Update shrimp.ts

Reviewed-on: #8

src/compiler/compiler.ts

@@ -303,7 +303,8 @@ export class Compiler {
         return instructions
       }
 
-      case terms.ThenBlock: {
+      case terms.ThenBlock:
+      case terms.SingleLineThenBlock: {
         const instructions = getAllChildren(node)
           .map((child) => this.#compileNode(child, input))
           .flat()
@@ -468,7 +469,11 @@ export class Compiler {
       }
 
       default:
-        throw new CompilerError(`Unsupported syntax node: ${node.type.name}`, node.from, node.to)
+        throw new CompilerError(
+          `Compiler doesn't know how to handle a "${node.type.name}" node.`,
+          node.from,
+          node.to
+        )
     }
   }
 }

src/compiler/tests/compiler.test.ts

@@ -85,6 +85,21 @@ describe('compiler', () => {
     expect(`bloop = do: 'bloop' end; bloop`).toEvaluateTo('bloop')
   })
 
+  test('function call with if statement and multiple expressions', () => {
+    expect(`
+      abc = do:
+        if false:
+          echo nope
+        end
+
+        true
+      end
+
+      abc
+      `)
+      .toEvaluateTo(true)
+  })
+
   test('simple conditionals', () => {
     expect(`(3 < 6)`).toEvaluateTo(true)
     expect(`(10 > 20)`).toEvaluateTo(false)
@@ -139,6 +154,10 @@ describe('compiler', () => {
       scattered
     end`).toEvaluateTo('dwarf')
   })
+
+  test('single line if', () => {
+    expect(`if 3 < 9: shire end`).toEvaluateTo('shire')
+  })
 })
 
 describe('errors', () => {

src/parser/shrimp.grammar

@@ -2,7 +2,7 @@
 
 @context trackScope from "./scopeTracker"
 
-@skip { space }
+@skip { space | comment }
 
 @top Program { item* }
 
@@ -18,6 +18,7 @@
   newlineOrSemicolon { "\n" | ";" }
   eof { @eof }
   space { " " | "\t" }
+  comment { "#" ![\n]* }
   leftParen { "(" }
   rightParen { ")" }
   colon[closedBy="end", @name="colon"] { ":" }
@@ -104,7 +105,7 @@ IfExpr {
 }
 
 singleLineIf {
-  @specialize[@name=keyword]<Identifier, "if"> (ConditionalOp | expression) colon ThenBlock { consumeToTerminator }
+  @specialize[@name=keyword]<Identifier, "if"> (ConditionalOp | expression) colon SingleLineThenBlock @specialize[@name=keyword]<Identifier, "end">
 }
 
 multilineIf {
@@ -123,6 +124,10 @@ ThenBlock {
   block
 }
 
+SingleLineThenBlock {
+  consumeToTerminator 
+}
+
 ConditionalOp {
   expression Eq expression |
   expression Neq expression |
@@ -199,5 +204,5 @@ expressionWithoutIdentifier {
 }
 
 block {
-  (consumeToTerminator newlineOrSemicolon)*
+  (consumeToTerminator? newlineOrSemicolon)*
 }

src/parser/shrimp.terms.ts

@@ -42,6 +42,7 @@ export const
   NamedArg = 40,
   NamedArgPrefix = 41,
   IfExpr = 43,
+  SingleLineThenBlock = 45,
   ThenBlock = 46,
   ElseIfExpr = 47,
   ElseExpr = 49,

