forked from GitHub/gf-core
a major refactoring in the C and the Haskell runtimes. Note incompatible change in the PGF format!!!
The following are the outcomes:
- Predef.nonExist is fully supported by both the Haskell and the C runtimes
- Predef.BIND is now an internal compiler defined token. For now
it behaves just as usual for the Haskell runtime, i.e. it generates &+.
However, the special treatment will let us to handle it properly in
the C runtime.
- This required a major change in the PGF format since both
nonExist and BIND may appear inside 'pre' and this was not supported
before.
This commit is contained in:
@@ -131,8 +131,11 @@ extern GU_DECLARE_TYPE(PgfCncCat, abstract);
|
||||
bool
|
||||
pgf_tokens_equal(PgfTokens* t1, PgfTokens* t2);
|
||||
|
||||
typedef GuSeq PgfSequence; // -> PgfSymbol
|
||||
typedef GuSeq PgfSequences;
|
||||
|
||||
typedef struct {
|
||||
PgfTokens* form;
|
||||
PgfSequence* form;
|
||||
/**< The form of this variant as a list of tokens. */
|
||||
|
||||
GuStrings* prefixes;
|
||||
@@ -175,7 +178,8 @@ typedef enum {
|
||||
PGF_SYMBOL_VAR,
|
||||
PGF_SYMBOL_KS,
|
||||
PGF_SYMBOL_KP,
|
||||
PGF_SYMBOL_NE
|
||||
PGF_SYMBOL_NE,
|
||||
PGF_SYMBOL_BIND
|
||||
} PgfSymbolTag;
|
||||
|
||||
typedef struct {
|
||||
@@ -186,14 +190,14 @@ typedef struct {
|
||||
typedef PgfSymbolIdx PgfSymbolCat, PgfSymbolLit, PgfSymbolVar;
|
||||
|
||||
typedef struct {
|
||||
PgfTokens* tokens;
|
||||
PgfToken token;
|
||||
} PgfSymbolKS;
|
||||
|
||||
typedef struct PgfSymbolKP
|
||||
/** A prefix-dependent symbol. The form that this symbol takes
|
||||
* depends on the form of a prefix of the following symbol. */
|
||||
{
|
||||
PgfTokens* default_form;
|
||||
PgfSequence* default_form;
|
||||
/**< Default form that this symbol takes if none of of the
|
||||
* variant forms is triggered. */
|
||||
|
||||
@@ -206,8 +210,8 @@ typedef struct PgfSymbolKP
|
||||
typedef struct {
|
||||
} PgfSymbolNE;
|
||||
|
||||
typedef GuSeq PgfSequence; // -> PgfSymbol
|
||||
typedef GuSeq PgfSequences;
|
||||
typedef struct {
|
||||
} PgfSymbolBIND;
|
||||
|
||||
typedef struct {
|
||||
PgfAbsFun* absfun;
|
||||
|
||||
@@ -116,18 +116,15 @@ typedef struct {
|
||||
} PgfBracketLznState;
|
||||
|
||||
static void
|
||||
pgf_bracket_lzn_symbol_tokens(PgfLinFuncs** funcs, PgfTokens* toks)
|
||||
pgf_bracket_lzn_symbol_token(PgfLinFuncs** funcs, PgfToken tok)
|
||||
{
|
||||
PgfBracketLznState* state = gu_container(funcs, PgfBracketLznState, funcs);
|
||||
|
||||
size_t len = gu_seq_length(toks);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
PgfParseNode* node = gu_new(PgfParseNode, state->pool);
|
||||
node->id = 100000 + gu_buf_length(state->leaves);
|
||||
node->parent = state->parent;
|
||||
node->label = gu_seq_get(toks, PgfToken, i);
|
||||
gu_buf_push(state->leaves, PgfParseNode*, node);
|
||||
}
|
||||
PgfParseNode* node = gu_new(PgfParseNode, state->pool);
|
||||
node->id = 100000 + gu_buf_length(state->leaves);
|
||||
node->parent = state->parent;
|
||||
node->label = tok;
|
||||
gu_buf_push(state->leaves, PgfParseNode*, node);
|
||||
}
|
||||
|
||||
static void
|
||||
@@ -214,7 +211,7 @@ pgf_bracket_lzn_end_phrase(PgfLinFuncs** funcs, PgfCId cat, int fid, int lindex,
|
||||
}
|
||||
|
||||
static PgfLinFuncs pgf_bracket_lin_funcs = {
|
||||
.symbol_tokens = pgf_bracket_lzn_symbol_tokens,
|
||||
.symbol_token = pgf_bracket_lzn_symbol_token,
|
||||
.expr_literal = pgf_bracket_lzn_expr_literal,
|
||||
.begin_phrase = pgf_bracket_lzn_begin_phrase,
|
||||
.end_phrase = pgf_bracket_lzn_end_phrase
|
||||
|
||||
