prelude tests

Reviewed-on: #10

.gitmodules

@@ -1,3 +0,0 @@
-[submodule "packages/ReefVM"]
-	path = packages/ReefVM
-	url = git@54.219.130.253:defunkt/ReefVM.git

bun.lock

@@ -9,7 +9,7 @@
         "bun-plugin-tailwind": "^0.0.15",
         "codemirror": "^6.0.2",
         "hono": "^4.9.8",
-        "reefvm": "workspace:*",
+        "reefvm": "git+https://git.nose.space/defunkt/reefvm",
         "tailwindcss": "^4.1.11",
       },
       "devDependencies": {
@@ -18,15 +18,6 @@
         "@types/bun": "latest",
       },
     },
-    "packages/ReefVM": {
-      "name": "reefvm",
-      "devDependencies": {
-        "@types/bun": "latest",
-      },
-      "peerDependencies": {
-        "typescript": "^5",
-      },
-    },
   },
   "packages": {
     "@codemirror/autocomplete": ["@codemirror/autocomplete@6.19.0", "", { "dependencies": { "@codemirror/language": "^6.0.0", "@codemirror/state": "^6.0.0", "@codemirror/view": "^6.17.0", "@lezer/common": "^1.0.0" } }, "sha512-61Hfv3cF07XvUxNeC3E7jhG8XNi1Yom1G0lRC936oLnlF+jrbrv8rc/J98XlYzcsAoTVupfsf5fLej1aI8kyIg=="],
@@ -71,7 +62,7 @@
 
     "hono": ["hono@4.9.8", "", {}, "sha512-JW8Bb4RFWD9iOKxg5PbUarBYGM99IcxFl2FPBo2gSJO11jjUDqlP1Bmfyqt8Z/dGhIQ63PMA9LdcLefXyIasyg=="],
 
-    "reefvm": ["reefvm@workspace:packages/ReefVM"],
+    "reefvm": ["reefvm@git+https://git.nose.space/defunkt/reefvm#97b6722a113417398a1c47d583bfe07a906f87a0", { "peerDependencies": { "typescript": "^5" } }, "97b6722a113417398a1c47d583bfe07a906f87a0"],
 
     "style-mod": ["style-mod@4.1.2", "", {}, "sha512-wnD1HyVqpJUI2+eKZ+eo1UwghftP6yuFheBqqe+bWCotBjC2K1YnteJILRMs3SM4V/0dLEW1SC27MWP5y+mwmw=="],
 
@@ -82,9 +73,5 @@
     "undici-types": ["undici-types@7.12.0", "", {}, "sha512-goOacqME2GYyOZZfb5Lgtu+1IDmAlAEu5xnD3+xTzS10hT0vzpf0SPjkXwAw9Jm+4n/mQGDP3LO8CPbYROeBfQ=="],
 
     "w3c-keyname": ["w3c-keyname@2.2.8", "", {}, "sha512-dpojBhNsCNN7T82Tm7k26A6G9ML3NkhDsnw9n/eoxSRlVBB4CEtIQ/KTCLI2Fwf3ataSXRhYFkQi3SlnFwPvPQ=="],
-
-    "reefvm/@types/bun": ["@types/bun@1.2.23", "", { "dependencies": { "bun-types": "1.2.23" } }, "sha512-le8ueOY5b6VKYf19xT3McVbXqLqmxzPXHsQT/q9JHgikJ2X22wyTW3g3ohz2ZMnp7dod6aduIiq8A14Xyimm0A=="],
-
-    "reefvm/@types/bun/bun-types": ["bun-types@1.2.23", "", { "dependencies": { "@types/node": "*" }, "peerDependencies": { "@types/react": "^19" } }, "sha512-R9f0hKAZXgFU3mlrA0YpE/fiDvwV0FT9rORApt2aQVWSuJDzZOyB5QLc0N/4HF57CS8IXJ6+L5E4W1bW6NS2Aw=="],
   }
 }

package.json

@@ -13,7 +13,7 @@
     "update-reef": "cd packages/ReefVM && git pull origin main"
   },
   "dependencies": {
-    "reefvm": "workspace:*",
+    "reefvm": "git+https://git.nose.space/defunkt/reefvm",
     "@codemirror/view": "^6.38.3",
     "@lezer/generator": "^1.8.0",
     "bun-plugin-tailwind": "^0.0.15",
@@ -31,4 +31,4 @@
     "singleQuote": true,
     "printWidth": 100
   }
-}
+}

packages/ReefVM

@@ -1 +0,0 @@
-Subproject commit 17d846b99910a46fc1c7ab98aa41ca8afbc14097

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/compiler/utils.ts

@@ -70,9 +70,9 @@ export const getFunctionDefParts = (node: SyntaxNode, input: string) => {
   }
 
   const paramNames = getAllChildren(paramsNode).map((param) => {
-    if (param.type.id !== terms.AssignableIdentifier) {
+    if (param.type.id !== terms.Identifier) {
       throw new CompilerError(
-        `FunctionDef params must be AssignableIdentifiers, got ${param.type.name}`,
+        `FunctionDef params must be Identifier, got ${param.type.name}`,
         param.from,
         param.to
       )
@@ -219,9 +219,9 @@ export const getDotGetParts = (node: SyntaxNode, input: string) => {
     )
   }
 
-  if (property.type.id !== terms.Identifier) {
+  if (property.type.id !== terms.Identifier && property.type.id !== terms.Number) {
     throw new CompilerError(
-      `DotGet property must be an Identifier, got ${property.type.name}`,
+      `DotGet property must be an Identifier or Number, got ${property.type.name}`,
       property.from,
       property.to
     )

src/parser/scopeTracker.ts

@@ -57,26 +57,30 @@ const readIdentifierText = (input: InputStream, start: number, end: number): str
   return text
 }
 