src/parser/shrimp.ts

@@ -7,11 +7,11 @@ import {highlighting} from "./highlight"
 const spec_Identifier = {__proto__:null,end:62, null:76, if:88, elseif:96, else:100}
 export const parser = LRParser.deserialize({
   version: 14,
-  states: ".jQVQbOOO!QOpO'#CqO#^QcO'#CuO$WQRO'#CvO$fQcO'#DmO$}QbO'#DtOOQ`'#Cx'#CxO%VQbO'#CtO%wOSO'#C|OOQa'#Dr'#DrO&VQcO'#DqOOQ`'#Dn'#DnO&nQbO'#DmO&|QbO'#EQOOQ`'#DX'#DXO'kQRO'#DaOOQ`'#Dm'#DmO'pQQO'#DlOOQ`'#Dl'#DlOOQ`'#Db'#DbQVQbOOO'xObO,59]OOQa'#Dq'#DqOOQ`'#Cs'#CsO(QQbO'#DUOOQ`'#Dp'#DpOOQ`'#Dc'#DcO([QbO,59[O&|QbO,59bO&|QbO,59bOOQ`'#Dd'#DdO(xQbO'#CyO)QQQO,5:`O)qQRO'#CvO*RQRO,59`O*dQRO,59`O*_QQO,59`O+_QQO,59`O+gQbO'#DOO+oQWO'#DPOOOO'#Dz'#DzOOOO'#Df'#DfO,TOSO,59hOOQa,59h,59hO,cQbO'#DgO,kQbO,59ZO,|QRO,5:lO-TQQO,5:lO-YQbO,59{OOQ`,5:W,5:WOOQ`-E7`-E7`OOQa1G.w1G.wOOQ`,59p,59pOOQ`-E7a-E7aOOQa1G.|1G.|O-dQcO1G.|OOQ`-E7b-E7bO.OQbO1G/zO&|QbO,59cO&|QbO,59cOOQa1G.z1G.zOOOO,59j,59jOOOO,59k,59kOOOO-E7d-E7dOOQa1G/S1G/SO!VQbO'#CuOOQ`,5:R,5:ROOQ`-E7e-E7eO.]QbO1G0WOOQ`1G/g1G/gO.jQbO7+%fO.oQbO7+%gOOQO1G.}1G.}O.|QRO1G.}OOQ`'#DZ'#DZOOQ`7+%r7+%rO/WQbO7+%sOOQ`<<IQ<<IQO/kQQO'#DeO/pQbO'#DwO0TQbO<<IROOQ`'#D['#D[O0YQbO<<I_OOQ`,5:P,5:POOQ`-E7c-E7cOOQ`AN>mAN>mO&|QbO'#D]OOQ`'#Dh'#DhO0eQbOAN>yO0pQQO'#D_OOQ`AN>yAN>yO0uQbOAN>yO0zQRO,59wO1RQQO,59wOOQ`-E7f-E7fOOQ`G24eG24eO1WQbOG24eO1]QQO,59yO1bQQO1G/cOOQ`LD*PLD*PO.oQbO1G/eO/WQbO7+$}OOQ`7+%P7+%POOQ`<<Hi<<Hi",
+  states: ".vQYQbOOO!TOpO'#CqO#aQcO'#CuO$ZQRO'#CvO$iQcO'#DnO%QQbO'#DuOOQ`'#Cx'#CxO%YQbO'#CtO%zOSO'#C|OOQa'#Ds'#DsO&YQcO'#DrOOQ`'#Do'#DoO&qQbO'#DnO'PQbO'#EROOQ`'#DX'#DXO'nQRO'#DaOOQ`'#Dn'#DnO'sQQO'#DmOOQ`'#Dm'#DmOOQ`'#Db'#DbQYQbOOO'{ObO,59]OOQa'#Dr'#DrOOQ`'#Cs'#CsO(TQbO'#DUOOQ`'#Dq'#DqOOQ`'#Dc'#DcO(_QbO,59[O'PQbO,59bO'PQbO,59bOOQ`'#Dd'#DdO({QbO'#CyO)TQQO,5:aO)tQRO'#CvO*UQRO,59`O*gQRO,59`O*bQQO,59`O+bQQO,59`O+jQbO'#DOO+rQWO'#DPOOOO'#D{'#D{OOOO'#Df'#DfO,WOSO,59hOOQa,59h,59hO,fQbO'#DgO,nQbO,59ZO-PQRO,5:mO-WQQO,5:mO-]QbO,59{OOQ`,5:X,5:XOOQ`-E7`-E7`OOQa1G.w1G.wOOQ`,59p,59pOOQ`-E7a-E7aOOQa1G.|1G.|O-gQcO1G.|OOQ`-E7b-E7bO.RQbO1G/{O'PQbO,59cO'PQbO,59cOOQa1G.z1G.zOOOO,59j,59jOOOO,59k,59kOOOO-E7d-E7dOOQa1G/S1G/SO!YQbO'#CuOOQ`,5:R,5:ROOQ`-E7e-E7eO.`QbO1G0XOOQ`1G/g1G/gO.mQbO7+%gO.rQbO7+%hOOQO1G.}1G.}O/SQRO1G.}OOQ`'#DZ'#DZO/^QbO7+%sO/cQbO7+%tOOQ`<<IR<<IROOQ`'#De'#DeO/yQQO'#DeO0OQbO'#DxO0fQbO<<ISOOQ`<<I_<<I_OOQ`'#D['#D[O0kQbO<<I`OOQ`,5:P,5:POOQ`-E7c-E7cOOQ`AN>nAN>nO'PQbO'#D]OOQ`'#Dh'#DhO0vQbOAN>zO1RQQO'#D_OOQ`AN>zAN>zO1WQbOAN>zO1]QRO,59wO1dQQO,59wOOQ`-E7f-E7fOOQ`G24fG24fO1iQbOG24fO1nQQO,59yO1sQQO1G/cOOQ`LD*QLD*QO.rQbO1G/eO/cQbO7+$}OOQ`7+%P7+%POOQ`<<Hi<<Hi",
