get rid of the destructive updates for seq_ids

This commit is contained in:
Krasimir Angelov
2022-01-10 10:27:09 +01:00
parent 19f7fb8d5e
commit d87b3ce166
12 changed files with 191 additions and 62 deletions

View File

@@ -130,7 +130,6 @@ struct PGF_INTERNAL_DECL PgfPResult {
typedef object PgfSymbol;
struct PGF_INTERNAL_DECL PgfSequence {
size_t seq_id;
size_t ref_count;
Vector<PgfSymbol> syms;

View File

@@ -827,16 +827,22 @@ void pgf_iter_lins(PgfDB *db, PgfConcrRevision cnc_revision,
}
PGF_API
void pgf_iter_sequences(PgfDB *db, PgfConcrRevision cnc_revision,
PgfSequenceItor *itor, PgfExn *err)
PgfPhrasetableIds *pgf_iter_sequences(PgfDB *db, PgfConcrRevision cnc_revision,
PgfSequenceItor *itor, PgfExn *err)
{
PGF_API_BEGIN {
DB_scope scope(db, READER_SCOPE);
ref<PgfConcr> concr = PgfDB::revision2concr(cnc_revision);
size_t next_id = 0;
phrasetable_iter(concr->phrasetable, itor, &next_id, err);
PgfPhrasetableIds *seq_ids = new PgfPhrasetableIds();
seq_ids->start(concr);
phrasetable_iter(concr->phrasetable, itor, seq_ids, err);
return seq_ids;
} PGF_API_END
return NULL;
}
PGF_API
@@ -863,7 +869,7 @@ size_t pgf_get_lin_get_prod_count(object o)
}
PGF_API
PgfText *pgf_print_lindef_internal(object o, size_t i)
PgfText *pgf_print_lindef_internal(PgfPhrasetableIds *seq_ids, object o, size_t i)
{
ref<PgfConcrLincat> lincat = o;
@@ -893,7 +899,7 @@ PgfText *pgf_print_lindef_internal(object o, size_t i)
printer.puts(",");
ref<PgfSequence> seq = *vector_elem(lincat->seqs, i*n_seqs + j);
printer.seq_id(seq->seq_id);
printer.seq_id(seq_ids, seq);
}
printer.puts("]");
@@ -902,7 +908,7 @@ PgfText *pgf_print_lindef_internal(object o, size_t i)
}
PGF_API
PgfText *pgf_print_linref_internal(object o, size_t i)
PgfText *pgf_print_linref_internal(PgfPhrasetableIds *seq_ids, object o, size_t i)
{
ref<PgfConcrLincat> lincat = o;
@@ -926,7 +932,7 @@ PgfText *pgf_print_linref_internal(object o, size_t i)
size_t n_seqs = lincat->fields->len;
ref<PgfSequence> seq = *vector_elem(lincat->seqs, lincat->n_lindefs*n_seqs+i);
printer.seq_id(seq->seq_id);
printer.seq_id(seq_ids, seq);
printer.puts("]");
@@ -934,7 +940,7 @@ PgfText *pgf_print_linref_internal(object o, size_t i)
}
PGF_API
PgfText *pgf_print_lin_internal(object o, size_t i)
PgfText *pgf_print_lin_internal(PgfPhrasetableIds *seq_ids, object o, size_t i)
{
ref<PgfConcrLin> lin = o;
ref<PgfDTyp> ty = lin->absfun->type;
@@ -975,7 +981,7 @@ PgfText *pgf_print_lin_internal(object o, size_t i)
printer.puts(",");
ref<PgfSequence> seq = *vector_elem(lin->seqs, i*n_seqs + j);
printer.seq_id(seq->seq_id);
printer.seq_id(seq_ids, seq);
}
printer.puts("]");
@@ -984,20 +990,25 @@ PgfText *pgf_print_lin_internal(object o, size_t i)
}
PGF_API
PgfText *pgf_print_sequence_internal(object o)
PgfText *pgf_print_sequence_internal(size_t seq_id, object o)
{
ref<PgfSequence> seq = o;
PgfInternalMarshaller m;
PgfPrinter printer(NULL,0,&m);
printer.seq_id(seq->seq_id);
printer.puts(" = ");
printer.nprintf(10,"S%zu = ", seq_id);
printer.sequence(seq);
return printer.get_text();
}
PGF_API_DECL
void pgf_release_phrasetable_ids(PgfPhrasetableIds *seq_ids)
{
delete seq_ids;
}
PGF_API
void pgf_check_expr(PgfDB *db, PgfRevision revision,
PgfExpr* pe, PgfType ty,
@@ -1578,7 +1589,6 @@ public:
throw pgf_error(builder_error_msg);
seq = PgfDB::malloc<PgfSequence>(n_syms*sizeof(PgfSymbol));
seq->seq_id = 0;
seq->ref_count = 1;
seq->syms.len = n_syms;
@@ -1680,7 +1690,6 @@ public:
throw pgf_error(builder_error_msg);
ref<PgfSequence> def = PgfDB::malloc<PgfSequence>(n_syms*sizeof(PgfSymbol));
def->seq_id = 0;
def->ref_count = 1;
def->syms.len = n_syms;
@@ -1707,7 +1716,6 @@ public:
throw pgf_error(builder_error_msg);
ref<PgfSequence> form = PgfDB::malloc<PgfSequence>(n_syms*sizeof(PgfSymbol));
form->seq_id = 0;
form->ref_count = 1;
form->syms.len = n_syms;

