All tests pass; lots of stuff is no longer const
This commit is contained in:
parent
b9a33f2c98
commit
d71215d494
24 changed files with 216 additions and 120 deletions
|
|
@ -15,21 +15,21 @@ const HParser* document = NULL;
|
|||
void init_parser(void)
|
||||
{
|
||||
// CORE
|
||||
const HParser *digit = h_ch_range(0x30, 0x39);
|
||||
const HParser *alpha = h_choice(h_ch_range(0x41, 0x5a), h_ch_range(0x61, 0x7a), NULL);
|
||||
HParser *digit = h_ch_range(0x30, 0x39);
|
||||
HParser *alpha = h_choice(h_ch_range(0x41, 0x5a), h_ch_range(0x61, 0x7a), NULL);
|
||||
|
||||
// AUX.
|
||||
const HParser *plus = h_ch('+');
|
||||
const HParser *slash = h_ch('/');
|
||||
const HParser *equals = h_ch('=');
|
||||
HParser *plus = h_ch('+');
|
||||
HParser *slash = h_ch('/');
|
||||
HParser *equals = h_ch('=');
|
||||
|
||||
const HParser *bsfdig = h_choice(alpha, digit, plus, slash, NULL);
|
||||
const HParser *bsfdig_4bit = h_in((uint8_t *)"AEIMQUYcgkosw048", 16);
|
||||
const HParser *bsfdig_2bit = h_in((uint8_t *)"AQgw", 4);
|
||||
const HParser *base64_3 = h_repeat_n(bsfdig, 4);
|
||||
const HParser *base64_2 = h_sequence(bsfdig, bsfdig, bsfdig_4bit, equals, NULL);
|
||||
const HParser *base64_1 = h_sequence(bsfdig, bsfdig_2bit, equals, equals, NULL);
|
||||
const HParser *base64 = h_sequence(h_many(base64_3),
|
||||
HParser *bsfdig = h_choice(alpha, digit, plus, slash, NULL);
|
||||
HParser *bsfdig_4bit = h_in((uint8_t *)"AEIMQUYcgkosw048", 16);
|
||||
HParser *bsfdig_2bit = h_in((uint8_t *)"AQgw", 4);
|
||||
HParser *base64_3 = h_repeat_n(bsfdig, 4);
|
||||
HParser *base64_2 = h_sequence(bsfdig, bsfdig, bsfdig_4bit, equals, NULL);
|
||||
HParser *base64_1 = h_sequence(bsfdig, bsfdig_2bit, equals, equals, NULL);
|
||||
HParser *base64 = h_sequence(h_many(base64_3),
|
||||
h_optional(h_choice(base64_2,
|
||||
base64_1, NULL)),
|
||||
NULL);
|
||||
|
|
|
|||
|
|
@ -22,7 +22,7 @@
|
|||
// They must be named act_<rulename>.
|
||||
///
|
||||
|
||||
const HParsedToken *act_bsfdig(const HParseResult *p)
|
||||
HParsedToken *act_bsfdig(const HParseResult *p)
|
||||
{
|
||||
HParsedToken *res = H_MAKE_UINT(0);
|
||||
|
||||
|
|
@ -53,7 +53,7 @@ H_ACT_APPLY(act_index0, h_act_index, 0);
|
|||
#define act_document act_index0
|
||||
|
||||
// General-form action to turn a block of base64 digits into bytes.
|
||||
const HParsedToken *act_base64_n(int n, const HParseResult *p)
|
||||
HParsedToken *act_base64_n(int n, const HParseResult *p)
|
||||
{
|
||||
HParsedToken *res = H_MAKE_SEQN(n);
|
||||
|
||||
|
|
@ -82,7 +82,7 @@ H_ACT_APPLY(act_base64_3, act_base64_n, 3);
|
|||
H_ACT_APPLY(act_base64_2, act_base64_n, 2);
|
||||
H_ACT_APPLY(act_base64_1, act_base64_n, 1);
|
||||
|
||||
const HParsedToken *act_base64(const HParseResult *p)
|
||||
HParsedToken *act_base64(const HParseResult *p)
|
||||
{
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used == 2);
|
||||
|
|
@ -96,7 +96,7 @@ const HParsedToken *act_base64(const HParseResult *p)
|
|||
h_seq_append(res, seq->elements[i]);
|
||||
|
||||
// append one trailing base64_2 or _1 block
|
||||
const HParsedToken *tok = h_seq_index(p->ast, 1);
|
||||
HParsedToken *tok = h_seq_index(p->ast, 1);
|
||||
if(tok->token_type == TT_SEQUENCE)
|
||||
h_seq_append(res, tok);
|
||||
|
||||
|
|
@ -108,7 +108,7 @@ const HParsedToken *act_base64(const HParseResult *p)
|
|||
// Set up the parser with the grammar to be recognized.
