Performance: Replace eval(parse()) with safer and faster alternatives

[anthropic-code-agent]

Description

Replace eval(parse()) patterns with safer and more performant alternatives. The eval(parse()) pattern is considered dangerous, slow, and hard to debug. Modern R provides better ways to handle dynamic evaluation.

Current Implementation Issues

R/MappedData.R - Multiple eval(parse()) occurrences

This is the most critical performance issue in the codebase. The pattern appears 10+ times:

Lines 336-342: Dynamic aesthetic assignment

eval(parse(text = paste0(
  "ggplot2::aes(",
  aestheticName, " = ", columnName,
  ")"
)))

• Very slow: parsing and evaluation on every call
• Security risk: potential code injection
• Hard to debug: errors show parsed text, not original code

Other occurrences in MappedData.R:

• Lines 386-392: Similar pattern
• Lines 405-411: Same inefficient pattern
• Lines 542-548: Repeated pattern
• Lines 601-607: Repeated pattern

R/MappedDataTimeProfile.R - Similar patterns

Similar eval(parse()) patterns for dynamic aesthetic creation.

Why eval(parse()) is Problematic

1. Performance: Text parsing is expensive
2. Security: Can execute arbitrary code
3. Debugging: Error messages are cryptic
4. Maintainability: Code is harder to understand
5. Type safety: No compile-time checking

Suggested Alternatives

Option 1: Use rlang (Recommended)

The package already uses rlang for tidy evaluation. Use rlang::sym() and rlang::expr():

Before:

eval(parse(text = paste0("ggplot2::aes(", aestheticName, " = ", columnName, ")")))

After:

aes_list <- list()
aes_list[[aestheticName]] <- rlang::sym(columnName)
do.call(ggplot2::aes, aes_list)

Or using tidy evaluation:

ggplot2::aes(!!rlang::sym(aestheticName) := !!rlang::sym(columnName))

Option 2: Use aes_string() (Deprecated but safer)

Note: aes_string() is deprecated in newer ggplot2, but still safer than eval(parse()):

ggplot2::aes_string(setNames(columnName, aestheticName))

Option 3: Build aesthetics as list

Build aesthetics programmatically:

aes_mapping <- list()
aes_mapping[[aestheticName]] <- as.name(columnName)
do.call(ggplot2::aes, aes_mapping)

Recommended Approach

For this codebase, use Option 1 with rlang since:

• Package already imports rlang
• Most modern and maintainable
• Best performance
• Type-safe
• Works with current ggplot2 versions

Expected Benefits

• Significant performance improvement (5-10x faster for aesthetic creation)
• Safer code (no code injection risk)
• Better error messages
• More maintainable code
• Type checking and IDE support

Implementation Notes

• Carefully test all aesthetic mappings after refactoring
• Ensure all plot types work correctly
• Document the new approach for future maintainers
• Consider adding helper functions to centralize aesthetic creation
• Follow OSP R coding standards: https://dev.open-systems-pharmacology.org/r-development-resources/coding_standards_r

Files to Modify

• R/MappedData.R (10+ occurrences - highest priority)
• R/MappedDataTimeProfile.R

Testing

• Run all existing plot tests
• Test each plot type with various aesthetic mappings
• Verify error messages are helpful
• Test edge cases (special characters in column names, etc.)
• Add performance benchmarks to measure improvement

References

• rlang documentation
• Why eval(parse()) is often suboptimal
• ggplot2 programming guide