View File

@@ -396,14 +396,16 @@ PGF_API_DECL
void pgf_iter_lins(PgfDB *db, PgfConcrRevision cnc_revision,
PgfItor *itor, PgfExn *err);
typedef struct PgfPhrasetableIds PgfPhrasetableIds;
typedef struct PgfSequenceItor PgfSequenceItor;
struct PgfSequenceItor {
void (*fn)(PgfSequenceItor* self, object value, PgfExn *err);
void (*fn)(PgfSequenceItor* self, size_t seq_id, object value, PgfExn *err);
};
PGF_API
void pgf_iter_sequences(PgfDB *db, PgfConcrRevision cnc_revision,
PgfSequenceItor *itor, PgfExn *err);
PgfPhrasetableIds *pgf_iter_sequences(PgfDB *db, PgfConcrRevision cnc_revision,
PgfSequenceItor *itor, PgfExn *err);
PGF_API_DECL
void pgf_get_lincat_counts_internal(object o, size_t *counts);
@@ -415,16 +417,19 @@ PGF_API_DECL
size_t pgf_get_lin_get_prod_count(object o);
PGF_API_DECL
PgfText *pgf_print_lindef_internal(object o, size_t i);
PgfText *pgf_print_lindef_internal(PgfPhrasetableIds *seq_ids, object o, size_t i);
PGF_API_DECL
PgfText *pgf_print_linref_internal(object o, size_t i);
PgfText *pgf_print_linref_internal(PgfPhrasetableIds *seq_ids, object o, size_t i);
PGF_API_DECL
PgfText *pgf_print_lin_internal(object o, size_t i);
PgfText *pgf_print_lin_internal(PgfPhrasetableIds *seq_ids, object o, size_t i);
PGF_API_DECL
PgfText *pgf_print_sequence_internal(object o);
PgfText *pgf_print_sequence_internal(size_t seq_id, object o);
PGF_API_DECL
void pgf_release_phrasetable_ids(PgfPhrasetableIds *seq_ids);
PGF_API_DECL
void pgf_check_expr(PgfDB *db, PgfRevision revision,