+let inParams = false
+
 export const trackScope = new ContextTracker<TrackerContext>({
   start: new TrackerContext(new Scope(null, new Set())),
 
   shift(context, term, stack, input) {
-    if (term !== terms.AssignableIdentifier) return context
+    if (term == terms.Do) inParams = true
 
-    const text = readIdentifierText(input, input.pos, stack.pos)
+    if (term === terms.AssignableIdentifier) {
-    return new TrackerContext(context.scope, [...context.pendingIds, text])
+      const text = readIdentifierText(input, input.pos, stack.pos)
+      return new TrackerContext(Scope.add(context.scope, text), context.pendingIds)
+    }
+
+    if (inParams && term === terms.Identifier) {
+      const text = readIdentifierText(input, input.pos, stack.pos)
+      return new TrackerContext(context.scope, [...context.pendingIds, text])
+    }
+
+    return context
   },
 
   reduce(context, term) {
-    // Add assignment variable to scope
-    if (term === terms.Assign) {
-      const varName = context.pendingIds.at(-1)
-      if (!varName) return context
-      return new TrackerContext(Scope.add(context.scope, varName), context.pendingIds.slice(0, -1))
-    }
-
-    // Push new scope and add all parameters
     if (term === terms.Params) {
+      inParams = false
       let newScope = context.scope.push()
       if (context.pendingIds.length > 0) {
         newScope = Scope.add(newScope, ...context.pendingIds)

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"] { ":" }
@@ -28,6 +29,7 @@
 }
 
 @external tokens tokenizer from "./tokenizer" { Identifier, AssignableIdentifier, Word, IdentifierBeforeDot }
+@external specialize {Identifier} specializeKeyword from "./tokenizer" { Do }
 
 @precedence {
   pipe @left,
@@ -46,7 +48,6 @@ item {
 consumeToTerminator {
   PipeExpr |
   ambiguousFunctionCall |
-  DotGet |
   IfExpr |
   FunctionDef |
   Assign |
@@ -63,7 +64,7 @@ pipeOperand {
 }
 
 FunctionCallOrIdentifier {
-  Identifier
+  DotGet | Identifier
 }
 
 ambiguousFunctionCall {
@@ -71,7 +72,7 @@ ambiguousFunctionCall {
 }
 
 FunctionCall {
-  Identifier arg+
+  (DotGet | Identifier | ParenExpr) arg+
 }
 
 arg {
@@ -92,11 +93,11 @@ FunctionDef {
 }
 
 singleLineFunctionDef {
-  @specialize[@name=keyword]<Identifier, "do"> Params colon consumeToTerminator @specialize[@name=keyword]<Identifier, "end">
+  Do Params colon consumeToTerminator @specialize[@name=keyword]<Identifier, "end">
 }
 
 multilineFunctionDef {
-  @specialize[@name=keyword]<Identifier, "do"> Params colon newlineOrSemicolon block @specialize[@name=keyword]<Identifier, "end">
+  Do Params colon newlineOrSemicolon block @specialize[@name=keyword]<Identifier, "end">
 }
 
 IfExpr {
@@ -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 |
@@ -135,7 +140,7 @@ ConditionalOp {
 }
 
 Params {
-  AssignableIdentifier*
+  Identifier*
 }
 
 Assign {
@@ -164,7 +169,7 @@ expression {
 
 @skip {} {
   DotGet {
-    IdentifierBeforeDot dot Identifier
+    IdentifierBeforeDot dot (Number | Identifier)
   }
 
   String { "'" stringContent* "'" }
@@ -199,5 +204,5 @@ expressionWithoutIdentifier {
 }
 
 block {
-  (consumeToTerminator newlineOrSemicolon)*
+  (consumeToTerminator? newlineOrSemicolon)*
 }

src/parser/shrimp.terms.ts

@@ -16,31 +16,33 @@ export const
   AssignableIdentifier = 14,
   Word = 15,
   IdentifierBeforeDot = 16,
-  Program = 17,
+  Do = 17,
-  PipeExpr = 18,
+  Program = 18,
-  FunctionCall = 19,
+  PipeExpr = 19,
-  PositionalArg = 20,
+  FunctionCall = 20,
-  ParenExpr = 21,
+  DotGet = 21,
-  FunctionCallOrIdentifier = 22,
+  Number = 22,
-  BinOp = 23,
+  ParenExpr = 23,
-  ConditionalOp = 24,
+  FunctionCallOrIdentifier = 24,
-  FunctionDef = 25,
+  BinOp = 25,
+  String = 26,
+  StringFragment = 27,
+  Interpolation = 28,
+  EscapeSeq = 29,
+  Boolean = 30,
+  Regex = 31,
+  Null = 32,
+  ConditionalOp = 33,
+  FunctionDef = 34,
+  Params = 35,
+  colon = 36,
   keyword = 50,
-  Params = 27,
+  PositionalArg = 38,
-  colon = 28,
-  String = 30,
-  StringFragment = 31,
-  Interpolation = 32,
-  EscapeSeq = 33,
-  Number = 34,
-  Boolean = 35,
-  Regex = 36,
-  Null = 37,
-  DotGet = 38,
   Underscore = 39,
   NamedArg = 40,
   NamedArgPrefix = 41,
   IfExpr = 43,
+  SingleLineThenBlock = 45,
   ThenBlock = 46,
   ElseIfExpr = 47,
   ElseExpr = 49,

