Fix some parser errors

README.md

@@ -32,6 +32,9 @@ here is a list of the features (or better, limitations) I want to introduce:
 - Polymorphic types à la Hindley Milner (probably unfeasible)
 - Automatic currying of functions, with optimised partial evaluation (unfeasible)
 - Delimited continuations (this is an overkill)
+- Easy syntax allowing easy kakoune support
 
 milly's full grammar is specified as a pair of lex and yacc files in `ref_parser/`.
 
+kakoune plugin for milly is in `extra/milly.kak`.
+

include/lexer.h

@@ -45,6 +45,9 @@ struct token {
     char *lexeme;
 };
 
+/* Gives a text description of a token type */
+const char *token_descr(enum token_type tok);
+
 struct location {
     size_t line;
     size_t col;

src/lexer.c

@@ -6,6 +6,73 @@
 #include <ctype.h>
 #include <stdnoreturn.h>
 
+const char *token_descr(enum token_type tok) {
+    switch (tok) {
+    case tok_ident:
+        return "identifier";
+    case tok_param_ident:
+        return "variable identifier";
+    case tok_int:
+        return "int literal";
+    case tok_string:
+        return "string literal";
+    case tok_arrow:
+        return "->";
+    case tok_backslash:
+        return "\\";
+    case tok_equal:
+        return "=";
+    case tok_left_paren:
+        return "(";
+    case tok_right_paren:
+        return ")";
+    case tok_left_square:
+        return "[";
+    case tok_right_square:
+        return "]";
+    case tok_left_brace:
+        return "{";
+    case tok_right_brace:
+        return "}";
+    case tok_left_angle_bracket:
+        return "<";
+    case tok_right_angle_bracket:
+        return ">";
+    case tok_comma:
+        return ",";
+    case tok_pipe:
+        return "|";
+    case tok_colon:
+        return ":";
+    case tok_true:
+        return "true";
+    case tok_false:
+        return "false";
+    case tok_case:
+        return "case";
+    case tok_let:
+        return "let";
+    case tok_in:
+        return "in";
+    case tok_match:
+        return "match";
+    case tok_of:
+        return "of";
+    case tok_def:
+        return "def";
+    case tok_datatype:
+        return "datatype";
+    case tok_alias:
+        return "alias";
+    case tok_typecheck:
+        return "typecheck";
+    case tok_eof:
+        return "end of file";
+    default:
+        return "unknown token";
+    }
+}
+
 /* Helper to report lexical errors */
 static noreturn void report_lex_error(struct lexer *lex, const char *fmt, ...) {
     if (fprintf(stderr, "Error at %ld:%ld ", lex->loc.line, lex->loc.col) < 0) {
@@ -78,35 +145,35 @@ struct reserved_symbol {
     enum token_type type;
 };
 
-/* These two arrays must be alpha sorted to be usable by bsearch */
+/* These two arrays must be sorted (according to ascii value) to be usable by bsearch */
 struct reserved_symbol punctuation[] = {
     { "(", tok_left_paren },
     { ")", tok_right_paren },
-    { "[", tok_left_square },
-    { "]", tok_right_square },
-    { "{", tok_left_brace },
-    { "}", tok_right_brace },
     { ",", tok_comma },
-    { "|", tok_pipe },
     { ":", tok_colon },
     { "<", tok_left_angle_bracket },
     { ">", tok_right_angle_bracket },
+    { "[", tok_left_square },
     { "\\", tok_backslash },
+    { "]", tok_right_square },
+    { "{", tok_left_brace },
+    { "|", tok_pipe },
+    { "}", tok_right_brace },
 };
 
 struct reserved_symbol keywords[] = {
-    { "=", tok_equal },
     { "->", tok_arrow },
-    { "true", tok_true },
-    { "false", tok_false },
+    { "=", tok_equal },
+    { "alias", tok_alias },
     { "case", tok_case },
-    { "let", tok_let },
+    { "datatype", tok_datatype },
+    { "def", tok_def },
+    { "false", tok_false },
     { "in", tok_in },
+    { "let", tok_let },
     { "match", tok_match },
     { "of", tok_of },
-    { "def", tok_def },
-    { "datatype", tok_datatype },
-    { "alias", tok_alias },
+    { "true", tok_true },
     { "typecheck", tok_typecheck },
 };
 