|
||||
///
|
||||
|
||||
const HParser *init_parser(void)
|
||||
HParser *init_parser(void)
|
||||
{
|
||||
// CORE
|
||||
H_RULE (digit, h_ch_range(0x30, 0x39));
|
||||
|
|
|
|||
|
|
@ -48,7 +48,7 @@ uint8_t bsfdig_value(const HParsedToken *p)
|
|||
// helper: append a byte value to a sequence
|
||||
#define seq_append_byte(res, b) h_seq_snoc(res, H_MAKE_UINT(b))
|
||||
|
||||
const HParsedToken *act_base64(const HParseResult *p)
|
||||
HParsedToken *act_base64(const HParseResult *p)
|
||||
{
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used == 2);
|
||||
|
|
|
|||
|
|
@ -86,7 +86,7 @@ void set_rdata(struct dns_rr *rr, HCountedArray *rdata) {
|
|||
}
|
||||
}
|
||||
|
||||
const HParsedToken* act_header(const HParseResult *p) {
|
||||
HParsedToken* act_header(const HParseResult *p) {
|
||||
HParsedToken **fields = h_seq_elements(p->ast);
|
||||
dns_header_t header_ = {
|
||||
.id = H_CAST_UINT(fields[0]),
|
||||
|
|
@ -109,7 +109,7 @@ const HParsedToken* act_header(const HParseResult *p) {
|
|||
return H_MAKE(dns_header_t, header);
|
||||
}
|
||||
|
||||
const HParsedToken* act_label(const HParseResult *p) {
|
||||
HParsedToken* act_label(const HParseResult *p) {
|
||||
dns_label_t *r = H_ALLOC(dns_label_t);
|
||||
|
||||
r->len = h_seq_len(p->ast);
|
||||
|
|
@ -121,7 +121,7 @@ const HParsedToken* act_label(const HParseResult *p) {
|
|||
return H_MAKE(dns_label_t, r);
|
||||
}
|
||||
|
||||
const HParsedToken* act_rr(const HParseResult *p) {
|
||||
HParsedToken* act_rr(const HParseResult *p) {
|
||||
dns_rr_t *rr = H_ALLOC(dns_rr_t);
|
||||
|
||||
rr->name = *H_FIELD(dns_domain_t, 0);
|
||||
|
|
@ -136,7 +136,7 @@ const HParsedToken* act_rr(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_t, rr);
|
||||
}
|
||||
|
||||
const HParsedToken* act_question(const HParseResult *p) {
|
||||
HParsedToken* act_question(const HParseResult *p) {
|
||||
dns_question_t *q = H_ALLOC(dns_question_t);
|
||||
HParsedToken **fields = h_seq_elements(p->ast);
|
||||
|
||||
|
|
@ -153,7 +153,7 @@ const HParsedToken* act_question(const HParseResult *p) {
|
|||
return H_MAKE(dns_question_t, q);
|
||||
}
|
||||
|
||||
const HParsedToken* act_message(const HParseResult *p) {
|
||||
HParsedToken* act_message(const HParseResult *p) {
|
||||
h_pprint(stdout, p->ast, 0, 2);
|
||||
dns_message_t *msg = H_ALLOC(dns_message_t);
|
||||
|
||||
|
|
|
|||
|
|
@ -18,8 +18,8 @@ bool validate_label(HParseResult *p) {
|
|||
|
||||
#define act_label h_act_flatten
|
||||
|
||||
const HParsedToken* act_domain(const HParseResult *p) {
|
||||
const HParsedToken *ret = NULL;
|
||||
HParsedToken* act_domain(const HParseResult *p) {
|
||||
HParsedToken *ret = NULL;
|
||||
char *arr = NULL;
|
||||
|
||||
switch(p->ast->token_type) {
|
||||
|
|
@ -56,8 +56,8 @@ const HParsedToken* act_domain(const HParseResult *p) {
|
|||
return ret;
|
||||
}
|
||||
|
||||
const HParser* init_domain() {
|
||||
static const HParser *ret = NULL;
|
||||
HParser* init_domain() {
|
||||
static HParser *ret = NULL;
|
||||
if (ret)
|
||||
return ret;
|
||||
|
||||
|
|
@ -76,8 +76,8 @@ const HParser* init_domain() {
|
|||
return ret;
|
||||
}
|
||||
|
||||
const HParser* init_character_string() {
|
||||
static const HParser *cstr = NULL;
|
||||
HParser* init_character_string() {
|
||||
static HParser *cstr = NULL;
|
||||
if (cstr)
|
||||
return cstr;
|
||||
|
||||
|
|
|
|||
|
|
@ -4,9 +4,9 @@
|
|||
#include "../src/hammer.h"
|
||||
#include "../src/glue.h"
|
||||
|
||||
const HParser* init_domain();
|
||||
const HParser* init_character_string();
|
||||
HParser* init_domain();
|
||||
HParser* init_character_string();
|
||||
|
||||
const HParsedToken* act_index0(const HParseResult *p);
|
||||
HParsedToken* act_index0(const HParseResult *p);
|
||||
|
||||
#endif
|
||||
|
|
|
|||
|
|
@ -17,7 +17,7 @@ bool validate_null(HParseResult *p) {
|
|||
return (65536 > p->ast->seq->used);
|
||||
}
|
||||
|
||||
const HParsedToken *act_null(const HParseResult *p) {
|
||||
HParsedToken *act_null(const HParseResult *p) {
|
||||
dns_rr_null_t *null = H_ALLOC(dns_rr_null_t);
|
||||
|
||||
size_t len = h_seq_len(p->ast);
|
||||
|
|
@ -28,7 +28,7 @@ const HParsedToken *act_null(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_null_t, null);
|
||||
}
|
||||
|
||||
const HParsedToken *act_txt(const HParseResult *p) {
|
||||
HParsedToken *act_txt(const HParseResult *p) {
|
||||
