C++ prt_error函数代码示例

int spellcheck_suggest(void * chk, char ***sug, const char * word)
{
    struct linkgrammar_aspell *aspell = (struct linkgrammar_aspell *)chk;
    if (!sug) {
        prt_error("Error: Aspell. Corrupt pointer.\n");
        return 0;
    }
    if (aspell && aspell->speller) {
        const AspellWordList *list = NULL;
        AspellStringEnumeration *elem = NULL;
        const char *aword = NULL;
        unsigned int size, i;
        char **array = NULL;

        list = aspell_speller_suggest(aspell->speller, word, -1);
        elem = aspell_word_list_elements(list);
        size = aspell_word_list_size(list);
        /* allocate an array of char* for returning back to link-parser
         */
        array = (char **)malloc(sizeof(char *) * size);
        if (!array) {
            prt_error("Error: Aspell. Out of memory.\n");
            delete_aspell_string_enumeration(elem);
            return 0;
        }
        i = 0;
        while ((aword = aspell_string_enumeration_next(elem)) != NULL) {
            array[i++] = strdup(aword);
        }
        delete_aspell_string_enumeration(elem);
        *sug = array;
        return size;
    }
    return 0;
}

/**
 * Upcase the first letter of the word.
 * XXX FIXME This works 'most of the time', but is not technically correct.
 * This is because towlower() and towupper() are locale dependent, and also
 * because the byte-counts might not match up, e.g. German ß and SS.
 * The correct long-term fix is to use ICU or glib g_utf8_strup(), etc.
 */
void upcase_utf8_str(char *to, const char * from, size_t usize)
{
	wchar_t c;
	int i, nbl, nbh;
	char low[MB_LEN_MAX];
	mbstate_t mbs;

	memset(&mbs, 0, sizeof(mbs));
	nbh = mbrtowc (&c, from, MB_CUR_MAX, &mbs);
	if (nbh < 0)
	{
		prt_error("Error: Invalid multi-byte string!");
		return;
	}
	c = towupper(c);
	nbl = wctomb_check(low, c);

	/* Check for error on an in-place copy */
	if ((nbh < nbl) && (to == from))
	{
		/* I'm to lazy to fix this */
		prt_error("Error: can't upcase multi-byte string!");
		return;
	}

	/* Upcase */
	for (i=0; i<nbl; i++) { to[i] = low[i]; }

	if ((nbh == nbl) && (to == from)) return;

	from += nbh;
	to += nbl;
	safe_strcpy(to, from, usize-nbl);
}

static char *
win32_getlocale (void)
{
	char lbuf[10];
	char locale[32];

	LCID lcid = GetThreadLocale();

	if (0 >= GetLocaleInfoA(lcid, LOCALE_SISO639LANGNAME, lbuf, sizeof(lbuf)))
	{
		prt_error("Error: GetLocaleInfoA LOCALE_SENGLISHLANGUAGENAME LCID=%d: "
		          "Error %d\n", (int)lcid, (int)GetLastError());
		return NULL;
	}
	strcpy(locale, lbuf);
	strcat(locale, "-");

	if (0 >= GetLocaleInfoA(lcid, LOCALE_SISO3166CTRYNAME, lbuf, sizeof(lbuf)))
	{
		prt_error("Error: GetLocaleInfoA LOCALE_SISO3166CTRYNAME LCID=%d: "
		          "Error %d\n", (int)lcid, (int)GetLastError());
		return NULL;
	}
	strcat(locale, lbuf);

	return strdup(locale);
}

/**
 * Ensure that the program's locale has a UTF-8 codeset.
 */
void set_utf8_program_locale(void)
{
#ifndef _WIN32
	/* The LG library doesn't use mbrtowc_l(), since it doesn't exist in
	 * the dynamic glibc (2.22). mbsrtowcs_l() could also be used, but for
	 * some reason it exists only in the static glibc.
	 * In order that mbrtowc() will work for any UTF-8 character, UTF-8
	 * codeset is ensured. */
	const char *codeset = nl_langinfo(CODESET);
	if (!strstr(codeset, "UTF") && !strstr(codeset, "utf"))
	{
		const char *locale = setlocale(LC_CTYPE, NULL);
		/* Avoid an initial spurious message. */
		if ((0 != strcmp(locale, "C")) && (0 != strcmp(locale, "POSIX")))
		{
			prt_error("Warning: Program locale \"%s\" (codeset %s) was not UTF-8; "
						 "force-setting to en_US.UTF-8\n", locale, codeset);
		}
		locale = setlocale(LC_CTYPE, "en_US.UTF-8");
		if (NULL == locale)
		{
			prt_error("Warning: Program locale en_US.UTF-8 could not be set; "
			          "force-setting to C.UTF-8\n");
			locale = setlocale(LC_CTYPE, "C.UTF-8");
			if (NULL == locale)
			{
				prt_error("Warning: Could not set a UTF-8 program locale; "
				          "program may malfunction\n");
			}
		}
	}
#endif /* !_WIN32 */
}

