circuitpython/py/parse.c

777 lines
30 KiB
C
Raw Normal View History

/*
* This file is part of the Micro Python project, http://micropython.org/
*
* The MIT License (MIT)
*
* Copyright (c) 2013-2015 Damien P. George
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include <stdbool.h>
2013-10-04 14:53:11 -04:00
#include <stdint.h>
#include <stdio.h>
#include <assert.h>
#include <string.h>
2013-10-04 14:53:11 -04:00
#include "py/nlr.h"
#include "py/lexer.h"
#include "py/parse.h"
#include "py/parsenum.h"
#include "py/smallint.h"
2013-10-04 14:53:11 -04:00
#define RULE_ACT_ARG_MASK (0x0f)
#define RULE_ACT_KIND_MASK (0x30)
#define RULE_ACT_ALLOW_IDENT (0x40)
#define RULE_ACT_ADD_BLANK (0x80)
2013-10-04 14:53:11 -04:00
#define RULE_ACT_OR (0x10)
#define RULE_ACT_AND (0x20)
#define RULE_ACT_LIST (0x30)
#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_RULE (0x3000)
2013-10-04 14:53:11 -04:00
#define ADD_BLANK_NODE(rule) ((rule->act & RULE_ACT_ADD_BLANK) != 0)
2013-10-04 14:53:11 -04:00
// (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 {
#define DEF_RULE(rule, comp, kind, ...) RULE_##rule,
#include "py/grammar.h"
2013-10-04 14:53:11 -04:00
#undef DEF_RULE
RULE_maximum_number_of,
RULE_string, // special node for non-interned string
RULE_bytes, // special node for non-interned bytes
RULE_const_object, // special node for a constant, generic Python object
2013-10-04 14:53:11 -04:00
};
#define ident (RULE_ACT_ALLOW_IDENT)
#define blank (RULE_ACT_ADD_BLANK)
2013-10-04 14:53:11 -04:00
#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)
#define tok(t) (RULE_ARG_TOK | MP_TOKEN_##t)
2013-10-04 14:53:11 -04:00
#define rule(r) (RULE_ARG_RULE | RULE_##r)
#define opt_rule(r) (RULE_ARG_OPT_RULE | RULE_##r)
#ifdef USE_RULE_NAME
#define DEF_RULE(rule, comp, kind, ...) static const rule_t rule_##rule = { RULE_##rule, kind, #rule, { __VA_ARGS__ } };
2013-10-04 14:53:11 -04:00
#else
#define DEF_RULE(rule, comp, kind, ...) static const rule_t rule_##rule = { RULE_##rule, kind, { __VA_ARGS__ } };
2013-10-04 14:53:11 -04:00
#endif
#include "py/grammar.h"
2013-10-04 14:53:11 -04:00
#undef or
#undef and
#undef list
#undef list_with_end
#undef tok
#undef rule
#undef opt_rule
#undef one_or_more
#undef DEF_RULE
STATIC const rule_t *rules[] = {
#define DEF_RULE(rule, comp, kind, ...) &rule_##rule,
#include "py/grammar.h"
2013-10-04 14:53:11 -04:00
#undef DEF_RULE
};
typedef struct _rule_stack_t {
mp_uint_t src_line : BITS_PER_WORD - 8; // maximum bits storing source line number
mp_uint_t rule_id : 8; // this must be large enough to fit largest rule number
mp_uint_t arg_i; // this dictates the maximum nodes in a "list" of things
2013-10-04 14:53:11 -04:00
} rule_stack_t;
typedef struct _parser_t {
bool had_memory_error;
2014-07-03 09:13:33 -04:00
mp_uint_t rule_stack_alloc;
mp_uint_t rule_stack_top;
2013-10-04 14:53:11 -04:00
