ts_locale.c File Reference

#include "postgres.h"
#include "catalog/pg_collation.h"
#include "common/string.h"
#include "storage/fd.h"
#include "tsearch/ts_locale.h"
#include "tsearch/ts_public.h"

Include dependency graph for ts_locale.c:

Go to the source code of this file.

Macros
#define	WC_BUF_LEN   3

Functions
static void	tsearch_readline_callback (void *arg)
int	t_isdigit (const char *ptr)
int	t_isspace (const char *ptr)
int	t_isalpha (const char *ptr)
int	t_isalnum (const char *ptr)
int	t_isprint (const char *ptr)
bool	tsearch_readline_begin (tsearch_readline_state *stp, const char *filename)
char *	tsearch_readline (tsearch_readline_state *stp)
void	tsearch_readline_end (tsearch_readline_state *stp)
char *	lowerstr (const char *str)
char *	lowerstr_with_len (const char *str, int len)

Macro Definition Documentation

WC_BUF_LEN

#define WC_BUF_LEN   3

Definition at line 34 of file ts_locale.c.

Function Documentation

lowerstr()

char* lowerstr

(

const char *

str

)

Definition at line 255 of file ts_locale.c.

{
    return lowerstr_with_len(str, strlen(str));
}

References lowerstr_with_len(), and generate_unaccent_rules::str.

Referenced by convertPgWchar(), dispell_init(), dsimple_init(), dsnowball_init(), dsynonym_init(), dxsyn_lexize(), lowerstr_ctx(), NIImportAffixes(), and read_dictionary().

lowerstr_with_len()

char* lowerstr_with_len	(	const char *	str,
		int	len
	)

Definition at line 268 of file ts_locale.c.

{
    char       *out;
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (len == 0)
        return pstrdup("");
 
    /*
     * Use wide char code only when max encoding length > 1 and ctype != C.
     * Some operating systems fail with multi-byte encodings and a C locale.
     * Also, for a C locale there is no need to process as multibyte. From
     * backend/utils/adt/oracle_compat.c Teodor
     */
    if (pg_database_encoding_max_length() > 1 && !database_ctype_is_c)
    {
        wchar_t    *wstr,
                   *wptr;
        int         wlen;
 
        /*
         * alloc number of wchar_t for worst case, len contains number of
         * bytes >= number of characters and alloc 1 wchar_t for 0, because
         * wchar2char wants zero-terminated string
         */
        wptr = wstr = (wchar_t *) palloc(sizeof(wchar_t) * (len + 1));
 
        wlen = char2wchar(wstr, len + 1, str, len, mylocale);
        Assert(wlen <= len);
 
        while (*wptr)
        {
            *wptr = towlower((wint_t) *wptr);
            wptr++;
        }
 
        /*
         * Alloc result string for worst case + '\0'
         */
        len = pg_database_encoding_max_length() * wlen + 1;
        out = (char *) palloc(len);
 
        wlen = wchar2char(out, wstr, len, mylocale);
 
        pfree(wstr);
 
        if (wlen < 0)
            ereport(ERROR,
                    (errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE),
                     errmsg("conversion from wchar_t to server encoding failed: %m")));
        Assert(wlen < len);
    }
    else
    {
        const char *ptr = str;
        char       *outptr;
 
        outptr = out = (char *) palloc(sizeof(char) * (len + 1));
        while ((ptr - str) < len && *ptr)
        {
            *outptr++ = tolower(TOUCHAR(ptr));
            ptr++;
        }
        *outptr = '\0';
    }
 
    return out;
}

References Assert(), char2wchar(), database_ctype_is_c, ereport, errcode(), errmsg(), ERROR, len, palloc(), pfree(), pg_database_encoding_max_length(), pstrdup(), generate_unaccent_rules::str, TOUCHAR, and wchar2char().

Referenced by dispell_lexize(), dsimple_lexize(), dsnowball_lexize(), dsynonym_lexize(), generate_trgm_only(), generate_wildcard_trgm(), and lowerstr().

t_isalnum()

int t_isalnum

(

const char *

ptr

)

Definition at line 82 of file ts_locale.c.