@@ -453,6 +453,50 @@ pgf_lzr_concretize(PgfConcr* concr, PgfExpr expr, GuPool* pool)
|
||||
return &lzn->en;
|
||||
}
|
||||
|
||||
void
|
||||
pgf_lzr_linearize_sequence(PgfConcr* concr,
|
||||
PgfCncTreeApp* fapp, PgfSequence* seq,
|
||||
PgfLinFuncs** fnsp)
|
||||
{
|
||||
size_t nsyms = gu_seq_length(seq);
|
||||
PgfSymbol* syms = gu_seq_data(seq);
|
||||
for (size_t i = 0; i < nsyms; i++) {
|
||||
PgfSymbol sym = syms[i];
|
||||
GuVariantInfo sym_i = gu_variant_open(sym);
|
||||
switch (sym_i.tag) {
|
||||
case PGF_SYMBOL_CAT:
|
||||
case PGF_SYMBOL_VAR:
|
||||
case PGF_SYMBOL_LIT: {
|
||||
PgfSymbolIdx* sidx = sym_i.data;
|
||||
gu_assert((unsigned) sidx->d < fapp->n_args);
|
||||
|
||||
PgfCncTree argf = fapp->args[sidx->d];
|
||||
pgf_lzr_linearize(concr, argf, sidx->r, fnsp);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* ks = sym_i.data;
|
||||
if ((*fnsp)->symbol_token) {
|
||||
(*fnsp)->symbol_token(fnsp, ks->token);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
// TODO: correct prefix-dependencies
|
||||
PgfSymbolKP* kp = sym_i.data;
|
||||
pgf_lzr_linearize_sequence(concr, fapp, kp->default_form, fnsp);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_NE: {
|
||||
// Nothing to be done here
|
||||
break;
|
||||
}
|
||||
default:
|
||||
gu_impossible();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
pgf_lzr_linearize(PgfConcr* concr, PgfCncTree ctree, size_t lin_idx, PgfLinFuncs** fnsp)
|
||||
{
|
||||
@@ -472,47 +516,9 @@ pgf_lzr_linearize(PgfConcr* concr, PgfCncTree ctree, size_t lin_idx, PgfLinFuncs
|
||||
}
|
||||
|
||||
gu_require(lin_idx < fun->n_lins);
|
||||
PgfSequence* seq = fun->lins[lin_idx];
|
||||
size_t nsyms = gu_seq_length(seq);
|
||||
PgfSymbol* syms = gu_seq_data(seq);
|
||||
for (size_t i = 0; i < nsyms; i++) {
|
||||
PgfSymbol sym = syms[i];
|
||||
GuVariantInfo sym_i = gu_variant_open(sym);
|
||||
switch (sym_i.tag) {
|
||||
case PGF_SYMBOL_CAT:
|
||||
case PGF_SYMBOL_VAR:
|
||||
case PGF_SYMBOL_LIT: {
|
||||
PgfSymbolIdx* sidx = sym_i.data;
|
||||
gu_assert((unsigned) sidx->d < fapp->n_args);
|
||||
|
||||
PgfCncTree argf = fapp->args[sidx->d];
|
||||
pgf_lzr_linearize(concr, argf, sidx->r, fnsp);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* ks = sym_i.data;
|
||||
if (fns->symbol_tokens) {
|
||||
fns->symbol_tokens(fnsp, ks->tokens);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
// TODO: correct prefix-dependencies
|
||||
PgfSymbolKP* kp = sym_i.data;
|
||||
if (fns->symbol_tokens) {
|
||||
fns->symbol_tokens(fnsp,
|
||||
kp->default_form);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_NE: {
|
||||
// Nothing to be done here
|
||||
break;
|
||||
}
|
||||
default:
|
||||
gu_impossible();
|
||||
}
|
||||
}
|
||||
PgfSequence* seq = fun->lins[lin_idx];
|
||||
pgf_lzr_linearize_sequence(concr, fapp, seq, fnsp);
|
||||
|
||||
if (fns->end_phrase) {
|
||||
fns->end_phrase(fnsp,
|
||||
@@ -572,22 +578,18 @@ struct PgfSimpleLin {
|
||||
};
|
||||
|
||||
static void
|
||||
pgf_file_lzn_symbol_tokens(PgfLinFuncs** funcs, PgfTokens* toks)
|
||||
pgf_file_lzn_symbol_token(PgfLinFuncs** funcs, PgfToken tok)
|
||||
{
|
||||
PgfSimpleLin* flin = gu_container(funcs, PgfSimpleLin, funcs);
|
||||
if (!gu_ok(flin->err)) {
|
||||
return;
|
||||
}
|
||||
size_t len = gu_seq_length(toks);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
if (flin->n_tokens > 0)
|
||||
gu_putc(' ', flin->out, flin->err);
|
||||
if (flin->n_tokens > 0)
|
||||
gu_putc(' ', flin->out, flin->err);
|
||||
|
||||
PgfToken tok = gu_seq_get(toks, PgfToken, i);
|
||||
gu_string_write(tok, flin->out, flin->err);