rule_stack_t *rule_stack;
2014-07-03 09:13:33 -04:00
mp_uint_t result_stack_alloc;
mp_uint_t result_stack_top;
mp_parse_node_t *result_stack;
mp_lexer_t *lexer;
2013-10-04 14:53:11 -04:00
} parser_t;
STATIC inline void memory_error(parser_t *parser) {
parser->had_memory_error = true;
}
2014-07-03 09:13:33 -04:00
STATIC void push_rule(parser_t *parser, mp_uint_t src_line, const rule_t *rule, mp_uint_t arg_i) {
if (parser->had_memory_error) {
return;
}
2013-10-04 14:53:11 -04:00
if (parser->rule_stack_top >= parser->rule_stack_alloc) {
rule_stack_t *rs = m_renew_maybe(rule_stack_t, parser->rule_stack, parser->rule_stack_alloc, parser->rule_stack_alloc + MICROPY_ALLOC_PARSE_RULE_INC, true);
if (rs == NULL) {
memory_error(parser);
return;
}
parser->rule_stack = rs;
parser->rule_stack_alloc += MICROPY_ALLOC_PARSE_RULE_INC;
2013-10-04 14:53:11 -04:00
}
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 14:53:11 -04:00
}
2014-07-03 09:13:33 -04:00
STATIC void push_rule_from_arg(parser_t *parser, mp_uint_t arg) {
2013-10-04 14:53:11 -04:00
assert((arg & RULE_ARG_KIND_MASK) == RULE_ARG_RULE || (arg & RULE_ARG_KIND_MASK) == RULE_ARG_OPT_RULE);
2014-07-03 09:13:33 -04:00
mp_uint_t rule_id = arg & RULE_ARG_ARG_MASK;
2013-10-04 14:53:11 -04:00
assert(rule_id < RULE_maximum_number_of);
push_rule(parser, parser->lexer->tok_line, rules[rule_id], 0);
2013-10-04 14:53:11 -04:00
}
2014-07-03 09:13:33 -04:00
STATIC void pop_rule(parser_t *parser, const rule_t **rule, mp_uint_t *arg_i, mp_uint_t *src_line) {
assert(!parser->had_memory_error);
2013-10-04 14:53:11 -04:00
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;
*src_line = parser->rule_stack[parser->rule_stack_top].src_line;
2013-10-04 14:53:11 -04:00
}
mp_parse_node_t mp_parse_node_new_leaf(mp_int_t kind, mp_int_t arg) {
if (kind == MP_PARSE_NODE_SMALL_INT) {
return (mp_parse_node_t)(kind | (arg << 1));
}
return (mp_parse_node_t)(kind | (arg << 4));
2013-10-04 14:53:11 -04:00
}
void mp_parse_node_free(mp_parse_node_t pn) {
if (MP_PARSE_NODE_IS_STRUCT(pn)) {
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t *)pn;
2014-07-03 09:13:33 -04:00
mp_uint_t n = MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
mp_uint_t rule_id = MP_PARSE_NODE_STRUCT_KIND(pns);
if (rule_id == RULE_string || rule_id == RULE_bytes) {
m_del(char, (char*)pns->nodes[0], (mp_uint_t)pns->nodes[1]);
} else if (rule_id == RULE_const_object) {
// don't free the const object since it's probably used by the compiled code
} else {
bool adjust = ADD_BLANK_NODE(rules[rule_id]);
if (adjust) {
n--;
}
for (mp_uint_t i = 0; i < n; i++) {
mp_parse_node_free(pns->nodes[i]);
}
if (adjust) {
n++;
}
}
m_del_var(mp_parse_node_struct_t, mp_parse_node_t, n, pns);
}
}
int mp_parse_node_extract_list(mp_parse_node_t *pn, mp_uint_t pn_kind, mp_parse_node_t **nodes) {
if (MP_PARSE_NODE_IS_NULL(*pn)) {