dns_rr_txt_t *txt = H_ALLOC(dns_rr_txt_t);
|
||||
|
||||
const HCountedArray *arr = H_CAST_SEQ(p->ast);
|
||||
|
|
@ -47,7 +47,7 @@ const HParsedToken *act_txt(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_txt_t, txt);
|
||||
}
|
||||
|
||||
const HParsedToken* act_cstr(const HParseResult *p) {
|
||||
HParsedToken* act_cstr(const HParseResult *p) {
|
||||
dns_cstr_t *cs = H_ALLOC(dns_cstr_t);
|
||||
|
||||
const HCountedArray *arr = H_CAST_SEQ(p->ast);
|
||||
|
|
@ -60,7 +60,7 @@ const HParsedToken* act_cstr(const HParseResult *p) {
|
|||
return H_MAKE(dns_cstr_t, cs);
|
||||
}
|
||||
|
||||
const HParsedToken* act_soa(const HParseResult *p) {
|
||||
HParsedToken* act_soa(const HParseResult *p) {
|
||||
dns_rr_soa_t *soa = H_ALLOC(dns_rr_soa_t);
|
||||
|
||||
soa->mname = *H_FIELD(dns_domain_t, 0);
|
||||
|
|
@ -74,7 +74,7 @@ const HParsedToken* act_soa(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_soa_t, soa);
|
||||
}
|
||||
|
||||
const HParsedToken* act_wks(const HParseResult *p) {
|
||||
HParsedToken* act_wks(const HParseResult *p) {
|
||||
dns_rr_wks_t *wks = H_ALLOC(dns_rr_wks_t);
|
||||
|
||||
wks->address = H_FIELD_UINT(0);
|
||||
|
|
@ -87,7 +87,7 @@ const HParsedToken* act_wks(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_wks_t, wks);
|
||||
}
|
||||
|
||||
const HParsedToken* act_hinfo(const HParseResult *p) {
|
||||
HParsedToken* act_hinfo(const HParseResult *p) {
|
||||
dns_rr_hinfo_t *hinfo = H_ALLOC(dns_rr_hinfo_t);
|
||||
|
||||
hinfo->cpu = *H_FIELD(dns_cstr_t, 0);
|
||||
|
|
@ -96,7 +96,7 @@ const HParsedToken* act_hinfo(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_hinfo_t, hinfo);
|
||||
}
|
||||
|
||||
const HParsedToken* act_minfo(const HParseResult *p) {
|
||||
HParsedToken* act_minfo(const HParseResult *p) {
|
||||
dns_rr_minfo_t *minfo = H_ALLOC(dns_rr_minfo_t);
|
||||
|
||||
minfo->rmailbx = *H_FIELD(dns_domain_t, 0);
|
||||
|
|
@ -105,7 +105,7 @@ const HParsedToken* act_minfo(const HParseResult *p) {
|
|||
return H_MAKE(dns_rr_minfo_t, minfo);
|
||||
}
|
||||
|
||||
const HParsedToken* act_mx(const HParseResult *p) {
|
||||
HParsedToken* act_mx(const HParseResult *p) {
|
||||
dns_rr_mx_t *mx = H_ALLOC(dns_rr_mx_t);
|
||||
|
||||
mx->preference = H_FIELD_UINT(0);
|
||||
|
|
@ -120,8 +120,8 @@ const HParsedToken* act_mx(const HParseResult *p) {
|
|||
///
|
||||
|
||||
#define RDATA_TYPE_MAX 16
|
||||
const HParser* init_rdata(uint16_t type) {
|
||||
static const HParser *parsers[RDATA_TYPE_MAX+1];
|
||||
HParser* init_rdata(uint16_t type) {
|
||||
static HParser *parsers[RDATA_TYPE_MAX+1];
|
||||
static int inited = 0;
|
||||
|
||||
if (type >= sizeof(parsers))
|
||||
|
|
|
|||
|
|
@ -3,6 +3,6 @@
|
|||
|
||||
#include "../src/hammer.h"
|
||||
|
||||
const HParser* init_rdata(uint16_t type);
|
||||
HParser* init_rdata(uint16_t type);
|
||||
|
||||
#endif
|
||||
|
|
|
|||
|
|
@ -53,7 +53,7 @@ void* h_rvm_run__m(HAllocator *mm__, HRVMProg *prog, const uint8_t* input, size_
|
|||
HRVMTrace **heads_p = a_new(HRVMTrace*, prog->length),
|
||||
**heads_n = a_new(HRVMTrace*, prog->length);
|
||||
|
||||
HRVMTrace *ret_trace;
|
||||
HRVMTrace *ret_trace = NULL;
|
||||
|
||||
uint8_t *insn_seen = a_new(uint8_t, prog->length); // 0 -> not seen, 1->processed, 2->queued
|
||||
HRVMThread *ip_queue = a_new(HRVMThread, prog->length);
|
||||
|
|
@ -61,6 +61,7 @@ void* h_rvm_run__m(HAllocator *mm__, HRVMProg *prog, const uint8_t* input, size_
|
|||
|
||||
|
||||
|
||||
|
||||
|
||||
#define THREAD ip_queue[ipq_top-1]
|
||||
#define PUSH_SVM(op_, arg_) do { \
|
||||
|
|
@ -102,15 +103,18 @@ void* h_rvm_run__m(HAllocator *mm__, HRVMProg *prog, const uint8_t* input, size_
|
|||
uint8_t hi, lo;
|
||||
uint16_t arg;
|
||||
while(ipq_top > 0) {
|
||||
if (insn_seen[THREAD.ip] == 1)
|
||||
if (insn_seen[THREAD.ip] == 1) {
|
||||
ipq_top--; // Kill thread.
|
||||
continue;
|
||||
}
|
||||
insn_seen[THREAD.ip] = 1;
|
||||
arg = prog->insns[THREAD.ip].arg;
|
||||
switch(prog->insns[THREAD.ip].op) {
|
||||
case RVM_ACCEPT:
|
||||
PUSH_SVM(SVM_ACCEPT, 0);
|
||||
ret_trace = THREAD.trace;
|
||||
goto run_trace;
|
||||
ipq_top--;
|
||||
goto next_insn;
|
||||
case RVM_MATCH:
|
||||
hi = (arg >> 8) & 0xff;
|
||||
lo = arg & 0xff;
|
||||
|
|
@ -163,10 +167,12 @@ void* h_rvm_run__m(HAllocator *mm__, HRVMProg *prog, const uint8_t* input, size_
|
|||
}
|
||||
// No accept was reached.