|
||||
|
||||
flin->n_tokens++;
|
||||
}
|
||||
gu_string_write(tok, flin->out, flin->err);
|
||||
|
||||
flin->n_tokens++;
|
||||
}
|
||||
|
||||
static void
|
||||
@@ -626,10 +628,10 @@ pgf_file_lzn_expr_literal(PgfLinFuncs** funcs, PgfLiteral lit)
|
||||
}
|
||||
|
||||
static PgfLinFuncs pgf_file_lin_funcs = {
|
||||
.symbol_tokens = pgf_file_lzn_symbol_tokens,
|
||||
.expr_literal = pgf_file_lzn_expr_literal,
|
||||
.begin_phrase = NULL,
|
||||
.end_phrase = NULL,
|
||||
.symbol_token = pgf_file_lzn_symbol_token,
|
||||
.expr_literal = pgf_file_lzn_expr_literal,
|
||||
.begin_phrase = NULL,
|
||||
.end_phrase = NULL,
|
||||
};
|
||||
|
||||
void
|
||||
|
||||
@@ -51,7 +51,7 @@ typedef struct PgfLinFuncs PgfLinFuncs;
|
||||
struct PgfLinFuncs
|
||||
{
|
||||
/// Output tokens
|
||||
void (*symbol_tokens)(PgfLinFuncs** self, PgfTokens* toks);
|
||||
void (*symbol_token)(PgfLinFuncs** self, PgfToken tok);
|
||||
|
||||
/// Output literal
|
||||
void (*expr_literal)(PgfLinFuncs** self, PgfLiteral lit);
|
||||
|
||||
@@ -42,7 +42,7 @@ pgf_match_string_lit(PgfConcr* concr, PgfItem* item, PgfToken tok,
|
||||
gu_new_variant(PGF_LITERAL_STR,
|
||||
PgfLiteralStr,
|
||||
&expr_lit->lit, pool);
|
||||
lit_str->val = gu_seq_get(sks->tokens, PgfToken, 0);
|
||||
lit_str->val = sks->token;
|
||||
|
||||
*out_ep = ep;
|
||||
accepted = false;
|
||||
@@ -80,10 +80,9 @@ pgf_match_int_lit(PgfConcr* concr, PgfItem* item, PgfToken tok,
|
||||
} else if (n_syms == 1) {
|
||||
PgfSymbolKS* sks =
|
||||
gu_variant_data(gu_seq_get(seq, PgfSymbol, 0));
|
||||
PgfToken tok = gu_seq_get(sks->tokens, PgfToken, 0);
|
||||
|
||||
int val;
|
||||
if (!gu_string_to_int(tok, &val)) {
|
||||
if (!gu_string_to_int(sks->token, &val)) {
|
||||
*out_ep = NULL;
|
||||
} else {
|
||||
PgfExprProb* ep = gu_new(PgfExprProb, pool);
|
||||
@@ -137,10 +136,9 @@ pgf_match_float_lit(PgfConcr* concr, PgfItem* item, PgfToken tok,
|
||||
} else if (n_syms == 1) {
|
||||
PgfSymbolKS* sks =
|
||||
gu_variant_data(gu_seq_get(seq, PgfSymbol, 0));
|
||||
PgfToken tok = gu_seq_get(sks->tokens, PgfToken, 0);
|
||||
|
||||
double val;
|
||||
if (!gu_string_to_double(tok, &val)) {
|
||||
if (!gu_string_to_double(sks->token, &val)) {
|
||||
*out_ep = NULL;
|
||||
} else {
|
||||
PgfExprProb* ep = gu_new(PgfExprProb, pool);
|
||||
@@ -209,9 +207,8 @@ pgf_match_name_lit(PgfConcr* concr, PgfItem* item, PgfToken tok,
|
||||
PgfSymbol sym = gu_seq_get(seq, PgfSymbol, i);
|
||||
gu_assert(gu_variant_tag(sym) == PGF_SYMBOL_KS);
|
||||
PgfSymbolKS* sks = gu_variant_data(sym);
|
||||
PgfToken tok = gu_seq_get(sks->tokens, PgfToken, 0);
|
||||
|
||||
gu_string_write(tok, out, err);
|
||||
gu_string_write(sks->token, out, err);
|
||||
}
|
||||
|
||||
PgfExprProb* ep = gu_new(PgfExprProb, pool);
|
||||
|
||||
@@ -133,8 +133,8 @@ struct PgfItem {
|
||||
PgfPArgs* args;
|
||||
PgfSymbol curr_sym;
|
||||
uint16_t seq_idx;
|
||||
uint8_t tok_idx;
|
||||
uint8_t alt;
|
||||
uint8_t alt_idx; // position in the pre alternative
|
||||
uint8_t alt; // the number of the alternative
|
||||
prob_t inside_prob;
|
||||
};
|
||||
|
||||
@@ -694,7 +694,7 @@ pgf_new_item(PgfItemConts* conts, PgfProduction prod,
|
||||
item->prod = prod;
|
||||
item->curr_sym = gu_null_variant;
|
||||
item->seq_idx = 0;
|
||||
item->tok_idx = 0;
|
||||
item->alt_idx = 0;
|
||||
item->alt = 0;
|
||||
|
||||
conts->ref_count++;
|
||||
@@ -758,8 +758,12 @@ pgf_item_update_arg(PgfItem* item, size_t d, PgfCCat *new_ccat,
|
||||
static void
|
||||
pgf_item_advance(PgfItem* item, GuPool* pool)
|
||||
{
|
||||
item->seq_idx++;
