summaryrefslogtreecommitdiff
path: root/src/runtime
diff options
context:
space:
mode:
authorkr.angelov <kr.angelov@gmail.com>2013-11-12 09:54:57 +0000
committerkr.angelov <kr.angelov@gmail.com>2013-11-12 09:54:57 +0000
commit0095119ec09b394332a23ab7cc16c0009c3f162a (patch)
treec2c19ef36c3c04ae608ae93a9131ee1e71957c98 /src/runtime
parentb111b476a75577a5b21d1d498975cbb2eeed9324 (diff)
added Predef.SOFT_BIND. This special token allows zero or more spaces between ordinary tokens. It is also used in the English RGL to attach the commas to the previous word.
Diffstat (limited to 'src/runtime')
-rw-r--r--src/runtime/c/pgf/data.h1
-rw-r--r--src/runtime/c/pgf/linearizer.c3
-rw-r--r--src/runtime/c/pgf/parser.c30
-rw-r--r--src/runtime/c/pgf/printer.c4
-rw-r--r--src/runtime/c/pgf/reader.c7
-rw-r--r--src/runtime/haskell/PGF/Binary.hs6
-rw-r--r--src/runtime/haskell/PGF/Data.hs1
-rw-r--r--src/runtime/haskell/PGF/Macros.hs2
-rw-r--r--src/runtime/haskell/PGF/Optimize.hs1
-rw-r--r--src/runtime/haskell/PGF/Parse.hs8
-rw-r--r--src/runtime/haskell/PGF/Printer.hs1
11 files changed, 58 insertions, 6 deletions
diff --git a/src/runtime/c/pgf/data.h b/src/runtime/c/pgf/data.h
index 9bc73dd0a..131201132 100644
--- a/src/runtime/c/pgf/data.h
+++ b/src/runtime/c/pgf/data.h
@@ -167,6 +167,7 @@ typedef enum {
PGF_SYMBOL_KS,
PGF_SYMBOL_KP,
PGF_SYMBOL_BIND,
+ PGF_SYMBOL_SOFT_BIND,
PGF_SYMBOL_NE
} PgfSymbolTag;
diff --git a/src/runtime/c/pgf/linearizer.c b/src/runtime/c/pgf/linearizer.c
index 698bab939..732bce75a 100644
--- a/src/runtime/c/pgf/linearizer.c
+++ b/src/runtime/c/pgf/linearizer.c
@@ -723,7 +723,8 @@ pgf_lzr_linearize_symbols(PgfConcr* concr, PgfCncTreeApp* fapp,
}
break;
}
- case PGF_SYMBOL_BIND: {
+ case PGF_SYMBOL_BIND:
+ case PGF_SYMBOL_SOFT_BIND: {
if ((*fnsp)->symbol_bind) {
(*fnsp)->symbol_bind(fnsp);
}
diff --git a/src/runtime/c/pgf/parser.c b/src/runtime/c/pgf/parser.c
index 46cfe5614..66d77175d 100644
--- a/src/runtime/c/pgf/parser.c
+++ b/src/runtime/c/pgf/parser.c
@@ -145,6 +145,7 @@ pgf_prev_extern_sym(PgfSymbol sym)
case PGF_SYMBOL_VAR:
return *((PgfSymbol*) (((PgfSymbolVar*) i.data)+1));
case PGF_SYMBOL_BIND:
+ case PGF_SYMBOL_SOFT_BIND:
return *((PgfSymbol*) (((PgfSymbolBIND*) i.data)+1));
case PGF_SYMBOL_NE:
return *((PgfSymbol*) (((PgfSymbolNE*) i.data)+1));
@@ -1137,6 +1138,10 @@ pgf_symbols_cmp(GuString* psent, size_t sent_len, BIND_TYPE* pbind, PgfSymbols*
*pbind = BIND_HARD;
break;
}
+ case PGF_SYMBOL_SOFT_BIND: {
+ *pbind = BIND_SOFT;
+ break;
+ }
case PGF_SYMBOL_NE: {
return -2;
}
@@ -1635,6 +1640,31 @@ pgf_parsing_symbol(PgfParsing* ps, PgfItem* item, PgfSymbol sym)
}
break;
}
+ case PGF_SYMBOL_SOFT_BIND: {
+ if (ps->before->start_offset == ps->before->end_offset) {
+ if (ps->before->needs_bind) {