/**
 * Assumes that the sentence expression lists have been generated.
 */
void prepare_to_parse(Sentence sent, Parse_Options opts)
{
	size_t i;

	build_sentence_disjuncts(sent, opts->disjunct_cost, opts);
	if (verbosity_level(D_PREP))
	{
		prt_error("Debug: After expanding expressions into disjuncts:\n");
		print_disjunct_counts(sent);
	}
	print_time(opts, "Built disjuncts");

	for (i=0; i<sent->length; i++)
	{
		sent->word[i].d = eliminate_duplicate_disjuncts(sent->word[i].d);

		/* Some long Russian sentences can really blow up, here. */
		if (resources_exhausted(opts->resources))
			return;
	}
	print_time(opts, "Eliminated duplicate disjuncts");

	if (verbosity_level(D_PREP))
	{
		prt_error("Debug: After expression pruning and duplicate elimination:\n");
		print_disjunct_counts(sent);
	}

	setup_connectors(sent);
}

/**
 * Read table of [link, domain type].
 * This tells us what domain type each link belongs to.
 * This lookup table *must* be defined in the knowledge file.
 */
static void read_starting_link_table(pp_knowledge *k)
{
  const char *p;
  const char label[] = "STARTING_LINK_TYPE_TABLE";
  int i, n_tokens;
  if (!pp_lexer_set_label(k->lt, label))
  {
    prt_error("Fatal error: post_process: Couldn't find starting link table %s",label);
    exit(1);
  }
  n_tokens = pp_lexer_count_tokens_of_label(k->lt);
  if (n_tokens %2)
  {
    prt_error("Fatal error: post_process: Link table must have format [<link> <domain name>]+");
    exit(1);
  }
  k->nStartingLinks = n_tokens/2;
  k->starting_link_lookup_table = (StartingLinkAndDomain*)
    xalloc((1+k->nStartingLinks)*sizeof(StartingLinkAndDomain));
  for (i=0; i<k->nStartingLinks; i++)
    {
      /* read the starting link itself */
      k->starting_link_lookup_table[i].starting_link =
	string_set_add(pp_lexer_get_next_token_of_label(k->lt),k->string_set);

      /* read the domain type of the link */
      p = pp_lexer_get_next_token_of_label(k->lt);
      check_domain_is_legal(p);
      k->starting_link_lookup_table[i].domain = (int) p[0];
    }

  /* end sentinel */
  k->starting_link_lookup_table[k->nStartingLinks].domain = -1;
}

int sentence_parse(Sentence sent, Parse_Options opts)
{
	int rc;

	sent->num_valid_linkages = 0;

	/* If the sentence has not yet been split, do so now.
	 * This is for backwards compatibility, for existing programs
	 * that do not explicitly call the splitter.
	 */
	if (0 == sent->length)
	{
		rc = sentence_split(sent, opts);
		if (rc) return -1;
	}
	else
	{
		/* During a panic parse, we enter here a second time, with leftover
		 * garbage. Free it. We really should make the code that is panicking
		 * do this free, but right now, they have no API for it, so we do it
		 * as a favor. XXX FIXME someday. */
		free_sentence_disjuncts(sent);
	}

	/* Check for bad sentence length */
	if (MAX_SENTENCE <= sent->length)
	{
		prt_error("Error: sentence too long, contains more than %d words\n",
			MAX_SENTENCE);
		return -2;
	}

	resources_reset(opts->resources);

	/* Expressions were set up during the tokenize stage.
	 * Prune them, and then parse.
	 */
	expression_prune(sent, opts);
	print_time(opts, "Finished expression pruning");
	if (opts->use_sat_solver)
	{
		sat_parse(sent, opts);
	}
	else
	{
		classic_parse(sent, opts);
	}
	print_time(opts, "Finished parse");

	if ((verbosity > 0) &&
	   (PARSE_NUM_OVERFLOW < sent->num_linkages_found))
	{
		prt_error("Warning: Combinatorial explosion! nulls=%zu cnt=%d\n"
			"Consider retrying the parse with the max allowed disjunct cost set lower.\n"
			"At the command line, use !cost-max\n",
			sent->null_count, sent->num_linkages_found);
	}
	return sent->num_valid_linkages;
}

int sentence_parse(Sentence sent, Parse_Options opts)
{
	int rc;

	verbosity = opts->verbosity;
	debug = opts->debug;
	test = opts->test;

	sent->num_valid_linkages = 0;