|
||||
pgf_item_set_curr_symbol(item, pool);
|
||||
if (GU_LIKELY(item->alt == 0)) {
|
||||
item->seq_idx++;
|
||||
pgf_item_set_curr_symbol(item, pool);
|
||||
}
|
||||
else
|
||||
item->alt_idx++;
|
||||
}
|
||||
|
||||
static void
|
||||
@@ -1133,8 +1137,7 @@ pgf_parsing_meta_scan(PgfParseState* before, PgfParseState* after,
|
||||
gu_alignof(PgfSymbolKS),
|
||||
&item->curr_sym, after->ps->pool);
|
||||
*((PgfSymbol*)(sks+1)) = prev;
|
||||
sks->tokens = gu_new_seq(PgfToken, 1, after->ps->pool);
|
||||
gu_seq_set(sks->tokens, PgfToken, 0, tok);
|
||||
sks->token = tok;
|
||||
|
||||
gu_buf_heap_push(before->agenda, &pgf_item_prob_order, &item);
|
||||
}
|
||||
@@ -1218,76 +1221,54 @@ pgf_parsing_symbol(PgfParseState* before, PgfParseState* after,
|
||||
case PGF_SYMBOL_KS: {
|
||||
if (after != NULL) {
|
||||
PgfSymbolKS* sks = gu_variant_data(sym);
|
||||
gu_assert(item->tok_idx < gu_seq_length(sks->tokens));
|
||||
PgfToken tok =
|
||||
gu_seq_get(sks->tokens, PgfToken, item->tok_idx++);
|
||||
if (item->tok_idx == gu_seq_length(sks->tokens)) {
|
||||
item->tok_idx = 0;
|
||||
pgf_item_advance(item, after->ps->pool);
|
||||
}
|
||||
pgf_parsing_add_transition(before, after, tok, item);
|
||||
pgf_item_advance(item, after->ps->pool);
|
||||
pgf_parsing_add_transition(before, after, sks->token, item);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
if (after != NULL) {
|
||||
PgfSymbolKP* skp = gu_variant_data(sym);
|
||||
size_t idx = item->tok_idx;
|
||||
uint8_t alt = item->alt;
|
||||
gu_assert(idx < gu_seq_length(skp->default_form));
|
||||
if (idx == 0) {
|
||||
PgfToken tok;
|
||||
|
||||
PgfSymbol sym;
|
||||
if (item->alt == 0) {
|
||||
PgfItem* new_item;
|
||||
|
||||
tok = gu_seq_get(skp->default_form, PgfToken, 0);
|
||||
|
||||
new_item = pgf_item_copy(item, after->ps->pool, after->ps);
|
||||
new_item->tok_idx++;
|
||||
if (new_item->tok_idx == gu_seq_length(skp->default_form)) {
|
||||
new_item->tok_idx = 0;
|
||||
pgf_item_advance(new_item, after->ps->pool);
|
||||
}
|
||||
pgf_parsing_add_transition(before, after, tok, new_item);
|
||||
new_item->alt = 1;
|
||||
new_item->alt_idx = 0;
|
||||
sym = gu_seq_get(skp->default_form, PgfSymbol, new_item->alt_idx);
|
||||
pgf_parsing_symbol(before, after, new_item, sym);
|
||||
|
||||
for (size_t i = 0; i < skp->n_forms; i++) {
|
||||
// XXX: do nubbing properly
|
||||
PgfTokens* toks = skp->forms[i].form;
|
||||
PgfTokens* toks2 = skp->default_form;
|
||||
bool skip = pgf_tokens_equal(toks, toks2);
|
||||
PgfSequence* syms = skp->forms[i].form;
|
||||
PgfSequence* syms2 = skp->default_form;
|
||||
bool skip = false; /*pgf_tokens_equal(toks, toks2);
|
||||
for (size_t j = 0; j < i; j++) {
|
||||
PgfTokens* toks2 = skp->forms[j].form;
|
||||
skip |= pgf_tokens_equal(toks, toks2);
|
||||
}
|
||||
}*/
|
||||
if (!skip) {
|
||||
tok = gu_seq_get(toks, PgfToken, 0);
|
||||
new_item = pgf_item_copy(item, after->ps->pool, after->ps);
|
||||
new_item->tok_idx++;
|
||||
new_item->alt = i;
|
||||
if (new_item->tok_idx == gu_seq_length(toks)) {
|
||||
new_item->tok_idx = 0;
|
||||
pgf_item_advance(new_item, after->ps->pool);
|
||||
}
|
||||
pgf_parsing_add_transition(before, after, tok, new_item);
|
||||
new_item->alt = i+2;
|
||||
new_item->alt_idx = 0;
|
||||
sym = gu_seq_get(syms, PgfSymbol, new_item->alt_idx);
|
||||
pgf_parsing_symbol(before, after, new_item, sym);
|
||||
}
|
||||
}
|
||||
} else if (alt == 0) {
|
||||
PgfToken tok =
|
||||
gu_seq_get(skp->default_form, PgfToken, idx);
|
||||
item->tok_idx++;
|
||||
if (item->tok_idx == gu_seq_length(skp->default_form)) {
