JNIEXPORT jobject JNICALL
Java_org_grammaticalframework_pgf_Lexicon_lookupWordPrefix
(JNIEnv* env, jclass clazz, jobject jconcr, jstring prefix)
{
GuPool* pool = gu_new_pool();
GuExn* err = gu_new_exn(pool);
GuEnum* en = pgf_lookup_word_prefix(get_ref(env, jconcr), j2gu_string(env, prefix, pool),
pool, err);
if (!gu_ok(err)) {
if (gu_exn_caught(err, PgfExn)) {
GuString msg = (GuString) gu_exn_caught_data(err);
throw_string_exception(env, "org/grammaticalframework/pgf/PGFError", msg);
} else {
throw_string_exception(env, "org/grammaticalframework/pgf/PGFError", "The lookup failed");
}
return NULL;
}
jclass iter_class = (*env)->FindClass(env, "org/grammaticalframework/pgf/FullFormIterator");
jmethodID iter_constrId = (*env)->GetMethodID(env, iter_class, "<init>", "(Lorg/grammaticalframework/pgf/Concr;JJ)V");
jobject iter = (*env)->NewObject(env, iter_class, iter_constrId, jconcr, p2l(pool), p2l(en));
return iter;
}
JNIEXPORT jobject JNICALL
Java_org_grammaticalframework_pgf_Completer_complete(JNIEnv* env, jclass clazz, jobject jconcr, jstring jstartCat, jstring js, jstring jprefix)
{
GuPool* pool = gu_new_pool();
GuString startCat = j2gu_string(env, jstartCat, pool);
GuString s = j2gu_string(env, js, pool);
GuString prefix = j2gu_string(env, jprefix, pool);
GuExn* parse_err = gu_new_exn(pool);
GuEnum* res =
pgf_complete(get_ref(env, jconcr), startCat, s, prefix, parse_err, pool);
if (!gu_ok(parse_err)) {
if (gu_exn_caught(parse_err, PgfExn)) {
GuString msg = (GuString) gu_exn_caught_data(parse_err);
throw_string_exception(env, "org/grammaticalframework/pgf/PGFError", msg);
} else if (gu_exn_caught(parse_err, PgfParseError)) {
GuString tok = (GuString) gu_exn_caught_data(parse_err);
throw_string_exception(env, "org/grammaticalframework/pgf/ParseError", tok);
}
gu_pool_free(pool);
return NULL;
}
jclass tokiter_class = (*env)->FindClass(env, "org/grammaticalframework/pgf/TokenIterator");
jmethodID constrId = (*env)->GetMethodID(env, tokiter_class, "<init>", "(JJ)V");
jobject jtokiter = (*env)->NewObject(env, tokiter_class, constrId, p2l(pool), p2l(res));
return jtokiter;
}
static void
gu_out_fini(GuFinalizer* self)
{
GuOut* out = gu_container(self, GuOut, fini);
if (gu_out_is_buffering(out)) {
GuPool* pool = gu_local_pool();
GuExn* err = gu_new_exn(pool);
gu_out_end_buf(out, err);
gu_pool_free(pool);
}
}
JNIEXPORT jobject JNICALL
Java_org_grammaticalframework_pgf_Parser_parseWithHeuristics
(JNIEnv* env, jclass clazz, jobject jconcr, jstring jstartCat, jstring js, jdouble heuristics, jlong callbacksRef, jobject jpool)
{
GuPool* pool = get_ref(env, jpool);
GuPool* out_pool = gu_new_pool();
GuString startCat = j2gu_string(env, jstartCat, pool);
GuString s = j2gu_string(env, js, pool);
GuExn* parse_err = gu_new_exn(pool);
GuEnum* res =
pgf_parse_with_heuristics(get_ref(env, jconcr), startCat, s, heuristics, l2p(callbacksRef), parse_err, pool, out_pool);
if (!gu_ok(parse_err)) {
if (gu_exn_caught(parse_err, PgfExn)) {
GuString msg = (GuString) gu_exn_caught_data(parse_err);
throw_string_exception(env, "org/grammaticalframework/pgf/PGFError", msg);
} else if (gu_exn_caught(parse_err, PgfParseError)) {
GuString tok = (GuString) gu_exn_caught_data(parse_err);
throw_string_exception(env, "org/grammaticalframework/pgf/ParseError", tok);
}
gu_pool_free(out_pool);
return NULL;
}
jfieldID refId = (*env)->GetFieldID(env, (*env)->GetObjectClass(env, jconcr), "gr", "Lorg/grammaticalframework/pgf/PGF;");
jobject jpgf = (*env)->GetObjectField(env, jconcr, refId);
jclass expiter_class = (*env)->FindClass(env, "org/grammaticalframework/pgf/ExprIterator");
jmethodID constrId = (*env)->GetMethodID(env, expiter_class, "<init>", "(Lorg/grammaticalframework/pgf/PGF;Lorg/grammaticalframework/pgf/Pool;JJ)V");
jobject jexpiter = (*env)->NewObject(env, expiter_class, constrId, jpgf, jpool, p2l(out_pool), p2l(res));
return jexpiter;
}
int main(int argc, char* argv[]) {
// Set the character locale, so we can produce proper output.