*nodes = NULL;
return 0;
} else if (MP_PARSE_NODE_IS_LEAF(*pn)) {
*nodes = pn;
return 1;
} else {
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)(*pn);
if (MP_PARSE_NODE_STRUCT_KIND(pns) != pn_kind) {
*nodes = pn;
return 1;
} else {
*nodes = pns->nodes;
return MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
}
}
}
#if MICROPY_DEBUG_PRINTERS
2014-07-03 09:13:33 -04:00
void mp_parse_node_print(mp_parse_node_t pn, mp_uint_t indent) {
if (MP_PARSE_NODE_IS_STRUCT(pn)) {
printf("[% 4d] ", (int)((mp_parse_node_struct_t*)pn)->source_line);
} else {
printf(" ");
}
2014-07-03 09:13:33 -04:00
for (mp_uint_t i = 0; i < indent; i++) {
2013-10-04 14:53:11 -04:00
printf(" ");
}
if (MP_PARSE_NODE_IS_NULL(pn)) {
2013-10-04 14:53:11 -04:00
printf("NULL\n");
} else if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
mp_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pn);
printf("int(" INT_FMT ")\n", arg);
} else if (MP_PARSE_NODE_IS_LEAF(pn)) {
mp_uint_t 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_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(" INT_FMT ")\n", arg); break;
2013-10-04 14:53:11 -04:00
default: assert(0);
}
} else {
// node must be a mp_parse_node_struct_t
mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)pn;
if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_string) {
printf("literal str(%.*s)\n", (int)pns->nodes[1], (char*)pns->nodes[0]);
} else if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_bytes) {
printf("literal bytes(%.*s)\n", (int)pns->nodes[1], (char*)pns->nodes[0]);
} else if (MP_PARSE_NODE_STRUCT_KIND(pns) == RULE_const_object) {
printf("literal const(%p)\n", (mp_obj_t)pns->nodes[0]);
} else {
2014-07-03 09:13:33 -04:00
mp_uint_t n = MP_PARSE_NODE_STRUCT_NUM_NODES(pns);
2013-10-04 14:53:11 -04:00
#ifdef USE_RULE_NAME
2014-07-03 09:13:33 -04:00
printf("%s(" UINT_FMT ") (n=" UINT_FMT ")\n", rules[MP_PARSE_NODE_STRUCT_KIND(pns)]->rule_name, (mp_uint_t)MP_PARSE_NODE_STRUCT_KIND(pns), n);
2013-10-04 14:53:11 -04:00
#else
2014-07-03 09:13:33 -04:00
printf("rule(" UINT_FMT ") (n=" UINT_FMT ")\n", (mp_uint_t)MP_PARSE_NODE_STRUCT_KIND(pns), n);
2013-10-04 14:53:11 -04:00
#endif
2014-07-03 09:13:33 -04:00
for (mp_uint_t i = 0; i < n; i++) {
mp_parse_node_print(pns->nodes[i], indent + 2);
}
2013-10-04 14:53:11 -04:00
}
}
}
#endif // MICROPY_DEBUG_PRINTERS
2013-10-04 14:53:11 -04:00
/*
STATIC void result_stack_show(parser_t *parser) {
2013-10-04 14:53:11 -04:00
printf("result stack, most recent first\n");
2014-07-03 09:13:33 -04:00
for (mp_int_t i = parser->result_stack_top - 1; i >= 0; i--) {
mp_parse_node_print(parser->result_stack[i], 0);
2013-10-04 14:53:11 -04:00
}
}
*/
STATIC mp_parse_node_t pop_result(parser_t *parser) {
if (parser->had_memory_error) {
return MP_PARSE_NODE_NULL;
}
2013-10-04 14:53:11 -04:00
assert(parser->result_stack_top > 0);
