craft-csv: a diacritic before another escape keeps it as written

An escape's diacritic took the next character as its letter even when that was the next escape (ESC N H ESC N H o gave ESC + U+0308 + NHo). A control character is no letter: the first escape is kept as written and the next one decodes on its own (ESC N H ö).
This commit is contained in:
Hamza
2026-11-16 19:07:27 +05:00
parent ab87149855
commit e096b5eb9b
2 changed files with 7 additions and 2 deletions
+3 -1
View File
@@ -355,7 +355,9 @@ fn decode_escapes(text: &str) -> String {
};
if let Some(&(_, mark, pairs)) = ISO6937_DIACRITICS.iter().find(|(d, _, _)| *d == code) {
// The letter it goes on: precomposed where Unicode has the pair, else letter and mark.
let Some(base) = ahead.next() else {
// The base is an ordinary character: another escape (or any control character) after a
// diacritic leaves this escape as written, and the next one is decoded on its own.
let Some(base) = ahead.next().filter(|b| !b.is_control()) else {
out.push(c);
continue;
};
+4 -1
View File
@@ -123,7 +123,7 @@ fn decodes_excels_character_escapes_in_text() {
let file = "ID;PWXL;N;E\r\nC;Y1;X1;K\"Gr\u{1b}NHo\u{1b}N{e f\u{1b}NHur K\u{1b}NHoln\"\r\nC;Y2;X1;K\"Plain ASCII 123\"\r\n\
C;Y3;X1;K\"\u{1b}NBe \u{1b}NAa \u{1b}NCa \u{1b}NKc \u{1b}NOs \u{1b}NMo \u{1b}NJa \u{1b}NDn\"\r\n\
C;Y4;X1;K\"\u{1b}NAa \u{1b}NAZ\"\r\nC;Y5;X1;K\"\u{1b}Nq \u{1b}NyStra\u{1b}N{e \u{1b}Na\u{1b}Ni\"\r\n\
C;Y6;X1;K\"\u{1b}NHq \u{1b}N\u{1b}NHo end\u{1b}N\"\r\nC;Y7;X1;K\"\u{1b}N\u{7f} \u{1b}NL \u{1b}X\"\r\nE\r\n";
C;Y6;X1;K\"\u{1b}NHq \u{1b}N\u{1b}NHo end\u{1b}N\"\r\nC;Y7;X1;K\"\u{1b}N\u{7f} \u{1b}NL \u{1b}X\"\r\nC;Y8;X1;K\"\u{1b}NH\u{1b}NHo \u{1b}NB\tx\"\r\nE\r\n";
let sh = read(file.as_bytes()).unwrap();
let value = |a: &str| sh.cells.get(&at(a)).map(|c| c.value.clone()).unwrap_or_default();
let text = |s: &str| Value::Text(s.into());
@@ -139,4 +139,7 @@ fn decodes_excels_character_escapes_in_text() {
assert_eq!(value("A6"), text("q\u{308} \u{1b}Nö end\u{1b}N"));
// Codes ISO 6937 doesn't define (0xFF, 0xCC) and an `ESC` without `N` stay as written.
assert_eq!(value("A7"), text("\u{1b}N\u{7f} \u{1b}NL \u{1b}X"));
// A diacritic followed by another escape (or a control character) is kept as written, and the
// next escape still decodes.
assert_eq!(value("A8"), text("\u{1b}NHö \u{1b}NB\tx"));
}