Example #1
0
int main(int argc, char* argv[]) {
	// Set the character locale, so we can produce proper output.
	setlocale(LC_CTYPE, "");

	if (argc != 2) {
		fprintf(stderr, "usage: %s pgf\n", argv[0]);
		return EXIT_FAILURE;
	}
	char* filename = argv[1];

	GuPool* pool = gu_new_pool();
	GuExn* err = gu_exn(NULL, type, pool);
	PgfPGF* pgf = pgf_read(filename, pool, err);
	int status = 0;
	if (!gu_ok(err)) {
		fprintf(stderr, "Reading PGF failed\n");
		status = 1;
		goto fail_read;
	}
	GuOut* out = gu_file_out(stdout, pool);
    pgf_print(pgf, out, err);
	gu_out_flush(out, err);
fail_read:
	gu_pool_free(pool);
	return status;
}
Example #2
0
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;
	}
Example #3
0
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;
}
Example #4
0
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;
    }