Use .data() instead of const_cast(.c_str()) to fix undefined behavior

[BYVoid]

Summary

Replace const_cast(str.c_str()) with C++17 non-const str.data() to eliminate undefined behavior when writing into std::string buffers. Also use .data() consistently for copy source operands, and cache by-value return results in local const std::string& to avoid redundant temporaries.

Why .data() instead of const_cast(.c_str())

In standard C++, std::string::c_str() returns a const char*. Writing through a const_cast-ed pointer obtained from .c_str() is undefined behavior — the implementation is not required to allow mutation through that pointer, and compilers may optimize assuming the pointed-to data is not modified.

C++17 added a non-const overload of std::string::data() that returns char*, specifically designed for cases where you need writable access to the internal buffer after resize(). This is the correct, well-defined way to obtain a mutable pointer to a std::string's storage.

Changes

File	Change
src/BinaryDict.cpp	const_cast(keyBuf.c_str()) / valueBuf.c_str() → .data() for writable buffers; cache Key()/Value() in const std::string&; use .data() in offset calculations and asserts
src/SerializedValues.cpp	const_cast(valueBuffer->c_str()) → valueBuffer->data(); use .data() consistently
src/UTF8Util.hpp	Replace strncpy(const_cast(newStr.c_str()), ...) with std::string(str, length) constructor
src/Converter.cpp	Use .data() instead of .c_str() for copy source
src/SimpleConverter.cpp	Use .data() instead of .c_str() for copy source
src/tools/CommandLine.cpp	Use .data() instead of .c_str() for copy source; add explicit #include

Test plan

• All 16 existing unit tests pass
• Build succeeds with no new warnings

https://claude.ai/code/session_01TeWHUg4WeAWkaumW2jWgzJ

[Copilot]

Pull request overview

This PR aims to remove unsafe C string copy APIs (strcpy/strncpy) by switching call sites to explicit-length copies (memcpy) and by simplifying UTF8Util::FromSubstr to a safer/cleaner std::string(ptr, len) construction.

Changes:

• Replace strcpy/strncpy with memcpy using explicit byte counts in multiple conversion/buffer-building paths.
• Simplify UTF8Util::FromSubstr implementation to avoid const_cast + strncpy.
• Add include where needed for memcpy usage (partially; one TU still appears to rely on transitive headers).

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
src/tools/CommandLine.cpp	Uses memcpy to restore UTF-8 remainder bytes between chunked reads.
src/UTF8Util.hpp	Simplifies substring creation to std::string(str, length).
src/SimpleConverter.cpp	Copies converted output into newly allocated C buffer via memcpy(len+1).
src/SerializedValues.cpp	Uses memcpy(len+1) when constructing serialized value buffer.
src/Converter.cpp	Switches from strcpy to memcpy(len+1) for caller-provided output buffer.
src/BinaryDict.cpp	Uses memcpy(len+1) when constructing key/value buffers for binary dict serialization.

---

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