setlocale(LC_CTYPE, "");
// Create the pool that is used to allocate everything
GuPool* pool = gu_new_pool();
int status = EXIT_SUCCESS;
if (argc < 4 || argc > 5) {
fprintf(stderr, "usage: %s pgf-file start-cat cnc-lang [heuristics]\n(0.0 <= heuristics < 1.0, default: 0.95)\n", argv[0]);
status = EXIT_FAILURE;
goto fail;
}
char* filename = argv[1];
GuString cat = argv[2];
GuString lang = argv[3];
double heuristics = 0.95;
if (argc == 5) {
heuristics = atof(argv[4]);
}
// Create an exception frame that catches all errors.
GuExn* err = gu_new_exn(pool);
clock_t start = clock();
// Read the PGF grammar.
PgfPGF* pgf = pgf_read(filename, pool, err);
// If an error occured, it shows in the exception frame
if (!gu_ok(err)) {
fprintf(stderr, "Reading PGF failed\n");
status = EXIT_FAILURE;
goto fail;
}
// Look up the source and destination concrete categories
PgfConcr* concr = pgf_get_language(pgf, lang);
if (!concr) {
fprintf(stderr, "Unknown language\n");
status = EXIT_FAILURE;
goto fail;
}
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
fprintf(stderr, "(%.0f ms) Ready to parse [heuristics=%.2f]!\n", 1000.0 * cpu_time_used, heuristics);
// Create an output stream for stdout
GuOut* out = gu_file_out(stdout, pool);
// We will keep the latest results in the 'ppool' and
// we will iterate over them by using 'result'.
GuPool* ppool = NULL;
// The interactive PARSING loop.
// XXX: This currently reads stdin directly, so it doesn't support
// encodings properly. TODO: use a locale reader for input
for (int ctr = 0; true; ctr++) {
// We release the last results
if (ppool != NULL) {
gu_pool_free(ppool);
ppool = NULL;
}
/* fprintf(stdout, "> "); */
/* fflush(stdout); */
char buf[4096];
char* line = fgets(buf, sizeof(buf), stdin);
if (line == NULL) {
if (ferror(stdin)) {
fprintf(stderr, "Input error\n");
status = EXIT_FAILURE;
}
break;
} else if (strcmp(line, "") == 0) {
// End nicely on empty input
break;
} else if (strcmp(line, "\n") == 0) {
// Empty line -> skip
continue;
}
// We create a temporary pool for translating a single
// sentence, so our memory usage doesn't increase over time.
ppool = gu_new_pool();
clock_t start = clock();
GuExn* parse_err = gu_new_exn(ppool);
PgfCallbacksMap* callbacks = pgf_new_callbacks_map(concr, ppool);
GuEnum* result = pgf_parse_with_heuristics(concr, cat, line, heuristics, callbacks, parse_err, ppool, ppool);
PgfExprProb* ep = NULL;
if (gu_ok(parse_err))
ep = gu_next(result, PgfExprProb*, ppool);
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
gu_printf(out, err, "%d (%.0f ms): ", ctr, 1000.0 * cpu_time_used);
if (ep != NULL) {
gu_printf(out, err, "[%.4f] (", ep->prob);
pgf_print_expr(ep->expr, NULL, 0, out, err);
gu_printf(out, err, ")\n");
} else {
gu_printf(out, err, "---\n");
}
gu_out_flush(out, err);
}
fail:
gu_pool_free(pool);
return status;
}
int main(int argc, char* argv[]) {
// Set the character locale, so we can produce proper output.
