2013-10-04 19:53:11 +01:00
|
|
|
#include <unistd.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <stdint.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <ctype.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <assert.h>
|
|
|
|
|
|
|
|
#include "misc.h"
|
2013-12-21 18:17:45 +00:00
|
|
|
#include "mpconfig.h"
|
2014-01-21 21:40:13 +00:00
|
|
|
#include "qstr.h"
|
2013-10-04 19:53:11 +01:00
|
|
|
#include "lexer.h"
|
|
|
|
#include "parse.h"
|
|
|
|
|
|
|
|
#define RULE_ACT_KIND_MASK (0xf0)
|
|
|
|
#define RULE_ACT_ARG_MASK (0x0f)
|
|
|
|
#define RULE_ACT_OR (0x10)
|
|
|
|
#define RULE_ACT_AND (0x20)
|
|
|
|
#define RULE_ACT_LIST (0x30)
|
|
|
|
|
|
|
|
#define RULE_ARG_BLANK (0x0000)
|
|
|
|
#define RULE_ARG_KIND_MASK (0xf000)
|
|
|
|
#define RULE_ARG_ARG_MASK (0x0fff)
|
|
|
|
#define RULE_ARG_TOK (0x1000)
|
|
|
|
#define RULE_ARG_RULE (0x2000)
|
|
|
|
#define RULE_ARG_OPT_TOK (0x3000)
|
|
|
|
#define RULE_ARG_OPT_RULE (0x4000)
|
|
|
|
|
|
|
|
// (un)comment to use rule names; for debugging
|
|
|
|
//#define USE_RULE_NAME (1)
|
|
|
|
|
|
|
|
typedef struct _rule_t {
|
|
|
|
byte rule_id;
|
|
|
|
byte act;
|
|
|
|
#ifdef USE_RULE_NAME
|
|
|
|
const char *rule_name;
|
|
|
|
#endif
|
|
|
|
uint16_t arg[];
|
|
|
|
} rule_t;
|
|
|
|
|
|
|
|
enum {
|
|
|
|
RULE_none = 0,
|
|
|
|
#define DEF_RULE(rule, comp, kind, arg...) RULE_##rule,
|
|
|
|
#include "grammar.h"
|
|
|
|
#undef DEF_RULE
|
|
|
|
RULE_maximum_number_of,
|
|
|
|
};
|
|
|
|
|
|
|
|
#define or(n) (RULE_ACT_OR | n)
|
|
|
|
#define and(n) (RULE_ACT_AND | n)
|
|
|
|
#define one_or_more (RULE_ACT_LIST | 2)
|
|
|
|
#define list (RULE_ACT_LIST | 1)
|
|
|
|
#define list_with_end (RULE_ACT_LIST | 3)
|
2013-12-21 18:17:45 +00:00
|
|
|
#define tok(t) (RULE_ARG_TOK | MP_TOKEN_##t)
|
2013-10-04 19:53:11 +01:00
|
|
|
#define rule(r) (RULE_ARG_RULE | RULE_##r)
|
2013-12-21 18:17:45 +00:00
|
|
|
#define opt_tok(t) (RULE_ARG_OPT_TOK | MP_TOKEN_##t)
|
2013-10-04 19:53:11 +01:00
|
|
|
#define opt_rule(r) (RULE_ARG_OPT_RULE | RULE_##r)
|
|
|
|
#ifdef USE_RULE_NAME
|
2013-10-19 18:28:01 +01:00
|
|
|
#define DEF_RULE(rule, comp, kind, arg...) static const rule_t rule_##rule = { RULE_##rule, kind, #rule, { arg } };
|
2013-10-04 19:53:11 +01:00
|
|
|
#else
|
2013-10-19 18:28:01 +01:00
|
|
|
#define DEF_RULE(rule, comp, kind, arg...) static const rule_t rule_##rule = { RULE_##rule, kind, { arg } };
|
2013-10-04 19:53:11 +01:00
|
|
|
#endif
|
|
|
|
#include "grammar.h"
|
|
|
|
#undef or
|
|
|
|
#undef and
|
|
|
|
#undef list
|
|
|
|
#undef list_with_end
|
|
|
|
#undef tok
|
|
|
|
#undef rule
|
|
|
|
#undef opt_tok
|
|
|
|
#undef opt_rule
|
|
|
|
#undef one_or_more
|
|
|
|
#undef DEF_RULE
|
|
|
|
|
2013-10-19 18:28:01 +01:00
|
|
|
static const rule_t *rules[] = {
|
2013-10-04 19:53:11 +01:00
|
|
|
NULL,
|
|
|
|
#define DEF_RULE(rule, comp, kind, arg...) &rule_##rule,
|
|
|
|
#include "grammar.h"
|
|
|
|
#undef DEF_RULE
|
|
|
|
};
|
|
|
|
|
|
|
|
typedef struct _rule_stack_t {
|
2014-01-18 23:24:36 +00:00
|
|
|
unsigned int src_line : 24;
|
|
|
|
unsigned int rule_id : 8;
|
2013-10-04 19:53:11 +01:00
|
|
|
int32_t arg_i; // what should be the size and signedness?
