gen_collate.c

一个嵌入式的C++库

								if (*t == '"') {
									assert(!t[1]);
									break;
								}
								n = is_ucode(t);
								assert(n);
								printf("%.*s", n, t);
							} while (1);
							printf("   collating-element %s\n", s);
						} else {
							printf("%6d: %s  (collating-symbol)\n", final_index, s);
						}
					}
				}
			}
		}
#endif
		++final_index;
	}

}

static int final_index_val0(const char *s)
{
	ENTRY *p;
	ENTRY e;
	e.key = (char *) s;

	if (!(p = hsearch(e, FIND))) {	/* not in the table */
		return 0;
	}

	return (int)(p->data);
}

static int final_index_val(const char *s)
{
	ENTRY *p;
	ENTRY e;
	e.key = (char *) s;

	if (!(p = hsearch(e, FIND))) {	/* not in the table */
		error_msg("can't find final index: %s", s);
	}

	return (int)(p->data);
}

static size_t num_tree_nodes;

static void count_nodes(const void *ptr, VISIT order, int level)
{
    if ((order == postorder) || (order == leaf))  {
		++num_tree_nodes;
    }
}

static size_t tnumnodes(const void *root)
{
	num_tree_nodes = 0;

	twalk(root, count_nodes);

	return num_tree_nodes;

}

static ll_item_t *find_wi_index(const char *sym, col_locale_t *cl)
{
	weighted_item_t w;
	ll_item_t l;
	void *p;

	w.symbol = sym;
	l.data = &w;
	l.data_type = DT_WEIGHTED;

	p = tfind(&l, &cl->root_wi_index, wi_index_cmp);

	if (p) {
		p = *(ll_item_t **)p;
	}

	return (ll_item_t *) p;
}

static void mark_reordered(const char *sym)
{
	ll_item_t *lli;

	lli = find_wi_index(sym, cur_base);

	if (lli) {
		if (!tsearch(lli, &cur_base->root_wi_index_reordered, wi_index_cmp)) {
			error_msg("OUT OF MEMORY!");
		}
	}
}

static ll_item_t *find_wi_index_reordered(const char *sym)
{
	weighted_item_t w;
	ll_item_t l;
	void *p;

	w.symbol = sym;
	l.data = &w;
	l.data_type = DT_WEIGHTED;

	p = tfind(&l, &cur_base->root_wi_index_reordered, wi_index_cmp);

	if (p) {
		p = *(ll_item_t **)p;
	}

	return (ll_item_t *) p;
}

static ll_item_t *init_comm_ptr(void)
{
	assert(cur_base);
	assert(cur_base->section_list);
	/* at the moment, only support one section in comm */
	assert(cur_base->section_list->next == NULL);

	comm_cur_ptr = ((section_t *)(cur_base->section_list->data))->itm_list;

	while (comm_cur_ptr && (comm_cur_ptr->data_type & DT_REORDER)) {
		comm_cur_ptr = comm_cur_ptr->next;
	}

#warning devel code
/* 	{ */
/* 		ll_item_t *p = comm_cur_ptr; */
/* 		fprintf(stderr, "init_comm_ptr\n"); */

/* 		while (p != comm_cur_ptr) { */
/* 			if (p->data_type & DT_WEIGHTED) { */
/* 				fprintf(stderr, "%s", ((weighted_item_t *)p)->symbol); */
/* 			} */
/* 			p = p->next; */
/* 		} */
/* 	} */

	assert(comm_cur_ptr);

/* 	fprintf(stderr, "init_comm_ptr -- %s %p %p %p %d\n", */
/* 			((weighted_item_t *)(comm_cur_ptr->data))->symbol, */
/* 			comm_cur_ptr, comm_cur_ptr->prev, comm_cur_ptr->next, */
/* 			ll_len(comm_cur_ptr)); */

	comm_prev_ptr = NULL;
	return comm_cur_ptr;
}

static ll_item_t *next_comm_ptr(void)
{
	/* at the moment, only support one section in comm */
	assert(cur_base->section_list->next == NULL);

	comm_prev_ptr = comm_cur_ptr;

    while (comm_cur_ptr && ((comm_cur_ptr = comm_cur_ptr->next) != NULL)) {
		if (!(comm_cur_ptr->data_type & DT_REORDER)) {
			break;
		}
	}

	return comm_cur_ptr;
}

static int dump_count;