|
||||
item->tok_idx = 0;
|
||||
pgf_item_advance(item, after->ps->pool);
|
||||
}
|
||||
pgf_parsing_add_transition(before, after, tok, item);
|
||||
} else {
|
||||
gu_assert(alt <= skp->n_forms);
|
||||
PgfTokens* toks = skp->forms[alt - 1].form;
|
||||
PgfToken tok = gu_seq_get(toks, PgfToken, idx);
|
||||
item->tok_idx++;
|
||||
if (item->tok_idx == gu_seq_length(toks)) {
|
||||
item->tok_idx = 0;
|
||||
PgfSequence* syms =
|
||||
(item->alt == 1) ? skp->default_form :
|
||||
skp->forms[item->alt-2].form;
|
||||
|
||||
if (item->alt_idx < gu_seq_length(syms)) {
|
||||
sym = gu_seq_get(syms, PgfSymbol, item->alt_idx);
|
||||
pgf_parsing_symbol(before, after, item, sym);
|
||||
} else {
|
||||
item->alt = 0;
|
||||
pgf_item_advance(item, after->ps->pool);
|
||||
gu_buf_heap_push(before->agenda, &pgf_item_prob_order, &item);
|
||||
}
|
||||
pgf_parsing_add_transition(before, after, tok, item);
|
||||
}
|
||||
}
|
||||
break;
|
||||
@@ -1357,7 +1338,7 @@ pgf_parsing_symbol(PgfParseState* before, PgfParseState* after,
|
||||
// XXX TODO proper support
|
||||
break;
|
||||
case PGF_SYMBOL_NE: {
|
||||
// Nothing to be done here
|
||||
pgf_item_free(before, after, item);
|
||||
break;
|
||||
}
|
||||
default:
|
||||
@@ -1450,8 +1431,7 @@ pgf_parsing_item(PgfParseState* before, PgfParseState* after, PgfItem* item)
|
||||
gu_alignof(PgfSymbolKS),
|
||||
&item->curr_sym, after->ps->pool);
|
||||
*((PgfSymbol*)(sks+1)) = prev;
|
||||
sks->tokens = gu_new_seq(PgfToken, 1, after->ps->pool);
|
||||
gu_seq_set(sks->tokens, PgfToken, 0, tok);
|
||||
sks->token = tok;
|
||||
|
||||
item->seq_idx++;
|
||||
pgf_parsing_add_transition(before, after, tok, item);
|
||||
@@ -1755,9 +1735,7 @@ typedef struct {
|
||||
} PgfPrefixTokenState;
|
||||
|
||||
static GuString
|
||||
pgf_get_tokens(PgfSequence* seq,
|
||||
uint16_t seq_idx, uint8_t tok_idx,
|
||||
GuPool* pool)
|
||||
pgf_get_tokens(PgfSequence* seq, uint16_t seq_idx, GuPool* pool)
|
||||
{
|
||||
GuPool* tmp_pool = gu_new_pool();
|
||||
GuExn* err = gu_new_exn(NULL, gu_kind(type), tmp_pool);
|
||||
@@ -1773,17 +1751,7 @@ pgf_get_tokens(PgfSequence* seq,
|
||||
switch (i.tag) {
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* symks = i.data;
|
||||
size_t len = gu_seq_length(symks->tokens);
|
||||
for (size_t i = tok_idx; i < len; i++) {
|
||||
if (i > 0) {
|
||||
gu_putc(' ', out, err);
|
||||
}
|
||||
|
||||
PgfToken tok = gu_seq_get(symks->tokens, PgfToken, i);
|
||||
gu_string_write(tok, out, err);
|
||||
}
|
||||
|
||||
tok_idx = 0;
|
||||
gu_string_write(symks->token, out, err);
|
||||
}
|
||||
default:
|
||||
goto end;
|
||||
@@ -1809,18 +1777,9 @@ pgf_prefix_match_token(PgfTokenState* ts0, PgfToken tok, PgfItem* item)
|
||||
PgfSequence* seq;
|
||||
pgf_item_sequence(item, &lin_idx, &seq, ts->pool);
|
||||
|
||||
uint16_t seq_idx = item->seq_idx;
|
||||
uint8_t tok_idx = item->tok_idx;
|
||||
|
||||
// go one token back
|
||||
if (tok_idx > 0)
|
||||
tok_idx--;
|
||||
else
|
||||
seq_idx--;
|
||||
|
||||
ts->tp = gu_new(PgfTokenProb, ts->pool);
|
||||
ts->tp->tok =
|
||||
pgf_get_tokens(seq, seq_idx, tok_idx, ts->pool);
|
||||
pgf_get_tokens(seq, item->seq_idx-1, ts->pool);
|
||||
ts->tp->cat = item->conts->ccat->cnccat->abscat->name;
|
||||
ts->tp->prob = item->inside_prob+item->conts->outside_prob;
|
||||
}
|
||||
@@ -2346,17 +2305,15 @@ pgf_morpho_iter(GuMapItor* fn, const void* key, void* value, GuExn* err)
|
||||
switch (i.tag) {
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* symks = i.data;
|
||||