|
||||
match_fail:
|
||||
h_delete_arena(arena);
|
||||
return NULL;
|
||||
if (ret_trace == NULL) {
|
||||
// No match found; definite failure.
|
||||
h_delete_arena(arena);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
run_trace:
|
||||
// Invert the direction of the trace linked list.
|
||||
|
||||
ret_trace = invert_trace(ret_trace);
|
||||
|
|
@ -213,8 +219,9 @@ HParseResult *run_trace(HAllocator *mm__, HRVMProg *orig_prog, HRVMTrace *trace,
|
|||
case SVM_ACTION:
|
||||
// Action should modify stack appropriately
|
||||
if (!orig_prog->actions[cur->arg].action(arena, &ctx, orig_prog->actions[cur->arg].env)) {
|
||||
|
||||
// action failed... abort somehow
|
||||
// TODO: Actually abort
|
||||
goto fail;
|
||||
}
|
||||
break;
|
||||
case SVM_CAPTURE:
|
||||
|
|
@ -243,7 +250,7 @@ HParseResult *run_trace(HAllocator *mm__, HRVMProg *orig_prog, HRVMTrace *trace,
|
|||
return res;
|
||||
}
|
||||
}
|
||||
|
||||
fail:
|
||||
h_delete_arena(arena);
|
||||
return NULL;
|
||||
}
|
||||
|
|
@ -308,20 +315,20 @@ bool h_svm_action_make_sequence(HArena *arena, HSVMContext *ctx, void* env) {
|
|||
HParsedToken *res = ctx->stack[ctx->stack_count - 1 - n_items];
|
||||
assert (res->token_type == TT_MARK);
|
||||
res->token_type = TT_SEQUENCE;
|
||||
|
||||
|
||||
HCountedArray *ret_carray = h_carray_new_sized(arena, n_items);
|
||||
res->seq = ret_carray;
|
||||
// res index and bit offset are the same as the mark.
|
||||
for (size_t i = 0; i < n_items; i++) {
|
||||
ret_carray->elements[i] = ctx->stack[ctx->stack_count - n_items + i];
|
||||
}
|
||||
ret_carray->used = n_items;
|
||||
ctx->stack_count -= n_items;
|
||||
return true;
|
||||
}
|
||||
|
||||
bool h_svm_action_clear_to_mark(HArena *arena, HSVMContext *ctx, void* env) {
|
||||
while (ctx->stack_count > 0) {
|
||||
if (ctx->stack[--ctx->stack_count]->token_type == TT_MARK)
|
||||
while (ctx->stack_count > 0) { if (ctx->stack[--ctx->stack_count]->token_type == TT_MARK)
|
||||
return true;
|
||||
}
|
||||
return false; // no mark found.
|
||||
|
|
|
|||
|
|
@ -9,7 +9,8 @@
|
|||
|
||||
HCountedArray *h_carray_new_sized(HArena * arena, size_t size) {
|
||||
HCountedArray *ret = h_arena_malloc(arena, sizeof(HCountedArray));
|
||||
assert(size > 0);
|
||||
if (size == 0)
|
||||
size = 1;
|
||||
ret->used = 0;
|
||||
ret->capacity = size;
|
||||
ret->arena = arena;
|
||||
|
|
|
|||
12
src/glue.c
12
src/glue.c
|
|
@ -5,7 +5,7 @@
|
|||
#include "parsers/parser_internal.h"
|
||||
|
||||
// Helper to build HAction's that pick one index out of a sequence.
|
||||
const HParsedToken *h_act_index(int i, const HParseResult *p)
|
||||
HParsedToken *h_act_index(int i, const HParseResult *p)
|
||||
{
|
||||
if(!p) return NULL;
|
||||
|
||||
|
|
@ -23,7 +23,7 @@ const HParsedToken *h_act_index(int i, const HParseResult *p)
|
|||
return tok->seq->elements[i];
|
||||
}
|
||||
|
||||
const HParsedToken *h_act_first(const HParseResult *p) {
|
||||
HParsedToken *h_act_first(const HParseResult *p) {
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used > 0);
|
||||
|
|
@ -31,7 +31,7 @@ const HParsedToken *h_act_first(const HParseResult *p) {
|
|||
return p->ast->seq->elements[0];
|
||||
}
|
||||
|
||||
const HParsedToken *h_act_second(const HParseResult *p) {
|
||||
HParsedToken *h_act_second(const HParseResult *p) {
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used > 0);
|
||||
|
|
@ -39,7 +39,7 @@ const HParsedToken *h_act_second(const HParseResult *p) {
|
|||
return p->ast->seq->elements[1];
|
||||
}
|
||||
|
||||
const HParsedToken *h_act_last(const HParseResult *p) {
|
||||
HParsedToken *h_act_last(const HParseResult *p) {
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used > 0);
|
||||
|
|
@ -59,7 +59,7 @@ static void act_flatten_(HCountedArray *seq, const HParsedToken *tok) {
|
|||
}
|
||||
}
|
||||
|
||||
const HParsedToken *h_act_flatten(const HParseResult *p) {
|
||||
HParsedToken *h_act_flatten(const HParseResult *p) {
|
||||
HCountedArray *seq = h_carray_new(p->arena);
|
||||
|
||||
act_flatten_(seq, p->ast);
|
||||
|
|
@ -72,7 +72,7 @@ const HParsedToken *h_act_flatten(const HParseResult *p) {
|
|||
return res;
|
||||
}
|
||||
|
||||
const HParsedToken *h_act_ignore(const HParseResult *p) {
|
||||
HParsedToken *h_act_ignore(const HParseResult *p) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
|
|||
24
src/glue.h
24
src/glue.h
|
|
@ -55,13 +55,13 @@
|
|||
//
|
||||
|
||||
|
||||
#define H_RULE(rule, def) const HParser *rule = def
|
||||
#define H_ARULE(rule, def) const HParser *rule = h_action(def, act_ ## rule)
|
||||
#define H_VRULE(rule, def) const HParser *rule = \
|
||||
#define H_RULE(rule, def) HParser *rule = def
|
||||
#define H_ARULE(rule, def) HParser *rule = h_action(def, act_ ## rule)
|
||||
#define H_VRULE(rule, def) HParser *rule = \
|
||||
h_attr_bool(def, validate_ ## rule)
|
||||
#define H_VARULE(rule, def) const HParser *rule = \
|
||||
#define H_VARULE(rule, def) HParser *rule = \
|
||||
h_attr_bool(h_action(def, act_ ## rule), validate_ ## rule)
|
||||
#define H_AVRULE(rule, def) const HParser *rule = \
|
||||
#define H_AVRULE(rule, def) HParser *rule = \
|
||||
h_action(h_attr_bool(def, validate_ ## rule), act_ ## rule)
|
||||
|
||||
|
||||
|
|
@ -88,17 +88,17 @@
|
|||
// action such as h_act_index.