src/parser/shrimp.ts

@@ -1,27 +1,27 @@
 // This file was generated by lezer-generator. You probably shouldn't edit it.
 import {LRParser, LocalTokenGroup} from "@lezer/lr"
 import {operatorTokenizer} from "./operatorTokenizer"
-import {tokenizer} from "./tokenizer"
+import {tokenizer, specializeKeyword} from "./tokenizer"
 import {trackScope} from "./scopeTracker"
 import {highlighting} from "./highlight"
-const spec_Identifier = {__proto__:null,do:52, end:58, null:74, if:88, elseif:96, else:100}
+const spec_Identifier = {__proto__:null,null:64, end:74, if:88, elseif:96, else:100}
 export const parser = LRParser.deserialize({
   version: 14,
-  states: ".jQVQbOOO#XQcO'#CrO$RQRO'#CsO$aQcO'#DmO$xQbO'#DsOOQ`'#Cu'#CuO%QQbO'#CqO%rOSO'#CzOOQa'#Dq'#DqO&QOpO'#DSO&VQcO'#DpOOQ`'#Dn'#DnO&nQbO'#DmO&|QbO'#EQOOQ`'#DX'#DXO'kQRO'#DaOOQ`'#Dm'#DmO'pQQO'#DlOOQ`'#Dl'#DlOOQ`'#Db'#DbQVQbOOOOQa'#Dp'#DpOOQ`'#Cp'#CpO'xQbO'#DUOOQ`'#Do'#DoOOQ`'#Dc'#DcO(SQbO,59ZO&|QbO,59_O&|QbO,59_OOQ`'#Dd'#DdO(pQbO'#CwO(xQQO,5:_O)iQRO'#CsO)yQRO,59]O*[QRO,59]O*VQQO,59]O+VQQO,59]O+_QbO'#C|O+gQWO'#C}OOOO'#Dy'#DyOOOO'#Df'#DfO+{OSO,59fOOQa,59f,59fO,ZO`O,59nO,`QbO'#DgO,eQbO,59YO,vQRO,5:lO,}QQO,5:lO-SQbO,59{OOQ`,5:W,5:WOOQ`-E7`-E7`OOQ`,59p,59pOOQ`-E7a-E7aOOQa1G.y1G.yO-^QcO1G.yOOQ`-E7b-E7bO-xQbO1G/yO&|QbO,59`O&|QbO,59`OOQa1G.w1G.wOOOO,59h,59hOOOO,59i,59iOOOO-E7d-E7dOOQa1G/Q1G/QOOQa1G/Y1G/YO!QQbO'#CrOOQ`,5:R,5:ROOQ`-E7e-E7eO.VQbO1G0WOOQ`1G/g1G/gO.dQbO7+%eO.iQbO7+%fOOQO1G.z1G.zO.vQRO1G.zOOQ`'#DZ'#DZOOQ`7+%r7+%rO/QQbO7+%sOOQ`<<IP<<IPO/eQQO'#DeO/jQbO'#DvO/}QbO<<IQOOQ`'#D['#D[O0SQbO<<I_OOQ`,5:P,5:POOQ`-E7c-E7cOOQ`AN>lAN>lO&|QbO'#D]OOQ`'#Dh'#DhO0_QbOAN>yO0jQQO'#D_OOQ`AN>yAN>yO0oQbOAN>yO0tQRO,59wO0{QQO,59wOOQ`-E7f-E7fOOQ`G24eG24eO1QQbOG24eO1VQQO,59yO1[QQO1G/cOOQ`LD*PLD*PO.iQbO1G/eO/QQbO7+$}OOQ`7+%P7+%POOQ`<<Hi<<Hi",
+  states: "/SQYQbOOO!TOpO'#CqO#aQcO'#CtO$ZOSO'#CvO%aQcO'#DsOOQa'#Ds'#DsO&gQcO'#DrO'OQRO'#CuO'^QcO'#DnO'uQbO'#D{OOQ`'#DO'#DOO'}QbO'#CsOOQ`'#Do'#DoO(oQbO'#DnO(}QbO'#EROOQ`'#DX'#DXO)lQRO'#DaOOQ`'#Dn'#DnO)qQQO'#DmOOQ`'#Dm'#DmOOQ`'#Db'#DbQYQbOOO)yObO,59]OOQa'#Dr'#DrOOQ`'#DS'#DSO*RQbO'#DUOOQ`'#EQ'#EQOOQ`'#Df'#DfO*]QbO,59[O*pQbO'#CxO*xQWO'#CyOOOO'#Du'#DuOOOO'#Dc'#DcO+^OSO,59bOOQa,59b,59bO(}QbO,59aO(}QbO,59aOOQ`'#Dd'#DdO+lQbO'#DPO+tQQO,5:gO+yQRO,59_O-`QRO'#CuO-pQRO,59_O-|QQO,59_O.RQQO,59_O.ZQbO'#DgO.fQbO,59ZO.wQRO,5:mO/OQQO,5:mO/TQbO,59{OOQ`,5:X,5:XOOQ`-E7`-E7`OOQa1G.w1G.wOOQ`,59p,59pOOQ`-E7d-E7dOOOO,59d,59dOOOO,59e,59eOOOO-E7a-E7aOOQa1G.|1G.|OOQa1G.{1G.{O/_QcO1G.{OOQ`-E7b-E7bO/yQbO1G0ROOQa1G.y1G.yO(}QbO,59iO(}QbO,59iO!YQbO'#CtO$iQbO'#CpOOQ`,5:R,5:ROOQ`-E7e-E7eO0WQbO1G0XOOQ`1G/g1G/gO0eQbO7+%mO0jQbO7+%nOOQO1G/T1G/TO0zQRO1G/TOOQ`'#DZ'#DZO1UQbO7+%sO1ZQbO7+%tOOQ`<<IX<<IXOOQ`'#De'#DeO1qQQO'#DeO1vQbO'#EOO2^QbO<<IYOOQ`<<I_<<I_OOQ`'#D['#D[O2cQbO<<I`OOQ`,5:P,5:POOQ`-E7c-E7cOOQ`AN>tAN>tO(}QbO'#D]OOQ`'#Dh'#DhO2nQbOAN>zO2yQQO'#D_OOQ`AN>zAN>zO3OQbOAN>zO3TQRO,59wO3[QQO,59wOOQ`-E7f-E7fOOQ`G24fG24fO3aQbOG24fO3fQQO,59yO3kQQO1G/cOOQ`LD*QLD*QO0jQbO1G/eO1ZQbO7+$}OOQ`7+%P7+%POOQ`<<Hi<<Hi",