+ PgfParseState* state =
+ pgf_new_parse_state(ps, ps->before->end_offset, BIND_HARD);
+ if (state != NULL) {
+ if (state->next == NULL) {
+ state->viterbi_prob =
+ item->inside_prob+item->conts->outside_prob;
+ }
+
+ pgf_item_advance(item, ps->pool);
+ gu_buf_heap_push(state->agenda, pgf_item_prob_order, &item);
+ } else {
+ pgf_item_free(ps, item);
+ }
+ } else {
+ pgf_item_free(ps, item);
+ }
+ } else {
+ pgf_item_advance(item, ps->pool);
+ gu_buf_heap_push(ps->before->agenda, pgf_item_prob_order, &item);
+ }
+ break;
+ }
default:
gu_impossible();
}
diff --git a/src/runtime/c/pgf/printer.c b/src/runtime/c/pgf/printer.c
index 78c2b74db..8b737266e 100644
--- a/src/runtime/c/pgf/printer.c
+++ b/src/runtime/c/pgf/printer.c
@@ -280,6 +280,10 @@ pgf_print_symbol(PgfSymbol sym, GuOut *out, GuExn *err)
gu_puts("BIND", out, err);
break;
}
+ case PGF_SYMBOL_SOFT_BIND: {
+ gu_puts("SOFT_BIND", out, err);
+ break;
+ }
default:
gu_impossible();
}
diff --git a/src/runtime/c/pgf/reader.c b/src/runtime/c/pgf/reader.c
index 477adc069..c40781e26 100644
--- a/src/runtime/c/pgf/reader.c
+++ b/src/runtime/c/pgf/reader.c
@@ -723,6 +723,13 @@ pgf_read_symbol(PgfReader* rdr)
gu_return_on_exn(rdr->err, gu_null_variant);
break;
}
+ case PGF_SYMBOL_SOFT_BIND: {
+ gu_new_variant(PGF_SYMBOL_SOFT_BIND,
+ PgfSymbolBIND,
+ &sym, rdr->opool);
+ gu_return_on_exn(rdr->err, gu_null_variant);
+ break;
+ }
default:
pgf_read_tag_error(rdr);
}
diff --git a/src/runtime/haskell/PGF/Binary.hs b/src/runtime/haskell/PGF/Binary.hs
index 2debcf12d..bf30e4506 100644
--- a/src/runtime/haskell/PGF/Binary.hs
+++ b/src/runtime/haskell/PGF/Binary.hs
@@ -152,7 +152,8 @@ instance Binary Symbol where
put (SymKS ts) = putWord8 3 >> put ts
put (SymKP d vs) = putWord8 4 >> put (d,vs)
put SymBIND = putWord8 5
- put SymNE = putWord8 6
+ put SymSOFT_BIND = putWord8 6
+ put SymNE = putWord8 7
get = do tag <- getWord8
case tag of
0 -> liftM2 SymCat get get
@@ -161,7 +162,8 @@ instance Binary Symbol where
3 -> liftM SymKS get
4 -> liftM2 (\d vs -> SymKP d vs) get get
5 -> return SymBIND
- 6 -> return SymNE
+ 6 -> return SymSOFT_BIND
+ 7 -> return SymNE
_ -> decodingError
instance Binary PArg where
diff --git a/src/runtime/haskell/PGF/Data.hs b/src/runtime/haskell/PGF/Data.hs
index f5797739f..3222867d2 100644
--- a/src/runtime/haskell/PGF/Data.hs
+++ b/src/runtime/haskell/PGF/Data.hs
@@ -62,6 +62,7 @@ data Symbol
| SymKS Token
| SymKP [Symbol] [([Symbol],[String])]
| SymBIND -- the special BIND token
+ | SymSOFT_BIND -- the special SOFT_BIND token
| SymNE -- non exist (this should be last constructor to simplify the binary search in the runtime)
deriving (Eq,Ord,Show)
data Production
diff --git a/src/runtime/haskell/PGF/Macros.hs b/src/runtime/haskell/PGF/Macros.hs