	/* If the sentence has not yet been split, do so now.
	 * This is for backwards compatibility, for existing programs
	 * that do not explicitly call the splitter.
	 */
	if (0 == sent->length)
	{
		rc = sentence_split(sent, opts);
		if (rc) return -1;
	}

	/* Check for bad sentence length */
	if (MAX_SENTENCE <= sent->length)
	{
		prt_error("Error: sentence too long, contains more than %d words\n",
			MAX_SENTENCE);
		return -2;
	}

	/* Initialize/free any leftover garbage */
	free_sentence_disjuncts(sent);  /* Is this really needed ??? */
	resources_reset_space(opts->resources);

	if (resources_exhausted(opts->resources))
		return 0;

	/* Expressions were previously set up during the tokenize stage. */
	expression_prune(sent);
	print_time(opts, "Finished expression pruning");
	if (opts->use_sat_solver)
	{
		sat_parse(sent, opts);
	}
	else
	{
		chart_parse(sent, opts);
	}
	print_time(opts, "Finished parse");

	if ((verbosity > 0) &&
	   (PARSE_NUM_OVERFLOW < sent->num_linkages_found))
	{
		prt_error("WARNING: Combinatorial explosion! nulls=%zu cnt=%d\n"
			"Consider retrying the parse with the max allowed disjunct cost set lower.\n"
			"At the command line, use !cost-max\n",
			sent->null_count, sent->num_linkages_found);
	}
	return sent->num_valid_linkages;
}

/**
 * Read in the whole stinkin file. This routine returns
 * malloced memory, which should be freed as soon as possible.
 */
char *get_file_contents(const char * dict_name)
{
	int fd;
	size_t tot_size;
	size_t tot_read = 0;
	struct stat buf;
	char * contents;

	/* On Windows, 'b' (binary mode) is mandatory, otherwise fstat file length
	 * is confused by crlf counted as one byte. POSIX systems just ignore it. */
	FILE *fp = dictopen(dict_name, "rb");

	if (fp == NULL)
		return NULL;

	/* Get the file size, in bytes. */
	fd = fileno(fp);
	fstat(fd, &buf);
	tot_size = buf.st_size;

	contents = (char *) malloc(sizeof(char) * (tot_size+7));

	/* Now, read the whole file.
	 * Normally, a single fread() call below reads the whole file. */
	while (1)
	{
		size_t read_size = fread(contents, 1, tot_size+7, fp);

		if (0 == read_size)
		{
			bool err = (0 != ferror(fp));

			if (err)
			{
				char errbuf[64];

				strerror_r(errno, errbuf, sizeof(errbuf));
				fclose(fp);
				prt_error("Error: %s: Read error (%s)\n", dict_name, errbuf);
				free(contents);
				return NULL;
			}
			fclose(fp);
			break;
		}
		tot_read += read_size;
	}

	if (tot_read > tot_size+6)
	{
		prt_error("Error: %s: File size is insane (%zu)!\n", dict_name, tot_size);
		free(contents);
		return NULL;
	}

	contents[tot_read] = '\0';
	return contents;
}

void  LGCheckGrammar::P_UNDEFINED ()
{
      int t, n = n_errors;

		if (n_terms > max_char_set+1)
		{
			strcpy (gft, ".lex");
			strcpy (grmfid, gdn);
			strcat (grmfid, gfn);
			strcat (grmfid, gft);

			char*  Input_start = input_start;
			char*  Input_end   = input_end;
			char** Line_ptr    = line_ptr;

			input_start = lex_input_start;
			input_end   = lex_input_end;
			line_ptr    = lex_line_ptr;

			for (t = max_char_set+1; t < n_terms; t++) // Skip <eof>
			{
				if (term_type[t] & LEXFILE)
				prt_error ("'%s' is listed in the .lex file, but not defined in the .lgr file", term_name[t], 0, term_line[t]);
			}

			strcpy (gft, ".lgr");
			strcpy (grmfid, gdn);
			strcat (grmfid, gfn);
			strcat (grmfid, gft);

			input_start = Input_start;
			input_end   = Input_end;
			line_ptr    = Line_ptr;											

			for (t = max_char_set+1; t < n_terms; t++) // Skip <eof>
			{
				if (!(term_type[t] & LEXFILE))
				{
					char code = charcode[*term_name[t]];
					if ( code == DIGIT || code == QUOTE) 
					{
						prt_error ("'%s' is not a predefined symbol", term_name[t], 0, term_line[t]);
						prt_log ("Predefined symbols are:\n");
						prt_log ("   0..31\n");
						for (int i = 32; i < 127; i++) prt_log ("   %s\n", term_name[i]);
						prt_log ("   127..255\n");
						LG::Terminate(95);
					}
					else
					{
						prt_error ("'%s' is not defined in the lexical grammar (.lgr file)", term_name[t], 0, term_line[t]);
					}
				}
			}
		}