@@ -124,6 +191,7 @@ void lex_punctuation(struct lexer *lex, struct token *out) {
     /* This can't fail because of the precondition to this function */
     struct reserved_symbol *r = bsearch(key, punctuation, num, elem_size, symbol_cmp);
     out->type = r->type;
+    advance(lex);
 }
 
 /* We assume that the curr char is already a digit */
@@ -171,6 +239,7 @@ static void lex_param_ident(struct lexer *lex, struct token *out) {
 static void lex_ident(struct lexer *lex, struct token *out) {
     /* Store ident in the internal buffer */
     store_char(lex, lex->cur);
+    advance(lex);
     while (is_ident_cont(lex->cur)) {
         /* Store ident in the internal buffer */
         store_char(lex, lex->cur);
@@ -276,6 +345,7 @@ keep_lexing:
 
     case '\'':
         *loc = lex->loc;
+        advance(lex);
         lex_param_ident(lex, out);
         break;
 

src/main.c

@@ -2,8 +2,18 @@
 #include "parser.h"
 
 int main(int argc, char *argv[]) {
+    if (argc != 2) {
+        printf("Usage: milly file.mil\n");
+        return 1;
+    }
+    FILE *input = fopen(argv[1], "r");
+    if (!input) {
+        fprintf(stderr, "Error: Could not open file %s.\n", argv[1]);
+        return 1;
+    }
     struct parser p;
-    init_parser(&p, stdin);
+    init_parser(&p, input);
     parse_program(&p);
+    fclose(input);
     return 0;
 }

src/parser.c

@@ -44,32 +44,36 @@ static noreturn void report_error(struct parser *p, const char *fmt, ...) {
     exit(EXIT_FAILURE);
 }
 
-static const char *token_descr(enum token_type t) {
-    return "another token";
-}
-
 /* Consume next token if it is of the required type, otherwise fail with error */
 static void expect(struct parser *p, enum token_type t) {
     if (cur_tok(p) != t) {
-        report_error(p, "Expected %s.\n", token_descr(t));
+        report_error(p, "Expected `%s`, but found `%s`.\n", token_descr(t), token_descr(cur_tok(p)));
     }
     consume(p);
 }
 
 /* Types */
-static struct type *parse_tuple_type(struct parser *p) {
+static struct type *parse_paren_type(struct parser *p) {
+    struct type *res;
     struct type *t = parse_type(p);
+    if (cur_tok(p) == tok_comma) {
+        consume(p);
         struct type_list_builder list = { NULL };
         type_list_append(&list, t);
-    expect(p, tok_comma);
         t = parse_type(p);
         type_list_append(&list, t);
-    while (cur_tok(p) != tok_eof && cur_tok(p) != tok_right_angle_bracket) {
+        while (cur_tok(p) != tok_eof && cur_tok(p) != tok_right_paren) {
             expect(p, tok_comma);
             t = parse_type(p);
             type_list_append(&list, t);
         }
-    return make_tuple_type(list.head);
+        res = make_tuple_type(list.head);
+    }
+    else {
+        res = t;
+    }
+    expect(p, tok_right_paren);
+    return res;
 }
 
 /* sets *is_ident to true if the result is a single type name */
@@ -79,14 +83,7 @@ static struct type *try_parse_atomic_type_piece(struct parser *p, bool *is_ident
     switch (cur_tok(p)) {
     case tok_left_paren:
         consume(p);
-        res = parse_type(p);
-        expect(p, tok_right_paren);
-        break;
-
-    case tok_left_angle_bracket:
-        consume(p);
-        res = parse_tuple_type(p);
-        expect(p, tok_right_angle_bracket);
+        res = parse_paren_type(p);
         break;
 
     case tok_param_ident:
@@ -177,7 +174,7 @@ static struct decl *parse_datatype_decl(struct parser *p) {
     struct type *ty;
     while (cur_tok(p) != tok_eof && cur_tok(p) != tok_right_brace) {
         if (cur_tok(p) != tok_ident) {
-            report_error(p, "Invalid datatype constructor, expected an identifier.\n");
+            report_error(p, "Invalid datatype constructor `%s`, expected an identifier.\n", token_descr(cur_tok(p)));
         }
         ctor_name = cur_lexeme(p);
         ty = parse_type(p);
@@ -264,7 +261,7 @@ struct decl *parse_decl(struct parser *p) {
         consume(p);
         return parse_value_or_func_decl(p);
     default:
-        report_error(p, "Declaration expected, invalid token.\n");
+        report_error(p, "Declaration expected, invalid token `%s`\n", token_descr(cur_tok(p)));
     }
 }