-  stateData: "1d~O!_OS~O]PO^_O_WO`XOjSOrWOsWOtWOuWO|]O!fUO!ibO!lVO~O]eO_WO`XOjSOrWOsWOtWOuWOwfOygO!fUO!lVOzfX!ifX!vfX!kfXmfX~OP!dXQ!dXR!dXS!dXT!dXU!dXV!dXW!dXX!dXY!dXZ!dX[!dX~P!QOPkOQkORlOSlO~OPkOQkORlOSlO!i!aX!v!aXm!aX~O^mOlkP~O]PO_WO`XOjSOrWOsWOtWOuWO!fUO!lVO~OowO!lzO!nuO!ovO~O!s{O~OP!dXQ!dXR!dXS!dX!i!aX!v!aXm!aX~Oz|O!i!aX!v!aXm!aX~O]eO_WO`XOrWOsWOtWOuWO!fUO!lVO~OV!QO~O!i!RO!v!RO~OjSOw!TO~P&|OjSOwfOygOzca!ica!vca!kcamca~P&|O^mOlkX~Ol!YO~OT![OU![OV!ZOW!ZOX!ZOY!ZOZ!ZO[!ZO~OPkOQkORlOSlO~P(}OPkOQkORlOSlO!k!]O~O!k!]OP!dXQ!dXR!dXS!dXT!dXU!dXV!dXW!dXX!dXY!dXZ!dX[!dX~Oz|O!k!]O~O]!^O!fUO~O!l!_O!n!_O!o!_O!p!_O!q!_O!r!_O~OowO!l!aO!nuO!ovO~O]!bO~O]!cO~Oz|O!iba!vba!kbamba~Ol!fO~P(}Ol!fO~O^_O|]O~P%QOPkOQkORgiSgi!igi!vgi!kgimgi~O^_O|]O!i!iO~P%QO^_O|]O!i!nO~P%QOm!oO~O^_O|]Om!jP~P%QO!khilhi~P(}O^_O|]Om!jP!Q!jP!S!jP~P%QO!i!uO~O^_O|]Om!jX!Q!jX!S!jX~P%QOm!wO~Om!|O!Q!xO!S!{O~Om#RO!Q!xO!S!{O~Ol#TO~Om#RO~Ol#UO~P(}Ol#UO~Om#VO~O!i#WO~O!i#XO~Ort~",
+  stateData: "3s~O!_OS!`OS~O]QO^`O_TO`POaXOfTOnTOoTOpTO|^O!eZO!hRO!qcO~O!dfO~O]gO_TO`POaXOfTOnTOoTOpTOwhOyiO!eZO!hROzhX!qhX!whX!shXuhX~OP!fXQ!fXR!fXS!fXT!fXU!fXV!fXW!fXX!fXY!fXZ!fX[!fX~P!YOkoO!hrO!jmO!knO~O]gO_TO`POaXOfTOnTOoTOpTOwhOyiO!eZO!hRO~OP!gXQ!gXR!gXS!gX!q!gX!w!gXT!gXU!gXV!gXW!gXX!gXY!gXZ!gX[!gX!s!gXu!gX~P$iOP!fXQ!fXR!fXS!fX!q!bX!w!bXu!bX~OPsOQsORtOStO~OPsOQsORtOStO!q!bX!w!bXu!bX~O]uOtsP~O]QO_TO`POaXOfTOnTOoTOpTO!eZO!hRO~Oz}O!q!bX!w!bXu!bX~O]gO_TO`POfTOnTOoTOpTO!eZO!hRO~OV!RO~O!q!SO!w!SO~O]!UOf!UO~OaXOw!VO~P(}Ozda!qda!wda!sdauda~P$iO]!XO!eZO~O!h!YO!j!YO!k!YO!l!YO!m!YO!n!YO~OkoO!h![O!jmO!knO~O]uOtsX~Ot!`O~O!s!aOP!fXQ!fXR!fXS!fXT!fXU!fXV!fXW!fXX!fXY!fXZ!fX[!fX~OT!cOU!cOV!bOW!bOX!bOY!bOZ!bO[!bO~OPsOQsORtOStO~P,tOPsOQsORtOStO!s!aO~Oz}O!s!aO~O]!dO`PO!eZO~Oz}O!qca!wca!scauca~Ot!hO~P,tOt!hO~O^`O|^O~P'}OPsOQsORiiSii!qii!wii!siiuii~O^`O|^O!q!kO~P'}O^`O|^O!q!pO~P'}Ou!qO~O^`O|^O!q!rOu!rP~P'}O!sqitqi~P,tOu!vO~O^`O|^O!q!rOu!rP!Q!rP!S!rP~P'}O!q!yO~O^`O|^O!q!rOu!rX!Q!rX!S!rX~P'}Ou!{O~Ou#QO!Q!|O!S#PO~Ou#VO!Q!|O!S#PO~Ot#XO~Ou#VO~Ot#YO~P,tOt#YO~Ou#ZO~O!q#[O~O!q#]O~Ofo~",
-  goto: "+q!vPPPPPPPPPPPPPPPPPP!w#W#f#k#W$V$l$xP%aPP%dP%{%{PPPP&PP#fPP&jP&v&y'SP'WP&j'^'d'k'q'z(Q(XPPP(_(c(w)Z)`*ZP*v*vP+XPP+aPPPPPP+e+ed`Od!Q!Y!f!i!n!q#W#XRsUiZOUd|!Q!Y!f!i!n!q#W#XVhPj!czWOPU]dgjkl!Q!Y!Z![!c!f!i!n!q!x#W#XR!^udROd!Q!Y!f!i!n!q#W#XQqUQ!VkR!WlQsUQ!P]Q!j![R#P!xd`Od!Q!Y!f!i!n!q#W#XUfPj!cQsUR!TgRoS{WOPU]dgjkl!Q!Y!Z![!c!f!i!n!q!x#W#XTwVydYOd!Q!Y!f!i!n!q#W#XgePU]gjkl!Z![!c!xe`Od!Q!Y!f!i!n!q#W#XR!m!fQ!t!nQ#Y#WR#Z#XT!y!t!zQ!}!tR#S!zQdOR!SdSjP!cR!UjQnSR!XnW!q!i!n#W#XR!v!qQyVR!`yS}[tR!e}Q!z!tR#Q!zTcOdSaOdQ!g!QQ!h!YQ!l!fZ!p!i!n!q#W#Xd[Od!Q!Y!f!i!n!q#W#XQtUR!d|ViPj!cdQOd!Q!Y!f!i!n!q#W#XUfPj!cQpUQ!O]Q!TgQ!VkQ!WlQ!j!ZQ!k![R#O!xdYOd!Q!Y!f!i!n!q#W#XdeP]gjkl!Z![!c!xRrUoTOPUdgj!Q!Y!c!f!i!n!q#W#XQ!r!iV!s!n#W#XTxVye^Od!Q!Y!f!i!n!q#W#X",
+  goto: ",`!wPPPPPPPPPPPPPPPPPPP!x#X#gP$V#X$x%_P%x%xPPP%|&Y&sPP&vP&vPP&}P'Z'^'gP'kP&}'q'w'}(T(^(g(nPPPP(t(x)^PP)p*mP+[PPPPP+`+`P+sP+{,S,SdaOe!R!`!h!k!p!t#[#]R{Zi[OZe}!R!`!h!k!p!t#[#]fQOZe!R!`!h!k!p!t#[#]hgQS^ilst!b!c!d!e!|R!d}fSOZe!R!`!h!k!p!t#[#]hTQS^ilst!b!c!d!e!|Q!XmR!e}dWOe!R!`!h!k!p!t#[#]QzZQ!]sR!^t!PTOQSZ^eilst!R!`!b!c!d!e!h!k!p!t!|#[#]ToRqQ{ZQ!Q^Q!l!cR#T!|daOe!R!`!h!k!p!t#[#]YhQSl!d!eQ{ZR!ViRwXZjQSl!d!eeaOe!R!`!h!k!p!t#[#]R!o!hQ!x!pQ#^#[R#_#]T!}!x#OQ#R!xR#W#OQeOR!TeQqRR!ZqQvXR!_vW!t!k!p#[#]R!z!tWlQS!d!eR!WlS!O]|R!g!OQ#O!xR#U#OTdOeSbOeQ!i!RQ!j!`Q!n!hZ!s!k!p!t#[#]d]Oe!R!`!h!k!p!t#[#]Q|ZR!f}dVOe!R!`!h!k!p!t#[#]YhQSl!d!eQyZQ!P^Q!ViQ!]sQ!^tQ!l!bQ!m!cR#S!|dUOe!R!`!h!k!p!t#[#]hgQS^ilst!b!c!d!e!|RxZTpRqsYOQSZeil!R!`!d!e!h!k!p!t#[#]Q!u!kV!w!p#[#]ZkQSl!d!ee_Oe!R!`!h!k!p!t#[#]",