|
|
|
|
} rule_stack_t;
|
|
|
|
|
|
|
|
typedef struct _parser_t {
|
|
|
|
uint rule_stack_alloc;
|
|
|
|
uint rule_stack_top;
|
|
|
|
rule_stack_t *rule_stack;
|
|
|
|
|
2014-01-12 13:55:24 +00:00
|
|
|
uint result_stack_alloc;
|
2013-10-04 19:53:11 +01:00
|
|
|
uint result_stack_top;
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t *result_stack;
|
2014-01-18 23:24:36 +00:00
|
|
|
|
|
|
|
mp_lexer_t *lexer;
|
2013-10-04 19:53:11 +01:00
|
|
|
} parser_t;
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
static void push_rule(parser_t *parser, int src_line, const rule_t *rule, int arg_i) {
|
2013-10-04 19:53:11 +01:00
|
|
|
if (parser->rule_stack_top >= parser->rule_stack_alloc) {
|
2013-12-29 19:33:23 +00:00
|
|
|
parser->rule_stack = m_renew(rule_stack_t, parser->rule_stack, parser->rule_stack_alloc, parser->rule_stack_alloc * 2);
|
2013-10-04 19:53:11 +01:00
|
|
|
parser->rule_stack_alloc *= 2;
|
|
|
|
}
|
2014-01-18 23:24:36 +00:00
|
|
|
rule_stack_t *rs = &parser->rule_stack[parser->rule_stack_top++];
|
|
|
|
rs->src_line = src_line;
|
|
|
|
rs->rule_id = rule->rule_id;
|
|
|
|
rs->arg_i = arg_i;
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static void push_rule_from_arg(parser_t *parser, uint arg) {
|
|
|
|
assert((arg & RULE_ARG_KIND_MASK) == RULE_ARG_RULE || (arg & RULE_ARG_KIND_MASK) == RULE_ARG_OPT_RULE);
|
|
|
|
uint rule_id = arg & RULE_ARG_ARG_MASK;
|
|
|
|
assert(rule_id < RULE_maximum_number_of);
|
2014-01-18 23:24:36 +00:00
|
|
|
push_rule(parser, mp_lexer_cur(parser->lexer)->src_line, rules[rule_id], 0);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
static void pop_rule(parser_t *parser, const rule_t **rule, uint *arg_i, uint *src_line) {
|
2013-10-04 19:53:11 +01:00
|
|
|
parser->rule_stack_top -= 1;
|
|
|
|
*rule = rules[parser->rule_stack[parser->rule_stack_top].rule_id];
|
|
|
|
*arg_i = parser->rule_stack[parser->rule_stack_top].arg_i;
|
2014-01-18 23:24:36 +00:00
|
|
|
*src_line = parser->rule_stack[parser->rule_stack_top].src_line;
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t mp_parse_node_new_leaf(machine_int_t kind, machine_int_t arg) {
|
|
|
|
return (mp_parse_node_t)(kind | (arg << 4));
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
//int num_parse_nodes_allocated = 0;
|
|
|
|
mp_parse_node_struct_t *parse_node_new_struct(int src_line, int rule_id, int num_args) {
|
2014-01-12 13:55:24 +00:00
|
|
|
mp_parse_node_struct_t *pn = m_new_obj_var(mp_parse_node_struct_t, mp_parse_node_t, num_args);
|
2014-01-18 23:24:36 +00:00
|
|
|
pn->source_line = src_line;
|
2013-10-04 19:53:11 +01:00
|
|
|
pn->kind_num_nodes = (rule_id & 0xff) | (num_args << 8);
|
2014-01-18 23:24:36 +00:00
|
|
|
//num_parse_nodes_allocated += 1;
|
2013-10-04 19:53:11 +01:00
|
|
|
return pn;
|
|
|
|
}
|
|
|
|
|
2014-01-19 11:48:48 +00:00
|
|
|
#if MICROPY_DEBUG_PRINTERS
|
|
|
|
void mp_parse_node_print(mp_parse_node_t pn, int indent) {
|
2014-01-18 23:24:36 +00:00
|
|
|
if (MP_PARSE_NODE_IS_STRUCT(pn)) {
|
|
|
|
printf("[% 4d] ", (int)((mp_parse_node_struct_t*)pn)->source_line);
|
|
|
|
} else {
|
|
|
|
printf(" ");
|
|
|
|
}
|
2013-10-04 19:53:11 +01:00
|
|
|
for (int i = 0; i < indent; i++) {
|
|
|
|
printf(" ");
|
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
if (MP_PARSE_NODE_IS_NULL(pn)) {
|
2013-10-04 19:53:11 +01:00
|
|
|
printf("NULL\n");
|
2013-12-21 18:17:45 +00:00
|
|
|
} else if (MP_PARSE_NODE_IS_LEAF(pn)) {
|
|
|
|
