feat(parser,lexer): start of actual parser

main
Jef Roosens 2024-03-10 21:29:37 +01:00
parent 35aae48a7f
commit 8c0105639f
Signed by: Jef Roosens
GPG Key ID: B75D4F293C7052DB
7 changed files with 207 additions and 3 deletions

View File

@ -8,7 +8,9 @@
#define MRK_AST_NODE_ARGS 1
typedef enum mrk_ast_node_type {
mrk_ast_node_type_header = 0,
mrk_ast_node_type_none = 0,
mrk_ast_node_type_header,
mrk_ast_node_type_sentence,
} mrk_ast_node_type;
typedef struct mrk_ast_node {
@ -17,7 +19,10 @@ typedef struct mrk_ast_node {
size_t len;
} children;
mrk_ast_node_type type;
void *args[MRK_AST_NODE_ARGS];
struct {
void *ptr;
size_t num;
} args[MRK_AST_NODE_ARGS];
} mrk_ast_node;
/**

View File

@ -30,6 +30,8 @@
typedef enum mrk_err {
mrk_err_ok = 0,
mrk_err_failed_alloc,
mrk_err_unexpected_token,
mrk_err_invalid,
} mrk_err;
#endif

View File

@ -37,6 +37,7 @@ typedef enum mrk_token_type {
mrk_token_type_word,
mrk_token_type_checked_box,
mrk_token_type_unchecked_box,
mrk_token_type_text,
} mrk_token_type;
typedef struct mrk_token {
@ -45,6 +46,17 @@ typedef struct mrk_token {
size_t end;
} mrk_token;
/**
* Return the length of the given token.
*/
size_t mrk_token_len(mrk_token t);
/**
* Return a pointer to the start of the lexeme in the initial buffer, returning
* the length of the lexeme.
*/
size_t mrk_token_lexeme(const char **out, mrk_lexer *lexer, mrk_token t);
/**
* Initialize a new lexer struct.
*/

View File

@ -5,6 +5,8 @@
#include "mrk/common.h"
#include "mrk/lexer.h"
#define MRK_MAX_HEADER_LEN 6
typedef struct mrk_parser mrk_parser;
/**

View File

@ -5,6 +5,10 @@
#include "mrk/parser.h"
#define MRK_PARSER_LOOKAHEAD_BUF_SIZE 4
#define MRK_PARSE_ERR(p, t, m) \
p->error.token = t; \
p->error.msg = m; \
return mrk_err_invalid
struct mrk_parser {
mrk_lexer *lexer;
@ -12,6 +16,10 @@ struct mrk_parser {
mrk_token buf[MRK_PARSER_LOOKAHEAD_BUF_SIZE];
size_t index;
} lookahead;
struct {
mrk_token token;
const char *msg;
} error;
};
/**
@ -34,4 +42,6 @@ void mrk_parser_advance(mrk_parser *parser);
mrk_err mrk_parser_parse_block(mrk_ast_node **out, mrk_parser *parser);
mrk_err mrk_parser_parse_header(mrk_ast_node **out, mrk_parser *parser);
#endif