-  nodeNames: "⚠ Star Slash Plus Minus And Or Eq Neq Lt Lte Gt Gte Identifier AssignableIdentifier Word IdentifierBeforeDot Program PipeExpr FunctionCall PositionalArg ParenExpr FunctionCallOrIdentifier BinOp ConditionalOp FunctionDef keyword Params colon keyword String StringFragment Interpolation EscapeSeq Number Boolean Regex Null DotGet 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 ParenExpr FunctionCallOrIdentifier BinOp String StringFragment Interpolation EscapeSeq Boolean Regex Null ConditionalOp FunctionDef Params colon keyword PositionalArg Underscore NamedArg NamedArgPrefix operator IfExpr keyword SingleLineThenBlock ThenBlock ElseIfExpr keyword ElseExpr keyword Assign",
-  maxTerm: 84,
+  maxTerm: 85,
   context: trackScope,
   nodeProps: [
-    ["closedBy", 28,"end"]
+    ["closedBy", 36,"end"]
   ],
   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;=`$Z<%l~#r~O#r~~<xS#wUoSOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rS$^P;=`<%l#r^$hUoS!_YOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU%RUoS!iQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#r~%jO!n~~%oO!l~U%vUoS!fQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU&aUoS!kQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU&xWoSOt#ruw#rx!Q#r!Q!['b![#O#r#P;'S#r;'S;=`$Z<%lO#rU'iYoSrQOt#ruw#rx!O#r!O!P(X!P!Q#r!Q!['b![#O#r#P;'S#r;'S;=`$Z<%lO#rU(^WoSOt#ruw#rx!Q#r!Q![(v![#O#r#P;'S#r;'S;=`$Z<%lO#rU(}WoSrQOt#ruw#rx!Q#r!Q![(v![#O#r#P;'S#r;'S;=`$Z<%lO#rU)lWoSOt#ruw#rx!P#r!P!Q*U!Q#O#r#P;'S#r;'S;=`$Z<%lO#rU*Z^oSOY+VYZ#rZt+Vtu,Yuw+Vwx,Yx!P+V!P!Q#r!Q!}+V!}#O0{#O#P.h#P;'S+V;'S;=`1|<%lO+VU+^^oStQOY+VYZ#rZt+Vtu,Yuw+Vwx,Yx!P+V!P!Q.}!Q!}+V!}#O0{#O#P.h#P;'S+V;'S;=`1|<%lO+VQ,_XtQOY,YZ!P,Y!P!Q,z!Q!},Y!}#O-i#O#P.h#P;'S,Y;'S;=`.w<%lO,YQ,}P!P!Q-QQ-VUtQ#Z#[-Q#]#^-Q#a#b-Q#g#h-Q#i#j-Q#m#n-QQ-lVOY-iZ#O-i#O#P.R#P#Q,Y#Q;'S-i;'S;=`.b<%lO-iQ.USOY-iZ;'S-i;'S;=`.b<%lO-iQ.eP;=`<%l-iQ.kSOY,YZ;'S,Y;'S;=`.w<%lO,YQ.zP;=`<%l,YU/SWoSOt#ruw#rx!P#r!P!Q/l!Q#O#r#P;'S#r;'S;=`$Z<%lO#rU/sboStQOt#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;=`$Z<%lO#rU1Q[oSOY0{YZ#rZt0{tu-iuw0{wx-ix#O0{#O#P.R#P#Q+V#Q;'S0{;'S;=`1v<%lO0{U1yP;=`<%l0{U2PP;=`<%l+VU2ZUoSlQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#r~2rO!o~U2yUoSwQOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU3bYoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#rU4XUyQoSOt#ruw#rx#O#r#P;'S#r;'S;=`$Z<%lO#rU4pZoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#U5c#U#o3]#o;'S#r;'S;=`$Z<%lO#rU5h[oSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#`3]#`#a6^#a#o3]#o;'S#r;'S;=`$Z<%lO#rU6c[oSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#g3]#g#h7X#h#o3]#o;'S#r;'S;=`$Z<%lO#rU7^[oSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#X3]#X#Y8S#Y#o3]#o;'S#r;'S;=`$Z<%lO#rU8ZYsQoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^9QY!pWoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^9wY!rWoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#o3]#o;'S#r;'S;=`$Z<%lO#r^:n[!qWoSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#f3]#f#g;d#g#o3]#o;'S#r;'S;=`$Z<%lO#rU;i[oSOt#ruw#rx!