-  stateData: "1j~O!_OS~O]QO^_O_XO`POaTOfXOtXOuXOvXO|]O!gVO!jbO!mWO~O!ceO~O]fO_XO`POaTOfXOtXOuXOvXOwgOyhO!gVO!mWOziX!jiX!viX!liXoiX~OP!eXQ!eXR!eXS!eXT!eXU!eXV!eXW!eXX!eXY!eXZ!eX[!eX~P!VOPlOQlORmOSmO~OPlOQlORmOSmO!j!aX!v!aXo!aX~O]nOnmP~O]QO_XO`POaTOfXOtXOuXOvXO!gVO!mWO~OqxO!m{O!ovO!pwO~OP!eXQ!eXR!eXS!eX!j!aX!v!aXo!aX~Oz|O!j!aX!v!aXo!aX~O]fO_XO`POfXOtXOuXOvXO!gVO!mWO~OV!QO~O!j!RO!v!RO~O]!TOf!TO~OaTOw!UO~P&|OaTOwgOyhOzda!jda!vda!ldaoda~P&|O]nOnmX~On!ZO~OT!]OU!]OV![OW![OX![OY![OZ![O[![O~OPlOQlORmOSmO~P)VOPlOQlORmOSmO!l!^O~O!l!^OP!eXQ!eXR!eXS!eXT!eXU!eXV!eXW!eXX!eXY!eXZ!eX[!eX~Oz|O!l!^O~O]!_O!gVO~O!m!`O!o!`O!p!`O!q!`O!r!`O!s!`O~OqxO!m!bO!ovO!pwO~O]!cO`PO~Oz|O!jca!vca!lcaoca~On!fO~P)VOn!fO~O^_O|]O~P%VOPlOQlORjiSji!jji!vji!ljioji~O^_O|]O!j!iO~P%VO^_O|]O!j!nO~P%VOo!oO~O^_O|]Oo!kP~P%VO!lkinki~P)VO^_O|]Oo!kP!Q!kP!S!kP~P%VO!j!uO~O^_O|]Oo!kX!Q!kX!S!kX~P%VOo!wO~Oo!|O!Q!xO!S!{O~Oo#RO!Q!xO!S!{O~On#TO~Oo#RO~On#UO~P)VOn#UO~Oo#VO~O!j#WO~O!j#XO~Ofu~",
+  stateData: "1{~O!_OS!`OS~O]QO^_O_XO`POaTOfXOtXOuXOvXO|]O!hVO!kbO!nWO~O!deO~O]fO_XO`POaTOfXOtXOuXOvXOwgOyhO!hVO!nWOziX!kiX!wiX!miXoiX~OP!fXQ!fXR!fXS!fXT!fXU!fXV!fXW!fXX!fXY!fXZ!fX[!fX~P!YOPlOQlORmOSmO~OPlOQlORmOSmO!k!bX!w!bXo!bX~O]nOnmP~O]QO_XO`POaTOfXOtXOuXOvXO!hVO!nWO~OqxO!n{O!pvO!qwO~OP!fXQ!fXR!fXS!fX!k!bX!w!bXo!bX~Oz|O!k!bX!w!bXo!bX~O]fO_XO`POfXOtXOuXOvXO!hVO!nWO~OV!QO~O!k!RO!w!RO~O]!TOf!TO~OaTOw!UO~P'POaTOwgOyhOzda!kda!wda!mdaoda~P'PO]nOnmX~On!ZO~OT!]OU!]OV![OW![OX![OY![OZ![O[![O~OPlOQlORmOSmO~P)YOPlOQlORmOSmO!m!^O~O!m!^OP!fXQ!fXR!fXS!fXT!fXU!fXV!fXW!fXX!fXY!fXZ!fX[!fX~Oz|O!m!^O~O]!_O!hVO~O!n!`O!p!`O!q!`O!r!`O!s!`O!t!`O~OqxO!n!bO!pvO!qwO~O]!cO`PO~Oz|O!kca!wca!mcaoca~On!fO~P)YOn!fO~O^_O|]O~P%YOPlOQlORjiSji!kji!wji!mjioji~O^_O|]O!k!iO~P%YO^_O|]O!k!nO~P%YOo!oO~O^_O|]O!k!pOo!lP~P%YO!mkinki~P)YOo!tO~O^_O|]O!k!pOo!lP!Q!lP!S!lP~P%YO!k!wO~O^_O|]O!k!pOo!lX!Q!lX!S!lX~P%YOo!yO~Oo#OO!Q!zO!S!}O~Oo#TO!Q!zO!S!}O~On#VO~Oo#TO~On#WO~P)YOn#WO~Oo#XO~O!k#YO~O!k#ZO~Ofu~",
-  goto: "+t!vPPPPPPPPPPPPPPPPPPP!w#W#fP$S$X#W$s%Y%f%}PP&QP&i&iPPPP$SPP&mP&y&|'VP'ZP&m'a'g'n't'}(T([PPP(b(f(zP)^)c*^P*y*yP+[PP+dPPPPP+h+hd`Od!Q!Z!f!i!n!q#W#XRtViZOVd|!Q!Z!f!i!n!q#W#XfQOVd!Q!Z!f!i!n!q#W#XdfQ]hklm![!]!c!xR!c|ViQk!czXOQV]dhklm!Q!Z![!]!c!f!i!n!q!x#W#XR!_vdSOd!Q!Z!f!i!n!q#W#XQrVQ!WlR!XmQtVQ!P]Q!j!]R#P!xd`Od!Q!Z!f!i!n!q#W#XUgQk!cQtVR!UhRpT{XOQV]dhklm!Q!Z![!]!c!f!i!n!q!x#W#XTxWze`Od!Q!Z!f!i!n!q#W#XR!m!fQ!t!nQ#Y#WR#Z#XT!y!t!zQ!}!tR#S!zQdOR!SdSkQ!cR!VkQoTR!YoW!q!i!n#W#XR!v!qQzWR!azS}[uR!e}Q!z!tR#Q!zTcOdSaOdQ!g!QQ!h!ZQ!l!fZ!p!i!n!q#W#Xd[Od!Q!Z!f!i!n!q#W#XQuVR!d|VjQk!cdROd!Q!Z!f!i!n!q#W#XUgQk!cQqVQ!O]Q!UhQ!WlQ!XmQ!j![Q!k!]R#O!xdYOd!Q!Z!f!i!n!q#W#XdfQ]hklm![!]!c!xRsVoUOQVdhk!Q!Z!c!f!i!n!q#W#XQ!r!iV!s!n#W#XTyWze^Od!Q!Z!f!i!n!q#W#X",
