lexing works

1 files changed, 153 insertions, 0 deletions

diff --git a/lexer.c b/lexer.c
new file mode 100644
index 0000000..bff9a40
--- /dev/null
+++ b/lexer.c
@@ -0,0 +1,153 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include "lexer.h"
+
+token_t init_token(tokenType_t type, void* value) {
+	token_t t;
+	t.type = type;
+	t.value = value;
+	return t;
+}
+linkList_t* init_linklist(void* value, linkList_t* next) {
+	linkList_t* ll = calloc(sizeof(linkList_t), 1);
+	ll->value = value;
+	ll->next = next;
+	return ll;
+}
+int advance_whitespace(char** linePointer) {
+	int64_t a = (int64_t)linePointer;
+	while (**linePointer == ' ' || **linePointer == '	') {
+		(*linePointer)++;
+	}
+	return (int)((int64_t)linePointer - a);
+}
+linkList_t* lex_line(char* line) {
+	char** p = malloc(sizeof(char*));
+	*p = line;
+	linkList_t* root = init_linklist(NULL, NULL);
+	linkList_t* ll = root;
+	while (1) {
+		token_t t = lex_first_token(p);
+		if (t.type == EOL) {
+			printf("eol\n");
+			break;
+		}
+		if (t.type == ERR) {
+			exit(1);
+		}
+		if (ll->value != NULL) {
+			ll->next = init_linklist(NULL, NULL);
+			ll = ll->next;
+		}
+		token_t* tp = malloc(sizeof(token_t));
+		printf("good\n");
+		*tp = t;
+		ll->value = tp;
+	}
+	free(p);
+	return root;
+}
+token_t lex_first_token(char** content) {
+	int m = 5;
+	char* s = malloc(sizeof(char) * m);
+	int c = 0;
+	advance_whitespace(content);
+	while (**content != '\n') {
+		s[c] = **content;
+		s[c+1] = '\0';
+		(*content)++;
+		printf("lexing: %c\n", s[c]);
+		/* check if we need to parse a string or an int */
+		if (s[c] >= 65 && s[c] <= 90) {
+			free(s);
+			(*content)--;
+			return lex_first_string(content);
+		}
+		else if (s[c] >= 48 && s[c] <= 57) {
+			free(s);
+			(*content)--;
+			return lex_first_int(content);
+		}
+		/* otherwise, check to see if a symbol has been matched */
+		if (strcmp(s, "(") == 0) {
+			free(s);
+			return init_token(OPEN_PAREN, NULL);
+		}
+		else if (strcmp(s, ")") == 0) {
+			free(s);
+			return init_token(CLOSE_PAREN, NULL);
+		}
+		else if (strcmp(s, "#") == 0) {
+			free(s);
+			return init_token(START_COMMENT, NULL);
+		}
+		else if (strcmp(s, "|") == 0) {
+			free(s);
+			return init_token(JUSTIFICATION, NULL);
+		}
+		else if (strcmp(s, "pr") == 0) {
+			free(s);
+			return init_token(PREMISE, NULL);
+		}
+		else if (strcmp(s, "as") == 0) {
+			free(s);
+			return init_token(ASSUMPTION, NULL);
+		}
+		else if (strcmp(s, "^") == 0) {
+			free(s);
+			return init_token(CONJUNCTION, NULL);
+		}
+		else if (strcmp(s, "v") == 0) {
+			free(s);
+			return init_token(DISJUNCTION, NULL);
+		}
+		else if (strcmp(s, "->") == 0) {
+			free(s);
+			return init_token(IMPLICATION, NULL);
+		}
+		else if (strcmp(s, "<-") == 0) {
+			free(s);
+			return init_token(BI_IMPLICATION, NULL);
+		}
+		else if (c <= m) {
+			printf("c: %d\n", c);
+			c++;
+			continue;
+		}
+		else {
+			free(s);
+			return init_token(ERR, NULL);
+		}
+	}
+	return init_token(EOL, NULL);
+}
+token_t lex_first_string(char** content) {
+	int m = 50;
+	char* s = malloc(sizeof(char) * m);
+	int c = 0;
+	advance_whitespace(content);
+	while (**content >= 65 && **content <= 90) {
+		s[c] = **content;
+		(*content)++;
+		c++;
+	}
+	s[c] = '\0';
+	return init_token(STRING, s);
+}
+token_t lex_first_int(char** content) {
+	int m = 10;
+	char* s = malloc(sizeof(char) * m);
+	int c = 0;
+	advance_whitespace(content);
+	while (**content >= 48 && **content <= 57) {
+		s[c] = **content;
+		(*content)++;
+		c++;
+	}
+	s[c] = '\0';
+	int* ip = malloc(sizeof(int));
+	*ip = atoi(s);
+	free(s);
+	return init_token(INT, (void*)ip);
+}