_#r!_!`4Q!`#O#r#P#T#r#T#i3]#i#j7X#j#o3]#o;'S#r;'S;=`$Z<%lO#rU<fUzQoSOt#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#zUkSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uS$aP;=`<%l#u^$kUkS!_YOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU%UUkS!qQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u^%oZkS!`YOY%hYZ#uZt%htu&buw%hwx&bx#O%h#O#P&b#P;'S%h;'S;=`&y<%lO%hY&gS!`YOY&bZ;'S&b;'S;=`&s<%lO&bY&vP;=`<%l&b^&|P;=`<%l%h~'UO!j~~'ZO!h~U'bUkS!eQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU'{UkS!sQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU(dWkSOt#uuw#ux!Q#u!Q![(|![#O#u#P;'S#u;'S;=`$^<%lO#uU)TYkSfQOt#uuw#ux!O#u!O!P)s!P!Q#u!Q![(|![#O#u#P;'S#u;'S;=`$^<%lO#uU)xWkSOt#uuw#ux!Q#u!Q![*b![#O#u#P;'S#u;'S;=`$^<%lO#uU*iWkSfQOt#uuw#ux!Q#u!Q![*b![#O#u#P;'S#u;'S;=`$^<%lO#uU+WWkSOt#uuw#ux!P#u!P!Q+p!Q#O#u#P;'S#u;'S;=`$^<%lO#uU+u^kSOY,qYZ#uZt,qtu-tuw,qwx-tx!P,q!P!Q#u!Q!},q!}#O2g#O#P0S#P;'S,q;'S;=`3h<%lO,qU,x^kSoQOY,qYZ#uZt,qtu-tuw,qwx-tx!P,q!P!Q0i!Q!},q!}#O2g#O#P0S#P;'S,q;'S;=`3h<%lO,qQ-yXoQOY-tZ!P-t!P!Q.f!Q!}-t!}#O/T#O#P0S#P;'S-t;'S;=`0c<%lO-tQ.iP!P!Q.lQ.qUoQ#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-tU0nWkSOt#uuw#ux!P#u!P!Q1W!Q#O#u#P;'S#u;'S;=`$^<%lO#uU1_bkSoQOt#uuw#ux#O#u#P#Z#u#Z#[1W#[#]#u#]#^1W#^#a#u#a#b1W#b#g#u#g#h1W#h#i#u#i#j1W#j#m#u#m#n1W#n;'S#u;'S;=`$^<%lO#uU2l[kSOY2gYZ#uZt2gtu/Tuw2gwx/Tx#O2g#O#P/m#P#Q,q#Q;'S2g;'S;=`3b<%lO2gU3eP;=`<%l2gU3kP;=`<%l,qU3uUkStQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u~4^O!k~U4eUkSwQOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU4|YkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#uU5sUyQkSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#uU6[ZkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#U6}#U#o4w#o;'S#u;'S;=`$^<%lO#uU7S[kSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#`4w#`#a7x#a#o4w#o;'S#u;'S;=`$^<%lO#uU7}[kSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#g4w#g#h8s#h#o4w#o;'S#u;'S;=`$^<%lO#uU8x[kSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#X4w#X#Y9n#Y#o4w#o;'S#u;'S;=`$^<%lO#uU9uYnQkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^:lY!lWkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^;cY!nWkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#o4w#o;'S#u;'S;=`$^<%lO#u^<Y[!mWkSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#f4w#f#g=O#g#o4w#o;'S#u;'S;=`$^<%lO#uU=T[kSOt#uuw#ux!_#u!_!`5l!`#O#u#P#T#u#T#i4w#i#j8s#j#o4w#o;'S#u;'S;=`$^<%lO#uU>QUzQkSOt#uuw#ux#O#u#P;'S#u;'S;=`$^<%lO#u~>iO!w~",
-  tokenizers: [operatorTokenizer, 1, 2, 3, tokenizer, new LocalTokenGroup("[~RP!O!PU~ZO!s~~", 11)],
+  tokenizers: [operatorTokenizer, 1, 2, 3, tokenizer, new LocalTokenGroup("[~RP!O!PU~ZO!d~~", 11)],
-  topRules: {"Program":[0,17]},
+  topRules: {"Program":[0,18]},
-  specialized: [{term: 13, get: (value: keyof typeof spec_Identifier) => spec_Identifier[value] || -1}],
+  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: 753
+  tokenPrec: 860
 })