+  goto: "+u!wPPPPPPPPPPPPPPPPPPP!x#X#gP$T$Y#X$t%Z%g&OPP&RP&j&jPPPP$TPP&nP&z&}'WP'[P&n'b'h'o'u(O(U(]PPPP(c(g({P)_)d*_P*z*zP+]PP+ePPPPP+i+id`Od!Q!Z!f!i!n!r#Y#ZRtViZOVd|!Q!Z!f!i!n!r#Y#ZfQOVd!Q!Z!f!i!n!r#Y#ZdfQ]hklm![!]!c!zR!c|ViQk!czXOQV]dhklm!Q!Z![!]!c!f!i!n!r!z#Y#ZR!_vdSOd!Q!Z!f!i!n!r#Y#ZQrVQ!WlR!XmQtVQ!P]Q!j!]R#R!zd`Od!Q!Z!f!i!n!r#Y#ZUgQk!cQtVR!UhRpT{XOQV]dhklm!Q!Z![!]!c!f!i!n!r!z#Y#ZTxWze`Od!Q!Z!f!i!n!r#Y#ZR!m!fQ!v!nQ#[#YR#]#ZT!{!v!|Q#P!vR#U!|QdOR!SdSkQ!cR!VkQoTR!YoW!r!i!n#Y#ZR!x!rQzWR!azS}[uR!e}Q!|!vR#S!|TcOdSaOdQ!g!QQ!h!ZQ!l!fZ!q!i!n!r#Y#Zd[Od!Q!Z!f!i!n!r#Y#ZQuVR!d|VjQk!cdROd!Q!Z!f!i!n!r#Y#ZUgQk!cQqVQ!O]Q!UhQ!WlQ!XmQ!j![Q!k!]R#Q!zdYOd!Q!Z!f!i!n!r#Y#ZdfQ]hklm![!]!c!zRsVoUOQVdhk!Q!Z!c!f!i!n!r#Y#ZQ!s!iV!u!n#Y#ZTyWze^Od!Q!Z!f!i!n!r#Y#Z",
-  nodeNames: "⚠ Star Slash Plus Minus And Or Eq Neq Lt Lte Gt Gte Identifier AssignableIdentifier Word IdentifierBeforeDot Do Program PipeExpr FunctionCall DotGet Number PositionalArg ParenExpr FunctionCallOrIdentifier BinOp ConditionalOp FunctionDef Params colon keyword String StringFragment Interpolation EscapeSeq Boolean Regex Null Underscore NamedArg NamedArgPrefix operator IfExpr keyword ThenBlock ThenBlock ElseIfExpr keyword ElseExpr keyword Assign",
+  nodeNames: "⚠ Star Slash Plus Minus And Or Eq Neq Lt Lte Gt Gte Identifier AssignableIdentifier Word IdentifierBeforeDot Do Program PipeExpr FunctionCall DotGet Number PositionalArg ParenExpr FunctionCallOrIdentifier BinOp ConditionalOp FunctionDef Params colon keyword String StringFragment Interpolation EscapeSeq Boolean Regex Null Underscore NamedArg NamedArgPrefix operator IfExpr keyword SingleLineThenBlock ThenBlock ElseIfExpr keyword ElseExpr keyword Assign",
-  maxTerm: 84,
+  maxTerm: 85,
   context: trackScope,
   nodeProps: [
     ["closedBy", 30,"end"]
@@ -19,9 +19,9 @@ export const parser = LRParser.deserialize({
   propSources: [highlighting],
   skippedNodes: [0],
   repeatNodeCount: 7,
-  tokenData: "<}~RyOX#rXY$aYZ$zZp#rpq$aqt#rtu%euw#rwx%jxy%oyz&Yz{#r{|&s|}#r}!O&s!O!P#r!P!Q)g!Q!['b![!]2S!]!^$z!^#O#r#O#P2m#P#R#r#R#S2r#S#T#r#T#Y3]#Y#Z4k#Z#b3]#b#c8y#c#f3]#f#g9p#g#h3]#h#i:g#i#o3]#o#p#r#p#q<_#q;'S#r;'S;=`$Z<%l~#r~O#r~~<xS#wUqSOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rS$^P;=`<%l#r^$hUqS!_YOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU%RUqS!jQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#r~%jO!o~~%oO!m~U%vUqS!gQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU&aUqS!lQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU&xWqSOt#ruw#rx!Q#r!Q!