View File

@@ -1,5 +1,77 @@
#include "data.h"
PgfPhrasetableIds::PgfPhrasetableIds()
{
next_id = 0;
n_pairs = 0;
pairs = NULL;
chains = NULL;
}
void PgfPhrasetableIds::start(ref<PgfConcr> concr)
{
next_id = 0;
n_pairs = namespace_size(concr->phrasetable);
size_t mem_size = sizeof(SeqIdPair)*n_pairs;
pairs = (SeqIdPair*) malloc(mem_size);
if (pairs == NULL)
throw pgf_systemerror(ENOMEM);
memset(pairs, 0, mem_size);
}
size_t PgfPhrasetableIds::add(ref<PgfSequence> seq)
{
size_t index = (seq.as_object() >> 4) % n_pairs;
if (pairs[index].seq == 0) {
pairs[index].seq = seq;
pairs[index].seq_id = next_id++;
return pairs[index].seq_id;
} else {
SeqIdChain *chain =
(SeqIdChain*) malloc(sizeof(SeqIdChain));
if (chain == NULL)
throw pgf_systemerror(ENOMEM);
chain->next = chains;
chain->chain = pairs[index].chain;
chain->seq = seq;
chain->seq_id = next_id++;
pairs[index].chain = chain;
chains = chain;
return chain->seq_id;
}
}
size_t PgfPhrasetableIds::get(ref<PgfSequence> seq)
{
size_t index = (seq.as_object() >> 4) % n_pairs;
if (pairs[index].seq == seq) {
return pairs[index].seq_id;
} else {
SeqIdChain *chain = pairs[index].chain;
while (chain != NULL) {
if (chain->seq == seq)
return chain->seq_id;
chain = chain->chain;
}
throw pgf_error("Can't find sequence id");
}
}
void PgfPhrasetableIds::end()
{
next_id = 0;
n_pairs = 0;
while (chains != NULL) {
SeqIdChain *next = chains->next;
free(chains);
chains = next;
}
free(pairs);
pairs = NULL;
}
static
int lparam_cmp(PgfLParam *p1, PgfLParam *p2)
{
@@ -213,21 +285,22 @@ ref<PgfSequence> phrasetable_get(PgfPhrasetable table, size_t seq_id)
}
PGF_INTERNAL
void phrasetable_iter(PgfPhrasetable table, PgfSequenceItor* itor, size_t *p_next_id, PgfExn *err)
void phrasetable_iter(PgfPhrasetable table, PgfSequenceItor* itor,
PgfPhrasetableIds *seq_ids, PgfExn *err)
{
if (table == 0)
return;
phrasetable_iter(table->left, itor, p_next_id, err);
phrasetable_iter(table->left, itor, seq_ids, err);
if (err->type != PGF_EXN_NONE)
return;
table->value->seq_id = (*p_next_id)++;
itor->fn(itor, table->value.as_object(), err);
size_t seq_id = seq_ids->add(table->value);
itor->fn(itor, seq_id, table->value.as_object(), err);
if (err->type != PGF_EXN_NONE)
return;
phrasetable_iter(table->right, itor, p_next_id, err);
phrasetable_iter(table->right, itor, seq_ids, err);
if (err->type != PGF_EXN_NONE)
return;
}

View File

@@ -5,6 +5,41 @@ class PgfSequence;
class PgfSequenceItor;
typedef ref<Node<PgfSequence>> PgfPhrasetable;
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wattributes"
struct PgfPhrasetableIds {
public:
PGF_INTERNAL_DECL PgfPhrasetableIds();
PGF_INTERNAL_DECL ~PgfPhrasetableIds() { end(); }
PGF_INTERNAL_DECL void start(ref<PgfConcr> concr);
PGF_INTERNAL_DECL size_t add(ref<PgfSequence> seq);
PGF_INTERNAL_DECL size_t get(ref<PgfSequence> seq);
PGF_INTERNAL_DECL void end();
private:
size_t next_id;
struct PGF_INTERNAL_DECL SeqIdChain;
struct PGF_INTERNAL_DECL SeqIdPair {
SeqIdChain *chain;
ref<PgfSequence> seq;
size_t seq_id;
};
struct PGF_INTERNAL_DECL SeqIdChain : public SeqIdPair {
SeqIdChain *next;
};
size_t n_pairs;
SeqIdPair *pairs;
SeqIdChain *chains;
};
#pragma GCC diagnostic pop
PGF_INTERNAL_DECL
PgfPhrasetable phrasetable_internalize(PgfPhrasetable table, ref<PgfSequence> *seq);
@@ -12,7 +47,8 @@ PGF_INTERNAL_DECL
ref<PgfSequence> phrasetable_get(PgfPhrasetable table, size_t seq_id);
PGF_INTERNAL_DECL
void phrasetable_iter(PgfPhrasetable table, PgfSequenceItor* itor, size_t *p_next_id, PgfExn *err);
void phrasetable_iter(PgfPhrasetable table, PgfSequenceItor* itor,
PgfPhrasetableIds *seq_ids, PgfExn *err);
PGF_INTERNAL_DECL
void phrasetable_release(PgfPhrasetable table);

View File

@@ -575,9 +575,9 @@ void PgfPrinter::sequence(ref<PgfSequence> seq)
}
}
void PgfPrinter::seq_id(size_t seq_id)
void PgfPrinter::seq_id(PgfPhrasetableIds *seq_ids, ref<PgfSequence> seq)
{
nprintf(5, "S%zu", seq_id);
nprintf(5, "S%zu", seq_ids->get(seq));
}
void PgfPrinter::free_ref(object x)

View File

@@ -56,7 +56,7 @@ public:
void lvar(size_t var);
void lparam(ref<PgfLParam> lparam);
void lvar_ranges(ref<Vector<PgfVariableRange>> vars);
void seq_id(size_t seqid);
void seq_id(PgfPhrasetableIds *seq_ids, ref<PgfSequence> seq);
void symbol(PgfSymbol sym);
void sequence(ref<PgfSequence> seq);