		FREE (term_type, n_terms);
}

static Sense * lg_corpus_senses(Corpus *corp,
                                const char * inflected_word,
                                const char * disjunct,
                                int wrd)
{
	double log_prob;
	const unsigned char *sense;
	Sense *sns, *head = NULL;
	int rc;

	/* Look up the disjunct in the database */
	rc = sqlite3_bind_text(corp->sense_query, 1,
		inflected_word, -1, SQLITE_STATIC);
	if (rc != SQLITE_OK)
	{
		prt_error("Error: SQLite can't bind word in sense query: rc=%d \n", rc);
		return NULL;
	}

	rc = sqlite3_bind_text(corp->sense_query, 2,
		disjunct, -1, SQLITE_STATIC);
	if (rc != SQLITE_OK)
	{
		prt_error("Error: SQLite can't bind disjunct in sense query: rc=%d \n", rc);
		return NULL;
	}

	rc = sqlite3_step(corp->sense_query);
	while (SQLITE_ROW == rc)
	{
		sense = sqlite3_column_text(corp->sense_query, 0);
		log_prob = sqlite3_column_double(corp->sense_query, 1);
		// printf ("Word=%s dj=%s sense=%s score=%f\n", 
		// 	inflected_word, disjunct, sense, log_prob);

		sns = (Sense *) malloc(sizeof(Sense));
		sns->next = head;
		head = sns;

		sns->inflected_word = inflected_word;
		sns->disjunct = disjunct;
		sns->sense = strdup(sense);
		sns->score = log_prob;
		sns->word = wrd;

		/* Get the next row, if any */
		rc = sqlite3_step(corp->sense_query);
	}

	/* Failure to do both a reset *and* a clear will cause subsequent
	 * binds tp fail. */
	sqlite3_reset(corp->sense_query);
	sqlite3_clear_bindings(corp->sense_query);

	return head;
}

/**
 * get_disjunct_score -- get log probability of observing disjunt.
 *
 * Given an "inflected" word and a disjunct, thris routine returns the
 * -log_2 conditional probability prob(d|w) of seeing the disjunct 'd'
 * given that the word 'w' was observed.  Here, "inflected word" means
 * the link-grammar dictionary entry, complete with its trailing period
 * and tag -- e.g. run.v or running.g -- everything after the dot is the
 * "inflection".
 */
static double get_disjunct_score(Corpus *corp,
                                 const char * inflected_word,
                                 const char * disjunct)
{
	double val;
	int rc;

	/* Look up the disjunct in the database */
	rc = sqlite3_bind_text(corp->rank_query, 1,
		inflected_word, -1, SQLITE_STATIC);
	if (rc != SQLITE_OK)
	{
		const char *errmsg = sqlite3_errmsg(corp->dbconn);
		prt_error("Error: SQLite can't bind word: rc=%d %s\n", rc, errmsg);
		return LOW_SCORE;
	}

	rc = sqlite3_bind_text(corp->rank_query, 2,
		disjunct, -1, SQLITE_STATIC);
	if (rc != SQLITE_OK)
	{
		const char *errmsg = sqlite3_errmsg(corp->dbconn);
		prt_error("Error: SQLite can't bind disjunct: rc=%d %s\n", rc, errmsg);
		return LOW_SCORE;
	}

	rc = sqlite3_step(corp->rank_query);
	if (rc != SQLITE_ROW)
	{
		val = LOW_SCORE;
#ifdef DEBUG
		printf ("Word=%s dj=%s not found in dict, assume score=%f\n",
			inflected_word, disjunct, val);
#endif
		if (rc < SQLITE_ROW)
		{
			const char *errmsg = sqlite3_errmsg(corp->dbconn);
			prt_error("Error: SQLite can't ifind word: rc=%d %s\n", rc, errmsg);
		}
	}
	else
	{
		val = sqlite3_column_double(corp->rank_query, 0);
		if (LOW_SCORE < val) val = LOW_SCORE;
#ifdef DEBUG
		printf ("Word=%s dj=%s score=%f\n", inflected_word, disjunct, val);
#endif
	}