#if 0
static void dump_section(section_t *s, int mask, col_locale_t *der)
{
	ll_item_t *lli;
	ll_item_t *lli0;
	weighted_item_t *w;
	weight_t *p;
	int i;

	lli0 = lli = s->itm_list;

	if (!lli0) {
		return;
	}

	do {
		if (!(lli->data_type & mask)) {
			lli = lli->next;
			continue;
		}
		if (lli->data_type & DT_WEIGHTED) {
			++dump_count;
			w = (weighted_item_t *)(lli->data);
			p = w->weight;
			printf("%6d: %s (%d) ", dump_count, w->symbol, p->num_weights);
			for (i = 0 ; i < p->num_weights ; i++) {
				if (p->rule[i] & R_FORWARD) {
					printf("F");
				}
				if (p->rule[i] & R_BACKWARD) {
					printf("B");
				}
				if (p->rule[i] & R_POSITION) {
					printf("P");
				}
				printf(",");
			}
			for (i = 0 ; i < p->num_weights ; i++) {
				printf("   %s", p->colitem[i]);
			}
			printf("\n");
		} else if (lli->data_type & (DT_SECTION|DT_REORDER)) {

			if (lli->data_type == DT_REORDER) {
				assert(der);
				if (strncmp(((section_t *)(lli->data))->name, der->name, strlen(der->name))) {
					lli = lli->next;
					continue;
				}
			}

			if (lli->data_type & DT_SECTION) {
				printf("SECTION -----------------\n");
			} else {
				printf("REORDER -----------------\n");
			}

			dump_section((section_t *)(lli->data), mask, der);
			printf("DONE --------------------\n");
		}
		lli = lli->next;
	} while (lli != lli0);
}
#else
static int in_reorder_section = 0;

static void dump_section(section_t *s, int mask, col_locale_t *der)
{
	ll_item_t *lli;
	ll_item_t *lli0;
	weighted_item_t *w;
	weight_t *p;
	int i;

	lli0 = lli = s->itm_list;

	if (!lli0) {
		return;
	}

	do {
		if (!(lli->data_type & mask)) {
			lli = lli->next;
			continue;
		}
		if (lli->data_type & DT_WEIGHTED) {
			++dump_count;
			w = (weighted_item_t *)(lli->data);
			p = w->weight;
#if 1
			if (in_reorder_section) {
				printf(" %p", w);
			}
#else
			printf("%6d: %s (%d) ", dump_count, w->symbol, p->num_weights);
			for (i = 0 ; i < p->num_weights ; i++) {
				if (p->rule[i] & R_FORWARD) {
					printf("F");
				}
				if (p->rule[i] & R_BACKWARD) {
					printf("B");
				}
				if (p->rule[i] & R_POSITION) {
					printf("P");
				}
				printf(",");
			}
			for (i = 0 ; i < p->num_weights ; i++) {
				printf("   %s", p->colitem[i]);
			}
			printf("\n");
#endif
		} else if (lli->data_type & (DT_SECTION|DT_REORDER)) {

			if (lli->data_type == DT_REORDER) {
				assert(der);
				if (strncmp(((section_t *)(lli->data))->name, der->name, strlen(der->name))) {
					lli = lli->next;
					continue;
				}
			}

			if (lli->data_type & DT_SECTION) {
/* 				printf("SECTION -----------------\n"); */
				assert(0);
			} else {
/* 				printf("REORDER -----------------\n"); */
				in_reorder_section = 1;
			}