size_t len = gu_seq_length(symks->tokens);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
if (pos >= gu_seq_length(clo->tokens))
|
||||
goto cont;
|
||||
|
||||
if (pos >= gu_seq_length(clo->tokens))
|
||||
goto cont;
|
||||
|
||||
PgfToken tok1 = gu_seq_get(symks->tokens, PgfToken, i);
|
||||
PgfToken tok2 = gu_seq_get(clo->tokens, PgfToken, pos++);
|
||||
|
||||
if (!gu_string_eq(tok1, tok2))
|
||||
goto cont;
|
||||
}
|
||||
PgfToken tok1 = symks->token;
|
||||
PgfToken tok2 = gu_seq_get(clo->tokens, PgfToken, pos++);
|
||||
|
||||
if (!gu_string_eq(tok1, tok2))
|
||||
goto cont;
|
||||
}
|
||||
default:
|
||||
continue;
|
||||
@@ -2443,7 +2400,7 @@ pgf_fullform_iter(GuMapItor* fn, const void* key, void* value, GuExn* err)
|
||||
PgfProductionApply* papp = i.data;
|
||||
|
||||
PgfSequence* seq = papp->fun->lins[cfc.lin_idx];
|
||||
GuString tokens = pgf_get_tokens(seq, 0, 0, st->pool);
|
||||
GuString tokens = pgf_get_tokens(seq, 0, st->pool);
|
||||
|
||||
// create a new production index with keys that
|
||||
// are multiword units
|
||||
@@ -2531,12 +2488,10 @@ pgf_fullform_get_analyses(PgfFullFormEntry* entry,
|
||||
|
||||
static void
|
||||
pgf_parser_index_token(PgfConcr* concr,
|
||||
PgfTokens* tokens,
|
||||
PgfToken tok,
|
||||
PgfCCat* ccat, size_t lin_idx, PgfProduction prod,
|
||||
GuPool *pool)
|
||||
{
|
||||
PgfToken tok = gu_seq_get(tokens, PgfToken, 0);
|
||||
|
||||
PgfProductionIdx* set =
|
||||
gu_map_get(concr->leftcorner_tok_idx, &tok, PgfProductionIdx*);
|
||||
if (set == NULL) {
|
||||
@@ -2570,6 +2525,47 @@ pgf_parser_index_epsilon(PgfConcr* concr,
|
||||
gu_buf_push(prods, PgfProduction, prod);
|
||||
}
|
||||
|
||||
static void
|
||||
pgf_parser_index_symbol(PgfConcr* concr, PgfSymbol sym,
|
||||
PgfCCat* ccat, size_t lin_idx, PgfProduction prod,
|
||||
GuPool *pool)
|
||||
{
|
||||
GuVariantInfo i = gu_variant_open(sym);
|
||||
switch (i.tag) {
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* sks = i.data;
|
||||
pgf_parser_index_token(concr,
|
||||
sks->token,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
PgfSymbolKP* skp = i.data;
|
||||
PgfSymbol sym =
|
||||
gu_seq_get(skp->default_form, PgfSymbol, 0);
|
||||
pgf_parser_index_symbol(concr, sym,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
for (size_t i = 0; i < skp->n_forms; i++) {
|
||||
sym = gu_seq_get(skp->forms[i].form, PgfSymbol, 0);
|
||||
pgf_parser_index_symbol(concr, sym,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_CAT:
|
||||
case PGF_SYMBOL_LIT:
|
||||
case PGF_SYMBOL_NE:
|
||||
case PGF_SYMBOL_VAR:
|
||||
// Nothing to be done here
|
||||
break;
|
||||
default:
|
||||
gu_impossible();
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
pgf_parser_index(PgfConcr* concr,
|
||||
PgfCCat* ccat, PgfProduction prod,
|
||||
@@ -2586,39 +2582,9 @@ pgf_parser_index(PgfConcr* concr,
|
||||
|
||||
PgfSequence* seq = papp->fun->lins[lin_idx];
|
||||
if (gu_seq_length(seq) > 0) {
|
||||
GuVariantInfo i = gu_variant_open(gu_seq_get(seq, PgfSymbol, 0));
|
||||
switch (i.tag) {
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* sks = i.data;
|
||||
pgf_parser_index_token(concr,
|
||||
sks->tokens,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
PgfSymbolKP* skp = i.data;
|
||||
pgf_parser_index_token(concr,
|
||||
skp->default_form,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
for (size_t i = 0; i < skp->n_forms; i++) {
|
||||
pgf_parser_index_token(concr,
|
||||
skp->forms[i].form,
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
}
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_CAT:
|
||||
case PGF_SYMBOL_LIT:
|
||||