View File

@@ -565,7 +565,6 @@ ref<PgfSequence> PgfReader::read_seq()
size_t n_syms = read_len();
ref<PgfSequence> seq = PgfDB::malloc<PgfSequence>(n_syms*sizeof(PgfSymbol));
seq->seq_id = 0;
seq->ref_count = 1;
seq->syms.len = n_syms;

View File

@@ -414,7 +414,7 @@ void PgfWriter::write_symbol(PgfSymbol sym)
void PgfWriter::write_seq(ref<PgfSequence> seq)
{
seq->seq_id = next_seq_id++;
seq_ids.add(seq);
write_vector(ref<Vector<PgfSymbol>>::from_ptr(&seq->syms), &PgfWriter::write_symbol);
}
@@ -444,7 +444,7 @@ void PgfWriter::write_printname(ref<PgfConcrPrintname> printname)
void PgfWriter::write_concrete(ref<PgfConcr> concr)
{
next_seq_id = 0;
seq_ids.start(concr);
write_name(&concr->name);
write_namespace<PgfFlag>(concr->cflags, &PgfWriter::write_flag);
@@ -452,6 +452,8 @@ void PgfWriter::write_concrete(ref<PgfConcr> concr)
write_namespace<PgfConcrLincat>(concr->lincats, &PgfWriter::write_lincat);
write_namespace<PgfConcrLin>(concr->lins, &PgfWriter::write_lin);
write_namespace<PgfConcrPrintname>(concr->printnames, &PgfWriter::write_printname);
seq_ids.end();
}
void PgfWriter::write_pgf(ref<PgfPGF> pgf)

View File

@@ -45,7 +45,7 @@ public:
void write_presult(ref<PgfPResult> linres);
void write_symbol(PgfSymbol sym);
void write_seq(ref<PgfSequence> seq);
void write_seq_id(ref<ref<PgfSequence>> r) { write_len((*r)->seq_id); };
void write_seq_id(ref<ref<PgfSequence>> r) { write_len(seq_ids.get(*r)); };
void write_lin(ref<PgfConcrLin> lin);
void write_printname(ref<PgfConcrPrintname> printname);
@@ -62,10 +62,10 @@ private:
void write_symbol(ref<PgfSymbol> r) { write_symbol(*r); };
void write_presult(ref<ref<PgfPResult>> r) { write_presult(*r); };
size_t next_seq_id;
FILE *out;
ref<PgfAbstr> abstract;
PgfPhrasetableIds seq_ids;
};
#endif