{
    int         clen = pg_mblen(ptr);
    wchar_t     character[WC_BUF_LEN];
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (clen == 1 || database_ctype_is_c)
        return isalnum(TOUCHAR(ptr));
 
    char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale);
 
    return iswalnum((wint_t) character[0]);
}

References char2wchar(), database_ctype_is_c, pg_mblen(), TOUCHAR, and WC_BUF_LEN.

Referenced by parse_or_operator().

t_isalpha()

int t_isalpha

(

const char *

ptr

)

Definition at line 67 of file ts_locale.c.

{
    int         clen = pg_mblen(ptr);
    wchar_t     character[WC_BUF_LEN];
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (clen == 1 || database_ctype_is_c)
        return isalpha(TOUCHAR(ptr));
 
    char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale);
 
    return iswalpha((wint_t) character[0]);
}

References char2wchar(), database_ctype_is_c, pg_mblen(), TOUCHAR, and WC_BUF_LEN.

Referenced by parse_affentry(), RS_compile(), and RS_isRegis().

t_isdigit()

int t_isdigit

(

const char *

ptr

)

Definition at line 37 of file ts_locale.c.

{
    int         clen = pg_mblen(ptr);
    wchar_t     character[WC_BUF_LEN];
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (clen == 1 || database_ctype_is_c)
        return isdigit(TOUCHAR(ptr));
 
    char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale);
 
    return iswdigit((wint_t) character[0]);
}

References char2wchar(), database_ctype_is_c, pg_mblen(), TOUCHAR, and WC_BUF_LEN.

Referenced by getNextFlagFromString(), gettoken_tsvector(), NISortDictionary(), parse_lquery(), and parse_phrase_operator().

t_isprint()

int t_isprint

(

const char *

ptr

)

Definition at line 97 of file ts_locale.c.

{
    int         clen = pg_mblen(ptr);
    wchar_t     character[WC_BUF_LEN];
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (clen == 1 || database_ctype_is_c)
        return isprint(TOUCHAR(ptr));
 
    char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale);
 
    return iswprint((wint_t) character[0]);
}

References char2wchar(), database_ctype_is_c, pg_mblen(), TOUCHAR, and WC_BUF_LEN.

Referenced by NIImportDictionary().

t_isspace()

int t_isspace

(

const char *

ptr

)

Definition at line 52 of file ts_locale.c.

{
    int         clen = pg_mblen(ptr);
    wchar_t     character[WC_BUF_LEN];
    pg_locale_t mylocale = 0;   /* TODO */
 
    if (clen == 1 || database_ctype_is_c)
        return isspace(TOUCHAR(ptr));
 
    char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale);
 
    return iswspace((wint_t) character[0]);
}

References char2wchar(), database_ctype_is_c, pg_mblen(), TOUCHAR, and WC_BUF_LEN.

Referenced by addCompoundAffixFlagValue(), find_word(), findwrd(), get_nextfield(), getNextFlagFromString(), gettoken_query(), gettoken_query_standard(), gettoken_query_websearch(), gettoken_tsvector(), initTrie(), NIImportAffixes(), NIImportDictionary(), NIImportOOAffixes(), NISortDictionary(), parse_affentry(), parse_or_operator(), readstoplist(), and thesaurusRead().

tsearch_readline()

char* tsearch_readline

(

tsearch_readline_state *

stp

)

Definition at line 159 of file ts_locale.c.

{
    char       *recoded;
 
    /* Advance line number to use in error reports */
    stp->lineno++;
 
    /* Clear curline, it's no longer relevant */
    if (stp->curline)
    {
        if (stp->curline != stp->buf.data)
            pfree(stp->curline);
        stp->curline = NULL;
    }
 
    /* Collect next line, if there is one */
    if (!pg_get_line_buf(stp->fp, &stp->buf))
        return NULL;
 
    /* Validate the input as UTF-8, then convert to DB encoding if needed */
    recoded = pg_any_to_server(stp->buf.data, stp->buf.len, PG_UTF8);
 
    /* Save the correctly-encoded string for possible error reports */
    stp->curline = recoded;     /* might be equal to buf.data */
 