['b![#O#r#P;'S#r;'S;=`$Z<%lO#rU'iYqSfQOt#ruw#rx!O#r!O!P(X!P!Q#r!Q!['b![#O#r#P;'S#r;'S;=`$Z<%lO#rU(^WqSOt#ruw#rx!Q#r!Q![(v![#O#r#P;'S#r;'S;=`$Z<%lO#rU(}WqSfQOt#ruw#rx!Q#r!Q![(v![#O#r#P;'S#r;'S;=`$Z<%lO#rU)lWqSOt#ruw#rx!P#r!P!Q*U!Q#O#r#P;'S#r;'S;=`$Z<%lO#rU*Z^qSOY+VYZ#rZt+Vtu,Yuw+Vwx,Yx!P+V!P!Q#r!Q!}+V!}#O0{#O#P.h#P;'S+V;'S;=`1|<%lO+VU+^^qSuQOY+VYZ#rZt+Vtu,Yuw+Vwx,Yx!P+V!P!Q.}!Q!}+V!}#O0{#O#P.h#P;'S+V;'S;=`1|<%lO+VQ,_XuQOY,YZ!P,Y!P!Q,z!Q!},Y!}#O-i#O#P.h#P;'S,Y;'S;=`.w<%lO,YQ,}P!P!Q-QQ-VUuQ#Z#[-Q#]#^-Q#a#b-Q#g#h-Q#i#j-Q#m#n-QQ-lVOY-iZ#O-i#O#P.R#P#Q,Y#Q;'S-i;'S;=`.b<%lO-iQ.USOY-iZ;'S-i;'S;=`.b<%lO-iQ.eP;=`<%l-iQ.kSOY,YZ;'S,Y;'S;=`.w<%lO,YQ.zP;=`<%l,YU/SWqSOt#ruw#rx!P#r!P!Q/l!Q#O#r#P;'S#r;'S;=`$Z<%lO#rU/sbqSuQOt#ruw#rx#O#r#P#Z#r#Z#[/l#[#]#r#]#^/l#^#a#r#a#b/l#b#g#r#g#h/l#h#i#r#i#j/l#j#m#r#m#n/l#n;'S#r;'S;=`$Z<%lO#rU1Q[qSOY0{YZ#rZt0{tu-iuw0{wx-ix#O0{#O#P.R#P#Q+V#Q;'S0{;'S;=`1v<%lO0{U1yP;=`<%l0{U2PP;=`<%l+VU2ZUqSnQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#r~2rO!p~U2yUqSwQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU3bYqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#rU4XUyQqSOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU4pZqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#U5c#U#o3]#o;'S#r;'S;=`$Z<%lO#rU5h[qSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#`3]#`#a6^#a#o3]#o;'S#r;'S;=`$Z<%lO#rU6c[qSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#g3]#g#h7X#h#o3]#o;'S#r;'S;=`$Z<%lO#rU7^[qSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#X3]#X#Y8S#Y#o3]#o;'S#r;'S;=`$Z<%lO#rU8ZYtQqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^9QY!qWqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^9wY!sWqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^:n[!rWqSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#f3]#f#g;d#g#o3]#o;'S#r;'S;=`$Z<%lO#rU;i[qSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#i3]#i#j7X#j#o3]#o;'S#r;'S;=`$Z<%lO#rU<fUzQqSOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#r~<}O!v~",
+  tokenData: ">i~RzOX#uXY$dYZ$}Zp#upq$dqs#ust%htu'Puw#uwx'Uxy'Zyz'tz{#u{|(_|}#u}!O(_!O!P#u!P!Q+R!Q![(|![!]3n!]!^$}!^#O#u#O#P4X#P#R#u#R#S4^#S#T#u#T#Y4w#Y#Z6V#Z#b4w#b#c:e#c#f4w#f#g;[#g#h4w#h#i<R#i#o4w#o#p#u#p#q=y#q;'S#u;'S;=`$^<%l~#u~O#u~~>dS#zUqSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uS$aP;=`<%l#u^$kUqS!_YOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU%UUqS!kQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u^%oZqS!`YOY%hYZ#uZt%htu&buw%hwx&bx#O%h#O#P&b#P;'S%h;'S;=`&y<%lO%hY&gS!