Splitting text

Both operations observe their text and arguments without consuming them. They return independently owned lines or pieces in Result[list[str], AllocError]. All list storage and long output strings are constructed fallibly. Failure reclaims the initialized result prefix and preserves the inputs.

Explicit separators

text.split(separator) requires one nonempty string, including a reference. A literal "" separator is a compile error, split requires a nonempty separator. A computed empty separator traps; this invalid-argument error is not an AllocError. Maximum-split arguments are unsupported.

Matches are literal, non-overlapping, and scanned left to right. Leading, trailing, and adjacent separators produce empty pieces. Empty input yields [""]; no match yields one piece containing the input’s contents. Multibyte separators and NUL work without normalization.

Whitespace

text.split() with no separator splits on runs of Unicode White_Space, using the same bundled property as isspace() and trimming. It omits leading, trailing, and repeated empty pieces; empty or whitespace-only text yields an empty list. NUL and U+200B are ordinary content. The output remains fallible and independently owned, like explicit-separator splitting.

Lines

text.splitlines() discards terminators. text.splitlines(True) retains each original terminator without translation; the optional argument is bool and positional.

Recognized boundaries are LF, CR, CRLF, vertical tab, form feed, U+001C–U+001E, U+0085, U+2028, and U+2029. Empty text yields no lines; interior blank lines are retained; a final terminator adds no extra empty line. This is broader than file universal-newline reading, which recognizes CR/LF only.