case PGF_SYMBOL_NE:
|
||||
case PGF_SYMBOL_VAR:
|
||||
// Nothing to be done here
|
||||
break;
|
||||
default:
|
||||
gu_impossible();
|
||||
}
|
||||
pgf_parser_index_symbol(concr, gu_seq_get(seq, PgfSymbol, 0),
|
||||
ccat, lin_idx, prod,
|
||||
pool);
|
||||
} else {
|
||||
pgf_parser_index_epsilon(concr,
|
||||
ccat, lin_idx, prod,
|
||||
|
||||
@@ -19,19 +19,14 @@ typedef struct {
|
||||
} PgfMetricsLznState;
|
||||
|
||||
static void
|
||||
pgf_metrics_lzn_symbol_tokens(PgfLinFuncs** funcs, PgfTokens* toks)
|
||||
pgf_metrics_lzn_symbol_token(PgfLinFuncs** funcs, PgfToken tok)
|
||||
{
|
||||
PgfMetricsLznState* state = gu_container(funcs, PgfMetricsLznState, funcs);
|
||||
|
||||
size_t len = gu_seq_length(toks);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
PgfToken tok = gu_seq_get(toks, PgfToken, i);
|
||||
|
||||
if (state->ps != NULL)
|
||||
state->ps = pgf_parser_next_state(state->ps, tok);
|
||||
|
||||
if (state->ps != NULL)
|
||||
state->ps = pgf_parser_next_state(state->ps, tok);
|
||||
|
||||
state->pos++;
|
||||
}
|
||||
state->pos++;
|
||||
}
|
||||
|
||||
static void
|
||||
@@ -128,17 +123,17 @@ pgf_metrics_lzn_end_phrase2(PgfLinFuncs** funcs, PgfCId cat, int fid, int lin_id
|
||||
}
|
||||
|
||||
static PgfLinFuncs pgf_metrics_lin_funcs1 = {
|
||||
.symbol_tokens = pgf_metrics_lzn_symbol_tokens,
|
||||
.expr_literal = pgf_metrics_lzn_expr_literal,
|
||||
.begin_phrase = pgf_metrics_lzn_begin_phrase,
|
||||
.end_phrase = pgf_metrics_lzn_end_phrase1
|
||||
.symbol_token = pgf_metrics_lzn_symbol_token,
|
||||
.expr_literal = pgf_metrics_lzn_expr_literal,
|
||||
.begin_phrase = pgf_metrics_lzn_begin_phrase,
|
||||
.end_phrase = pgf_metrics_lzn_end_phrase1
|
||||
};
|
||||
|
||||
static PgfLinFuncs pgf_metrics_lin_funcs2 = {
|
||||
.symbol_tokens = pgf_metrics_lzn_symbol_tokens,
|
||||
.expr_literal = pgf_metrics_lzn_expr_literal,
|
||||
.begin_phrase = pgf_metrics_lzn_begin_phrase,
|
||||
.end_phrase = pgf_metrics_lzn_end_phrase2
|
||||
.symbol_token = pgf_metrics_lzn_symbol_token,
|
||||
.expr_literal = pgf_metrics_lzn_expr_literal,
|
||||
.begin_phrase = pgf_metrics_lzn_begin_phrase,
|
||||
.end_phrase = pgf_metrics_lzn_end_phrase2
|
||||
};
|
||||
|
||||
bool
|
||||
|
||||
@@ -196,19 +196,16 @@ pgf_print_cncfun(PgfCncFun *cncfun, PgfSequences* sequences,
|
||||
}
|
||||
|
||||
static void
|
||||
pgf_print_tokens(PgfTokens* tokens, GuOut *out, GuExn *err)
|
||||
pgf_print_token(PgfToken tok, GuOut *out, GuExn *err)
|
||||
{
|
||||
gu_putc('"', out, err);
|
||||
size_t n_toks = gu_seq_length(tokens);
|
||||
for (size_t i = 0; i < n_toks; i++) {
|
||||
if (i > 0) gu_putc(' ', out, err);
|
||||
|
||||
PgfToken tok = gu_seq_get(tokens, PgfToken, i);
|
||||
gu_string_write(tok, out, err);
|
||||
}
|
||||
gu_string_write(tok, out, err);
|
||||
gu_putc('"', out, err);
|
||||
}
|
||||
|
||||
static void
|
||||
pgf_print_sequence(PgfSequence* seq, GuOut *out, GuExn *err);
|
||||
|
||||
void
|
||||
pgf_print_symbol(PgfSymbol sym, GuOut *out, GuExn *err)
|
||||
{
|
||||
@@ -220,18 +217,18 @@ pgf_print_symbol(PgfSymbol sym, GuOut *out, GuExn *err)
|
||||
}
|
||||
case PGF_SYMBOL_KS: {
|
||||
PgfSymbolKS* sks = gu_variant_data(sym);
|
||||
pgf_print_tokens(sks->tokens, out, err);
|
||||
pgf_print_token(sks->token, out, err);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
PgfSymbolKP* skp = gu_variant_data(sym);
|
||||
|
||||
gu_puts("pre {", out, err);
|
||||
pgf_print_tokens(skp->default_form, out, err);
|
||||
pgf_print_sequence(skp->default_form, out, err);
|
||||
|
||||
for (size_t i = 0; i < skp->n_forms; i++) {
