jedib0t / go-pretty / 41903829197

100%

main: 100%

Pull #447

github

text: end CSI sequences at their final byte, not only at "m"

EscSeqParser.isEscapeStopRune() ended a CSI sequence at "m" only, so every
other CSI sequence stayed open and the text after it was consumed as part of
the sequence, even though a CSI sequence ends at its first final byte in the
range 0x40-0x7E (ECMA-48, section 5.4).

text.Cursor and text.EraseLine emit exactly such sequences ("\x1b[A",
"\x1b[5B", "\x1b[K", ...):

    StripEscape(EraseLine.Sprint() + "Ghost")                   -> "" (want "Ghost")
    StringWidthWithoutEscSequences(CursorUp.Sprintn(3) + "abc") -> 0 (want 3)
    LongestLineLen(EraseLine.Sprint() + "abc\ndef")             -> 0 (want 3)
    Trim(EraseLine.Sprint() + "abcdef", 3)                      -> "\x1b[Kabcdef"

Pad, Snip, RuneCount, Widen and InsertEveryN inherited the miscount.

A CSI sequence now ends at the first byte in 0x40-0x7E that follows the "ESC ["
introducer, and ParseSeq() ignores the CSI sequences that are not SGR
("...m"): their parameter bytes are not attribute codes, so "\x1b[2;3H"
positions the cursor instead of turning on dim, and it leaves the attributes
in force untouched.

7 of 7 new or added lines in 1 file covered. (100.0%)

5170 of 5170 relevant lines covered (100.0%)

1.21 hits per line