setlocale(LC_CTYPE, "");
// Create the pool that is used to allocate everything
GuPool* pool = gu_new_pool();
int status = EXIT_SUCCESS;
if (argc < 5 || argc > 6) {
fprintf(stderr, "usage: %s pgf cat from-lang to-lang [probs-file]\n", argv[0]);
status = EXIT_FAILURE;
goto fail;
}
char* filename = argv[1];
GuString cat = gu_str_string(argv[2], pool);
GuString from_lang = gu_str_string(argv[3], pool);
GuString to_lang = gu_str_string(argv[4], pool);
// Create an exception frame that catches all errors.
GuExn* err = gu_new_exn(NULL, gu_kind(type), pool);
// Read the PGF grammar.
PgfPGF* pgf = pgf_read(filename, pool, err);
// If an error occured, it shows in the exception frame
if (!gu_ok(err)) {
fprintf(stderr, "Reading PGF failed\n");
status = EXIT_FAILURE;
goto fail;
}
if (argc == 6) {
char* meta_probs_filename = argv[5];
pgf_load_meta_child_probs(pgf, meta_probs_filename, pool, err);
if (!gu_ok(err)) {
fprintf(stderr, "Loading meta child probs failed\n");
status = EXIT_FAILURE;
goto fail;
}
}
// Look up the source and destination concrete categories
PgfConcr* from_concr = pgf_get_language(pgf, from_lang);
PgfConcr* to_concr = pgf_get_language(pgf, to_lang);
if (!from_concr || !to_concr) {
fprintf(stderr, "Unknown language\n");
status = EXIT_FAILURE;
goto fail_concr;
}
// Register a callback for the literal category Symbol
pgf_parser_add_literal(from_concr, gu_str_string("Symb", pool),
&pgf_nerc_literal_callback);
// Create an output stream for stdout
GuOut* out = gu_file_out(stdout, pool);
// Locale-encoding writers are currently unsupported
// GuWriter* wtr = gu_locale_writer(out, pool);
// Use a writer with hard-coded utf-8 encoding for now.
GuWriter* wtr = gu_new_utf8_writer(out, pool);
// We will keep the latest results in the 'ppool' and
// we will iterate over them by using 'result'.
GuPool* ppool = NULL;
GuEnum* result = NULL;
// The interactive translation loop.
// XXX: This currently reads stdin directly, so it doesn't support
// encodings properly. TODO: use a locale reader for input
while (true) {
fprintf(stdout, "> ");
fflush(stdout);
char buf[4096];
char* line = fgets(buf, sizeof(buf), stdin);
if (line == NULL) {
if (ferror(stdin)) {
fprintf(stderr, "Input error\n");
status = EXIT_FAILURE;
}
break;
} else if (strcmp(line, "") == 0) {
// End nicely on empty input
break;
} else if (strcmp(line, "\n") == 0) {
// Empty line -> show the next tree for the last sentence
if (result != NULL) {
clock_t start = clock();
PgfExprProb* ep = gu_next(result, PgfExprProb*, ppool);
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
printf("%.2f sec\n", cpu_time_used);
// The enumerator will return a null variant at the
// end of the results.
if (ep == NULL) {
goto fail_parse;
}
print_result(ep, to_concr, wtr, err, ppool);
}
continue;
}
// We release the last results
if (ppool != NULL) {
gu_pool_free(ppool);
ppool = NULL;
result = NULL;
}
// We create a temporary pool for translating a single
// sentence, so our memory usage doesn't increase over time.
