emileclarkb/hammer
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

support naming of generating functions, and 0-weighting of symbols

Browse source
This commit is contained in:
Mikael Vejdemo-Johansson 2015-04-08 15:29:32 +02:00
parent 1e95eac6cd
commit 24e9e9de7f
2 changed files with 45 additions and 159 deletions
Download patch file Download diff file Expand all files Collapse all files

152
examples/explore_singular.c
View file

@ -1,152 +0,0 @@
//
// Created by Mikael Vejdemo Johansson on 4/7/15.
//
// Intention: read in a parser, generate the system of equations for its
// generating functions
//
#include <inttypes.h>
#include "../src/backends/contextfree.h"
#include "../src/backends/lr.h"
#include <stdio.h>
static const char *nonterminal_name(const HCFGrammar *g, const HCFChoice *nt)
{
static char buf[16] = {0}; // 14 characters in base 26 are enough for 64 bits
// find nt's number in g
size_t n = (uintptr_t)h_hashtable_get(g->nts, nt);
// NB the start symbol (number 0) is always "A".
int i;
for(i=14; i>=0 && (n>0 || i==14); i--) {
buf[i] = 'A' + n%26;
n = n/26; // shift one digit
}
return buf+i+1;
}
void readsequence(FILE *file, uint32_t *count, uint32_t *length,
const HCFGrammar *g, const HCFSequence *seq) {
// tally up numbers of choices, and lengths of emitted strings.
// Immediately emit any nonterminals encountered.
HCFChoice** x = seq->items;
if (*x == NULL) {
return;
} else {
fprintf(file, "1");
HCharset cs;
unsigned int i, cscount=0;
for(; *x; x++) {
switch((*x)->type) {
case HCF_CHAR:
(*length)++;
break;
case HCF_END:
break;
case HCF_CHARSET:
cs = (*x)->charset;
for(i=0; i<256; i++) {
if (charset_isset(cs, i)) {
cscount++;
}
}
*count *= cscount;
break;
default: // HCF_CHOICE, non-terminal symbol
fprintf(file, "*%s(t)", nonterminal_name(g, *x));
break;
}
}
}
}
// For each nt in g->nts
// For each choice in nt->key->seq
// For all elements in sequence
// Accumulate counts
// Accumulate string lengths
// Emit count*t^length
void h_pprint_gfeqns(FILE *file, const HCFGrammar *g) {
if (g->nts->used < 1) {
return;
}
// determine maximum string length of symbol names
int len;
size_t s;
for(len=1, s=26; s < g->nts->used; len++, s*=26);
// iterate over g->nts
size_t i;
HHashTableEntry *hte;
for(i=0; i < g->nts->capacity; i++) {
for(hte = &g->nts->contents[i]; hte; hte = hte->next) {
if (hte->key == NULL) {
continue;
}
const HCFChoice *nt = hte->key;
fprintf(file, "%s(t) = ", nonterminal_name(g, nt));
for(HCFSequence **seq = nt->seq; *seq; seq++) {
if (seq != nt->seq) {
fprintf(file, " + ");
}
uint32_t count=1, length=0;
readsequence(file, &count, &length, g, *seq);
if(count == 1) {
if(length == 1) {
fprintf(file, "*t");
}
if(length > 1) {
fprintf(file, "*t^%d", length);
}
} else if(count > 1) {
if(length == 0) {
fprintf(file, "*%d", count);
}
if(length == 1) {
fprintf(file, "*%d*t", count);
}
if (length > 1) {
fprintf(file, "*%d*t^%d", count, length);
}
}
}
fprintf(file, "\n");
}
}
}
int main(int argc, char **argv)
{
HAllocator *mm__ = &system_allocator;
HParser *n = h_ch('n');
HParser *E = h_indirect();
HParser *T = h_choice(h_sequence(h_ch('('), E, h_ch(')'), NULL), n, NULL);
HParser *E_ = h_choice(h_sequence(E, h_ch('-'), T, NULL), T, NULL);
h_bind_indirect(E, E_);
HParser *p = E;
HCFGrammar *g = h_cfgrammar_(mm__, h_desugar_augmented(mm__, p));
if (g == NULL) {
fprintf(stderr, "h_cfgrammar failed\n");
return 1;
}
printf("\n==== Generating functions ====\n");
h_pprint_gfeqns(stdout, g);
printf("\n==== Grammar ====\n");
h_pprint_grammar(stdout, g, 0);
}

52
examples/ties.c
View file

@ -10,8 +10,30 @@
#include "../src/backends/lr.h"
#include <stdio.h>
static const char *nonterminal_name(const HCFGrammar *g, const HCFChoice *nt)
{
HAllocator *mm__;
// If a parser has user_data set, the generating function systems will try
// to interpret it as a string:
//
// If this string for an h_ch starts with the character 0, then that character
// will have weight 0 in the generating function.
//
// Use the remaining string to set the preferred name of that parser in the
// generating function.
static const char *nonterminal_name(const HCFGrammar *g, const HCFChoice *nt) {
if(nt->user_data != NULL) {
if(*(char*)(nt->user_data) != '0') {
// user_data is a non-empty string
return nt->user_data;
} else {
return nt->user_data+1;
}
}
static char buf[16] = {0}; // 14 characters in base 26 are enough for 64 bits
// find nt's number in g
@ -38,13 +60,16 @@ void readsequence(FILE *file, uint32_t *count, uint32_t *length,
if (*x == NULL) {
return;
} else {
char has_user_data = (*x)->user_data != NULL && *(char*)(*x)->user_data != 0;
fprintf(file, "1");
HCharset cs;
unsigned int i, cscount=0;
for(; *x; x++) {
switch((*x)->type) {
case HCF_CHAR:
(*length)++;
if(!(has_user_data && *(char*)(*x)->user_data == '0')) {
(*length)++;
}
break;
case HCF_END:
break;
@ -182,7 +207,21 @@ HParser* finkmaoTW() {
h_epsilon_p(),
NULL);
h_bind_indirect(pairstar, pstar_);
return h_sequence(prefix, pairstar, tuck, NULL);
HParser* tie = h_sequence(prefix, pairstar, tuck, NULL);
h_desugar_augmented(mm__, tie);
T->desugared->user_data = "T";
W->desugared->user_data = "W";
U->desugared->user_data = "0U";
prefix->desugared->user_data = "prefix";
pair->desugared->user_data = "pair";
tuck->desugared->user_data = "tuck";
pstar_->desugared->user_data = "pairstar";
tie->desugared->user_data = "tie";
return tie;
}
HParser* depth1TW() {
@ -341,9 +380,8 @@ HParser* depthNTW() {
}
int main(int argc, char **argv)
{
HAllocator *mm__ = &system_allocator;
int main(int argc, char **argv) {
mm__ = &system_allocator;
HParser *p = finkmaoTW();
HCFGrammar *g = h_cfgrammar_(mm__, h_desugar_augmented(mm__, p));