int arg = MP_PARSE_NODE_LEAF_ARG(pn);
|
|
|
|
switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
|
|
|
|
case MP_PARSE_NODE_ID: printf("id(%s)\n", qstr_str(arg)); break;
|
|
|
|
case MP_PARSE_NODE_SMALL_INT: printf("int(%d)\n", arg); break;
|
|
|
|
case MP_PARSE_NODE_INTEGER: printf("int(%s)\n", qstr_str(arg)); break;
|
|
|
|
case MP_PARSE_NODE_DECIMAL: printf("dec(%s)\n", qstr_str(arg)); break;
|
|
|
|
case MP_PARSE_NODE_STRING: printf("str(%s)\n", qstr_str(arg)); break;
|
|
|
|
case MP_PARSE_NODE_BYTES: printf("bytes(%s)\n", qstr_str(arg)); break;
|
|
|
|
case MP_PARSE_NODE_TOKEN: printf("tok(%d)\n", arg); break;
|
2013-10-04 19:53:11 +01:00
|
|
|
default: assert(0);
|
|
|
|
}
|
|
|
|
} else {
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_struct_t *pns2 = (mp_parse_node_struct_t*)pn;
|
2013-10-04 19:53:11 +01:00
|
|
|
int n = pns2->kind_num_nodes >> 8;
|
|
|
|
#ifdef USE_RULE_NAME
|
2013-12-21 18:17:45 +00:00
|
|
|
printf("%s(%d) (n=%d)\n", rules[MP_PARSE_NODE_STRUCT_KIND(pns2)]->rule_name, MP_PARSE_NODE_STRUCT_KIND(pns2), n);
|
2013-10-04 19:53:11 +01:00
|
|
|
#else
|
2013-12-21 18:17:45 +00:00
|
|
|
printf("rule(%u) (n=%d)\n", (uint)MP_PARSE_NODE_STRUCT_KIND(pns2), n);
|
2013-10-04 19:53:11 +01:00
|
|
|
#endif
|
|
|
|
for (int i = 0; i < n; i++) {
|
2014-01-19 11:48:48 +00:00
|
|
|
mp_parse_node_print(pns2->nodes[i], indent + 2);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2014-01-19 11:48:48 +00:00
|
|
|
#endif // MICROPY_DEBUG_PRINTERS
|
2013-10-04 19:53:11 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
static void result_stack_show(parser_t *parser) {
|
|
|
|
printf("result stack, most recent first\n");
|
|
|
|
for (int i = parser->result_stack_top - 1; i >= 0; i--) {
|
2014-01-19 11:48:48 +00:00
|
|
|
mp_parse_node_print(parser->result_stack[i], 0);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
static mp_parse_node_t pop_result(parser_t *parser) {
|
2013-10-04 19:53:11 +01:00
|
|
|
assert(parser->result_stack_top > 0);
|
|
|
|
return parser->result_stack[--parser->result_stack_top];
|
|
|
|
}
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
static mp_parse_node_t peek_result(parser_t *parser, int pos) {
|
2013-10-04 19:53:11 +01:00
|
|
|
assert(parser->result_stack_top > pos);
|
|
|
|
return parser->result_stack[parser->result_stack_top - 1 - pos];
|
|
|
|
}
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
static void push_result_node(parser_t *parser, mp_parse_node_t pn) {
|
2014-01-12 13:55:24 +00:00
|
|
|
if (parser->result_stack_top >= parser->result_stack_alloc) {
|
|
|
|
parser->result_stack = m_renew(mp_parse_node_t, parser->result_stack, parser->result_stack_alloc, parser->result_stack_alloc * 2);
|
|
|
|
parser->result_stack_alloc *= 2;
|
|
|
|
}
|
2013-10-04 19:53:11 +01:00
|
|
|
parser->result_stack[parser->result_stack_top++] = pn;
|
|
|
|
}
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
static void push_result_token(parser_t *parser, const mp_lexer_t *lex) {
|
|
|
|
const mp_token_t *tok = mp_lexer_cur(lex);
|
|
|
|
mp_parse_node_t pn;
|
|
|
|
if (tok->kind == MP_TOKEN_NAME) {
|
2014-01-21 21:40:13 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_ID, qstr_from_strn(tok->str, tok->len));
|
2013-12-21 18:17:45 +00:00
|
|
|
} else if (tok->kind == MP_TOKEN_NUMBER) {
|
2013-10-04 19:53:11 +01:00
|
|
|
bool dec = false;
|
|
|
|
bool small_int = true;
|
2014-01-11 00:33:29 +00:00
|
|
|
machine_int_t int_val = 0;
