From fa6b2a24aad096824790501550623a7c3c2418c1 Mon Sep 17 00:00:00 2001
From: Michael Williamson <mike@zwobble.org>
Date: Sat, 30 Sep 2023 10:28:05 +0100
Subject: Parse sum types

---
 .../ast/untyped/UntypedSumTypeDefinitionNode.java  |  4 +-
 .../java/org/shedlang/compiler/parser/Parser.java  | 53 ++++++++++++++++++++++
 .../org/shedlang/compiler/parser/TokenType.java    |  1 +
 .../org/shedlang/compiler/parser/Tokeniser.java    |  1 +
 4 files changed, 58 insertions(+), 1 deletion(-)

(limited to 'src/main/java')

diff --git a/src/main/java/org/shedlang/compiler/ast/untyped/UntypedSumTypeDefinitionNode.java b/src/main/java/org/shedlang/compiler/ast/untyped/UntypedSumTypeDefinitionNode.java
index 3a49f46e..681c4506 100644
--- a/src/main/java/org/shedlang/compiler/ast/untyped/UntypedSumTypeDefinitionNode.java
+++ b/src/main/java/org/shedlang/compiler/ast/untyped/UntypedSumTypeDefinitionNode.java
@@ -2,9 +2,11 @@ package org.shedlang.compiler.ast.untyped;
 
 import org.shedlang.compiler.sources.Source;
 
+import java.util.List;
+
 public record UntypedSumTypeDefinitionNode(
     String name,
-    UntypedTypeLevelExpressionNode variants,
+    List<UntypedTypeLevelExpressionNode> variants,
     Source source
 ) implements UntypedNamespaceStatementNode {
 }
diff --git a/src/main/java/org/shedlang/compiler/parser/Parser.java b/src/main/java/org/shedlang/compiler/parser/Parser.java
index e3c78d07..84900688 100644
--- a/src/main/java/org/shedlang/compiler/parser/Parser.java
+++ b/src/main/java/org/shedlang/compiler/parser/Parser.java
@@ -429,6 +429,13 @@ public class Parser {
         return new UntypedIntLiteralNode(value, source);
     }
 
+    /**
+     * Parse zero, one, or many instances of {@code T} using {@code parseOne}.
+     * Each instance is expected to be separated by tokens consumed by
+     * {@code parseSeparator}. Parsing stops either if {@code parseSeparator}
+     * returns false, or if {@code stop} returns false. Trailing separators
+     * are permitted.
+     */
     private static <T> List<T> parseMany(
         BooleanSupplier stop,
         Supplier<T> parseOne,
@@ -447,6 +454,27 @@ public class Parser {
         return values;
     }
 
+    /**
+     * Parse one or more instances of {@code T} using {@code parseOne}. Each
+     * instance is expected to be separated by tokens consumed by
+     * {@code parseSeparator}. Parsing stops if {@code parseSeparator} returns
+     * false. Trailing separators are not permitted.
+     */
+    private static <T> List<T> parseOneOrMoreNoTrailingSeparator(
+        Supplier<T> parseOne,
+        BooleanSupplier parseSeparator
+    ) {
+        var values = new ArrayList<T>();
+
+        while (true) {
+            var value = parseOne.get();
+            values.add(value);
+            if (!parseSeparator.getAsBoolean()) {
+                return values;
+            }
+        }
+    }
+
     static UntypedNamespaceStatementNode parseNamespaceStatement(TokenIterator<TokenType> tokens) {
         return switch (tokens.peek().tokenType()) {
             case KEYWORD_FUN ->
@@ -455,6 +483,9 @@ public class Parser {
             case KEYWORD_STRUCT ->
                 parseStructDefinition(tokens);
 
+            case KEYWORD_SUMTYPE ->
+                parseSumTypeDefinition(tokens);
+
             case KEYWORD_VAL ->
                 parseAssign(tokens);
 
@@ -603,6 +634,28 @@ public class Parser {
         return parseExpression(tokens, Optional.of(parentPrecedence));
     }
 
+    private static UntypedNamespaceStatementNode parseSumTypeDefinition(TokenIterator<TokenType> tokens) {
+        var start = tokens.position();
+
+        tokens.skip(TokenType.KEYWORD_SUMTYPE);
+
+        var name = tokens.nextValue(TokenType.IDENTIFIER);
+
+        tokens.skip(TokenType.SYMBOL_EQUALS);
+
+        var variants = parseOneOrMoreNoTrailingSeparator(
+            () -> parseTopLevelTypeLevelExpression(tokens),
+            () -> tokens.trySkip(TokenType.SYMBOL_BAR)
+        );
+
+        tokens.skip(TokenType.SYMBOL_SEMICOLON);
+
+        var end = tokens.position();
+        var source = start.to(end);
+
+        return new UntypedSumTypeDefinitionNode(name, variants, source);
+    }
+
     static UntypedExpressionNode parseTopLevelExpression(TokenIterator<TokenType> tokens) {
         return parseExpression(tokens, Optional.empty());
     }
diff --git a/src/main/java/org/shedlang/compiler/parser/TokenType.java b/src/main/java/org/shedlang/compiler/parser/TokenType.java
index ec4fd332..9c8183dc 100644
--- a/src/main/java/org/shedlang/compiler/parser/TokenType.java
+++ b/src/main/java/org/shedlang/compiler/parser/TokenType.java
@@ -24,6 +24,7 @@ public enum TokenType {
 
     SYMBOL_BANG,
     SYMBOL_BANG_EQUALS,
+    SYMBOL_BAR,
     SYMBOL_BRACE_OPEN,
     SYMBOL_BRACE_CLOSE,
     SYMBOL_COLON,
diff --git a/src/main/java/org/shedlang/compiler/parser/Tokeniser.java b/src/main/java/org/shedlang/compiler/parser/Tokeniser.java
index f33e4d53..7659d717 100644
--- a/src/main/java/org/shedlang/compiler/parser/Tokeniser.java
+++ b/src/main/java/org/shedlang/compiler/parser/Tokeniser.java
@@ -43,6 +43,7 @@ public class Tokeniser {
             RegexTokeniser.rule(TokenType.SYMBOL_HASH_PAREN_OPEN, "#\\("),
 
             RegexTokeniser.rule(TokenType.SYMBOL_BANG, "!"),
+            RegexTokeniser.rule(TokenType.SYMBOL_BAR, "\\|"),
             RegexTokeniser.rule(TokenType.SYMBOL_BRACE_OPEN, "\\{"),
             RegexTokeniser.rule(TokenType.SYMBOL_BRACE_CLOSE, "\\}"),
             RegexTokeniser.rule(TokenType.SYMBOL_COLON, ":"),
-- 
cgit v1.2.3