|
||||
//
|
||||
|
||||
const HParsedToken *h_act_index(int i, const HParseResult *p);
|
||||
const HParsedToken *h_act_first(const HParseResult *p);
|
||||
const HParsedToken *h_act_second(const HParseResult *p);
|
||||
const HParsedToken *h_act_last(const HParseResult *p);
|
||||
const HParsedToken *h_act_flatten(const HParseResult *p);
|
||||
const HParsedToken *h_act_ignore(const HParseResult *p);
|
||||
HParsedToken *h_act_index(int i, const HParseResult *p);
|
||||
HParsedToken *h_act_first(const HParseResult *p);
|
||||
HParsedToken *h_act_second(const HParseResult *p);
|
||||
HParsedToken *h_act_last(const HParseResult *p);
|
||||
HParsedToken *h_act_flatten(const HParseResult *p);
|
||||
HParsedToken *h_act_ignore(const HParseResult *p);
|
||||
|
||||
// Define 'myaction' as a specialization of 'paction' by supplying the leading
|
||||
// parameters.
|
||||
#define H_ACT_APPLY(myaction, paction, ...) \
|
||||
const HParsedToken *myaction(const HParseResult *p) { \
|
||||
HParsedToken *myaction(const HParseResult *p) { \
|
||||
return paction(__VA_ARGS__, p); \
|
||||
}
|
||||
|
||||
|
|
|
|||
12
src/hammer.h
12
src/hammer.h
|
|
@ -111,7 +111,7 @@ typedef struct HBitWriter_ HBitWriter;
|
|||
* say, structs) and stuff values for them into the void* in the
|
||||
* tagged union in HParsedToken.
|
||||
*/
|
||||
typedef const HParsedToken* (*HAction)(const HParseResult *p);
|
||||
typedef HParsedToken* (*HAction)(const HParseResult *p);
|
||||
|
||||
/**
|
||||
* Type of a boolean attribute-checking function, used in the
|
||||
|
|
@ -605,11 +605,11 @@ void h_bit_writer_free(HBitWriter* w);
|
|||
|
||||
// General-purpose actions for use with h_action
|
||||
// XXX to be consolidated with glue.h when merged upstream
|
||||
const HParsedToken *h_act_first(const HParseResult *p);
|
||||
const HParsedToken *h_act_second(const HParseResult *p);
|
||||
const HParsedToken *h_act_last(const HParseResult *p);
|
||||
const HParsedToken *h_act_flatten(const HParseResult *p);
|
||||
const HParsedToken *h_act_ignore(const HParseResult *p);
|
||||
HParsedToken *h_act_first(const HParseResult *p);
|
||||
HParsedToken *h_act_second(const HParseResult *p);
|
||||
HParsedToken *h_act_last(const HParseResult *p);
|
||||
HParsedToken *h_act_flatten(const HParseResult *p);
|
||||
HParsedToken *h_act_ignore(const HParseResult *p);
|
||||
|
||||
// {{{ Benchmark functions
|
||||
HAMMER_FN_DECL(HBenchmarkResults *, h_benchmark, HParser* parser, HParserTestcase* testcases);
|
||||
|
|
|
|||
|
|
@ -1,3 +1,4 @@
|
|||
#include <assert.h>
|
||||
#include "parser_internal.h"
|
||||
|
||||
typedef struct {
|
||||
|
|
@ -45,9 +46,35 @@ static bool action_isValidCF(void *env) {
|
|||
return a->p->vtable->isValidCF(a->p->env);
|
||||
}
|
||||
|
||||
static bool h_svm_action_action(HArena *arena, HSVMContext *ctx, void* arg) {
|
||||
HParseResult res;
|
||||
HAction action = arg;
|
||||
assert(ctx->stack_count >= 1);
|
||||
if (ctx->stack[ctx->stack_count-1]->token_type != TT_MARK) {
|
||||
assert(ctx->stack_count >= 2 && ctx->stack[ctx->stack_count-2]->token_type == TT_MARK);
|
||||
res.ast = ctx->stack[ctx->stack_count-2] = ctx->stack[ctx->stack_count-1];
|
||||
ctx->stack_count--;
|
||||
// mark replaced.