`YOY&bZ;'S&b;'S;=`&s<%lO&bY&vP;=`<%l&b^&|P;=`<%l%h~'UO!p~~'ZO!n~U'bUqS!hQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU'{UqS!mQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU(dWqSOt#uuw#ux!Q#u!Q![(|![#O#u#P;'S#u;'S;=`$^<%lO#uU)TYqSfQOt#uuw#ux!O#u!O!P)s!P!Q#u!Q![(|![#O#u#P;'S#u;'S;=`$^<%lO#uU)xWqSOt#uuw#ux!Q#u!Q![*b![#O#u#P;'S#u;'S;=`$^<%lO#uU*iWqSfQOt#uuw#ux!Q#u!Q![*b![#O#u#P;'S#u;'S;=`$^<%lO#uU+WWqSOt#uuw#ux!P#u!P!Q+p!Q#O#u#P;'S#u;'S;=`$^<%lO#uU+u^qSOY,qYZ#uZt,qtu-tuw,qwx-tx!P,q!P!Q#u!Q!},q!}#O2g#O#P0S#P;'S,q;'S;=`3h<%lO,qU,x^qSuQOY,qYZ#uZt,qtu-tuw,qwx-tx!P,q!P!Q0i!Q!},q!}#O2g#O#P0S#P;'S,q;'S;=`3h<%lO,qQ-yXuQOY-tZ!P-t!P!Q.f!Q!}-t!}#O/T#O#P0S#P;'S-t;'S;=`0c<%lO-tQ.iP!P!Q.lQ.qUuQ#Z#[.l#]#^.l#a#b.l#g#h.l#i#j.l#m#n.lQ/WVOY/TZ#O/T#O#P/m#P#Q-t#Q;'S/T;'S;=`/|<%lO/TQ/pSOY/TZ;'S/T;'S;=`/|<%lO/TQ0PP;=`<%l/TQ0VSOY-tZ;'S-t;'S;=`0c<%lO-tQ0fP;=`<%l-tU0nWqSOt#uuw#ux!P#u!P!Q1W!Q#O#u#P;'S#u;'S;=`$^<%lO#uU1_bqSuQOt#uuw#ux#O#u#P#Z#u#Z#[1W#[#]#u#]#^1W#^#a#u#a#b1W#b#g#u#g#h1W#h#i#u#i#j1W#j#m#u#m#n1W#n;'S#u;'S;=`$^<%lO#uU2l[qSOY2gYZ#uZt2gtu/Tuw2gwx/Tx#O2g#O#P/m#P#Q,q#Q;'S2g;'S;=`3b<%lO2gU3eP;=`<%l2gU3kP;=`<%l,qU3uUqSnQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u~4^O!q~U4eUqSwQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU4|YqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#uU5sUyQqSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU6[ZqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#U6}#U#o4w#o;'S#u;'S;=`$^<%lO#uU7S[qSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#`4w#`#a7x#a#o4w#o;'S#u;'S;=`$^<%lO#uU7}[qSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#g4w#g#h8s#h#o4w#o;'S#u;'S;=`$^<%lO#uU8x[qSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#X4w#X#Y9n#Y#o4w#o;'S#u;'S;=`$^<%lO#uU9uYtQqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^:lY!rWqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^;cY!tWqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^<Y[!sWqSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#f4w#f#g=O#g#o4w#o;'S#u;'S;=`$^<%lO#uU=T[qSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#i4w#i#j8s#j#o4w#o;'S#u;'S;=`$^<%lO#uU>QUzQqSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u~>iO!w~",
-  tokenizers: [operatorTokenizer, 1, 2, 3, tokenizer, new LocalTokenGroup("[~RP!O!PU~ZO!c~~", 11)],
+  tokenizers: [operatorTokenizer, 1, 2, 3, tokenizer, new LocalTokenGroup("[~RP!O!PU~ZO!d~~", 11)],
   topRules: {"Program":[0,18]},
   specialized: [{term: 13, get: (value: any, stack: any) => (specializeKeyword(value, stack) << 1), external: specializeKeyword},{term: 13, get: (value: keyof typeof spec_Identifier) => spec_Identifier[value] || -1}],
-  tokenPrec: 759
+  tokenPrec: 776
 })