	/* Failure to do both a reset *and* a clear will cause subsequent
	 * binds to fail. */
	sqlite3_reset(corp->rank_query);
	sqlite3_clear_bindings(corp->rank_query);
	return val;
}

static void read_contains_rules(pp_knowledge *k, const char *label,
				pp_rule **rules, int *nRules)
{
  /* Reading the 'contains_one_rules' and reading the
     'contains_none_rules' into their respective arrays */
  int n_commas, n_tokens, i, r;
  const char *p;
  const char **tokens;
  if (!pp_lexer_set_label(k->lt, label)) {
      *nRules = 0;
      if (verbosity>0) printf("PP warning: Not using any %s rules\n", label);
  }
  else {
    n_commas = pp_lexer_count_commas_of_label(k->lt);
    *nRules = (n_commas + 1)/3;
  }
  *rules = (pp_rule*) xalloc ((1+*nRules)*sizeof(pp_rule));
  for (r=0; r<*nRules; r++)
    {
      /* first read link */
      tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
      if (n_tokens>1)
      {
        prt_error("Fatal Error: post_process: Invalid syntax in %s (rule %i)",label,r+1);
        exit(1);
      }
      (*rules)[r].selector = string_set_add(tokens[0], k->string_set);

      /* read link set */
      tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
      (*rules)[r].link_set = pp_linkset_open(n_tokens);
      (*rules)[r].link_set_size = n_tokens;
      (*rules)[r].link_array = (const char **) xalloc((1+n_tokens)*sizeof(const char*));
      for (i=0; i<n_tokens; i++)
      {
        p = string_set_add(tokens[i], k->string_set);
        pp_linkset_add((*rules)[r].link_set, p);
        (*rules)[r].link_array[i] = p;
      }
      (*rules)[r].link_array[i]=0; /* NULL-terminator */

      /* read error message */
      tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
      if (n_tokens>1)
      {
        prt_error("Fatal Error: post_process: Invalid syntax in %s (rule %i)",label,r+1);
        exit(1);
      }
      (*rules)[r].msg = string_set_add(tokens[0], k->string_set);
    }

  /* sentinel entry */
  (*rules)[*nRules].msg = 0;
}

static bool read_form_a_cycle_rules(pp_knowledge *k, const char *label)
{
  size_t n_commas, n_tokens;
  size_t r, i;
  pp_linkset *lsHandle;
  const char **tokens;
  if (!pp_lexer_set_label(k->lt, label)) {
      k->n_form_a_cycle_rules = 0;
      if (verbosity_level(+D_PPK))
          prt_error("Warning: File %s: Not using any 'form a cycle' rules\n",
                    k->path);
  }
  else {
    n_commas = pp_lexer_count_commas_of_label(k->lt);
    k->n_form_a_cycle_rules = (n_commas + 1)/2;
  }
  k->form_a_cycle_rules=
    (pp_rule*) malloc ((1+k->n_form_a_cycle_rules)*sizeof(pp_rule));
  for (r=0; r<k->n_form_a_cycle_rules; r++)
    {
      /* read link set */
      tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
      if (n_tokens <= 0)
      {
        prt_error("Error: File %s: Syntax error\n", k->path);
        return false;
      }
      lsHandle = pp_linkset_open(n_tokens);
      for (i=0; i<n_tokens; i++)
          pp_linkset_add(lsHandle,string_set_add(tokens[i], k->string_set));
      k->form_a_cycle_rules[r].link_set = lsHandle;

      /* read error message */
      tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
      if (n_tokens > 1)
      {
         prt_error("Error: File %s: Invalid syntax (rule %zu of %s)\n",
                   k->path, r+1,label);
         return false;
      }
      k->form_a_cycle_rules[r].msg = string_set_add(tokens[0], k->string_set);
      k->form_a_cycle_rules[r].use_count = 0;
    }

  /* sentinel entry */
  k->form_a_cycle_rules[k->n_form_a_cycle_rules].msg = 0;
  k->form_a_cycle_rules[k->n_form_a_cycle_rules].use_count = 0;

  return true;
}

static void prt_regerror(const char *msg, const Regex_node *re, int rc,
                         int erroffset)
{
#if HAVE_PCRE2_H
	PCRE2_UCHAR errbuf[ERRBUFFLEN];
	pcre2_get_error_message(rc, errbuf, ERRBUFFLEN);
#else
	char errbuf[ERRBUFFLEN];
	regerror(rc, re->re, errbuf, ERRBUFFLEN);
#endif /* HAVE_PCRE2_H */

	prt_error("Error: %s: \"%s\" (%s", msg, re->pattern, re->name);
	if (-1 != erroffset) prt_error(" at %d", erroffset);
	prt_error("): %s (%d)\n", errbuf, rc);
}