Static library for dynamic memory and Unicode strings manipulation for AmigaOS/M68k

String functions use suffixes for text encoding used:

• W - UTF-16
• L - UTF-32
• U - UTF-8

No suffix means functions for 8-bit encodings (codepages). Sometimes such functions also work for UTF-8 (for example StrSize()), then a macro with U suffix is provided, to make the function set orthogonal.

• AllocVecPooled() - pooled allocation with size remembering
• FreeVecPooled() - freeing memory allocated with the above
• CloneMem() - a copy of memory block in newly allocated memory
• CloneMemPooled() - the same as above but into memory pool

• StrSize[UWL]() - string size in bytes including zero terminator
• StrCount[UWL]() - number of characters in a string, without terminator

• StrCopy[UWL]() - plain copy, may be used for concatenation
• StrClone[UWL]() - a copy in newly allocated memory
• StrClonePooled[UWL]() - a copy in memory allocated from a pool
• StrJoin[UWL]() - joins multiple strings into newly allocated memory
• StrJoinPooled[UWL]() - joins multiple strings into memory allocated from a pool

• Utf8Size() - byte size of UTF-8 encoded character
• Utf8Dec() - single character validating UTF-8 decoder
• Utf8Enc() - single character validating UTF-8 encoder

To speed up the code, arguments are passed in CPU registers (An for pointers, Dn for other data). As a rule, there is no argument validation (e. g. NULL checking). Passing valid arguments is a duty of calling code, error reporting on such low level is ineffective. The only exception is validation of string encoding (UTF-8 rules, UTF-16 surrogate pairs, UTF-32 range), which is done in functions which analyse strings character by character.

Where UTF-16 and UTF-32 endianness is significant, big endian (M680x0 native) is assumed, even for strings containing BOM. Automatic endianness detection via BOM is only supported by StrConvert(). Some simple functions work for both endians, it is documented per function. In general case little endian strings have to be converted before processing.

Each function is documented in its source file, in standard Amiga AutoDoc format. The makefile provides doc target, which generates AmigaGuide and plaintext autodoc file using RoboDoc tool, available on Aminet.

GCC at the moment. VBCC is planned as well. Of course these functions may be used in assembler programs by just including their source, or linking.

libustring-m68k does not depend on any library other than system ones (exec.library at the moment). It does not depend on the standard C library and may be used in programs compiled with -nostdlib option.