correctness bug.

Accessibility

No accessibility concerns. The widgets already have proper aria-label attributes set from the element label. The query param binding feature is a backend/URL concern that doesn't alter the widget's DOM structure or interactive behavior.

Recommendations

1. Add URL param clearing for dynamic bounds in _datetime_input: Add the remove_param logic to _datetime_input when value_needs_reset occurs with bind="query-params", matching the pattern in _date_input. This ensures consistent behavior across all bound time widgets. The fix would be adding lines after get_session_state().reset_state_value(key, current_value) in _datetime_input:

if bind == "query-params":
    with get_session_state().query_params() as qp:
        qp.remove_param(str(key))

Verdict

APPROVED: Well-implemented feature that follows established patterns with comprehensive test coverage. The missing URL param clearing for the dynamic bounds edge case (recommendation #1) is a minor inconsistency that could be addressed as a follow-up.

This is an automated AI review by opus-4.6-thinking.