From 17d6e6a265c44569f4533e12cc04442013ab3b3e Mon Sep 17 00:00:00 2001 From: c+1 Date: Fri, 12 May 2023 11:49:12 -0400 Subject: nothing workds help --- src/include/token.h | 6 ++++ src/lexer.c | 88 ++++++++++++++++++++++++++++++++++++++++++++--------- src/main.c | 30 +++++++++++++++++- 3 files changed, 108 insertions(+), 16 deletions(-) (limited to 'src') diff --git a/src/include/token.h b/src/include/token.h index 4ca8356..43d868c 100644 --- a/src/include/token.h +++ b/src/include/token.h @@ -17,6 +17,12 @@ typedef struct TOKEN_STRUC { TOKEN_RORD, // ')' TOKEN_DIRECTIVE_DELIM, // '#' TOKEN_DIRECTIVE, // #DIRECTIVE; + TOKEN_FN_APPLY, // '.' + TOKEN_LIST_DELIM, // ',' + TOKEN_MODULE_MEMBER_DELIM, // ':' + TOKEN_BLOCK_DELIM_START, // '{' + TOKEN_BLOCK_DELIM_END, // '}' + TOKEN_ANY, // '_' TOKEN_EOF, // '\0' } type; diff --git a/src/lexer.c b/src/lexer.c index 77de3e4..d8534c2 100644 --- a/src/lexer.c +++ b/src/lexer.c @@ -19,10 +19,9 @@ lexer_t* lexer_init(char* content) { } void lexer_next(lexer_t* lexer) { - // if not at end of line or file if (lexer->c != '\0' && lexer->i < strlen(lexer->content)) { - lexer->i += 1; // advance index 1 character - lexer->c = lexer->content[lexer->i]; // set character w/ index + lexer->i += 1; + lexer->c = lexer->content[lexer->i]; } } @@ -37,20 +36,33 @@ void lexer_pass(lexer_t* lexer) { token_t* lexer_get_next_token(lexer_t* lexer) { while (lexer->c != '\0' && lexer->i < strlen(lexer->content)) { - if (lexer->c == ' ' || - lexer->c == '\t' || - lexer->c == '\n') { - lexer_pass(lexer); + if ( + lexer->c == ' ' || + lexer->c == '\t' || + lexer->c == '\n' + ) { + lexer_pass(lexer); } - if (isalnum(lexer->c)) + if ( + isalnum(lexer->c) || + lexer->c == '_' + ) { return lexer_get_keyword(lexer); + } switch (lexer->c) { - case '"': + case '\'': return lexer_get_string(lexer); break; case '[': return lexer_get_comment(lexer); break; + case ']': return lexer_next_token( + lexer, + token_init( + TOKEN_COMM_DELIM_END, + lexer_get_c_as_string(lexer) + ) + ); break; case '=': return lexer_get_def_const(lexer); break; case '-': @@ -76,8 +88,45 @@ token_t* lexer_get_next_token(lexer_t* lexer) { lexer_get_c_as_string(lexer) ) ); break; - case '#': + case '#': + return lexer_get_directive(lexer); break; + case '.': return lexer_next_token( + lexer, + token_init( + TOKEN_FN_APPLY, + lexer_get_c_as_string(lexer) + ) + ); break; + case ',': return lexer_next_token( + lexer, + token_init( + TOKEN_LIST_DELIM, + lexer_get_c_as_string(lexer) + ) + ); break; + case ':': return lexer_next_token( + lexer, + token_init( + TOKEN_MODULE_MEMBER_DELIM, + lexer_get_c_as_string(lexer) + ) + ); break; + case '{': return lexer_next_token( + lexer, + token_init( + TOKEN_BLOCK_DELIM_START, + lexer_get_c_as_string(lexer) + ) + ); break; + case '}': return lexer_next_token( + lexer, + token_init( + TOKEN_BLOCK_DELIM_END, + lexer_get_c_as_string(lexer) + ) + ); break; default: + fputs("Unrecognized token.", stderr); exit(1); } } @@ -91,7 +140,7 @@ token_t* lexer_get_string(lexer_t* lexer) { char* str_so_far = calloc(1, sizeof(char)); str_so_far[0] = '\0'; - while (lexer->c != '"') { + while (lexer->c != '\'') { char* current = lexer_get_c_as_string(lexer); str_so_far = realloc( str_so_far, @@ -110,14 +159,23 @@ token_t* lexer_get_string(lexer_t* lexer) { token_t* lexer_get_comment(lexer_t* lexer) { lexer_next(lexer); + char* comment_so_far = calloc(1, sizeof(char)); + while (lexer->c != ']') { - lexer_next(lexer); // don't need to keep track of comments - // for now. might change this later. + char* current = lexer_get_c_as_string(lexer); + comment_so_far = realloc( + comment_so_far, + (strlen(comment_so_far) + strlen(current) * sizeof(char)) + ); + + strcat(comment_so_far, current); + + lexer_next(lexer); } lexer_next(lexer); // skip over closing ] - return token_init(TOKEN_COMM, lexer_get_c_as_string(lexer)); + return token_init(TOKEN_COMM, comment_so_far); } token_t* lexer_get_def_const(lexer_t* lexer) { @@ -165,7 +223,7 @@ token_t* lexer_get_keyword(lexer_t* lexer) { char* str_so_far = calloc(1, sizeof(char)); str_so_far[0] = '\0'; - while (isalnum(lexer->c)) { + while (isalnum(lexer->c) || (lexer->c != '.' && lexer->c != ',')) { char* current = lexer_get_c_as_string(lexer); str_so_far = realloc(str_so_far, (strlen(str_so_far) + strlen(current) * sizeof(char))); strcat(str_so_far, current); diff --git a/src/main.c b/src/main.c index 2d741e4..a463a7e 100644 --- a/src/main.c +++ b/src/main.c @@ -6,10 +6,38 @@ int main(int argc, char* argv[]) { + FILE *fsource; + long fsource_size; + char *source; + + fsource = fopen ("examples/hello.halk", "rb"); + if (!fsource) { fputs("Source file not found.", stderr); exit(1); }; + + fseek(fsource, 0L, SEEK_END); + fsource_size = ftell(fsource); + rewind(fsource); + + source = calloc(1, fsource_size + 1); + if (!source) { + fclose(fsource); + fputs("Memory allocation faled.", stderr); + exit(1); + } + + if (1 != fread(source, fsource_size, 1, fsource)) { + fclose(fsource); + free(source); + fputs("Could not read source file.", stderr); + exit(1); + } + lexer_t* lexer = lexer_init( - "[a simple test script]\nlet it = \"Hello, World!\";\nsay it;" + source ); + fclose(fsource); + free(source); + printf("\n=== INPUT =======\n%s\n=== END INPUT ===\n", lexer->content); token_t* token = NULL; -- cgit v1.2.3