src/parser/tests/basics.test.ts

@@ -30,6 +30,204 @@ describe('Identifier', () => {
       FunctionCallOrIdentifier
         Identifier moo-😊-34`)
   })
+
+  test('parses mathematical unicode symbols like 𝜋 as identifiers', () => {
+    expect('𝜋').toMatchTree(`
+      FunctionCallOrIdentifier
+        Identifier 𝜋`)
+  })
+})
+
+describe('Unicode Symbol Support', () => {
+  describe('Emoji (currently supported)', () => {
+    test('Basic Emoticons (U+1F600-U+1F64F)', () => {
+      expect('😀').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 😀`)
+
+      expect('😊-counter').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 😊-counter`)
+    })
+
+    test('Miscellaneous Symbols and Pictographs (U+1F300-U+1F5FF)', () => {
+      expect('🌍').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🌍`)
+
+      expect('🔥-handler').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🔥-handler`)
+    })
+
+    test('Transport and Map Symbols (U+1F680-U+1F6FF)', () => {
+      expect('🚀').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🚀`)
+
+      expect('🚀-launch').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🚀-launch`)
+    })
+
+    test('Regional Indicator Symbols / Flags (U+1F1E6-U+1F1FF)', () => {
+      // Note: Flags are typically two regional indicators combined
+      expect('🇺').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🇺`)
+    })
+
+    test('Supplemental Symbols and Pictographs (U+1F900-U+1F9FF)', () => {
+      expect('🤖').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🤖`)
+
+      expect('🦀-lang').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 🦀-lang`)
+    })
+
+    test('Dingbats (U+2700-U+27BF)', () => {
+      expect('✂').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ✂`)
+
+      expect('✨-magic').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ✨-magic`)
+    })
+
+    test('Miscellaneous Symbols (U+2600-U+26FF)', () => {
+      expect('⚡').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ⚡`)
+
+      expect('☀-bright').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ☀-bright`)
+    })
+  })
+
+  describe('Greek Letters (not currently supported)', () => {
+    test('Greek lowercase alpha α (U+03B1)', () => {
+      expect('α').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier α`)
+    })
+
+    test('Greek lowercase beta β (U+03B2)', () => {
+      expect('β').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier β`)
+    })
+
+    test('Greek lowercase lambda λ (U+03BB)', () => {
+      expect('λ').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier λ`)
+    })
+
+    test('Greek lowercase pi π (U+03C0)', () => {
+      // Note: This is different from mathematical pi 𝜋
+      expect('π').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier π`)
+    })
+  })
+
+  describe('Mathematical Alphanumeric Symbols (not currently supported)', () => {
+    test('Mathematical italic small pi 𝜋 (U+1D70B)', () => {
+      expect('𝜋').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 𝜋`)
+    })
+
+    test('Mathematical bold small x 𝐱 (U+1D431)', () => {
+      expect('𝐱').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 𝐱`)
+    })
+
+    test('Mathematical script capital F 𝓕 (U+1D4D5)', () => {
+      expect('𝓕').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 𝓕`)
+    })
+  })
+
+  describe('Mathematical Operators (not currently supported)', () => {
+    test('Infinity symbol ∞ (U+221E)', () => {
+      expect('∞').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ∞`)
+    })
+
+    test('Sum symbol ∑ (U+2211)', () => {
+      expect('∑').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ∑`)
+    })
+
+    test('Integral symbol ∫ (U+222B)', () => {
+      expect('∫').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ∫`)
+    })
+  })
+
+  describe('Superscripts and Subscripts (not currently supported)', () => {
+    test('Superscript two ² (U+00B2)', () => {
+      expect('x²').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier x²`)
+    })
+
+    test('Subscript two ₂ (U+2082)', () => {
+      expect('h₂o').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier h₂o`)
+    })
+  })
+
+  describe('Arrows (not currently supported)', () => {
+    test('Rightward arrow → (U+2192)', () => {
+      expect('→').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier →`)
+    })
+
+    test('Leftward arrow ← (U+2190)', () => {
+      expect('←').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ←`)
+    })
+
+    test('Double rightward arrow ⇒ (U+21D2)', () => {
+      expect('⇒').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier ⇒`)
+    })
+  })
+
+  describe('CJK Symbols (not currently supported)', () => {
+    test('Hiragana あ (U+3042)', () => {
+      expect('あ').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier あ`)
+    })
+
+    test('Katakana カ (U+30AB)', () => {
+      expect('カ').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier カ`)
+    })
+
+    test('CJK Unified Ideograph 中 (U+4E2D)', () => {
+      expect('中').toMatchTree(`
+        FunctionCallOrIdentifier
+          Identifier 中`)
+    })
+  })
 })
 
 describe('Parentheses', () => {
@@ -349,3 +547,27 @@ describe('DotGet whitespace sensitivity', () => {
     expect('readme.txt').toMatchTree(`Word readme.txt`)
   })
 })
+
+describe('Comments', () => {
+  test('are barely there', () => {
+    expect(`x = 5 # one banana\ny = 2 # two bananas`).toMatchTree(`
+      Assign
+        AssignableIdentifier x
+        Eq =
+        Number 5
+      Assign
+        AssignableIdentifier y
+        Eq =
+        Number 2`)
+
+    expect('# some comment\nbasename = 5 # very astute\n basename / prop\n# good info').toMatchTree(`
+      Assign
+        AssignableIdentifier basename
+        Eq =
+        Number 5
+      BinOp
+        Identifier basename
+        Slash /
+        Identifier prop`)
+  })
+})

src/parser/tests/control-flow.test.ts

@@ -4,7 +4,7 @@ import '../shrimp.grammar' // Importing this so changes cause it to retest!
 