|
||||
gu_puts("; ", out, err);
|
||||
pgf_print_tokens(skp->forms[i].form, out, err);
|
||||
pgf_print_sequence(skp->forms[i].form, out, err);
|
||||
gu_puts(" / ", out, err);
|
||||
|
||||
size_t n_prefixes = gu_seq_length(skp->forms[i].prefixes);
|
||||
@@ -262,16 +259,18 @@ pgf_print_symbol(PgfSymbol sym, GuOut *out, GuExn *err)
|
||||
gu_puts("nonExist", out, err);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_BIND: {
|
||||
gu_puts("BIND", out, err);
|
||||
break;
|
||||
}
|
||||
default:
|
||||
gu_impossible();
|
||||
}
|
||||
}
|
||||
|
||||
static void
|
||||
pgf_print_sequence(size_t seqid, PgfSequence* seq, GuOut *out, GuExn *err)
|
||||
pgf_print_sequence(PgfSequence* seq, GuOut *out, GuExn *err)
|
||||
{
|
||||
gu_printf(out,err," S%d := ", seqid);
|
||||
|
||||
int n_syms = gu_seq_length(seq);
|
||||
for (int i = 0; i < n_syms; i++) {
|
||||
if (i > 0) gu_putc(' ', out, err);
|
||||
@@ -279,8 +278,6 @@ pgf_print_sequence(size_t seqid, PgfSequence* seq, GuOut *out, GuExn *err)
|
||||
PgfSymbol sym = gu_seq_get(seq, PgfSymbol, i);
|
||||
pgf_print_symbol(sym, out, err);
|
||||
}
|
||||
|
||||
gu_putc('\n', out, err);
|
||||
}
|
||||
|
||||
static void
|
||||
@@ -342,7 +339,10 @@ pgf_print_concrete(PgfCId cncname, PgfConcr* concr,
|
||||
size_t n_seqs = gu_seq_length(concr->sequences);
|
||||
for (size_t i = 0; i < n_seqs; i++) {
|
||||
PgfSequence* seq = gu_seq_get(concr->sequences, PgfSequence*, i);
|
||||
pgf_print_sequence(i, seq, out, err);
|
||||
|
||||
gu_printf(out,err," S%d := ", i);
|
||||
pgf_print_sequence(seq, out, err);
|
||||
gu_putc('\n', out, err);
|
||||
}
|
||||
|
||||
gu_puts(" categories\n", out, err);
|
||||
|
||||
@@ -586,27 +586,13 @@ pgf_read_printnames(PgfReader* rdr)
|
||||
return printnames;
|
||||
}
|
||||
|
||||
static PgfTokens*
|
||||
pgf_read_tokens(PgfReader* rdr)
|
||||
{
|
||||
size_t len = pgf_read_len(rdr);
|
||||
gu_return_on_exn(rdr->err, NULL);
|
||||
|
||||
PgfTokens* tokens = gu_new_seq(PgfToken, len, rdr->opool);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
PgfToken token = pgf_read_string(rdr);
|
||||
gu_return_on_exn(rdr->err, NULL);
|
||||
|
||||
gu_seq_set(tokens, PgfToken, i, token);
|
||||
}
|
||||
|
||||
return tokens;
|
||||
}
|
||||
static PgfSequence*
|
||||
pgf_read_sequence(PgfReader* rdr);
|
||||
|
||||
static void
|
||||
pgf_read_alternative(PgfReader* rdr, PgfAlternative* alt)
|
||||
{
|
||||
alt->form = pgf_read_tokens(rdr);
|
||||
alt->form = pgf_read_sequence(rdr);
|
||||
gu_return_on_exn(rdr->err,);
|
||||
|
||||
size_t n_prefixes = pgf_read_len(rdr);
|
||||
@@ -672,12 +658,12 @@ pgf_read_symbol(PgfReader* rdr)
|
||||
gu_new_variant(PGF_SYMBOL_KS,
|
||||
PgfSymbolKS,
|
||||
&sym, rdr->opool);
|
||||
sym_ks->tokens = pgf_read_tokens(rdr);
|
||||
sym_ks->token = pgf_read_string(rdr);
|
||||
gu_return_on_exn(rdr->err, gu_null_variant);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_KP: {
|
||||
PgfTokens* default_form = pgf_read_tokens(rdr);
|
||||
PgfSequence* default_form = pgf_read_sequence(rdr);
|
||||
gu_return_on_exn(rdr->err, gu_null_variant);
|
||||
|
||||
size_t n_forms = pgf_read_len(rdr);
|
||||
@@ -703,6 +689,13 @@ pgf_read_symbol(PgfReader* rdr)
|
||||
gu_return_on_exn(rdr->err, gu_null_variant);
|
||||
break;
|
||||
}
|
||||
case PGF_SYMBOL_BIND: {
|
||||
gu_new_variant(PGF_SYMBOL_BIND,
|
||||
PgfSymbolBIND,
|
||||
&sym, rdr->opool);
|
||||
gu_return_on_exn(rdr->err, gu_null_variant);
|
||||
break;
|
||||
}
|
||||
default:
|
||||
pgf_read_tag_error(rdr);
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user