From ce169026b329a35e57f45b0c57b32e706fa4b9e2 Mon Sep 17 00:00:00 2001
From: Alice Carroll <git@alice-carroll.pet>
Date: Mon, 7 Jul 2025 10:59:50 +0300
Subject: [PATCH] feat: non-breaking hyphen shorthand

Closes #6355
---
 crates/typst-syntax/src/ast.rs              |   1 +
 crates/typst-syntax/src/lexer.rs            |   3 ++-
 tests/ref/shorthand-nbsp-and-shy-hyphen.png | Bin 1356 -> 2047 bytes
 tests/suite/layout/inline/linebreak.typ     |   2 +-
 tests/suite/syntax/shorthand.typ            |   2 +-
 tools/support/typst.tmLanguage.json         |   4 ++++
 6 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/crates/typst-syntax/src/ast.rs b/crates/typst-syntax/src/ast.rs
index 547d53cd8..c51eaeec8 100644
--- a/crates/typst-syntax/src/ast.rs
+++ b/crates/typst-syntax/src/ast.rs
@@ -595,6 +595,7 @@ impl Shorthand<'_> {
         ("--", '\u{2013}'),
         ("---", '\u{2014}'),
         ("-?", '\u{00AD}'),
+        ("-!", '\u{2011}'),
     ];
 
