diff options
Diffstat (limited to 'py/parse.c')
-rw-r--r-- | py/parse.c | 44 |
1 files changed, 22 insertions, 22 deletions
diff --git a/py/parse.c b/py/parse.c index d3786ba956..1d3badbc3f 100644 --- a/py/parse.c +++ b/py/parse.c @@ -189,8 +189,8 @@ static void push_result_token(parser_t *parser, const mp_lexer_t *lex) { if (tok->kind == MP_TOKEN_NAME) { pn = mp_parse_node_new_leaf(MP_PARSE_NODE_ID, qstr_from_strn_copy(tok->str, tok->len)); } else if (tok->kind == MP_TOKEN_NUMBER) { - bool dec = false; - bool small_int = true; + MP_BOOL dec = MP_FALSE; + MP_BOOL small_int = MP_TRUE; int int_val = 0; int len = tok->len; const char *str = tok->str; @@ -219,10 +219,10 @@ static void push_result_token(parser_t *parser, const mp_lexer_t *lex) { } else if (base == 16 && 'A' <= str[i] && str[i] <= 'F') { int_val = base * int_val + str[i] - 'A' + 10; } else if (str[i] == '.' || str[i] == 'e' || str[i] == 'E' || str[i] == 'j' || str[i] == 'J') { - dec = true; + dec = MP_TRUE; break; } else { - small_int = false; + small_int = MP_FALSE; break; } } @@ -269,11 +269,11 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { push_rule(parser, rules[top_level_rule], 0); uint n, i; - bool backtrack = false; + MP_BOOL backtrack = MP_FALSE; const rule_t *rule; mp_token_kind_t tok_kind; - bool emit_rule; - bool had_trailing_sep; + MP_BOOL emit_rule; + MP_BOOL had_trailing_sep; for (;;) { next_rule: @@ -298,7 +298,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { if (i > 0 && !backtrack) { goto next_rule; } else { - backtrack = false; + backtrack = MP_FALSE; } for (; i < n - 1; ++i) { switch (rule->arg[i] & RULE_ARG_KIND_MASK) { @@ -322,7 +322,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { push_result_token(parser, lex); mp_lexer_to_next(lex); } else { - backtrack = true; + backtrack = MP_TRUE; goto next_rule; } } else { @@ -338,7 +338,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { if ((rule->arg[i - 1] & RULE_ARG_KIND_MASK) == RULE_ARG_OPT_RULE) { // an optional rule that failed, so continue with next arg push_result_node(parser, MP_PARSE_NODE_NULL); - backtrack = false; + backtrack = MP_FALSE; } else { // a mandatory rule that failed, so propagate backtrack if (i > 1) { @@ -369,7 +369,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { goto syntax_error; } else { // this rule failed, so backtrack - backtrack = true; + backtrack = MP_TRUE; goto next_rule; } } @@ -395,12 +395,12 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { // count number of arguments for the parse_node i = 0; - emit_rule = false; + emit_rule = MP_FALSE; for (int x = 0; x < n; ++x) { if ((rule->arg[x] & RULE_ARG_KIND_MASK) == RULE_ARG_TOK) { tok_kind = rule->arg[x] & RULE_ARG_ARG_MASK; if (tok_kind >= MP_TOKEN_NAME) { - emit_rule = true; + emit_rule = MP_TRUE; } if (tok_kind == MP_TOKEN_NAME) { // only tokens which were names are pushed to stack @@ -414,19 +414,19 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { // always emit these rules, even if they have only 1 argument if (rule->rule_id == RULE_expr_stmt || rule->rule_id == RULE_yield_stmt) { - emit_rule = true; + emit_rule = MP_TRUE; } // never emit these rules if they have only 1 argument // NOTE: can't put atom_paren here because we need it to distinguisg, for example, [a,b] from [(a,b)] // TODO possibly put varargslist_name, varargslist_equal here as well if (rule->rule_id == RULE_else_stmt || rule->rule_id == RULE_testlist_comp_3b || rule->rule_id == RULE_import_as_names_paren || rule->rule_id == RULE_typedargslist_name || rule->rule_id == RULE_typedargslist_colon || rule->rule_id == RULE_typedargslist_equal || rule->rule_id == RULE_dictorsetmaker_colon || rule->rule_id == RULE_classdef_2 || rule->rule_id == RULE_with_item_as || rule->rule_id == RULE_assert_stmt_extra || rule->rule_id == RULE_as_name || rule->rule_id == RULE_raise_stmt_from || rule->rule_id == RULE_vfpdef) { - emit_rule = false; + emit_rule = MP_FALSE; } // always emit these rules, and add an extra blank node at the end (to be used by the compiler to store data) if (rule->rule_id == RULE_funcdef || rule->rule_id == RULE_classdef || rule->rule_id == RULE_comp_for || rule->rule_id == RULE_lambdef || rule->rule_id == RULE_lambdef_nocond) { - emit_rule = true; + emit_rule = MP_TRUE; push_result_node(parser, MP_PARSE_NODE_NULL); i += 1; } @@ -465,14 +465,14 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { // n=3 is: item (sep item)* [sep] if (backtrack) { list_backtrack: - had_trailing_sep = false; + had_trailing_sep = MP_FALSE; if (n == 2) { if (i == 1) { // fail on item, first time round; propagate backtrack goto next_rule; } else { // fail on item, in later rounds; finish with this rule - backtrack = false; + backtrack = MP_FALSE; } } else { if (i == 1) { @@ -482,15 +482,15 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { // fail on item, in later rounds; have eaten tokens so can't backtrack if (n == 3) { // list allows trailing separator; finish parsing list - had_trailing_sep = true; - backtrack = false; + had_trailing_sep = MP_TRUE; + backtrack = MP_FALSE; } else { // list doesn't allowing trailing separator; fail goto syntax_error; } } else { // fail on separator; finish parsing list - backtrack = false; + backtrack = MP_FALSE; } } } else { @@ -510,7 +510,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) { } else { // couldn't get element of list i += 1; - backtrack = true; + backtrack = MP_TRUE; goto list_backtrack; } break; |