return parser->result_stack[--parser->result_stack_top];
}
2014-07-03 09:13:33 -04:00
STATIC mp_parse_node_t peek_result(parser_t *parser, mp_uint_t pos) {
if (parser->had_memory_error) {
return MP_PARSE_NODE_NULL;
}
2013-10-04 14:53:11 -04:00
assert(parser->result_stack_top > pos);
return parser->result_stack[parser->result_stack_top - 1 - pos];
}
STATIC void push_result_node(parser_t *parser, mp_parse_node_t pn) {
if (parser->had_memory_error) {
return;
}
if (parser->result_stack_top >= parser->result_stack_alloc) {
mp_parse_node_t *stack = m_renew_maybe(mp_parse_node_t, parser->result_stack, parser->result_stack_alloc, parser->result_stack_alloc + MICROPY_ALLOC_PARSE_RESULT_INC, true);
if (stack == NULL) {
memory_error(parser);
return;
}
parser->result_stack = stack;
parser->result_stack_alloc += MICROPY_ALLOC_PARSE_RESULT_INC;
}
2013-10-04 14:53:11 -04:00
parser->result_stack[parser->result_stack_top++] = pn;
}
STATIC mp_parse_node_t make_node_string_bytes(parser_t *parser, mp_uint_t src_line, mp_uint_t rule_kind, const char *str, mp_uint_t len) {
mp_parse_node_struct_t *pn = m_new_obj_var_maybe(mp_parse_node_struct_t, mp_parse_node_t, 2);
if (pn == NULL) {
memory_error(parser);
return MP_PARSE_NODE_NULL;
}
pn->source_line = src_line;
pn->kind_num_nodes = rule_kind | (2 << 8);
char *p = m_new(char, len);
memcpy(p, str, len);
pn->nodes[0] = (mp_int_t)p;
pn->nodes[1] = len;
return (mp_parse_node_t)pn;
}
STATIC mp_parse_node_t make_node_const_object(parser_t *parser, mp_uint_t src_line, mp_obj_t obj) {
mp_parse_node_struct_t *pn = m_new_obj_var_maybe(mp_parse_node_struct_t, mp_parse_node_t, 1);
if (pn == NULL) {
memory_error(parser);
return MP_PARSE_NODE_NULL;
}
pn->source_line = src_line;
pn->kind_num_nodes = RULE_const_object | (1 << 8);
pn->nodes[0] = (mp_uint_t)obj;
return (mp_parse_node_t)pn;
}
STATIC void push_result_token(parser_t *parser) {
mp_parse_node_t pn;
mp_lexer_t *lex = parser->lexer;
if (lex->tok_kind == MP_TOKEN_NAME) {
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_ID, qstr_from_strn(lex->vstr.buf, lex->vstr.len));
} else if (lex->tok_kind == MP_TOKEN_INTEGER) {
mp_obj_t o = mp_parse_num_integer(lex->vstr.buf, lex->vstr.len, 0, lex);
if (MP_OBJ_IS_SMALL_INT(o)) {
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_SMALL_INT, MP_OBJ_SMALL_INT_VALUE(o));
2013-10-04 14:53:11 -04:00
} else {
pn = make_node_const_object(parser, lex->tok_line, o);
2013-10-04 14:53:11 -04:00
}
} else if (lex->tok_kind == MP_TOKEN_FLOAT_OR_IMAG) {
mp_obj_t o = mp_parse_num_decimal(lex->vstr.buf, lex->vstr.len, true, false, lex);
pn = make_node_const_object(parser, lex->tok_line, o);
} else if (lex->tok_kind == MP_TOKEN_STRING || lex->tok_kind == MP_TOKEN_BYTES) {
// Don't automatically intern all strings/bytes. doc strings (which are usually large)
// will be discarded by the compiler, and so we shouldn't intern them.