     /// Get the shorthanded character.
diff --git a/crates/typst-syntax/src/lexer.rs b/crates/typst-syntax/src/lexer.rs
index 82f65cd36..59a89d138 100644
--- a/crates/typst-syntax/src/lexer.rs
+++ b/crates/typst-syntax/src/lexer.rs
@@ -191,6 +191,7 @@ impl Lexer<'_> {
             '-' if self.s.eat_if("--") => SyntaxKind::Shorthand,
             '-' if self.s.eat_if('-') => SyntaxKind::Shorthand,
             '-' if self.s.eat_if('?') => SyntaxKind::Shorthand,
+            '-' if self.s.eat_if('!') => SyntaxKind::Shorthand,
             '-' if self.s.at(char::is_numeric) => SyntaxKind::Shorthand,
             '*' if !self.in_word() => SyntaxKind::Star,
             '_' if !self.in_word() => SyntaxKind::Underscore,
@@ -528,7 +529,7 @@ impl Lexer<'_> {
             match s.eat() {
                 Some(' ') if s.at(char::is_alphanumeric) => {}
                 Some('/') if !s.at(['/', '*']) => {}
-                Some('-') if !s.at(['-', '?']) => {}
+                Some('-') if !s.at(['-', '?', '!']) => {}
                 Some('.') if !s.at("..") => {}
                 Some('h') if !s.at("ttp://") && !s.at("ttps://") => {}
                 Some('@') if !s.at(is_valid_in_label_literal) => {}
diff --git a/tests/ref/shorthand-nbsp-and-shy-hyphen.png b/tests/ref/shorthand-nbsp-and-shy-hyphen.png
index e6984bb79d8bdca473675cca0edc62fb3569de2b..ad5a92a132ae4bca883e0d2c73d7774bdfeddedd 100644
GIT binary patch
delta 2046
zcmV<a2LbrZ3jYs~7k^I(000001#n&)000NZNkl<Zc-rmOcT*I09>?)Kh?}{)f9@u6
zbTo;MeIo&kiY+P;)U|M{5ygUXDt1H^!3GK_g1CU9v<0PiL5k7@RGJ6U%q~&{frVYl
z*7rGcb8~K(aRVyOCA^=&Yv%XI=llDD&(s&=@HjjU|Nh{r0)O6@B2|uI(zFkSt<G*-
zc4m3B4ezV(`&taE`v$<ffAsJJLY>K}+su0t)q!CG03D-;A<`A<Hh79^c@*KThQ`&I
zy8`KP7bkGu-_1r2Ao6m`?cWh%ql<|Oq%iTQ76FcDn}yAZNBy=u<>B1}-hHEkzgNax
z)|lq3j|438Fn`0YmL=60)@t|kzPvn@s7f<af68sJ);ecqT{wpd5Y$G1m2~i{uDla0
z<WeG|gGFQ?e`49vJOnK7LBOnTj>cpk$2VLS8Ko7~VXAOXM(o_utj`Iwy~j5eDIk<-
zOex=qFmuoAk;CFcP$4AsfNElw`Ze%F8<TLjiM1@{0DqoA0HT_%jW|V9X{zlSn1{1O
zKSe&HmX$9Qvnk%f#d<X(hMQfERugo%<J$4=tb-w1sM()Tx+h<K)jt<cC;6!}df<+)
zS8u12+W@zm5(vH#>7MS;Q0VyB@n+n<`kRN_PVV{n$mbXk>0xkFeg;@85=Rt{Z}tBp
ztSV7pq<@Yl3`<e0|3LW7tXhneKA&Hw-WvYA-zaSKG#F`M0l%((JH8PZhgIT>l^5<@
zyga0j@hYRw*Q*9eXg<L0z_05F43a>KB#Hb8H{krS7u$-f9Gc6n_;t10r8&NZ_xl0R
z7wF!k_6bhM+dO#vt|H@b#_6^fG3#{~{9fdBSby_#i!?L#bot+apg64SBz_!@D7UMc
z+-=#J3p6ig)1&@i<qE2nVH4Cfu%I>!0K%>{zB-eIRY{c{UeUBYml}Zw?N)RqyHjGL
z5=VW<L_OHuDz5}=N)grlGllqr7hTJr>GJ0U0IY2|vNcLkCWNoBea^tt@2UVmzG<Ce
zS${nqH~s@B`KB3x6Z8Qg$<73cZ-frb;fHLb{>OO(IQy`)<fbE@5%-1kYGXC#0QXJX
zm--_AR@wd|Hcyb7Hq7i%Hv+M^r5p=jM_HK&pD;aO1Wvwkbuif?z(ZeiD0WBr%bgyb
z{f_S0fMV~Zv%y^V5~FiH5y!FccmTXet$%?Mm#eRu?9LIhr8Ro5Y&8%W;Hy)gzZS*y
z9d%W9kB!3sNoV+wUZ;(AK(@?(H9ncYV|d^t26c8Ytf5BUpRRV~u>Ts~O|w?au)-Jd
z{wn`Ga#$+FAn{Fx)w={E2}Taf)=LgImme(acFA!pJ@bq}|I4RzJd)%MoGx+m0e|QW
zIlY9QYvN{9;kUbq_AdgUEzwh3f2iirebixBzZ?ZTy(a21Kb3KXmErv}`U1Ch+g8`_
zZkzt*)PCh61+YdXTf!=N&rI#1!h5#$I~BPA+!*jk!X&zHAxHkjlu^J!3LSa4EmMY2
zhJ8k0FfJKB2ea<0F2*6M1Nl69mw)Id!J?)qS(%l3yOpm1XbC*vgM|_SQ?;Xc>P9Q_
z9Khr}2pd`5>URUD<O~~BI`?QcJ_p@jR5CNZZ9_1$td#zO{#cg*1+a85+>7}=Ok?q5
z{Y34E;ed5*mqeYoKLnx6sl=f@`}=L3HT#_+<#x{P>-~V(UBOnBwfp@Z@PB0E)ryC1
zSEP@30$$#}yNfO3iY<yNMZVh-Mmr!uz&a_vt@_-Lwp{i(&^+bx*iWV&P|177kHlER
zZ@@5}oZ$@)U{Kcw!x_VN$o?3h3+3JV%#wJw>OZMJ6lN|Mmgx*2#h}w^7&MO&!P$~L
z_jbAhBl^2Mf4PtZz>N4i>VF#V!YI-u?K}%BBFp*p`K$sz#3jMLP<lHu7;ufha*I&U
zp?t3_Nvc?diS$eX#7lasd}~~v8--gY<%L~hRd3}=Y+TysfgUC){V!-I$5%_DWacVQ
z%-_OKZ|dZ;30Tv#p|;%~3k~4v(7Ld<$%LYO*@T+A+c61iUC5<^qJK@A32_xRM&W@4
zlsd>PV1mw}lm&xf_C5$`S<GCp^ovZUC~UGmYo)ms1Vnq`lJU|?w+6s94ZA8fJPAT+
zfp!umdIZb~kIY7Z#?&aRFsFF;(3e`xOm-Jzt^g45;4CxsI7FeX-HC}d{kYFQ^s|WY
z<V33K^u2a~lgZdVqknK=3d#k%S<YMy)3GW<%M6C*8HK~l8a*sur#pD%x32PRFpC6O
z?tfDCcy1{5mL3_MYa^{=k8RM}oQ)Bb?A|OF9#*c{b)*8ih0dKTJittc@~<KGjpg&w
zN@wJkT6*Uf&&p3Uss8u_5|>NoKK`2~1e6wIvRsMV3hopG0Dm-zdLix0mtz6b`3wJR
z0FX6wj30^L3TrV;Gd8etK@0{J{C#X-+!HWt&KSUjK?=?#P!d`}!>y+Pq}`eZDUn_9
zD7KICu#!&!KTLb%^cL>f(ibY*qCOf9oUG`Y%!Q<0v+jgY>2l2)b>OFf$qZig7RG9K
zm!B_X*1$9nFn4{39qB(^4++I~bL*#o{UzN$;vUzaxU=`}9`g18U$-)q#TD38Y)^*P
zj?@k4KRzt6zcZ&~m0kXUt2w%Z=Z>`k_tssRomuk_|BF;C^hmAQDUOqF82{dm!{hKc
c{D+1A4QE8ItBFBzG5`Po07*qoM6N<$f|CCM00000