View File

@ -129,6 +129,126 @@ void mrk_lexer_emit(mrk_token *out, mrk_lexer *lexer, mrk_token_type type) {
lexer->token.emitted = true;
}
void mrk_lexer_lex_special(mrk_token *out, mrk_lexer *lexer) {
char c = mrk_lexer_advance(lexer);
switch (c) {
// All these characters have multiple meanings depending on their location
// in the file and how many there are, so the lexer can only match them as
// one or more grouped characters
case '#':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_pounds);
break;
case '`':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_backticks);
break;
case '-':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_dashes);
break;
case '_':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_underscores);
break;
case '*':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_stars);
break;
case '=':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_equals);
break;
case '\t':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_tabs);
break;
case '>':
mrk_lexer_advance_eq(lexer, c);
mrk_lexer_emit(out, lexer, mrk_token_type_right_angle_brackets);
break;
case '!':
mrk_lexer_emit(out, lexer, mrk_token_type_bang);
break;
case '[':
// Checkboxes for lists are lexed separately to simplify the parser later
// on
if (mrk_lexer_peek_str(lexer, " ]")) {
mrk_lexer_advance_n(lexer, 2);
mrk_lexer_emit(out, lexer, mrk_token_type_unchecked_box);
} else if (mrk_lexer_peek_str(lexer, "x]")) {
mrk_lexer_advance_n(lexer, 2);
mrk_lexer_emit(out, lexer, mrk_token_type_checked_box);
} else {
mrk_lexer_emit(out, lexer, mrk_token_type_left_bracket);
}
break;
case ']':
mrk_lexer_emit(out, lexer, mrk_token_type_right_bracket);
break;
case '(':
mrk_lexer_emit(out, lexer, mrk_token_type_left_paren);
break;
case ')':
mrk_lexer_emit(out, lexer, mrk_token_type_right_paren);
break;
case '\\':
if (mrk_lexer_peek(lexer) == '\n') {
mrk_lexer_emit(out, lexer, mrk_token_type_line_break);
} else {
mrk_lexer_emit(out, lexer, mrk_token_type_backslash);
}
break;
// Two consecutive newlines constitute a blank line, otherwise they're
// ignored as whitespace
case '\n':
if (mrk_lexer_peek(lexer) == '\n') {
mrk_lexer_advance(lexer);
mrk_lexer_emit(out, lexer, mrk_token_type_blank_line);
} else {
mrk_lexer_emit(out, lexer, mrk_token_type_newline);
}
break;
case ' ': {
// Either a double space or a line break
if (mrk_lexer_peek_str(lexer, " \n")) {
mrk_lexer_advance_n(lexer, 2);
mrk_lexer_emit(out, lexer, mrk_token_type_line_break);
} else {
mrk_lexer_advance_eq(lexer, ' ');
mrk_lexer_emit(out, lexer, mrk_token_type_spaces);
}
} break;
default: {
// Match ordered list headers
if (isdigit(c)) {
mrk_lexer_advance(lexer);
while (isdigit(mrk_lexer_peek(lexer))) {
mrk_lexer_advance(lexer);
}
if (mrk_lexer_peek(lexer) == '.') {
mrk_lexer_advance(lexer);
mrk_lexer_emit(out, lexer, mrk_token_type_dotted_number);
}
// Doesn't end with a dot, so it's just a word that happens to start
// with a number
else {
mrk_lexer_advance_word(lexer);
mrk_lexer_emit(out, lexer, mrk_token_type_word);
}
}
// Any other special scenarios we simply parse as a word
else {
mrk_lexer_advance_word(lexer);
mrk_lexer_emit(out, lexer, mrk_token_type_word);
}
} break;
}
}
mrk_lexer_err mrk_lexer_next(mrk_token *out, mrk_lexer *lexer) {
if (mrk_lexer_done(lexer)) {
return mrk_lexer_err_done;
@ -258,3 +378,11 @@ mrk_lexer_err mrk_lexer_next(mrk_token *out, mrk_lexer *lexer) {
return lexer->token.emitted ? mrk_lexer_err_ok : mrk_lexer_err_done;
}
size_t mrk_token_len(mrk_token t) { return t.end - t.start; }
size_t mrk_token_lexeme(const char **out, mrk_lexer *lexer, mrk_token t) {
*out = lexer->buf.s + t.start;
return t.end - t.start;
}

View File

@ -26,4 +26,49 @@ mrk_err mrk_parser_parse(mrk_ast_node **out, mrk_parser *parser) {
return mrk_err_ok;
}
mrk_err mrk_parser_parse_block(mrk_ast_node **out, mrk_parser *parser) {}
mrk_err mrk_parser_parse_block(mrk_ast_node **out, mrk_parser *parser) {
mrk_token t = mrk_parser_peek(parser);
mrk_err (*parse_fn)(mrk_ast_node **, mrk_parser *) = NULL;
switch (t.type) {
case mrk_token_type_pounds: {
parse_fn = mrk_parser_parse_header;
break;
}
}
if (parse_fn == NULL) {
MRK_PARSE_ERR(parser, t, "Unexpected token.");
}
return parse_fn(out, parser);
}
/* mrk_err mrk_parser_parse_ */
mrk_err mrk_parser_parse_header(mrk_ast_node **out, mrk_parser *parser) {
mrk_token t = mrk_parser_peek(parser);
if (mrk_token_len(t) > MRK_MAX_HEADER_LEN) {
MRK_PARSE_ERR(parser, t, "Headers can be at most 6 levels deep.");
}
mrk_parser_advance(parser);
mrk_ast_node *header;
MRK_RES(mrk_ast_node_init(&header));
header->args[0].num = mrk_token_len(t);
// Headers are blocks of their own, so they're delimited by blank lines
while (!mrk_parser_done(parser) &&
(t = mrk_parser_peek(parser)).type != mrk_token_type_blank_line) {
switch (t.type) { /* case */ }
}
// Skip blank line
mrk_parser_advance(parser);
*out = header;
return mrk_err_ok;
}