qstr qst = MP_QSTR_NULL;
if (lex->vstr.len <= MICROPY_ALLOC_PARSE_INTERN_STRING_LEN) {
// intern short strings
qst = qstr_from_strn(lex->vstr.buf, lex->vstr.len);
} else {
// check if this string is already interned
qst = qstr_find_strn(lex->vstr.buf, lex->vstr.len);
}
if (qst != MP_QSTR_NULL) {
// qstr exists, make a leaf node
pn = mp_parse_node_new_leaf(lex->tok_kind == MP_TOKEN_STRING ? MP_PARSE_NODE_STRING : MP_PARSE_NODE_BYTES, qst);
} else {
// not interned, make a node holding a pointer to the string/bytes data
pn = make_node_string_bytes(parser, lex->tok_line, lex->tok_kind == MP_TOKEN_STRING ? RULE_string : RULE_bytes, lex->vstr.buf, lex->vstr.len);
}
2013-10-04 14:53:11 -04:00
} else {
pn = mp_parse_node_new_leaf(MP_PARSE_NODE_TOKEN, lex->tok_kind);
2013-10-04 14:53:11 -04:00
}
push_result_node(parser, pn);
}
2014-07-03 09:13:33 -04:00
STATIC void push_result_rule(parser_t *parser, mp_uint_t src_line, const rule_t *rule, mp_uint_t num_args) {
mp_parse_node_struct_t *pn = m_new_obj_var_maybe(mp_parse_node_struct_t, mp_parse_node_t, num_args);
if (pn == NULL) {
memory_error(parser);
return;
}
pn->source_line = src_line;
pn->kind_num_nodes = (rule->rule_id & 0xff) | (num_args << 8);
2014-07-03 09:13:33 -04:00
for (mp_uint_t i = num_args; i > 0; i--) {
2013-10-04 14:53:11 -04:00
pn->nodes[i - 1] = pop_result(parser);
}
push_result_node(parser, (mp_parse_node_t)pn);
2013-10-04 14:53:11 -04:00
}
mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind) {
// initialise parser and allocate memory for its stacks
parser_t parser;
parser.had_memory_error = false;
parser.rule_stack_alloc = MICROPY_ALLOC_PARSE_RULE_INIT;
parser.rule_stack_top = 0;
parser.rule_stack = m_new_maybe(rule_stack_t, parser.rule_stack_alloc);
2013-10-04 14:53:11 -04:00
parser.result_stack_alloc = MICROPY_ALLOC_PARSE_RESULT_INIT;
parser.result_stack_top = 0;
parser.result_stack = m_new_maybe(mp_parse_node_t, parser.result_stack_alloc);
2013-10-04 14:53:11 -04:00
parser.lexer = lex;
// check if we could allocate the stacks
if (parser.rule_stack == NULL || parser.result_stack == NULL) {
goto memory_error;
}
// work out the top-level rule to use, and push it on the stack
2014-07-03 09:13:33 -04:00
mp_uint_t top_level_rule;
2013-10-18 14:58:12 -04:00
switch (input_kind) {
case MP_PARSE_SINGLE_INPUT: top_level_rule = RULE_single_input; break;
2014-01-15 17:14:03 -05:00
case MP_PARSE_EVAL_INPUT: top_level_rule = RULE_eval_input; break;
2013-10-18 14:58:12 -04:00
default: top_level_rule = RULE_file_input;
}
push_rule(&parser, lex->tok_line, rules[top_level_rule], 0);
2013-10-04 14:53:11 -04:00
// parse!
