From f181a8f04dfdfd8829861e0d0d549f39e40081e6 Mon Sep 17 00:00:00 2001
From: Marvin Borner
Date: Thu, 13 May 2021 12:03:43 +0200
Subject: New layout

---
 src/tokenize.c | 175 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 175 insertions(+)
 create mode 100644 src/tokenize.c

(limited to 'src/tokenize.c')

diff --git a/src/tokenize.c b/src/tokenize.c
new file mode 100644
index 0000000..3424454
--- /dev/null
+++ b/src/tokenize.c
@@ -0,0 +1,175 @@
+#include <assert.h>
+#include <ctype.h>
+#include <log.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <tokenize.h>
+
+static char next_non_alnum(struct ctx *ctx, size_t start)
+{
+	for (size_t i = start; i < ctx->size; i++)
+		if (!isalnum(ctx->data[i]))
+			return ctx->data[i];
+
+	errln(ctx, "Unexpected end of buffer");
+}
+
+static bool peek_to_is_alnum(struct ctx *ctx, size_t start, char ch)
+{
+	for (size_t i = start; i < ctx->size; i++) {
+		char cur = ctx->data[i];
+
+		if (cur == ch || cur == ';' || cur == ')')
+			return true;
+
+		if (!isalnum(cur))
+			return false;
+	}
+
+	errln(ctx, "Unexpected end of buffer");
+}
+
+static size_t peek_alnum_to(struct ctx *ctx, size_t start, char ch)
+{
+	for (size_t i = start; i < ctx->size; i++) {
+		char cur = ctx->data[i];
+
+		if (cur == ch || cur == ';' || cur == ')')
+			return i;
+
+		if (!isalnum(cur))
+			errln(ctx, "'%c' is not alpha-numeric", cur);
+	}
+
+	errln(ctx, "Unexpected end of buffer");
+}
+
+static size_t peek_special_to(struct ctx *ctx, size_t start, char ch)
+{
+	for (size_t i = start; i < ctx->size; i++) {
+		char cur = ctx->data[i];
+
+		if (cur == ch || cur == ';' || cur == ')')
+			return i;
+
+		if (isalnum(cur) || cur < '!' || cur > '~')
+			errln(ctx, "'%c' is not special", cur);
+	}
+
+	errln(ctx, "Unexpected end of buffer");
+}
+
+static size_t peek_to(struct ctx *ctx, size_t start, char ch)
+{
+	for (size_t i = start; i < ctx->size; i++) {
+		char cur = ctx->data[i];
+
+		if (cur == ch || cur == ';' || cur == ')')
+			return i;
+	}
+
+	errln(ctx, "Unexpected end of buffer");
+}
+
+static void token_add(struct ctx *ctx, enum token_type type, size_t start, size_t end)
+{
+	struct token token = { 0 };
+	token.type = type;
+	token.start = start;
+	token.end = end;
+
+	assert(++ctx->token_count < TOKENS_MAX);
+	ctx->tokens[ctx->token_count] = token;
+
+	if (type == NEWLINE) {
+		ctx->line++;
+		ctx->column = 0;
+	} else {
+		ctx->column += end - start;
+	}
+}
+
+static void token_print(struct ctx *ctx, struct token *token)
+{
+	assert(token->type != UNKNOWN);
+
+	printf("[token type=%d] '", token->type);
+	if (token->type == NEWLINE || token->type == END) {
+		printf("' (Unprintable)\n");
+		return;
+	}
+
+	for (size_t i = token->start; i < token->end; i++)
+		printf("%c", ctx->data[i]);
+	printf("'\n");
+}
+
+void tokens_print(struct ctx *ctx)
+{
+	for (size_t i = 1; i < ctx->token_count; i++)
+		token_print(ctx, &ctx->tokens[i]);
+}
+
+void tokenize(struct ctx *ctx)
+{
+	for (size_t i = 0; i < ctx->size; i++) {
+		const char cur = ctx->data[i];
+
+		switch (cur) {
+		case '\0':
+			token_add(ctx, END, i, i + 1);
+			return;
+		case '\n':
+			token_add(ctx, NEWLINE, i, i + 1);
+			continue;
+		case ';':
+			token_add(ctx, EOL, i, i + 1);
+			continue;
+		case '(':
+			token_add(ctx, LPAREN, i, i + 1);
+			continue;
+		case ')':
+			token_add(ctx, RPAREN, i, i + 1);
+			continue;
+		case '=':
+			token_add(ctx, EQUAL, i, i + 1);
+			continue;
+		case ' ':
+			ctx->column++;
+			continue;
+		default:
+			break;
+		}
+
+		if (next_non_alnum(ctx, i) == ':') { // Type with param identifier
+			size_t start_param = peek_alnum_to(ctx, i, ':') + 1;
+			size_t end_param;
+			if (peek_to_is_alnum(ctx, start_param, ' ')) {
+				end_param = peek_alnum_to(ctx, start_param, ' ');
+			} else { // Unnamed identifier ('_')
+				end_param = peek_to(ctx, start_param, ' ');
+				if (end_param - start_param != 1 || ctx->data[start_param] != '_')
+					errln(ctx, "Invalid param identifier");
+			}
+
+			token_add(ctx, TYPE, i, start_param - 1);
+			token_add(ctx, TYPEDELIM, start_param - 1, start_param);
+			token_add(ctx, PARAM, start_param, end_param);
+
+			i = end_param - 1;
+			continue;
+		}
+
+		if (peek_to_is_alnum(ctx, i, ' ')) { // General identifier
+			size_t end_ident = peek_alnum_to(ctx, i, ' ');
+			token_add(ctx, IDENT, i, end_ident);
+			i = end_ident - 1;
+		} else { // Special/custom operator
+			size_t end_operator = peek_special_to(ctx, i, ' ');
+			token_add(ctx, OPERATOR, i, end_operator);
+			i = end_operator - 1;
+		}
+	}
+
+	context_rewind(ctx);
+}
-- 
cgit v1.2.3