index ce75b1c91..0e73180d5 100644
--- a/src/runtime/haskell/PGF/Macros.hs
+++ b/src/runtime/haskell/PGF/Macros.hs
@@ -160,6 +160,7 @@ data BracketedTokn
| LeafKS Token
| LeafNE
| LeafBIND
+ | LeafSOFT_BIND
| LeafKP [BracketedTokn] [([BracketedTokn],[String])]
deriving Eq
@@ -222,6 +223,7 @@ computeSeq filter seq args = concatMap compute seq
compute (SymKS t) = [LeafKS t]
compute SymNE = [LeafNE]
compute SymBIND = [LeafKS "&+"]
+ compute SymSOFT_BIND = []
compute (SymKP syms alts) = [LeafKP (concatMap compute syms) [(concatMap compute syms,cs) | (syms,cs) <- alts]]
getArg d r
diff --git a/src/runtime/haskell/PGF/Optimize.hs b/src/runtime/haskell/PGF/Optimize.hs
index bfc12e097..ab831704f 100644
--- a/src/runtime/haskell/PGF/Optimize.hs
+++ b/src/runtime/haskell/PGF/Optimize.hs
@@ -228,6 +228,7 @@ splitLexicalRules cnc p_prods =
[seq2prefix (syms1 ++ syms) | (syms1,ps) <- alts])
seq2prefix (SymNE :syms) = TrieMap.empty
seq2prefix (SymBIND :syms) = TrieMap.fromList [wf ["&+"]]
+ seq2prefix (SymSOFT_BIND :syms) = TrieMap.fromList [wf []]
updateConcrete abs cnc =
let p_prods0 = filterProductions IntMap.empty IntSet.empty (productions cnc)
diff --git a/src/runtime/haskell/PGF/Parse.hs b/src/runtime/haskell/PGF/Parse.hs
index 9c69940ef..3f5dc313a 100644
--- a/src/runtime/haskell/PGF/Parse.hs
+++ b/src/runtime/haskell/PGF/Parse.hs
@@ -309,10 +309,12 @@ process flit ftok cnc (item@(Active j ppos funid seqid args key0):items) acc cha
SymNE -> process flit ftok cnc items acc chart
SymBIND -> let !acc' = ftok_ ["&+"] (Active j (ppos+1) funid seqid args key0) acc
in process flit ftok cnc items acc' chart
+ SymSOFT_BIND->process flit ftok cnc ((Active j (ppos+1) funid seqid args key0):items) acc chart
SymKP syms vars
- -> let to_tok (SymKS t) = [t]
- to_tok SymBIND = ["&+"]
- to_tok _ = []
+ -> let to_tok (SymKS t) = [t]
+ to_tok SymBIND = ["&+"]
+ to_tok SymSOFT_BIND = []
+ to_tok _ = []
!acc' = foldl (\acc syms -> ftok_ (concatMap to_tok syms) (Active j (ppos+1) funid seqid args key0) acc) acc
(syms:[syms' | (syms',_) <- vars])
diff --git a/src/runtime/haskell/PGF/Printer.hs b/src/runtime/haskell/PGF/Printer.hs
index d3a5ea1d9..4945667f4 100644
--- a/src/runtime/haskell/PGF/Printer.hs
+++ b/src/runtime/haskell/PGF/Printer.hs
@@ -92,6 +92,7 @@ ppSymbol (SymVar d r) = char '<' <> int d <> comma <> char '$' <> int r <> char
ppSymbol (SymKS t) = doubleQuotes (text t)
ppSymbol SymNE = text "nonExist"
ppSymbol SymBIND = text "BIND"
+ppSymbol SymSOFT_BIND = text "SOFT_BIND"
ppSymbol (SymKP syms alts) = text "pre" <+> braces (hsep (punctuate semi (hsep (map ppSymbol syms) : map ppAlt alts)))
ppAlt (syms,ps) = hsep (map ppSymbol syms) <+> char '/' <+> hsep (map (doubleQuotes . text) ps)