 describe('if/elseif/else', () => {
   test('parses single line if', () => {
-    expect(`if y = 1: 'cool'`).toMatchTree(`
+    expect(`if y = 1: 'cool' end`).toMatchTree(`
       IfExpr
         keyword if
         ConditionalOp
@@ -12,12 +12,13 @@ describe('if/elseif/else', () => {
           Eq =
           Number 1
         colon :
-        ThenBlock
+        SingleLineThenBlock
           String
             StringFragment cool
+        keyword end
     `)
 
-    expect('a = if x: 2').toMatchTree(`
+    expect('a = if x: 2 end').toMatchTree(`
       Assign
         AssignableIdentifier a
         Eq =
@@ -25,8 +26,9 @@ describe('if/elseif/else', () => {
           keyword if
           Identifier x
           colon :
-          ThenBlock
+          SingleLineThenBlock
             Number 2
+          keyword end
     `)
   })
 
@@ -138,7 +140,7 @@ describe('if/elseif/else', () => {
   })
 
   test('does not parse identifiers that start with if', () => {
-    expect('iffy = if true: 2').toMatchTree(`
+    expect('iffy = if true: 2 end').toMatchTree(`
       Assign
         AssignableIdentifier iffy
         Eq =
@@ -146,8 +148,9 @@ describe('if/elseif/else', () => {
           keyword if
           Boolean true
           colon :
-          ThenBlock
+          SingleLineThenBlock
             Number 2
+          keyword end
     `)
   })
 })

src/parser/tests/multiline.test.ts

@@ -71,4 +71,20 @@ end
         keyword end
     `)
   })
+
+  test('multiline with empty lines', () => {
+    expect(`
+      do:
+          2
+   
+      end
+      `).toMatchTree(`
+      FunctionDef
+        keyword do
+        Params 
+        colon :
+        Number 2
+        keyword end
+    `)
+  })
 })

src/parser/tokenizer.ts

@@ -19,7 +19,7 @@ export const tokenizer = new ExternalTokenizer(
     // Don't consume things that start with - or + followed by a digit (negative/positive numbers)
     if ((ch === 45 /* - */ || ch === 43) /* + */ && isDigit(input.peek(1))) return
 
-    const isValidStart = isLowercaseLetter(ch) || isEmoji(ch)
+    const isValidStart = isLowercaseLetter(ch) || isEmojiOrUnicode(ch)
     const canBeWord = stack.canShift(Word)
 
     // Consume all word characters, tracking if it remains a valid identifier
@@ -111,8 +111,8 @@ const consumeWordToken = (
       if (!isWordChar(nextCh)) break
     }
 
-    // Track identifier validity: must be lowercase, digit, dash, or emoji
+    // Track identifier validity: must be lowercase, digit, dash, or emoji/unicode
-    if (!isLowercaseLetter(ch) && !isDigit(ch) && ch !== 45 /* - */ && !isEmoji(ch)) {
+    if (!isLowercaseLetter(ch) && !isDigit(ch) && ch !== 45 /* - */ && !isEmojiOrUnicode(ch)) {
       if (!canBeWord) break
       isValidIdentifier = false
     }
@@ -222,7 +222,7 @@ const getFullCodePoint = (input: InputStream, pos: number): number => {
   return ch
 }
 
-const isEmoji = (ch: number): boolean => {
+const isEmojiOrUnicode = (ch: number): boolean => {
   return (
     // Basic Emoticons
     (ch >= 0x1f600 && ch <= 0x1f64f) ||
@@ -247,7 +247,25 @@ const isEmoji = (ch: number): boolean => {
     // Additional miscellaneous items
     (ch >= 0x238c && ch <= 0x2454) ||
     // Combining Diacritical Marks for Symbols
-    (ch >= 0x20d0 && ch <= 0x20ff)
+    (ch >= 0x20d0 && ch <= 0x20ff) ||
+    // Latin-1 Supplement (includes ², ³, ¹ and other special chars)
+    (ch >= 0x00a0 && ch <= 0x00ff) ||
+    // Greek and Coptic (U+0370-U+03FF)
+    (ch >= 0x0370 && ch <= 0x03ff) ||
+    // Mathematical Alphanumeric Symbols (U+1D400-U+1D7FF)
+    (ch >= 0x1d400 && ch <= 0x1d7ff) ||
+    // Mathematical Operators (U+2200-U+22FF)
+    (ch >= 0x2200 && ch <= 0x22ff) ||
+    // Superscripts and Subscripts (U+2070-U+209F)
+    (ch >= 0x2070 && ch <= 0x209f) ||
+    // Arrows (U+2190-U+21FF)
+    (ch >= 0x2190 && ch <= 0x21ff) ||
+    // Hiragana (U+3040-U+309F)
+    (ch >= 0x3040 && ch <= 0x309f) ||
+    // Katakana (U+30A0-U+30FF)
+    (ch >= 0x30a0 && ch <= 0x30ff) ||
+    // CJK Unified Ideographs (U+4E00-U+9FFF)
+    (ch >= 0x4e00 && ch <= 0x9fff)
   )
 }