|
||||
} else {
|
||||
res.ast = NULL;
|
||||
}
|
||||
res.arena = arena;
|
||||
|
||||
HParsedToken *tok = action(&res);
|
||||
if (tok != NULL)
|
||||
ctx->stack[ctx->stack_count-1] = tok;
|
||||
else
|
||||
ctx->stack_count--;
|
||||
return true; // action can't fail
|
||||
}
|
||||
|
||||
static bool action_ctrvm(HRVMProg *prog, void* env) {
|
||||
HParseAction *a = (HParseAction*)env;
|
||||
return a->p->vtable->compile_to_rvm(prog, a->p->env);
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
if (!h_compile_regex(prog, a->p))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_action, a->action));
|
||||
return true;
|
||||
}
|
||||
|
||||
static const HParserVtable action_vt = {
|
||||
|
|
|
|||
|
|
@ -1,3 +1,4 @@
|
|||
#include <assert.h>
|
||||
#include "parser_internal.h"
|
||||
|
||||
typedef struct {
|
||||
|
|
@ -48,9 +49,30 @@ static HCFChoice* desugar_ab(HAllocator *mm__, void *env) {
|
|||
return ret;
|
||||
}
|
||||
|
||||
static bool h_svm_action_attr_bool(HArena *arena, HSVMContext *ctx, void* arg) {
|
||||
HParseResult res;
|
||||
HPredicate pred = arg;
|
||||
assert(ctx->stack_count >= 1);
|
||||
if (ctx->stack[ctx->stack_count-1]->token_type != TT_MARK) {
|
||||
assert(ctx->stack_count >= 2 && ctx->stack[ctx->stack_count-2]->token_type == TT_MARK);
|
||||
ctx->stack_count--;
|
||||
res.ast = ctx->stack[ctx->stack_count-1] = ctx->stack[ctx->stack_count];
|
||||
// mark replaced.
|
||||
} else {
|
||||
ctx->stack_count--;
|
||||
res.ast = NULL;
|
||||
}
|
||||
res.arena = arena;
|
||||
return pred(&res);
|
||||
}
|
||||
|
||||
static bool ab_ctrvm(HRVMProg *prog, void *env) {
|
||||
HAttrBool *ab = (HAttrBool*)env;
|
||||
return h_compile_regex(prog, ab->p);
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
if (!h_compile_regex(prog, ab->p))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_attr_bool, ab->pred));
|
||||
return true;
|
||||
}
|
||||
|
||||
static const HParserVtable attr_bool_vt = {
|
||||
|
|
|
|||
|
|
@ -45,10 +45,10 @@ static HParsedToken *reshape_bits(const HParseResult *p, bool signedp) {
|
|||
|
||||
return ret;
|
||||
}
|
||||
static const HParsedToken *reshape_bits_unsigned(const HParseResult *p) {
|
||||
static HParsedToken *reshape_bits_unsigned(const HParseResult *p) {
|
||||
return reshape_bits(p, false);
|
||||
}
|
||||
static const HParsedToken *reshape_bits_signed(const HParseResult *p) {
|
||||
static HParsedToken *reshape_bits_signed(const HParseResult *p) {
|
||||
return reshape_bits(p, true);
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -65,6 +65,7 @@ static bool choice_ctrvm(HRVMProg *prog, void* env) {
|
|||
gotos[i] = h_rvm_insert_insn(prog, RVM_GOTO, 65535);
|
||||
h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
|
||||
}
|
||||
h_rvm_insert_insn(prog, RVM_MATCH, 0x00FF); // fail.
|
||||
uint16_t jump = h_rvm_get_ip(prog);
|
||||
for (size_t i=0; i<s->len; ++i) {
|
||||
h_rvm_patch_arg(prog, gotos[i], jump);
|
||||
|
|
|
|||
|
|
@ -47,7 +47,7 @@ static bool h_svm_action_pop(HArena *arena, HSVMContext *ctx, void* arg) {
|
|||
|
||||
static bool ignore_ctrvm(HRVMProg *prog, void *env) {
|
||||
HParser *p = (HParser*)env;
|
||||
h_compile_regex(prog, p->env);
|
||||
h_compile_regex(prog, p);
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_pop, NULL));
|
||||
return true;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,3 +1,4 @@
|
|||
#include <assert.h>
|
||||
#include "parser_internal.h"
|
||||
|
||||
// TODO: split this up.