|
2013-10-04 19:53:11 +01:00
|
|
|
int len = tok->len;
|
|
|
|
const char *str = tok->str;
|
|
|
|
int base = 10;
|
|
|
|
int i = 0;
|
|
|
|
if (len >= 3 && str[0] == '0') {
|
|
|
|
if (str[1] == 'o' || str[1] == 'O') {
|
|
|
|
// octal
|
|
|
|
base = 8;
|
|
|
|
i = 2;
|
|
|
|
} else if (str[1] == 'x' || str[1] == 'X') {
|
|
|
|
// hexadecimal
|
|
|
|
base = 16;
|
|
|
|
i = 2;
|
|
|
|
} else if (str[1] == 'b' || str[1] == 'B') {
|
|
|
|
// binary
|
|
|
|
base = 2;
|
|
|
|
i = 2;
|
|
|
|
}
|
|
|
|
}
|
2014-01-11 00:33:29 +00:00
|
|
|
bool overflow = false;
|
2013-10-04 19:53:11 +01:00
|
|
|
for (; i < len; i++) {
|
2014-01-11 00:33:29 +00:00
|
|
|
machine_int_t old_val = int_val;
|
2013-12-30 18:23:50 +00:00
|
|
|
if (unichar_isdigit(str[i]) && str[i] - '0' < base) {
|
2013-10-04 19:53:11 +01:00
|
|
|
int_val = base * int_val + str[i] - '0';
|
|
|
|
} else if (base == 16 && 'a' <= str[i] && str[i] <= 'f') {
|
|
|
|
int_val = base * int_val + str[i] - 'a' + 10;
|
2013-12-29 13:03:49 +00:00
|
|
|
} else if (base == 16 && 'A' <= str[i] && str[i] <= 'F') {
|
2013-10-04 19:53:11 +01:00
|
|
|
int_val = base * int_val + str[i] - 'A' + 10;
|
2013-11-02 19:47:57 +00:00
|
|
|
} else if (str[i] == '.' || str[i] == 'e' || str[i] == 'E' || str[i] == 'j' || str[i] == 'J') {
|
2013-10-04 19:53:11 +01:00
|
|
|
dec = true;
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
small_int = false;
|
|
|
|
break;
|
|
|
|
}
|
2014-01-11 00:33:29 +00:00
|
|
|
if (int_val < old_val) {
|
|
|
|
// If new value became less than previous, it's overflow
|
|
|
|
overflow = true;
|
|
|
|
} else if ((old_val ^ int_val) & WORD_MSBIT_HIGH) {
|
|
|
|
// If signed number changed sign - it's overflow
|
|
|
|
overflow = true;
|
|
|
|
}
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
if (dec) {
|
2014-01-21 21:40:13 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_DECIMAL, qstr_from_strn(str, len));
|
2014-01-11 00:33:29 +00:00
|
|
|
} else if (small_int && !overflow && MP_FIT_SMALL_INT(int_val)) {
|
2013-12-21 18:17:45 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_SMALL_INT, int_val);
|
2013-10-04 19:53:11 +01:00
|
|
|
} else {
|
2014-01-21 21:40:13 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_INTEGER, qstr_from_strn(str, len));
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
} else if (tok->kind == MP_TOKEN_STRING) {
|
2014-01-21 21:40:13 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_STRING, qstr_from_strn(tok->str, tok->len));
|
2013-12-21 18:17:45 +00:00
|
|
|
} else if (tok->kind == MP_TOKEN_BYTES) {
|
2014-01-21 21:40:13 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_BYTES, qstr_from_strn(tok->str, tok->len));
|
2013-10-04 19:53:11 +01:00
|
|
|
} else {
|
2013-12-21 18:17:45 +00:00
|
|
|
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_TOKEN, tok->kind);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
push_result_node(parser, pn);
|
|
|
|
}
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
static void push_result_rule(parser_t *parser, int src_line, const rule_t *rule, int num_args) {
|
|
|
|
mp_parse_node_struct_t *pn = parse_node_new_struct(src_line, rule->rule_id, num_args);
|
2013-10-04 19:53:11 +01:00
|
|
|
for (int i = num_args; i > 0; i--) {
|
|
|
|
pn->nodes[i - 1] = pop_result(parser);
|
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
push_result_node(parser, (mp_parse_node_t)pn);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
|
2014-01-15 21:23:31 +00:00
|
|
|
mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr *exc_id_out, const char **exc_msg_out) {
|
2014-01-12 13:55:24 +00:00
|
|
|
|
|
|
|
// allocate memory for the parser and its stacks
|
|
|
|
|
|
|
|
parser_t *parser = m_new_obj(parser_t);
|
|
|
|
|
2013-10-04 19:53:11 +01:00
|
|
|
parser->rule_stack_alloc = 64;
|
|
|
|
parser->rule_stack_top = 0;
|
|
|
|
parser->rule_stack = m_new(rule_stack_t, parser->rule_stack_alloc);
|
|
|
|
|
2014-01-12 13:55:24 +00:00
|
|
|
parser->result_stack_alloc = 64;
|
2013-10-04 19:53:11 +01:00
|
|
|
parser->result_stack_top = 0;
|
2014-01-12 13:55:24 +00:00
|
|
|
parser->result_stack = m_new(mp_parse_node_t, parser->result_stack_alloc);
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
parser->lexer = lex;
|
|
|
|
|
2014-01-12 13:55:24 +00:00
|
|
|
// work out the top-level rule to use, and push it on the stack
|
2013-10-18 19:58:12 +01:00
|
|
|
int top_level_rule;
|
|
|
|
switch (input_kind) {
|
2013-12-21 18:17:45 +00:00
|
|
|
case MP_PARSE_SINGLE_INPUT: top_level_rule = RULE_single_input; break;
|
2014-01-15 22:14:03 +00:00
|
|
|
case MP_PARSE_EVAL_INPUT: top_level_rule = RULE_eval_input; break;
|
2013-10-18 19:58:12 +01:00
|
|
|
default: top_level_rule = RULE_file_input;
|
|
|
|
}
|
2014-01-18 23:24:36 +00:00
|
|
|
push_rule(parser, mp_lexer_cur(lex)->src_line, rules[top_level_rule], 0);
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2014-01-12 13:55:24 +00:00
|
|
|
// parse!
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
uint n, i; // state for the current rule
|
|
|
|
uint rule_src_line; // source line for the first token matched by the current rule
|
2013-10-04 19:53:11 +01:00
|
|
|
bool backtrack = false;
|
2014-01-18 23:24:36 +00:00
|
|
|
const rule_t *rule = NULL;
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_token_kind_t tok_kind;
|
2013-10-04 19:53:11 +01:00
|
|
|
bool emit_rule;
|
|
|
|
bool had_trailing_sep;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
next_rule:
|
|
|
|
if (parser->rule_stack_top == 0) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2014-01-18 23:24:36 +00:00
|
|
|
pop_rule(parser, &rule, &i, &rule_src_line);
|
2013-10-04 19:53:11 +01:00
|
|
|
n = rule->act & RULE_ACT_ARG_MASK;
|
|
|
|
|
|
|
|
/*
|
|
|
|
// debugging
|
|
|
|
printf("depth=%d ", parser->rule_stack_top);
|
|
|
|
for (int j = 0; j < parser->rule_stack_top; ++j) {
|
|
|
|
printf(" ");
|
|
|
|
}
|
|
|
|
printf("%s n=%d i=%d bt=%d\n", rule->rule_name, n, i, backtrack);
|
|
|
|
*/
|
|
|
|
|
|
|
|
switch (rule->act & RULE_ACT_KIND_MASK) {
|
|
|
|
case RULE_ACT_OR:
|
|
|
|
if (i > 0 && !backtrack) {
|
|
|
|
goto next_rule;
|
|
|
|
} else {
|
|
|
|
backtrack = false;
|
|
|
|
}
|
|
|
|
for (; i < n - 1; ++i) {
|
|
|
|
switch (rule->arg[i] & RULE_ARG_KIND_MASK) {
|
|
|
|
case RULE_ARG_TOK:
|
2013-12-21 18:17:45 +00:00
|
|
|
if (mp_lexer_is_kind(lex, rule->arg[i] & RULE_ARG_ARG_MASK)) {
|
2013-10-04 19:53:11 +01:00
|
|
|
push_result_token(parser, lex);
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_lexer_to_next(lex);
|
2013-10-04 19:53:11 +01:00
|
|
|
goto next_rule;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case RULE_ARG_RULE:
|
2014-01-18 23:24:36 +00:00
|
|
|
push_rule(parser, rule_src_line, rule, i + 1); // save this or-rule
|
|
|
|
push_rule_from_arg(parser, rule->arg[i]); // push child of or-rule
|
2013-10-04 19:53:11 +01:00
|
|
|
goto next_rule;
|
|
|
|
default:
|
|
|
|
assert(0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if ((rule->arg[i] & RULE_ARG_KIND_MASK) == RULE_ARG_TOK) {
|
2013-12-21 18:17:45 +00:00
|
|
|
if (mp_lexer_is_kind(lex, rule->arg[i] & RULE_ARG_ARG_MASK)) {
|
2013-10-04 19:53:11 +01:00
|
|
|
push_result_token(parser, lex);
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_lexer_to_next(lex);
|
2013-10-04 19:53:11 +01:00
|
|
|
} else {
|
|
|
|
backtrack = true;
|
|
|
|
goto next_rule;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
push_rule_from_arg(parser, rule->arg[i]);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case RULE_ACT_AND:
|
|
|
|
|
|
|
|
// failed, backtrack if we can, else syntax error
|
|
|
|
if (backtrack) {
|
|
|
|
assert(i > 0);
|
|
|
|
if ((rule->arg[i - 1] & RULE_ARG_KIND_MASK) == RULE_ARG_OPT_RULE) {
|
|
|
|
// an optional rule that failed, so continue with next arg
|
2013-12-21 18:17:45 +00:00
|
|
|
push_result_node(parser, MP_PARSE_NODE_NULL);
|
2013-10-04 19:53:11 +01:00
|
|
|
backtrack = false;
|
|
|
|
} else {
|
|
|
|
// a mandatory rule that failed, so propagate backtrack
|
|
|
|
if (i > 1) {
|
|
|
|
// already eaten tokens so can't backtrack
|
|
|
|
goto syntax_error;
|
|
|
|
} else {
|
|
|
|
goto next_rule;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// progress through the rule
|
|
|
|
for (; i < n; ++i) {
|
|
|
|
switch (rule->arg[i] & RULE_ARG_KIND_MASK) {
|
|
|
|
case RULE_ARG_TOK:
|
|
|
|
// need to match a token
|
|
|
|
tok_kind = rule->arg[i] & RULE_ARG_ARG_MASK;
|
2013-12-21 18:17:45 +00:00
|
|
|
if (mp_lexer_is_kind(lex, tok_kind)) {
|
2013-10-04 19:53:11 +01:00
|
|
|
// matched token
|
2013-12-21 18:17:45 +00:00
|
|
|
if (tok_kind == MP_TOKEN_NAME) {
|
2013-10-04 19:53:11 +01:00
|
|
|
push_result_token(parser, lex);
|
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_lexer_to_next(lex);
|
2013-10-04 19:53:11 +01:00
|
|
|
} else {
|
|
|
|
// failed to match token
|
|
|
|
if (i > 0) {
|
|
|
|
// already eaten tokens so can't backtrack
|
|
|
|
goto syntax_error;
|
|
|
|
} else {
|
|
|
|
// this rule failed, so backtrack
|
|
|
|
backtrack = true;
|
|
|
|
goto next_rule;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case RULE_ARG_RULE:
|
|
|
|
case RULE_ARG_OPT_RULE:
|
2014-01-18 23:24:36 +00:00
|
|
|
push_rule(parser, rule_src_line, rule, i + 1); // save this and-rule
|
|
|
|
push_rule_from_arg(parser, rule->arg[i]); // push child of and-rule
|
2013-10-04 19:53:11 +01:00
|
|
|
goto next_rule;
|
|
|
|
default:
|
|
|
|
assert(0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(i == n);
|
|
|
|
|
|
|
|
// matched the rule, so now build the corresponding parse_node
|
|
|
|
|
|
|
|
// count number of arguments for the parse_node
|
|
|
|
i = 0;
|
|
|
|
emit_rule = 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;
|
2013-12-21 18:17:45 +00:00
|
|
|
if (tok_kind >= MP_TOKEN_NAME) {
|
2013-10-04 19:53:11 +01:00
|
|
|
emit_rule = true;
|
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
if (tok_kind == MP_TOKEN_NAME) {
|
2013-10-04 19:53:11 +01:00
|
|
|
// only tokens which were names are pushed to stack
|
|
|
|
i += 1;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// rules are always pushed
|
|
|
|
i += 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// 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;
|
|
|
|
}
|
|
|
|
|
|
|
|
// 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)]
|
2013-10-06 00:28:28 +01:00
|
|
|
// 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) {
|
2013-10-04 19:53:11 +01:00
|
|
|
emit_rule = 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;
|
2013-12-21 18:17:45 +00:00
|
|
|
push_result_node(parser, MP_PARSE_NODE_NULL);
|
2013-10-04 19:53:11 +01:00
|
|
|
i += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
int num_not_nil = 0;
|
|
|
|
for (int x = 0; x < i; ++x) {
|
2013-12-21 18:17:45 +00:00
|
|
|
if (peek_result(parser, x) != MP_PARSE_NODE_NULL) {
|
2013-10-04 19:53:11 +01:00
|
|
|
num_not_nil += 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
//printf("done and %s n=%d i=%d notnil=%d\n", rule->rule_name, n, i, num_not_nil);
|
|
|
|
if (emit_rule) {
|
2014-01-18 23:24:36 +00:00
|
|
|
push_result_rule(parser, rule_src_line, rule, i);
|
2013-10-04 19:53:11 +01:00
|
|
|
} else if (num_not_nil == 0) {
|
2014-01-18 23:24:36 +00:00
|
|
|
push_result_rule(parser, rule_src_line, rule, i); // needed for, eg, atom_paren, testlist_comp_3b
|
2013-10-04 19:53:11 +01:00
|
|
|
//result_stack_show(parser);
|
|
|
|
//assert(0);
|
|
|
|
} else if (num_not_nil == 1) {
|
|
|
|
// single result, leave it on stack
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t pn = MP_PARSE_NODE_NULL;
|
2013-10-04 19:53:11 +01:00
|
|
|
for (int x = 0; x < i; ++x) {
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t pn2 = pop_result(parser);
|
|
|
|
if (pn2 != MP_PARSE_NODE_NULL) {
|
2013-10-04 19:53:11 +01:00
|
|
|
pn = pn2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
push_result_node(parser, pn);
|
|
|
|
} else {
|
2014-01-18 23:24:36 +00:00
|
|
|
push_result_rule(parser, rule_src_line, rule, i);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case RULE_ACT_LIST:
|
|
|
|
// n=2 is: item item*
|
|
|
|
// n=1 is: item (sep item)*
|
|
|
|
// n=3 is: item (sep item)* [sep]
|
|
|
|
if (backtrack) {
|
|
|
|
list_backtrack:
|
|
|
|
had_trailing_sep = 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;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (i == 1) {
|
|
|
|
// fail on item, first time round; propagate backtrack
|
|
|
|
goto next_rule;
|
|
|
|
} else if ((i & 1) == 1) {
|
|
|
|
// 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;
|
|
|
|
} else {
|
|
|
|
// list doesn't allowing trailing separator; fail
|
|
|
|
goto syntax_error;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// fail on separator; finish parsing list
|
|
|
|
backtrack = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
for (;;) {
|
|
|
|
uint arg = rule->arg[i & 1 & n];
|
|
|
|
switch (arg & RULE_ARG_KIND_MASK) {
|
|
|
|
case RULE_ARG_TOK:
|
2013-12-21 18:17:45 +00:00
|
|
|
if (mp_lexer_is_kind(lex, arg & RULE_ARG_ARG_MASK)) {
|
2013-10-04 19:53:11 +01:00
|
|
|
if (i & 1 & n) {
|
|
|
|
// separators which are tokens are not pushed to result stack
|
|
|
|
} else {
|
|
|
|
push_result_token(parser, lex);
|
|
|
|
}
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_lexer_to_next(lex);
|
2013-10-04 19:53:11 +01:00
|
|
|
// got element of list, so continue parsing list
|
|
|
|
i += 1;
|
|
|
|
} else {
|
|
|
|
// couldn't get element of list
|
|
|
|
i += 1;
|
|
|
|
backtrack = true;
|
|
|
|
goto list_backtrack;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case RULE_ARG_RULE:
|
2014-01-18 23:24:36 +00:00
|
|
|
push_rule(parser, rule_src_line, rule, i + 1); // save this list-rule
|
|
|
|
push_rule_from_arg(parser, arg); // push child of list-rule
|
2013-10-04 19:53:11 +01:00
|
|
|
goto next_rule;
|
|
|
|
default:
|
|
|
|
assert(0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
assert(i >= 1);
|
|
|
|
|
|
|
|
// compute number of elements in list, result in i
|
|
|
|
i -= 1;
|
|
|
|
if ((n & 1) && (rule->arg[1] & RULE_ARG_KIND_MASK) == RULE_ARG_TOK) {
|
|
|
|
// don't count separators when they are tokens
|
|
|
|
i = (i + 1) / 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (i == 1) {
|
|
|
|
// list matched single item
|
|
|
|
if (had_trailing_sep) {
|
|
|
|
// if there was a trailing separator, make a list of a single item
|
2014-01-18 23:24:36 +00:00
|
|
|
push_result_rule(parser, rule_src_line, rule, i);
|
2013-10-04 19:53:11 +01:00
|
|
|
} else {
|
|
|
|
// just leave single item on stack (ie don't wrap in a list)
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
//printf("done list %s %d %d\n", rule->rule_name, n, i);
|
2014-01-18 23:24:36 +00:00
|
|
|
push_result_rule(parser, rule_src_line, rule, i);
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
assert(0);
|
|
|
|
}
|
|
|
|
}
|
2013-10-09 15:09:52 +01:00
|
|
|
|
|
|
|
// check we are at the end of the token stream
|
2013-12-21 18:17:45 +00:00
|
|
|
if (!mp_lexer_is_kind(lex, MP_TOKEN_END)) {
|
2013-10-09 15:09:52 +01:00
|
|
|
goto syntax_error;
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|
2013-10-09 15:09:52 +01:00
|
|
|
|
2013-10-04 19:53:11 +01:00
|
|
|
//printf("--------------\n");
|
|
|
|
//result_stack_show(parser);
|
2014-01-12 13:55:24 +00:00
|
|
|
//printf("rule stack alloc: %d\n", parser->rule_stack_alloc);
|
|
|
|
//printf("result stack alloc: %d\n", parser->result_stack_alloc);
|
2013-10-04 19:53:11 +01:00
|
|
|
//printf("number of parse nodes allocated: %d\n", num_parse_nodes_allocated);
|
2014-01-12 13:55:24 +00:00
|
|
|
|
|
|
|
// get the root parse node that we created
|
|
|
|
assert(parser->result_stack_top == 1);
|
|
|
|
mp_parse_node_t result = parser->result_stack[0];
|
|
|
|
|
|
|
|
finished:
|
|
|
|
// free the memory that we don't need anymore
|
|
|
|
m_del(rule_stack_t, parser->rule_stack, parser->rule_stack_alloc);
|
|
|
|
m_del(mp_parse_node_t, parser->result_stack, parser->result_stack_alloc);
|
|
|
|
m_del_obj(parser_t, parser);
|
|
|
|
|
|
|
|
// return the result
|
|
|
|
return result;
|
2013-10-04 19:53:11 +01:00
|
|
|
|
|
|
|
syntax_error:
|
2013-12-21 18:17:45 +00:00
|
|
|
if (mp_lexer_is_kind(lex, MP_TOKEN_INDENT)) {
|
2014-01-15 21:23:31 +00:00
|
|
|
*exc_id_out = MP_QSTR_IndentationError;
|
|
|
|
*exc_msg_out = "unexpected indent";
|
2013-12-21 18:17:45 +00:00
|
|
|
} else if (mp_lexer_is_kind(lex, MP_TOKEN_DEDENT_MISMATCH)) {
|
2014-01-15 21:23:31 +00:00
|
|
|
*exc_id_out = MP_QSTR_IndentationError;
|
|
|
|
*exc_msg_out = "unindent does not match any outer indentation level";
|
2013-10-09 15:09:52 +01:00
|
|
|
} else {
|
2014-01-15 21:23:31 +00:00
|
|
|
*exc_id_out = MP_QSTR_SyntaxError;
|
|
|
|
*exc_msg_out = "invalid syntax";
|
2013-10-04 19:53:11 +01:00
|
|
|
#ifdef USE_RULE_NAME
|
2014-01-15 21:23:31 +00:00
|
|
|
// debugging: print the rule name that failed and the token
|
|
|
|
mp_lexer_show_error_pythonic(lex, rule->rule_name);
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_token_show(mp_lexer_cur(lex));
|
2014-01-15 21:23:31 +00:00
|
|
|
#endif
|
2013-10-09 15:09:52 +01:00
|
|
|
}
|
2014-01-12 13:55:24 +00:00
|
|
|
result = MP_PARSE_NODE_NULL;
|
|
|
|
goto finished;
|
2013-10-04 19:53:11 +01:00
|
|
|
}
|