ppool = gu_new_pool();
GuReader *rdr =
gu_string_reader(gu_str_string(line, ppool), ppool);
PgfLexer *lexer =
pgf_new_lexer(rdr, ppool);
clock_t start = clock();
GuEnum* result =
pgf_parse(from_concr, cat, lexer, ppool);
if (result == NULL) {
PgfToken tok =
pgf_lexer_current_token(lexer);
if (gu_string_eq(tok, gu_empty_string))
gu_puts("Couldn't begin parsing", wtr, err);
else {
gu_puts("Unexpected token: \"", wtr, err);
gu_string_write(tok, wtr, err);
gu_puts("\"\n", wtr, err);
}
goto fail_parse;
}
PgfExprProb* ep = gu_next(result, PgfExprProb*, ppool);
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
printf("%.2f sec\n", cpu_time_used);
// The enumerator will return null at the end of the results.
if (ep == NULL) {
goto fail_parse;
}
print_result(ep, to_concr, wtr, err, ppool);
continue;
fail_parse:
// Free all resources allocated during parsing and linearization
gu_pool_free(ppool);
ppool = NULL;
result = NULL;
}
int main(int argc, char* argv[]) {
// Set the character locale, so we can produce proper output.
setlocale(LC_CTYPE, "");
// Create the pool that is used to allocate everything
GuPool* pool = gu_new_pool();
int status = EXIT_SUCCESS;
if (argc < 5) {
fprintf(stderr, "usage: %s pgf cat from-lang to-lang\n", argv[0]);
status = EXIT_FAILURE;
goto fail;
}
GuString filename = argv[1];
GuString cat = argv[2];
GuString from_lang = argv[3];
GuString to_lang = argv[4];
// Create an exception frame that catches all errors.
GuExn* err = gu_new_exn(pool);
// Read the PGF grammar.
PgfPGF* pgf = pgf_read(filename, pool, err);
// If an error occured, it shows in the exception frame
if (!gu_ok(err)) {
fprintf(stderr, "Reading PGF failed\n");
status = EXIT_FAILURE;
goto fail;
}
// Look up the source and destination concrete categories
PgfConcr* from_concr = pgf_get_language(pgf, from_lang);
PgfConcr* to_concr = pgf_get_language(pgf, to_lang);
if (!from_concr || !to_concr) {
fprintf(stderr, "Unknown language\n");
status = EXIT_FAILURE;
goto fail_concr;
}
// Register a callback for the literal category Symbol
PgfCallbacksMap* callbacks =
pgf_new_callbacks_map(from_concr, pool);
pgf_callbacks_map_add_literal(from_concr, callbacks,
"PN", &pgf_nerc_literal_callback);
pgf_callbacks_map_add_literal(from_concr, callbacks,
"Symb", &pgf_unknown_literal_callback);
// Create an output stream for stdout
GuOut* out = gu_file_out(stdout, pool);
// We will keep the latest results in the 'ppool' and
// we will iterate over them by using 'result'.
GuPool* ppool = NULL;
GuEnum* result = NULL;
// The interactive translation loop.
// XXX: This currently reads stdin directly, so it doesn't support
// encodings properly. TODO: use a locale reader for input
while (true) {
fprintf(stdout, "> ");
fflush(stdout);
char buf[4096];
char* line = fgets(buf, sizeof(buf), stdin);
if (line == NULL) {
if (ferror(stdin)) {
fprintf(stderr, "Input error\n");
status = EXIT_FAILURE;
}
break;
} else if (strcmp(line, "") == 0) {
// End nicely on empty input
break;
} else if (strcmp(line, "\n") == 0) {
// Empty line -> show the next tree for the last sentence
if (result != NULL) {
clock_t start = clock();
PgfExprProb* ep = gu_next(result, PgfExprProb*, ppool);
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
printf("%.2f sec\n", cpu_time_used);
// The enumerator will return a null variant at the
// end of the results.
if (ep == NULL) {
goto fail_parse;
}
print_result(ep, to_concr, out, err, ppool);
}
continue;
}
// We release the last results
if (ppool != NULL) {
gu_pool_free(ppool);
ppool = NULL;
result = NULL;
}
// We create a temporary pool for translating a single
// sentence, so our memory usage doesn't increase over time.