|
||||
|
|
@ -14,7 +15,7 @@ static HParseResult *parse_many(void* env, HParseState *state) {
|
|||
HInputStream bak;
|
||||
while (env_->min_p || env_->count > count) {
|
||||
bak = state->input_stream;
|
||||
if (count > 0) {
|
||||
if (count > 0 && env_->sep != NULL) {
|
||||
HParseResult *sep = h_do_parse(env_->sep, state);
|
||||
if (!sep)
|
||||
goto err0;
|
||||
|
|
@ -47,13 +48,15 @@ static HParseResult *parse_many(void* env, HParseState *state) {
|
|||
static bool many_isValidRegular(void *env) {
|
||||
HRepeat *repeat = (HRepeat*)env;
|
||||
return (repeat->p->vtable->isValidRegular(repeat->p->env) &&
|
||||
repeat->sep->vtable->isValidRegular(repeat->sep->env));
|
||||
(repeat->sep == NULL ||
|
||||
repeat->sep->vtable->isValidRegular(repeat->sep->env)));
|
||||
}
|
||||
|
||||
static bool many_isValidCF(void *env) {
|
||||
HRepeat *repeat = (HRepeat*)env;
|
||||
return (repeat->p->vtable->isValidCF(repeat->p->env) &&
|
||||
repeat->sep->vtable->isValidCF(repeat->sep->env));
|
||||
(repeat->sep == NULL ||
|
||||
repeat->sep->vtable->isValidCF(repeat->sep->env)));
|
||||
}
|
||||
|
||||
static HCFChoice* desugar_many(HAllocator *mm__, void *env) {
|
||||
|
|
@ -70,7 +73,9 @@ static HCFChoice* desugar_many(HAllocator *mm__, void *env) {
|
|||
-> \epsilon
|
||||
*/
|
||||
|
||||
HCFChoice *sep = h_desugar(mm__, repeat->sep);
|
||||
HParser *epsilon = h_epsilon_p__m(mm__);
|
||||
|
||||
HCFChoice *sep = h_desugar(mm__, (repeat->sep != NULL) ? repeat->sep : epsilon);
|
||||
HCFChoice *a = h_desugar(mm__, repeat->p);
|
||||
HCFChoice *ma = h_new(HCFChoice, 1);
|
||||
HCFChoice *mar = h_new(HCFChoice, 1);
|
||||
|
|
@ -119,24 +124,56 @@ static HCFChoice* desugar_many(HAllocator *mm__, void *env) {
|
|||
|
||||
static bool many_ctrvm(HRVMProg *prog, void *env) {
|
||||
HRepeat *repeat = (HRepeat*)env;
|
||||
// FIXME: Implement clear_to_mark
|
||||
uint16_t clear_to_mark = h_rvm_create_action(prog, h_svm_action_clear_to_mark, NULL);
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
// TODO: implement min and max properly. Right now, it's always min==0, max==inf
|
||||
uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
|
||||
if (!h_compile_regex(prog, repeat->p))
|
||||
return false;
|
||||
if (repeat->sep != NULL) {
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
if (!h_compile_regex(prog, repeat->sep))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, clear_to_mark);
|
||||
}
|
||||
h_rvm_insert_insn(prog, RVM_GOTO, insn);
|
||||
h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
|
||||
// TODO: implement min & max properly. Right now, it's always
|
||||
// max==inf, min={0,1}
|
||||
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
|
||||
return true;
|
||||
// Structure:
|
||||
// Min == 0:
|
||||
// FORK end // if Min == 0
|
||||
// GOTO mid
|
||||
// nxt: <SEP>
|
||||
// mid: <ELEM>
|
||||
// FORK nxt
|
||||
// end:
|
||||
|
||||
if (repeat->min_p) {
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
assert(repeat->count < 2); // TODO: The other cases should be supported later.
|
||||
uint16_t end_fork;
|
||||
if (repeat->count == 0)
|
||||
end_fork = h_rvm_insert_insn(prog, RVM_FORK, 0xFFFF);
|
||||
uint16_t goto_mid = h_rvm_insert_insn(prog, RVM_GOTO, 0xFFFF);
|
||||
uint16_t nxt = h_rvm_get_ip(prog);
|
||||
if (repeat->sep != NULL) {
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
if (!h_compile_regex(prog, repeat->sep))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, clear_to_mark);
|
||||
}
|
||||
h_rvm_patch_arg(prog, goto_mid, h_rvm_get_ip(prog));
|
||||
if (!h_compile_regex(prog, repeat->p))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_FORK, nxt);
|
||||
h_rvm_patch_arg(prog, end_fork, h_rvm_get_ip(prog));
|
||||
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
|
||||
return true;
|
||||
} else {
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
for (size_t i = 0; i < repeat->count; i++) {
|
||||
if (repeat->sep != NULL && i != 0) {
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
if (!h_compile_regex(prog, repeat->sep))
|
||||
return false;
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, clear_to_mark);
|
||||
}
|
||||
if (!h_compile_regex(prog, repeat->p))
|
||||
return false;
|
||||
}
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
static const HParserVtable many_vt = {
|
||||
|
|
@ -153,7 +190,7 @@ HParser* h_many(const HParser* p) {
|
|||
HParser* h_many__m(HAllocator* mm__, const HParser* p) {
|
||||
HRepeat *env = h_new(HRepeat, 1);
|
||||
env->p = p;
|
||||
env->sep = h_epsilon_p__m(mm__);
|
||||
env->sep = NULL;
|
||||
env->count = 0;
|
||||
env->min_p = true;
|
||||
return h_new_parser(mm__, &many_vt, env);
|
||||
|
|
@ -165,7 +202,7 @@ HParser* h_many1(const HParser* p) {
|
|||
HParser* h_many1__m(HAllocator* mm__, const HParser* p) {
|
||||
HRepeat *env = h_new(HRepeat, 1);
|
||||
env->p = p;
|
||||
env->sep = h_epsilon_p__m(mm__);
|
||||
env->sep = NULL;
|
||||
env->count = 1;
|
||||