View File

@@ -223,16 +223,19 @@ showPGF p =
modifyIORef ref (\doc -> doc $$ text def)
ppConcr name c = unsafePerformIO $ do
doc <- prepareSequences c -- run first to update all seq_id
(seq_ids,doc3) <- prepareSequences c -- run first to update all seq_id
doc1 <- ppLincats seq_ids c
doc2 <- ppLins seq_ids c
pgf_release_phrasetable_ids seq_ids
return (text "concrete" <+> text name <+> char '{' $$
nest 2 (ppLincats c $$
ppLins c $$
nest 2 (doc1 $$
doc2 $$
(text "sequences" <+> char '{' $$
nest 2 doc $$
nest 2 doc3 $$
char '}')) $$
char '}')
ppLincats c = unsafePerformIO $ do
ppLincats seq_ids c = do
ref <- newIORef empty
(allocaBytes (#size PgfItor) $ \itor ->
bracket (wrapItorCallback (getLincats ref)) freeHaskellFunPtr $ \fptr ->
@@ -259,15 +262,15 @@ showPGF p =
char ']')
modifyIORef ref $ (\doc -> doc $$ def)
forM_ (init [0..n_lindefs]) $ \i -> do
def <- bracket (pgf_print_lindef_internal val i) free $ \c_text -> do
def <- bracket (pgf_print_lindef_internal seq_ids val i) free $ \c_text -> do
fmap text (peekText c_text)
modifyIORef ref (\doc -> doc $$ text "lindef" <+> def)
forM_ (init [0..n_linrefs]) $ \i -> do
def <- bracket (pgf_print_linref_internal val i) free $ \c_text -> do
def <- bracket (pgf_print_linref_internal seq_ids val i) free $ \c_text -> do
fmap text (peekText c_text)
modifyIORef ref $ (\doc -> doc $$ text "linref" <+> def)
ppLins c = unsafePerformIO $ do
ppLins seq_ids c = do
ref <- newIORef empty
(allocaBytes (#size PgfItor) $ \itor ->
bracket (wrapItorCallback (getLins ref)) freeHaskellFunPtr $ \fptr ->
@@ -280,23 +283,24 @@ showPGF p =
getLins ref itor key val exn = do
n_prods <- pgf_get_lin_get_prod_count val
forM_ (init [0..n_prods]) $ \i -> do
def <- bracket (pgf_print_lin_internal val i) free $ \c_text -> do
def <- bracket (pgf_print_lin_internal seq_ids val i) free $ \c_text -> do
fmap text (peekText c_text)
modifyIORef ref (\doc -> doc $$ text "lin" <+> def)
return ()
prepareSequences c = do
ref <- newIORef empty
(allocaBytes (#size PgfSequenceItor) $ \itor ->
bracket (wrapSequenceItorCallback (getSequences ref)) freeHaskellFunPtr $ \fptr ->
withForeignPtr (c_revision c) $ \c_revision -> do
(#poke PgfSequenceItor, fn) itor fptr
withPgfExn "showPGF" (pgf_iter_sequences (a_db p) c_revision itor))
readIORef ref
seq_ids <- (allocaBytes (#size PgfSequenceItor) $ \itor ->
bracket (wrapSequenceItorCallback (getSequences ref)) freeHaskellFunPtr $ \fptr ->
withForeignPtr (c_revision c) $ \c_revision -> do
(#poke PgfSequenceItor, fn) itor fptr
withPgfExn "showPGF" (pgf_iter_sequences (a_db p) c_revision itor))
doc <- readIORef ref
return (seq_ids, doc)
where
getSequences :: IORef Doc -> SequenceItorCallback
getSequences ref itor val exn = do
def <- bracket (pgf_print_sequence_internal val) free $ \c_text -> do
getSequences ref itor seq_id val exn = do
def <- bracket (pgf_print_sequence_internal seq_id val) free $ \c_text -> do
fmap text (peekText c_text)
modifyIORef ref $ (\doc -> doc $$ def)

View File

@@ -46,6 +46,7 @@ data PgfLinBuilderIface
data PgfLinearizationOutputIface
data PgfGraphvizOptions
data PgfSequenceItor
data PgfPhrasetableIds
type Wrapper a = a -> IO (FunPtr a)
type Dynamic a = FunPtr a -> a
@@ -111,25 +112,27 @@ foreign import ccall pgf_iter_lincats :: Ptr PgfDB -> Ptr Concr -> Ptr PgfItor -
foreign import ccall pgf_iter_lins :: Ptr PgfDB -> Ptr Concr -> Ptr PgfItor -> Ptr PgfExn -> IO ()
type SequenceItorCallback = Ptr PgfSequenceItor -> Ptr () -> Ptr PgfExn -> IO ()
type SequenceItorCallback = Ptr PgfSequenceItor -> CSize -> Ptr () -> Ptr PgfExn -> IO ()
foreign import ccall "wrapper" wrapSequenceItorCallback :: Wrapper SequenceItorCallback
foreign import ccall pgf_iter_sequences :: Ptr PgfDB -> Ptr Concr -> Ptr PgfSequenceItor -> Ptr PgfExn -> IO ()
foreign import ccall pgf_iter_sequences :: Ptr PgfDB -> Ptr Concr -> Ptr PgfSequenceItor -> Ptr PgfExn -> IO (Ptr PgfPhrasetableIds)
foreign import ccall pgf_get_lincat_counts_internal :: Ptr () -> Ptr CSize -> IO ()
foreign import ccall pgf_get_lincat_field_internal :: Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_lindef_internal :: Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_lindef_internal :: Ptr PgfPhrasetableIds -> Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_linref_internal :: Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_linref_internal :: Ptr PgfPhrasetableIds -> Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_get_lin_get_prod_count :: Ptr () -> IO CSize
foreign import ccall pgf_print_lin_internal :: Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_lin_internal :: Ptr PgfPhrasetableIds -> Ptr () -> CSize -> IO (Ptr PgfText)
foreign import ccall pgf_print_sequence_internal :: Ptr () -> IO (Ptr PgfText)
foreign import ccall pgf_print_sequence_internal :: CSize -> Ptr () -> IO (Ptr PgfText)
foreign import ccall pgf_release_phrasetable_ids :: Ptr PgfPhrasetableIds -> IO ()
type ItorCallback = Ptr PgfItor -> Ptr PgfText -> Ptr () -> Ptr PgfExn -> IO ()