2014-07-03 09:13:33 -04:00
mp_uint_t n, i; // state for the current rule
mp_uint_t rule_src_line; // source line for the first token matched by the current rule
2013-10-04 14:53:11 -04:00
bool backtrack = false;
const rule_t *rule = NULL;
2013-10-04 14:53:11 -04:00
for (;;) {
next_rule:
if (parser.rule_stack_top == 0 || parser.had_memory_error) {
2013-10-04 14:53:11 -04:00
break;
}
pop_rule(&parser, &rule, &i, &rule_src_line);
2013-10-04 14:53:11 -04:00
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) {
2013-10-04 14:53:11 -04:00
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; ++i) {
uint16_t kind = rule->arg[i] & RULE_ARG_KIND_MASK;
if (kind == RULE_ARG_TOK) {
if (lex->tok_kind == (rule->arg[i] & RULE_ARG_ARG_MASK)) {
push_result_token(&parser);
mp_lexer_to_next(lex);
2013-10-04 14:53:11 -04:00
goto next_rule;
}
2013-10-04 14:53:11 -04:00
} else {
assert(kind == RULE_ARG_RULE);
if (i + 1 < n) {
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 14:53:11 -04:00
goto next_rule;
}
}
backtrack = true;
2013-10-04 14:53:11 -04:00
break;
case RULE_ACT_AND: {
2013-10-04 14:53:11 -04:00
// 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
push_result_node(&parser, MP_PARSE_NODE_NULL);
2013-10-04 14:53:11 -04:00
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: {
2013-10-04 14:53:11 -04:00
// need to match a token
mp_token_kind_t tok_kind = rule->arg[i] & RULE_ARG_ARG_MASK;
if (lex->tok_kind == tok_kind) {
2013-10-04 14:53:11 -04:00
// matched token
if (tok_kind == MP_TOKEN_NAME) {
push_result_token(&parser);
2013-10-04 14:53:11 -04:00
}
mp_lexer_to_next(lex);
2013-10-04 14:53:11 -04:00
} 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;
}
2013-10-04 14:53:11 -04:00
case RULE_ARG_RULE:
case RULE_ARG_OPT_RULE:
rule_and_no_other_choice:
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 14:53:11 -04:00
goto next_rule;
default:
assert(0);
goto rule_and_no_other_choice; // to help flow control analysis
2013-10-04 14:53:11 -04:00
}
}
assert(i == n);
// matched the rule, so now build the corresponding parse_node
// count number of arguments for the parse_node
i = 0;
bool emit_rule = false;
2014-07-03 09:13:33 -04:00
for (mp_uint_t x = 0; x < n; ++x) {
2013-10-04 14:53:11 -04:00
if ((rule->arg[x] & RULE_ARG_KIND_MASK) == RULE_ARG_TOK) {
mp_token_kind_t tok_kind = rule->arg[x] & RULE_ARG_ARG_MASK;
if (tok_kind >= MP_TOKEN_NAME) {
2013-10-04 14:53:11 -04:00
emit_rule = true;
}
if (tok_kind == MP_TOKEN_NAME) {
2013-10-04 14:53:11 -04:00
// only tokens which were names are pushed to stack
i += 1;
}
} else {
// rules are always pushed
i += 1;
}
}
unix-cpy: Remove unix-cpy. It's no longer needed. unix-cpy was originally written to get semantic equivalent with CPython without writing functional tests. When writing the initial implementation of uPy it was a long way between lexer and functional tests, so the half-way test was to make sure that the bytecode was correct. The idea was that if the uPy bytecode matched CPython 1-1 then uPy would be proper Python if the bytecodes acted correctly. And having matching bytecode meant that it was less likely to miss some deep subtlety in the Python semantics that would require an architectural change later on. But that is all history and it no longer makes sense to retain the ability to output CPython bytecode, because: 1. It outputs CPython 3.3 compatible bytecode. CPython's bytecode changes from version to version, and seems to have changed quite a bit in 3.5. There's no point in changing the bytecode output to match CPython anymore. 2. uPy and CPy do different optimisations to the bytecode which makes it harder to match. 3. The bytecode tests are not run. They were never part of Travis and are not run locally anymore. 4. The EMIT_CPYTHON option needs a lot of extra source code which adds heaps of noise, especially in compile.c. 5. Now that there is an extensive test suite (which tests functionality) there is no need to match the bytecode. Some very subtle behaviour is tested with the test suite and passing these tests is a much better way to stay Python-language compliant, rather than trying to match CPy bytecode.