    /*
     * We always return a freshly pstrdup'd string.  This is clearly necessary
     * if pg_any_to_server() returned buf.data, and we need a second copy even
     * if encoding conversion did occur.  The caller is entitled to pfree the
     * returned string at any time, which would leave curline pointing to
     * recycled storage, causing problems if an error occurs after that point.
     * (It's preferable to return the result of pstrdup instead of the output
     * of pg_any_to_server, because the conversion result tends to be
     * over-allocated.  Since callers might save the result string directly
     * into a long-lived dictionary structure, we don't want it to be a larger
     * palloc chunk than necessary.  We'll reclaim the conversion result on
     * the next call.)
     */
    return pstrdup(recoded);
}

References tsearch_readline_state::buf, tsearch_readline_state::curline, StringInfoData::data, tsearch_readline_state::fp, StringInfoData::len, tsearch_readline_state::lineno, pfree(), pg_any_to_server(), pg_get_line_buf(), PG_UTF8, and pstrdup().

Referenced by dsynonym_init(), initTrie(), NIImportAffixes(), NIImportDictionary(), NIImportOOAffixes(), read_dictionary(), readstoplist(), and thesaurusRead().

tsearch_readline_begin()

bool tsearch_readline_begin	(	tsearch_readline_state *	stp,
		const char *	filename
	)

Definition at line 136 of file ts_locale.c.

{
    if ((stp->fp = AllocateFile(filename, "r")) == NULL)
        return false;
    stp->filename = filename;
    stp->lineno = 0;
    initStringInfo(&stp->buf);
    stp->curline = NULL;
    /* Setup error traceback support for ereport() */
    stp->cb.callback = tsearch_readline_callback;
    stp->cb.arg = (void *) stp;
    stp->cb.previous = error_context_stack;
    error_context_stack = &stp->cb;
    return true;
}

References AllocateFile(), ErrorContextCallback::arg, tsearch_readline_state::buf, ErrorContextCallback::callback, tsearch_readline_state::cb, tsearch_readline_state::curline, error_context_stack, filename, tsearch_readline_state::filename, tsearch_readline_state::fp, initStringInfo(), tsearch_readline_state::lineno, ErrorContextCallback::previous, and tsearch_readline_callback().

Referenced by dsynonym_init(), initTrie(), NIImportAffixes(), NIImportDictionary(), NIImportOOAffixes(), read_dictionary(), readstoplist(), and thesaurusRead().

tsearch_readline_callback()

static void tsearch_readline_callback ( void *  arg )

static

Definition at line 227 of file ts_locale.c.

{
    tsearch_readline_state *stp = (tsearch_readline_state *) arg;
 
    /*
     * We can't include the text of the config line for errors that occur
     * during tsearch_readline() itself.  The major cause of such errors is
     * encoding violations, and we daren't try to print error messages
     * containing badly-encoded data.
     */
    if (stp->curline)
        errcontext("line %d of configuration file \"%s\": \"%s\"",
                   stp->lineno,
                   stp->filename,
                   stp->curline);
    else
        errcontext("line %d of configuration file \"%s\"",
                   stp->lineno,
                   stp->filename);
}

References arg, tsearch_readline_state::curline, errcontext, tsearch_readline_state::filename, and tsearch_readline_state::lineno.

Referenced by tsearch_readline_begin().

tsearch_readline_end()

void tsearch_readline_end

(

tsearch_readline_state *

stp

)

Definition at line 204 of file ts_locale.c.

{
    /* Suppress use of curline in any error reported below */
    if (stp->curline)
    {
        if (stp->curline != stp->buf.data)
            pfree(stp->curline);
        stp->curline = NULL;
    }
 
    /* Release other resources */
    pfree(stp->buf.data);
    FreeFile(stp->fp);
 
    /* Pop the error context stack */
    error_context_stack = stp->cb.previous;
}

References tsearch_readline_state::buf, tsearch_readline_state::cb, tsearch_readline_state::curline, StringInfoData::data, error_context_stack, tsearch_readline_state::fp, FreeFile(), pfree(), and ErrorContextCallback::previous.

Referenced by dsynonym_init(), initTrie(), NIImportAffixes(), NIImportDictionary(), NIImportOOAffixes(), read_dictionary(), readstoplist(), and thesaurusRead().