			dump_section((section_t *)(lli->data), mask, der);
/* 			printf("DONE --------------------\n"); */
			printf("\n");
			in_reorder_section = 0;
		}
		lli = lli->next;
	} while (lli != lli0);
}
#endif

static void dump_weights(const char *name)
{
	ll_item_t *lli;
	col_locale_t *base;
	col_locale_t *der;
	col_locale_t cl;
	void *p;

	assert(name);

	if (!*name) {				/* use last */
		base = cur_base;
		der = cur_derived;
	} else {
		cl.name = (char *) name;
		if (!(p = tfind(&cl, &root_col_locale, col_locale_cmp))) {
			error_msg("unknown locale: %s", name);
		}
		base = *((col_locale_t **) p);
		der = NULL;
		if (base->base_locale) { /* oops... really derived */
			der = base;
			base = der->base_locale;
		}
	}

	dump_count = 0;

	if (base) {
/* 		printf("BASE - %s\n", base->name); */
		for (lli = base->section_list ; lli ; lli = lli->next) {
/* 			printf("SECTION %s\n", ((section_t *)(lli->data))->name); */
			dump_section((section_t *)(lli->data), ~0, der);
		}
	}

	assert(der != base);

	if (der) {
/* 		printf("DERIVED - %s\n", der->name); */
		for (lli = der->section_list ; lli ; lli = lli->next) {
			if (lli->data_type == DT_SECTION) {
				dump_section((section_t *)(lli->data), DT_WEIGHTED, der);
			}
		}
	}
/* 	printf("DONE\n"); */
}

static void print_starter_node(const void *ptr, VISIT order, int level)
{
    if (order == postorder || order == leaf)  {
		fprintf(stderr, "   %s\n", *(const char **) ptr);
    }
}

static void finalize_base(void)
{
	ll_item_t *s;
	ll_item_t *h;
	ll_item_t *lli;
	ll_item_t *h2;
	ll_item_t *l2;
	ll_item_t *cli;
	ll_item_t *rli = NULL;
	weighted_item_t *w;
	weight_t *p;
	int i, n, mr, r, mi;
	col_locale_t *cl;
	void *mm;

	int num_invariant = 0;
	int num_varying = 0;
	int max_weight;
	int index2weight_len_inc = 1;

	assert(cur_base);
	assert(base_locale_len+1 < BASE_LOCALE_LEN);

	base_locale_array[base_locale_len].name = cur_base->name;
	base_locale_array[base_locale_len].num_weights = 1;
	base_locale_array[base_locale_len].index2weight_offset = index2weight_len;
	base_locale_array[base_locale_len].index2ruleidx_offset = index2ruleidx_len;
	if (!strcmp(cur_base->name,"ja_JP") || !strcmp(cur_base->name,"ko_KR")) {
#warning fix the index2weight check!!
		index2weight_len_inc = 0;
	}
/* 	printf("%s -- index2weight_len = %d\n", cur_base->name, index2weight_len); */

	if (!hcreate(30000)) {
		error_msg("OUT OF MEMORY!");
	}

	/* first pass ... set the fixed indexes */
	final_index = i = 1;
	mr = 0;
	for (s = cur_base->section_list ; s ; s = s->next) {
#if 1
		if (s->data_type & DT_REORDER) { /* a reordered section */
			fprintf(stderr, "pass1: reordered section %s - xxx\n", ((section_t *)(s->data))->name);
			lli = ((section_t *)(s->data))->itm_list;
			r = 0;
			if (lli) {
/* 				r = ll_len( ((section_t *)(lli->data))->itm_list ); */
				r = ll_len(lli) + 1;
			}
			if (r > mr) {
				mr = r;
			}
			fprintf(stderr, "pass1: reordered section %s - %d\n", ((section_t *)(s->data))->name, r);
			continue;
		}
#endif
		h = lli = ((section_t *)(s->data))->itm_list;
		if (!lli) {
			continue;
		}
		do {
			if (lli->data_type & DT_RANGE) {
				i += mr;
				mr = 0;
#warning check ko_kR and 9
/* 				++i; */
				lli->idx = i;
				assert(!rli);
				rli = lli;
				fprintf(stderr, "range pre = %d  after = ", i);
				i += ((range_item_t *)(lli->data))->length + 1;
#warning check ko_kR and 9
/* 				++i; */
				fprintf(stderr, "%d\n", i);
				if (!index2weight_len_inc) { /* ko_KR hack */
					final_index += ((range_item_t *)(lli->data))->length + 1;
				}
/* 				add_final_col_index("RANGE"); */
			} else if (lli->data_type & DT_WEIGHTED) {
				i += mr;
				mr = 0;
				w = (weighted_item_t *)(lli->data);
				if (find_wi_index_reordered(w->symbol)) { /* reordered symbol so skip on first pass */
					++num_varying;
					++i;
					continue;
				}
				++num_invariant;
				index2weight_buffer[index2weight_len] = lli->idx = i++;
				index2weight_len += index2weight_len_inc;
				add_final_col_index(w->symbol);
				