2015-08-14 07:24:11 -04:00
#if !MICROPY_ENABLE_DOC_STRING
// this code discards lonely statements, such as doc strings
if (input_kind != MP_PARSE_SINGLE_INPUT && rule->rule_id == RULE_expr_stmt && peek_result(&parser, 0) == MP_PARSE_NODE_NULL) {
mp_parse_node_t p = peek_result(&parser, 1);
if ((MP_PARSE_NODE_IS_LEAF(p) && !MP_PARSE_NODE_IS_ID(p)) || MP_PARSE_NODE_IS_STRUCT_KIND(p, RULE_string)) {
pop_result(&parser); // MP_PARSE_NODE_NULL
mp_parse_node_free(pop_result(&parser)); // RULE_string
push_result_rule(&parser, rule_src_line, rules[RULE_pass_stmt], 0);
break;
}
}
unix-cpy: Remove unix-cpy. It's no longer needed. unix-cpy was originally written to get semantic equivalent with CPython without writing functional tests. When writing the initial implementation of uPy it was a long way between lexer and functional tests, so the half-way test was to make sure that the bytecode was correct. The idea was that if the uPy bytecode matched CPython 1-1 then uPy would be proper Python if the bytecodes acted correctly. And having matching bytecode meant that it was less likely to miss some deep subtlety in the Python semantics that would require an architectural change later on. But that is all history and it no longer makes sense to retain the ability to output CPython bytecode, because: 1. It outputs CPython 3.3 compatible bytecode. CPython's bytecode changes from version to version, and seems to have changed quite a bit in 3.5. There's no point in changing the bytecode output to match CPython anymore. 2. uPy and CPy do different optimisations to the bytecode which makes it harder to match. 3. The bytecode tests are not run. They were never part of Travis and are not run locally anymore. 4. The EMIT_CPYTHON option needs a lot of extra source code which adds heaps of noise, especially in compile.c. 5. Now that there is an extensive test suite (which tests functionality) there is no need to match the bytecode. Some very subtle behaviour is tested with the test suite and passing these tests is a much better way to stay Python-language compliant, rather than trying to match CPy bytecode.
2015-08-14 07:24:11 -04:00
#endif
2013-10-04 14:53:11 -04:00
// 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;
}
// if a rule has the RULE_ACT_ALLOW_IDENT bit set then this
// rule should not be emitted if it has only 1 argument
// NOTE: can't set this flag for atom_paren because we need it
// to distinguish, for example, [a,b] from [(a,b)]
// TODO possibly set for: varargslist_name, varargslist_equal
if (rule->act & RULE_ACT_ALLOW_IDENT) {
2013-10-04 14:53:11 -04:00
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 (ADD_BLANK_NODE(rule)) {
2013-10-04 14:53:11 -04:00
emit_rule = true;
push_result_node(&parser, MP_PARSE_NODE_NULL);
2013-10-04 14:53:11 -04:00
i += 1;
}
2014-07-03 09:13:33 -04:00
mp_uint_t num_not_nil = 0;
for (mp_uint_t x = 0; x < i; ++x) {
if (peek_result(&parser, x) != MP_PARSE_NODE_NULL) {
2013-10-04 14:53:11 -04:00
num_not_nil += 1;
}
}
if (emit_rule) {
push_result_rule(&parser, rule_src_line, rule, i);
2013-10-04 14:53:11 -04:00
} else if (num_not_nil == 0) {
push_result_rule(&parser, rule_src_line, rule, i); // needed for, eg, atom_paren, testlist_comp_3b
2013-10-04 14:53:11 -04:00
} else if (num_not_nil == 1) {
// single result, leave it on stack
mp_parse_node_t pn = MP_PARSE_NODE_NULL;
2014-07-03 09:13:33 -04:00
for (mp_uint_t x = 0; x < i; ++x) {
mp_parse_node_t pn2 = pop_result(&parser);
if (pn2 != MP_PARSE_NODE_NULL) {
2013-10-04 14:53:11 -04:00
pn = pn2;
}
}
push_result_node(&parser, pn);