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', () => {
@@ -299,10 +497,10 @@ describe('Assign', () => {
         AssignableIdentifier add
         Eq =
         FunctionDef
-          keyword do
+          Do do
           Params
-            AssignableIdentifier a
+            Identifier a
-            AssignableIdentifier b
+            Identifier b
           colon :
           BinOp
             Identifier a
@@ -319,9 +517,10 @@ describe('DotGet whitespace sensitivity', () => {
         AssignableIdentifier basename
         Eq =
         Number 5
-      DotGet
+      FunctionCallOrIdentifier
-        IdentifierBeforeDot basename
+        DotGet
-        Identifier prop`)
+          IdentifierBeforeDot basename
+          Identifier prop`)
   })
 
   test('space before dot - NOT DotGet, parses as division', () => {
@@ -348,3 +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/dot-get.test.ts

@@ -20,22 +20,24 @@ describe('DotGet', () => {
         AssignableIdentifier obj
         Eq =
         Number 5
-      DotGet
+      FunctionCallOrIdentifier
-        IdentifierBeforeDot obj
+        DotGet
-        Identifier prop
+          IdentifierBeforeDot obj
+          Identifier prop
     `)
   })
 
   test('function parameters are in scope within function body', () => {
     expect('do config: config.path end').toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier config
+          Identifier config
         colon :
-        DotGet
+        FunctionCallOrIdentifier
-          IdentifierBeforeDot config
+          DotGet
-          Identifier path
+            IdentifierBeforeDot config
+            Identifier path
         keyword end
     `)
   })
@@ -43,13 +45,14 @@ describe('DotGet', () => {
   test('parameters out of scope outside function', () => {
     expect('do x: x.prop end; x.prop').toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
         colon :
-        DotGet
+        FunctionCallOrIdentifier
-          IdentifierBeforeDot x
+          DotGet
-          Identifier prop
+            IdentifierBeforeDot x
+            Identifier prop
         keyword end
       Word x.prop
     `)
@@ -61,17 +64,19 @@ describe('DotGet', () => {
   y.bar
 end`).toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
-          AssignableIdentifier y
+          Identifier y
         colon :
-        DotGet
+        FunctionCallOrIdentifier
-          IdentifierBeforeDot x
+          DotGet
-          Identifier foo
+            IdentifierBeforeDot x
-        DotGet
+            Identifier foo
-          IdentifierBeforeDot y
+        FunctionCallOrIdentifier
-          Identifier bar
+          DotGet
+            IdentifierBeforeDot y
+            Identifier bar
         keyword end
     `)
   })
@@ -82,21 +87,23 @@ end`).toMatchTree(`
   do y: y.inner end
 end`).toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
         colon :
-        DotGet
+        FunctionCallOrIdentifier
-          IdentifierBeforeDot x
-          Identifier outer
-        FunctionDef
-          keyword do
-          Params
-            AssignableIdentifier y
-          colon :
           DotGet
-            IdentifierBeforeDot y
+            IdentifierBeforeDot x
-            Identifier inner
+            Identifier outer
+        FunctionDef
+          Do do
+          Params
+            Identifier y
+          colon :
+          FunctionCallOrIdentifier
+            DotGet
+              IdentifierBeforeDot y
+              Identifier inner
           keyword end
         keyword end
     `)
@@ -117,6 +124,62 @@ end`).toMatchTree(`
     `)
   })
 
+  test('dot get works as bare function', () => {
+    expect('io = dict print=echo; io.print').toMatchTree(`
+      Assign
+        AssignableIdentifier io
+        Eq =
+        FunctionCall
+          Identifier dict
+          NamedArg
+            NamedArgPrefix print=
+            Identifier echo
+      FunctionCallOrIdentifier
+        DotGet
+          IdentifierBeforeDot io
+          Identifier print
+    `)
+  })
+
+  test('dot get works as function w/ args', () => {
+    expect('io = dict print=echo; io.print heya').toMatchTree(`
+      Assign
+        AssignableIdentifier io
+        Eq =
+        FunctionCall
+          Identifier dict
+          NamedArg
+            NamedArgPrefix print=
+            Identifier echo
+      FunctionCall
+        DotGet
+          IdentifierBeforeDot io
+          Identifier print
+        PositionalArg
+          Identifier heya
+    `)
+  })
+
+  test('dot get works as function in parens', () => {
+    expect('io = dict print=echo; (io.print heya)').toMatchTree(`
+      Assign
+        AssignableIdentifier io
+        Eq =
+        FunctionCall
+          Identifier dict
+          NamedArg
+            NamedArgPrefix print=
+            Identifier echo
+      ParenExpr
+        FunctionCall
+          DotGet
+            IdentifierBeforeDot io
+            Identifier print
+          PositionalArg
+            Identifier heya
+    `)
+  })
+
   test('mixed file paths and dot get', () => {
     expect('config = 42; cat readme.txt; echo config.path').toMatchTree(`
       Assign
@@ -145,4 +208,70 @@ end`).toMatchTree(`
         PositionalArg
           Identifier prop`)
   })
+
+  test('readme.1 is Word when readme not in scope', () => {
+    expect('readme.1').toMatchTree(`Word readme.1`)
+  })
+
+  test('readme.1 is Word when used in function', () => {
+    expect('echo readme.1').toMatchTree(`
+      FunctionCall
+        Identifier echo
+        PositionalArg
+          Word readme.1`)
+  })
+
+  test('obj.1 is DotGet when obj is assigned', () => {
+    expect('obj = 5; obj.1').toMatchTree(`
+      Assign
+        AssignableIdentifier obj
+        Eq =
+        Number 5
+      FunctionCallOrIdentifier
+        DotGet
+          IdentifierBeforeDot obj
+          Number 1
+    `)
+  })
+
+  test('obj.1 arg is DotGet when obj is assigned', () => {
+    expect('obj = 5; obj.1').toMatchTree(`
+      Assign
+        AssignableIdentifier obj
+        Eq =
+        Number 5
+      FunctionCallOrIdentifier
+        DotGet
+          IdentifierBeforeDot obj
+          Number 1
+    `)
+  })
+
+  test('dot get index works as function w/ args', () => {
+    expect(`io = list (do x: echo x end); io.0 heya`).toMatchTree(`
+      Assign
+        AssignableIdentifier io
+        Eq =
+        FunctionCall
+          Identifier list
+          PositionalArg
+            ParenExpr
+              FunctionDef
+                Do do
+                Params
+                  Identifier x
+                colon :
+                FunctionCall
+                  Identifier echo
+                  PositionalArg
+                    Identifier x
+                keyword end
+      FunctionCall
+        DotGet
+          IdentifierBeforeDot io
+          Number 0
+        PositionalArg
+          Identifier heya
+    `)
+  })
 })