env->min_p = true;
|
||||
return h_new_parser(mm__, &many_vt, env);
|
||||
|
|
@ -177,7 +214,7 @@ HParser* h_repeat_n(const HParser* p, const size_t n) {
|
|||
HParser* h_repeat_n__m(HAllocator* mm__, const HParser* p, const size_t n) {
|
||||
HRepeat *env = h_new(HRepeat, 1);
|
||||
env->p = p;
|
||||
env->sep = h_epsilon_p__m(mm__);
|
||||
env->sep = NULL;
|
||||
env->count = n;
|
||||
env->min_p = false;
|
||||
return h_new_parser(mm__, &many_vt, env);
|
||||
|
|
@ -222,7 +259,7 @@ static HParseResult* parse_length_value(void *env, HParseState *state) {
|
|||
// TODO: allocate this using public functions
|
||||
HRepeat repeat = {
|
||||
.p = lv->value,
|
||||
.sep = h_epsilon_p(),
|
||||
.sep = NULL,
|
||||
.count = len->ast->uint,
|
||||
.min_p = false
|
||||
};
|
||||
|
|
|
|||
|
|
@ -22,7 +22,7 @@ static bool opt_isValidCF(void *env) {
|
|||
return p->vtable->isValidCF(p->env);
|
||||
}
|
||||
|
||||
static const HParsedToken* reshape_optional(const HParseResult *p) {
|
||||
static HParsedToken* reshape_optional(const HParseResult *p) {
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
assert(p->ast->seq->used > 0);
|
||||
|
|
@ -83,7 +83,7 @@ static bool opt_ctrvm(HRVMProg *prog, void* env) {
|
|||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
uint16_t insn = h_rvm_insert_insn(prog, RVM_FORK, 0);
|
||||
HParser *p = (HParser*) env;
|
||||
if (!h_compile_regex(prog, p->env))
|
||||
if (!h_compile_regex(prog, p))
|
||||
return false;
|
||||
h_rvm_patch_arg(prog, insn, h_rvm_get_ip(prog));
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_optional, NULL));
|
||||
|
|
|
|||
|
|
@ -43,7 +43,7 @@ static bool sequence_isValidCF(void *env) {
|
|||
return true;
|
||||
}
|
||||
|
||||
static const HParsedToken *reshape_sequence(const HParseResult *p) {
|
||||
static HParsedToken *reshape_sequence(const HParseResult *p) {
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
|
||||
|
|
@ -86,7 +86,7 @@ static bool sequence_ctrvm(HRVMProg *prog, void *env) {
|
|||
HSequence *s = (HSequence*)env;
|
||||
h_rvm_insert_insn(prog, RVM_PUSH, 0);
|
||||
for (size_t i=0; i<s->len; ++i) {
|
||||
if (!s->p_array[i]->vtable->compile_to_rvm(prog, s->p_array[i]))
|
||||
if (!s->p_array[i]->vtable->compile_to_rvm(prog, s->p_array[i]->env))
|
||||
return false;
|
||||
}
|
||||
h_rvm_insert_insn(prog, RVM_ACTION, h_rvm_create_action(prog, h_svm_action_make_sequence, NULL));
|
||||
|
|
|
|||
|
|
@ -20,7 +20,7 @@ static HParseResult* parse_token(void *env, HParseState *state) {
|
|||
}
|
||||
|
||||
|
||||
static const HParsedToken *reshape_token(const HParseResult *p) {
|
||||
static HParsedToken *reshape_token(const HParseResult *p) {
|
||||
// fetch sequence of uints from p
|
||||
assert(p->ast);
|
||||
assert(p->ast->token_type == TT_SEQUENCE);
|
||||
|
|
|
|||
|
|
@ -15,9 +15,9 @@ static void test_end(void) {
|
|||
}
|
||||
|
||||
static void test_example_1(void) {
|
||||
const HParser *c = h_many(h_ch('x'));
|
||||
const HParser *q = h_sequence(c, h_ch('y'), NULL);
|
||||
const HParser *p = h_choice(q, h_end_p(), NULL);
|
||||
HParser *c = h_many(h_ch('x'));
|
||||
HParser *q = h_sequence(c, h_ch('y'), NULL);
|
||||
HParser *p = h_choice(q, h_end_p(), NULL);
|
||||
HCFGrammar *g = h_cfgrammar(&system_allocator, p);
|
||||
|
||||
g_check_nonterminal(g, c);
|
||||
|
|
|
|||
|
|
@ -162,7 +162,7 @@ static void test_middle(gconstpointer backend) {
|
|||
|
||||
#include <ctype.h>
|
||||
|
||||
const HParsedToken* upcase(const HParseResult *p) {
|
||||
HParsedToken* upcase(const HParseResult *p) {
|
||||
switch(p->ast->token_type) {
|
||||
case TT_SEQUENCE:
|
||||
{
|
||||
|
|
@ -180,17 +180,17 @@ const HParsedToken* upcase(const HParseResult *p) {
|
|||
}
|
||||
}
|
||||
ret->seq = seq;
|
||||
return (const HParsedToken*)ret;
|
||||
return ret;
|
||||
}
|
||||
case TT_UINT:
|
||||
{
|
||||
HParsedToken *ret = a_new_(p->arena, HParsedToken, 1);
|
||||
ret->token_type = TT_UINT;
|
||||
ret->uint = toupper(p->ast->uint);
|
||||
return (const HParsedToken*)ret;
|
||||
return ret;
|
||||
}
|
||||
default:
|
||||
return p->ast;
|
||||
return (HParsedToken*)p->ast;
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -526,6 +526,7 @@ void register_parser_tests(void) {
|
|||
g_test_add_data_func("/core/parser/regex/choice", GINT_TO_POINTER(PB_REGULAR), test_choice);
|
||||
g_test_add_data_func("/core/parser/regex/many", GINT_TO_POINTER(PB_REGULAR), test_many);
|
||||
g_test_add_data_func("/core/parser/regex/many1", GINT_TO_POINTER(PB_REGULAR), test_many1);
|
||||
g_test_add_data_func("/core/parser/regex/repeat_n", GINT_TO_POINTER(PB_REGULAR), test_repeat_n);
|
||||
g_test_add_data_func("/core/parser/regex/optional", GINT_TO_POINTER(PB_REGULAR), test_optional);
|
||||
g_test_add_data_func("/core/parser/regex/sepBy", GINT_TO_POINTER(PB_REGULAR), test_sepBy);
|
||||
g_test_add_data_func("/core/parser/regex/sepBy1", GINT_TO_POINTER(PB_REGULAR), test_sepBy1);
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue