Parser: rules

nois-lang · Jun 8, 2023 · ffb37fc · ffb37fc
1 parent ddea266
commit ffb37fc
Show file tree

Hide file tree

Showing 7 changed files with 435 additions and 76 deletions.
diff --git a/data/hello.no b/data/hello.no
@@ -1,3 +1,3 @@
-let main = (): Unit {
+fn main(): Unit {
     print("Hello, World!")
 }
diff --git a/nois.bnf b/nois.bnf
@@ -1,20 +1,27 @@
 module              ::= statement*
                     ;
-statement           ::= variable-def | type-def | return-stmt | expr
+statement           ::= var-def | fn-def | type-def | return-stmt | expr
                     ;
-  variable-def      ::= LET-KEYWORD IDENTIFIER EQUALS expr
+  var-def           ::= LET-KEYWORD IDENTIFIER type-annot? EQUALS expr
                     ;
-  type-def          ::= TYPE-KEYWORD IDENTIFIER TODO
+  fn-def            ::= FN-KEYWORD IDENTIFIER O-PAREN params? C-PAREN type-annot? block?
+                    ;
+  type-def          ::= TYPE-KEYWORD type-expr (constr-params | constr-list)
+                    ;
+    constr-params   ::= O-PAREN params? C-PAREN
+                    ;
+    constr-list     ::= O-BRACE (constructor (COMMA constructor)* COMMA?)? C-BRACE
+                    ;
+      constructor   ::= IDENTIFIER constr-params?
                     ;
   return-stmt       ::= RETURN-KEYWORD expr?
                     ;
   expr              ::= sub-expr (infix-op sub-expr)*
                     ;
-    sub-expr        ::= prefix-op operand
-                    | operand postfix-op?
+    sub-expr        ::= prefix-op operand | operand postfix-op?
                     ;
       operand       ::= if-expr
-                    | fn-expr
+                    | lambda-expr
                     | O-PAREN expr C-PAREN
                     | STRING
                     | CHAR
@@ -49,22 +56,25 @@ statement           ::= variable-def | type-def | return-stmt | expr
 
     postfix-op      ::= call-op
                     ;
-      call-op       ::= O-PAREN args C-PAREN
+      call-op       ::= args
+                    ;
+        args        ::= O-PAREN (expr (COMMA expr)*)? COMMA? C-PAREN
+                    ;
+    lambda-expr     ::= lambda-params type-annot? block
                     ;
-        args        ::= expr (COMMA expr)* COMMA?
+      lambda-params ::= PIPE (param (COMMA param)*)? COMMA? PIPE
                     ;
-    fn-expr         ::= PIPE params PIPE COLON type-expr block
+    params          ::= O-PAREN (param (COMMA param)*)? COMMA? C-PAREN
                     ;
-      params        ::= param (COMMA param)* COMMA?
+        param       ::= IDENTIFIER type-annot?
                     ;
-        param       ::= IDENTIFIER COLON type-expr
+block               ::= O-BRACE statement* C-BRACE | O-BRACE C-BRACE
                     ;
-block               ::= O-BRACE statement* C-BRACE
-                    | O-BRACE C-BRACE
+type-annot          ::= COLON type-expr
                     ;
-type-expr           ::= IDENTIFIER (O-ANGLE type-params C-ANGLE)?
+  type-expr         ::= IDENTIFIER type-params?
                     ;
-  type-params       ::= type-expr (COMMA type-expr)* COMMA?
+    type-params     ::= O-ANGLE (type-expr (COMMA type-expr)* COMMA?)? C-ANGLE
                     ;
 if-expr             ::= IF-KEYWORD expr block (ELSE-KEYWORD block)?
                     ;
diff --git a/src/error.ts b/src/error.ts
@@ -1,19 +1,22 @@
-import { indexToLocation, LocationRange, prettyIndex, prettyLocation } from './location'
+import { indexToLocation, prettyIndex, prettyLocation } from './location'
 import { Source } from './source'
 import { ParseToken, TokenKind } from './lexer/lexer'
 
-export interface SyntaxErrorInfo {
+export interface SyntaxError {
     expected: TokenKind[],
-    got: TokenKind,
-    location: LocationRange
+    got: ParseToken,
+    message?: string
 }
 
 export const prettyLexerError = (token: ParseToken): string => {
     return `lexer error: unknown token \`${token.value}\``
 }
 
-export const prettySyntaxError = (error: SyntaxErrorInfo): string => {
-    return `syntax error: expected \`${error.expected}\`, got \`${error.got}\``
+export const prettySyntaxError = (error: SyntaxError): string => {
+    if (error.message) {
+        return `syntax error: ${error.message}`
+    }
+    return `syntax error: expected \`${error.expected}\`, got \`${error.got.kind}\``
 }
 
 export const prettySourceMessage = (message: string, index: number, source: Source): string => {

diff --git a/src/index.ts b/src/index.ts
@@ -1,6 +1,8 @@
 import { tokenize } from './lexer/lexer'
 import { readFileSync } from 'fs'
 import { join, resolve } from 'path'
+import { compactNode, parseModule, Parser } from './parser/parser'
+import { prettySourceMessage, prettySyntaxError } from './error'
 
 
 const version = JSON.parse(readFileSync(join(__dirname, '..', 'package.json')).toString()).version
@@ -18,10 +20,13 @@ if (!path) {
 const source = { str: readFileSync(resolve(path)).toString(), filename: path }
 
 const tokens = tokenize(source.str)
-// const token = parse(tokens)
-// if ('expected' in token) {
-//     console.error(prettySourceMessage(prettySyntaxError(token), token.location.start, source))
-//     process.exit(1)
-// }
-//
-// console.dir(compactToken(flattenToken(token)), { depth: null, colors: true, compact: true })
+const parser = new Parser(tokens)
+parseModule(parser)
+const root = parser.buildTree()
+
+for (const error of parser.errors) {
+    console.error(prettySourceMessage(prettySyntaxError(error), error.got.location.start, source))
+    process.exit(1)
+}
+
+console.dir(compactNode(root), { depth: null, colors: true, compact: true })
diff --git a/src/lexer/lexer.ts b/src/lexer/lexer.ts
@@ -6,6 +6,7 @@ export const lexerTokenKinds = <const>[
     'kind-keyword',
     'impl-keyword',
     'let-keyword',
+    'fn-keyword',
     'if-keyword',
     'else-keyword',
     'return-keyword',
@@ -61,6 +62,7 @@ export const constTokenKindMap: Map<TokenKind, string> = new Map([
     ['return-keyword', 'return'],
     ['impl-keyword', 'impl'],
     ['let-keyword', 'let'],
+    ['fn-keyword', 'fn'],
 
     ['o-paren', '('],
     ['c-paren', ')'],

diff --git a/src/parser/parser.spec.ts b/src/parser/parser.spec.ts
@@ -1,4 +1,30 @@
+import { compactNode, parseModule, Parser } from './parser'
+import { tokenize } from '../lexer/lexer'
+import { expect } from '@jest/globals'
+
 describe('parser', () => {
-    it('buildTree', () => {
+
+    const parse = (code) => {
+        const p = new Parser(tokenize(code))
+        parseModule(p)
+        const tree = p.buildTree()
+        return { tree: compactNode(tree), errors: p.errors }
+    }
+
+    it('parse fn-def empty', () => {
+        const { tree, errors } = parse('fn main() {}')
+        expect(errors.length).toEqual(0)
+        expect(tree).toEqual({
+            'module': [{
+                'statement': [{
+                    'fn-def': [
+                        { 'fn-keyword': 'fn' },
+                        { 'identifier': 'main' },
+                        { 'o-paren': '(' }, { 'c-paren': ')' },
+                        { 'block': [{ 'o-brace': '{' }, { 'c-brace': '}' }] }
+                    ]
+                }]
+            }]
+        })
     })
 })