delta 1350
zcmV-M1-bhF56lXX7k@Pf00000ugP${000FONkl<Zc-rmN=}!{|7{~GbC3|7!GUrS@
zX6h#M*mTM?9;iiS^<1478x|2k1cWj`<PhX)xyDhB4UqdBh#*s_iXsCFXi8Cz3RFsg
z7TWKY>=ne=0EtG!_wDoI^X~V2#5?Mv1$Y5ofImFA`3`SNUw^D)dg_w-!n^$rZ1u6t
za^OvktZid@s-*y-hPlIkk>WwM>JV>Cq+&(_05fxkA^i%d_6LdDs8xiTplesfNjbw(
zSswmTz79G7QHc93!-*sZb9Sbj`h3?Z1W5ceLQTGF<lzV0{^MNe$lTzyqq)J8Hr4j&
zz@{3$LEOgoPk+knOw!Vkni`|1d2*%2%XhrZu6m^^jq`U9v?*Y>dVJAf%_Y9&T0V1w
zMO3-n-|<{C0oV>BVAT+=*;Q^tq;;ofPKi|5=%t0oJf3ST)$(J1vUF1s2)ie3XgZ0o
z;!OYSVcB_jB&7J9aq*z3AK1{#r#L2H*Ijgupf~_g>wn+@+_QLWGc`E&6&@LuCAt-D
zY;V_p$=$;Qc9@totnqYxLY&LM4Y#36=cFlUb;@fzQ)dc~x<!z?&dbLFBu9k2(z^SO
zk?5XBd~CY3)UmT(a!-<w`)fynOYh|~M=rdJ0a3ev@8~;#r$jPE{lZqikFc>pkCg`Q
z!_4$|Er0WbU#@DyN~P<WndxX+i(!5+A{AB|+sMp1ej-AFb=W8id>oJ*7(8Lg4pA}^
z5z<OVN*%!O^3PSs<1(NqqkuYvFW`T%{rKHxM~(7&<Y1q;MDic(6bZnHT>8`$7GFx}
zx_JAOe_MykynEZT?acv^?KJ~-9C@y;$nkFa7=MV%#ZHwGiHo91-2B<l4(%;Kvzddn
zhWOEKG;hTrZg6a4TPgsg-stM_sbqIUvzU-9dRx?1;BoObZK0F~2cs-=4T~)xPHcJ%
zI5Z+s{ZhnD=fEpX56yL}V*vJcUO1YmR|;Vf$66R$wx$^XbO-n8w{~zf_J6n_qQnZE
zXMc%FD)b;u7Mn7$nhjN`quh82c!aT~AwfcrqFl(Ushr7m0OdA)TO!bXG1~V74)@X7
zbgmpWbpbi}7F~?tgi<LYEY$*7feWvPj~8x=xoT-WpL3$=+0R$C28px^(4W2L7f*DE
z4D*#sxm--U2|x(Nt<c~Z-t$!a8`+0nWPhFQ-UXz`M3_zfH!_Jnp=uW2vkn6k_%NUP
zOXfNthu8NkOy&m+k3C~r+5ujZv4iOiXy*$@?U{C&+$-ZX6|H1irakfVh1Ei)B}?FS
z37B4V(tP1=`wSONle2Qjvs%*V^ALGdu$P(IjG8g;2H!A%rrabDpicEO>m%jQ0Do$e
zw<mEgI?YULMh^Wc0i$z(m$zmHvs1}+wx=0ZjL44;9edJos(1NIFQ?J(^gt*(hY;Y{
zQoz2x(N$xhk`S2Ga}dRa$sK(^R`}FZ8Rh^Bsh4p3cttnDZcdlU<MAvDbFM5m2IeAC
z0m?+%2MBl|2<8*m^$4QEiAl=5M1TL9JKHPE2ik2snLXUKn;J*>+8=~n?0OLyi+f?2
z14gqnivulSu;pZ<c?D&YTeN~s*6HBpAp)WvZbo@ryKD+WnPq=#zj?NB%)Z`hA}!L}
zAoTRgpE%=WIOZ|w<esh*d-T~y0;dw)19EMZp2<->b&4(qEY6pGoF)|Bfj?TpPsQo|
z6GOTD&m0FN4%nxD`^$B@8Z=9S-#%;sjJn~4C$Ru8!0!+IF9u>)utd`_^#A|>07*qo
IM6N<$f~udI>Hq)$

diff --git a/tests/suite/layout/inline/linebreak.typ b/tests/suite/layout/inline/linebreak.typ
index 86a900252..7de01490a 100644
--- a/tests/suite/layout/inline/linebreak.typ
+++ b/tests/suite/layout/inline/linebreak.typ
@@ -10,7 +10,7 @@ Supercalifragilisticexpialidocious Expialigoricmetrioxidation.
 
 --- linebreak-hyphen-nbsp ---
 // Test for non-breaking space and hyphen.
-There are non\u{2011}breaking~characters.
+There are non-!breaking~characters.
 
 --- linebreak-narrow-nbsp ---
 // Test for narrow non-breaking space.
diff --git a/tests/suite/syntax/shorthand.typ b/tests/suite/syntax/shorthand.typ
index 7d1782b9f..696a794e0 100644
--- a/tests/suite/syntax/shorthand.typ
+++ b/tests/suite/syntax/shorthand.typ
@@ -1,7 +1,7 @@
 // Test shorthands for unicode codepoints.
 
 --- shorthand-nbsp-and-shy-hyphen ---
-The non-breaking space~does work, soft-?hyphen also does.
+The non-breaking space~does work, soft-?hyphen does, and non-!breaking hyphen also does.
 
 --- shorthand-nbsp-width ---
 // Make sure non-breaking and normal space always
diff --git a/tools/support/typst.tmLanguage.json b/tools/support/typst.tmLanguage.json
index 104df060e..a8e85bc91 100644
--- a/tools/support/typst.tmLanguage.json
+++ b/tools/support/typst.tmLanguage.json
@@ -42,6 +42,10 @@
           "name": "punctuation.definition.nonbreaking-space.typst",
           "match": "~"
         },
+        {
+          "name": "punctuation.definition.nonbreaking-hyphen.typst",
+          "match": "-!"
+        },
         {
           "name": "punctuation.definition.shy.typst",
           "match": "-\\?"