			} else {
				assert(lli->data_type & DT_REORDER);
				r = ll_len( ((section_t *)(lli->data))->itm_list );
#warning check ko_kR and 9
				if (r > mr) {
					mr = r;
				}
/* 				r = 0; */
			}
		} while ((lli = lli->next) != h);
	}

	/* second pass ... set the reordered indexes */
	mi = i + mr;
	mr = i = 0;
	for (s = cur_base->section_list ; s ; s = s->next) {
		h = lli = ((section_t *)(s->data))->itm_list;
		if (!lli) {
			continue;
		}
		do {
			if (lli->data_type & DT_RANGE) {
				i += mr;
				mr = 0;
				i = lli->idx + ((range_item_t *)(lli->data))->length + 1;
#warning check
			} else if ((lli->data_type & DT_WEIGHTED) && !(s->data_type & DT_REORDER)) {
				i += mr;
				mr = 0;
				w = (weighted_item_t *)(lli->data);
				if (find_wi_index_reordered(w->symbol) /* reordered symbol skipped on first pass */
#if 0
					|| (s->data_type & DT_REORDER) /* or in a reordered section */
#endif
					) {
					assert(!(s->data_type & DT_REORDER));
					index2weight_buffer[index2weight_len] = lli->idx = ++i;
					index2weight_len += index2weight_len_inc;
					add_final_col_index(w->symbol);

/* 					fprintf(stdout, "%11s: r %6d %6d %s\n", */
/* 							cur_base->name, lli->idx, final_index_val(w->symbol), w->symbol); */

					continue;
				}
				i = lli->idx;

/* 				fprintf(stdout, "%11s: w %6d %6d %s\n", */
/* 						cur_base->name, lli->idx, final_index_val(w->symbol), w->symbol); */

			} else {
/* 				fprintf(stderr, "section: %s  %d  %d\n", ((section_t *)(s->data))->name, */
/* 						s->data_type, lli->data_type); */
/* 					assert(!(s->data_type & DT_REORDER)); */
/* 				assert(lli->data_type & DT_REORDER); */
#if 1
				if (s->data_type & DT_REORDER) {
					h2 = l2 = lli;
					if (!h2) {
						continue;
					}
				} else {
					assert(s->data_type & DT_SECTION);
					h2 = l2 = ((section_t *)(lli->data))->itm_list;
					if (!h2) {
						continue;
					}
				}


#else
				h2 = l2 = ((section_t *)(lli->data))->itm_list;
				if (!h2) {
					continue;
				}
#endif
				r = 0;
				do {
					assert(l2->data_type & DT_WEIGHTED);
					++r;
					l2->idx = i + r;

/* 					fprintf(stdout, "%s: R %6d        %s\n", */
/* 							((section_t *)(lli->data))->name, l2->idx, ((weighted_item_t *)(l2->data))->symbol); */

				} while ((l2 = l2->next) != h2);
				if (r > mr) {
					mr = r;
				}
			}
		} while ((lli = ll