2013-10-04 14:53:11 -04:00
} else {
push_result_rule(&parser, rule_src_line, rule, i);
2013-10-04 14:53:11 -04:00
}
break;
}
2013-10-04 14:53:11 -04:00
case RULE_ACT_LIST: {
2013-10-04 14:53:11 -04:00
// n=2 is: item item*
// n=1 is: item (sep item)*
// n=3 is: item (sep item)* [sep]
bool had_trailing_sep;
2013-10-04 14:53:11 -04:00
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 (;;) {
2014-07-03 09:13:33 -04:00
mp_uint_t arg = rule->arg[i & 1 & n];
2013-10-04 14:53:11 -04:00
switch (arg & RULE_ARG_KIND_MASK) {
case RULE_ARG_TOK:
if (lex->tok_kind == (arg & RULE_ARG_ARG_MASK)) {
2013-10-04 14:53:11 -04:00
if (i & 1 & n) {
// separators which are tokens are not pushed to result stack
} else {
push_result_token(&parser);
2013-10-04 14:53:11 -04:00
}
mp_lexer_to_next(lex);
2013-10-04 14:53:11 -04:00
// 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:
rule_list_no_other_choice:
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 14:53:11 -04:00
goto next_rule;
default:
assert(0);
goto rule_list_no_other_choice; // to help flow control analysis
2013-10-04 14:53:11 -04:00
}
}
}
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
push_result_rule(&parser, rule_src_line, rule, i);
2013-10-04 14:53:11 -04:00
} else {
// just leave single item on stack (ie don't wrap in a list)
}
} else {
push_result_rule(&parser, rule_src_line, rule, i);
2013-10-04 14:53:11 -04:00
}
break;
}
2013-10-04 14:53:11 -04:00
default:
assert(0);
}
}
mp_obj_t exc;
mp_parse_node_t result;
// check if we had a memory error
if (parser.had_memory_error) {
memory_error:
exc = mp_obj_new_exception_msg(&mp_type_MemoryError,
"parser could not allocate enough memory");
result = MP_PARSE_NODE_NULL;
goto finished;
}
// check we are at the end of the token stream
if (lex->tok_kind != MP_TOKEN_END) {
goto syntax_error;
2013-10-04 14:53:11 -04:00
}
// check that parsing resulted in a parse node (can fail on empty input)
if (parser.result_stack_top == 0) {
goto syntax_error;
}
2013-10-04 14:53:11 -04:00
//result_stack_show(parser);
//printf("rule stack alloc: %d\n", parser.rule_stack_alloc);
//printf("result stack alloc: %d\n", parser.result_stack_alloc);
2013-10-04 14:53:11 -04:00
//printf("number of parse nodes allocated: %d\n", num_parse_nodes_allocated);
// get the root parse node that we created
assert(parser.result_stack_top == 1);
exc = MP_OBJ_NULL;
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);
// we also free the lexer on behalf of the caller (see below)
if (exc != MP_OBJ_NULL) {
// had an error so raise the exception
// add traceback to give info about file name and location
// we don't have a 'block' name, so just pass the NULL qstr to indicate this
mp_obj_exception_add_traceback(exc, lex->source_name, lex->tok_line, MP_QSTR_NULL);
mp_lexer_free(lex);
nlr_raise(exc);
} else {
mp_lexer_free(lex);
return result;
}
2013-10-04 14:53:11 -04:00
syntax_error:
if (lex->tok_kind == MP_TOKEN_INDENT) {
exc = mp_obj_new_exception_msg(&mp_type_IndentationError,
"unexpected indent");
} else if (lex->tok_kind == MP_TOKEN_DEDENT_MISMATCH) {
exc = mp_obj_new_exception_msg(&mp_type_IndentationError,
"unindent does not match any outer indentation level");
} else {
exc = mp_obj_new_exception_msg(&mp_type_SyntaxError,
"invalid syntax");
2013-10-04 14:53:11 -04:00
#ifdef USE_RULE_NAME
// debugging: print the rule name that failed and the token
printf("rule: %s\n", rule->rule_name);
#if MICROPY_DEBUG_PRINTERS
mp_lexer_show_token(lex);
#endif
#endif
}
result = MP_PARSE_NODE_NULL;
goto finished;
2013-10-04 14:53:11 -04:00
}