src/parser/tests/functions.test.ts

@@ -60,7 +60,7 @@ describe('Do', () => {
   test('parses function no parameters', () => {
     expect('do: 1 end').toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params 
         colon :
         Number 1
@@ -70,9 +70,9 @@ describe('Do', () => {
   test('parses function with single parameter', () => {
     expect('do x: x + 1 end').toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
         colon :
         BinOp
           Identifier x
@@ -84,10 +84,10 @@ describe('Do', () => {
   test('parses function with multiple parameters', () => {
     expect('do x y: x * y end').toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
-          AssignableIdentifier y
+          Identifier y
         colon :
         BinOp
           Identifier x
@@ -102,10 +102,10 @@ describe('Do', () => {
   x + 9
 end`).toMatchTree(`
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
-          AssignableIdentifier y
+          Identifier y
         colon :
         BinOp
           Identifier x
@@ -124,9 +124,9 @@ end`).toMatchTree(`
         AssignableIdentifier fnnn
         Eq =
         FunctionDef
-          keyword do
+          Do do
           Params
-            AssignableIdentifier x
+            Identifier x
           colon :
           FunctionCallOrIdentifier
             Identifier x
@@ -139,12 +139,29 @@ end`).toMatchTree(`
         AssignableIdentifier enddd
         Eq =
         FunctionDef
-          keyword do
+          Do do
           Params
-            AssignableIdentifier x
+            Identifier x
           colon :
           FunctionCallOrIdentifier
             Identifier x
           keyword end`)
   })
+
+  test('can call a function returned by a parens expression', () => {
+    expect('(do x: x end) 5').toMatchTree(`
+      FunctionCall
+        ParenExpr
+          FunctionDef
+            Do do
+            Params
+              Identifier x
+            colon :
+            FunctionCallOrIdentifier
+              Identifier x
+            keyword end
+        PositionalArg
+          Number 5
+    `)
+  })
 })

src/parser/tests/multiline.test.ts

@@ -24,10 +24,10 @@ describe('multiline', () => {
         AssignableIdentifier add
         Eq =
         FunctionDef
-          keyword do
+          Do do
           Params
-            AssignableIdentifier a
+            Identifier a
-            AssignableIdentifier b
+            Identifier b
           colon :
           Assign
             AssignableIdentifier result
@@ -61,14 +61,30 @@ end
       Number 3
 
       FunctionDef
-        keyword do
+        Do do
         Params
-          AssignableIdentifier x
+          Identifier x
-          AssignableIdentifier y
+          Identifier y
         colon :
         FunctionCallOrIdentifier
           Identifier x
         keyword end
     `)
   })
+
+  test('multiline with empty lines', () => {
+    expect(`
+      do:
+          2
+   
+      end
+      `).toMatchTree(`
+      FunctionDef
+        Do do
+        Params 
+        colon :
+        Number 2
+        keyword end
+    `)
+  })
 })

src/parser/tests/pipes.test.ts

@@ -75,13 +75,27 @@ describe('pipe expressions', () => {
           Identifier each
           PositionalArg
             FunctionDef
-              keyword do
+              Do do
               Params
-                AssignableIdentifier x
+                Identifier x
               colon :
               FunctionCallOrIdentifier
                 Identifier x
               keyword end
     `)
   })
+
+  test(`double trouble (do keyword isn't over eager)`, () => {
+    expect(`
+      double 2 | double`).toMatchTree(`
+      PipeExpr
+        FunctionCall
+          Identifier double
+          PositionalArg
+            Number 2
+        operator |
+        FunctionCallOrIdentifier
+          Identifier double
+      `)
+  })
 })

src/parser/tokenizer.ts

@@ -1,5 +1,10 @@
 import { ExternalTokenizer, InputStream, Stack } from '@lezer/lr'
-import { Identifier, AssignableIdentifier, Word, IdentifierBeforeDot } from './shrimp.terms'
+import { Identifier, AssignableIdentifier, Word, IdentifierBeforeDot, Do } from './shrimp.terms'
+
+// doobie doobie do (we need the `do` keyword to know when we're defining params)
+export function specializeKeyword(ident: string) {
+  return ident === 'do' ? Do : -1
+}
 
 // The only chars that can't be words are whitespace, apostrophes, closing parens, and EOF.
 
@@ -14,7 +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
@@ -106,8 +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
     }
@@ -217,7 +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) ||
@@ -242,7 +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)
   )
 }