ppool = gu_new_pool();
clock_t start = clock();
GuExn* parse_err = gu_new_exn(ppool);
result =
pgf_parse_with_heuristics(from_concr, cat, line,
-1, callbacks,
parse_err, ppool, ppool);
if (!gu_ok(parse_err)) {
if (gu_exn_caught(parse_err, PgfExn)) {
GuString msg = gu_exn_caught_data(parse_err);
gu_string_write(msg, out, err);
gu_putc('\n', out, err);
} else if (gu_exn_caught(parse_err, PgfParseError)) {
gu_puts("Unexpected token: \"", out, err);
GuString tok = gu_exn_caught_data(parse_err);
gu_string_write(tok, out, err);
gu_puts("\"\n", out, err);
}
goto fail_parse;
}
PgfExprProb* ep = gu_next(result, PgfExprProb*, ppool);
clock_t end = clock();
double cpu_time_used = ((double) (end - start)) / CLOCKS_PER_SEC;
printf("%.2f sec\n", cpu_time_used);
// The enumerator will return null at the end of the results.
if (ep == NULL) {
goto fail_parse;
}
print_result(ep, to_concr, out, err, ppool);
continue;
fail_parse:
// Free all resources allocated during parsing and linearization
gu_pool_free(ppool);
ppool = NULL;
result = NULL;
}
int main(int argc, char* argv[]) {
// Set the character locale, so we can produce proper output.
setlocale(LC_CTYPE, "");
// Create the pool that is used to allocate everything
GuPool* pool = gu_new_pool();
int status = EXIT_SUCCESS;
if (argc != 4) {
fprintf(stderr, "usage: %s pgf cat from_lang\n", argv[0]);
status = EXIT_FAILURE;
goto fail;
}
char* filename = argv[1];
GuString cat = gu_str_string(argv[2], pool);
GuString from_lang = gu_str_string(argv[3], pool);
// Create an exception frame that catches all errors.
GuExn* err = gu_new_exn(NULL, gu_kind(type), pool);
// Read the PGF grammar.
PgfPGF* pgf = pgf_read(filename, pool, err);
// If an error occured, it shows in the exception frame
if (!gu_ok(err)) {
fprintf(stderr, "Reading PGF failed\n");
status = EXIT_FAILURE;
goto fail;
}
pgf_load_meta_child_probs(pgf, "../../../treebanks/PennTreebank/ParseEngAbs3.probs", pool, err);
if (!gu_ok(err)) {
fprintf(stderr, "Loading meta child probs failed\n");
status = EXIT_FAILURE;
goto fail;
}
// Look up the source and destination concrete categories
PgfConcr* from_concr = pgf_get_language(pgf, from_lang);
if (!from_concr) {
fprintf(stderr, "Unknown language\n");
status = EXIT_FAILURE;
goto fail_concr;
}
// Register a callback for the literal category Symbol
pgf_parser_add_literal(from_concr, gu_str_string("Symb", pool),
&pgf_nerc_literal_callback);
// We will keep the latest results in the 'ppool' and
// we will iterate over them by using 'result'.
GuPool* ppool = NULL;
// The interactive translation loop.
// XXX: This currently reads stdin directly, so it doesn't support
// encodings properly. TODO: use a locale reader for input
while (true) {
char buf[4096];
char* line = fgets(buf, sizeof(buf), stdin);
if (line == NULL) {
if (ferror(stdin)) {
fprintf(stderr, "Input error\n");
status = EXIT_FAILURE;
}
break;
} else if (strcmp(line, "") == 0) {
// End nicely on empty input
break;
}
// We create a temporary pool for translating a single
// sentence, so our memory usage doesn't increase over time.
ppool = gu_new_pool();
GuReader *rdr =
gu_string_reader(gu_str_string(line, ppool), ppool);
PgfLexer *lexer =
pgf_new_simple_lexer(rdr, ppool);
pgf_print_chunks(from_concr, cat, lexer, ppool);
// Free all resources allocated during parsing and linearization
gu_pool_free(ppool);
}
fail_concr:
fail:
gu_pool_free(pool);
return status;
}
