zip_utf-8.c File Reference

#include "zipint.h"
#include <stdlib.h>

Go to the source code of this file.

Macros
#define	UTF_8_LEN_2_MASK   0xe0
#define	UTF_8_LEN_2_MATCH   0xc0
#define	UTF_8_LEN_3_MASK   0xf0
#define	UTF_8_LEN_3_MATCH   0xe0
#define	UTF_8_LEN_4_MASK   0xf8
#define	UTF_8_LEN_4_MATCH   0xf0
#define	UTF_8_CONTINUE_MASK   0xc0
#define	UTF_8_CONTINUE_MATCH   0x80

Functions
zip_encoding_type_t	_zip_guess_encoding (zip_string_t *str, zip_encoding_type_t expected_encoding)
static zip_uint32_t	_zip_unicode_to_utf8_len (zip_uint32_t codepoint)
static zip_uint32_t	_zip_unicode_to_utf8 (zip_uint32_t codepoint, zip_uint8_t *buf)
zip_uint8_t *	_zip_cp437_to_utf8 (const zip_uint8_t *const _cp437buf, zip_uint32_t len, zip_uint32_t *utf8_lenp, zip_error_t *error)

Variables
static const zip_uint16_t	_cp437_to_unicode [256]

Macro Definition Documentation

UTF_8_CONTINUE_MASK

#define UTF_8_CONTINUE_MASK   0xc0

Definition at line 95 of file zip_utf-8.c.

UTF_8_CONTINUE_MATCH

#define UTF_8_CONTINUE_MATCH   0x80

Definition at line 96 of file zip_utf-8.c.

UTF_8_LEN_2_MASK

#define UTF_8_LEN_2_MASK   0xe0

Definition at line 89 of file zip_utf-8.c.

UTF_8_LEN_2_MATCH

#define UTF_8_LEN_2_MATCH   0xc0

Definition at line 90 of file zip_utf-8.c.

UTF_8_LEN_3_MASK

#define UTF_8_LEN_3_MASK   0xf0

Definition at line 91 of file zip_utf-8.c.

UTF_8_LEN_3_MATCH

#define UTF_8_LEN_3_MATCH   0xe0

Definition at line 92 of file zip_utf-8.c.

UTF_8_LEN_4_MASK

#define UTF_8_LEN_4_MASK   0xf8

Definition at line 93 of file zip_utf-8.c.

UTF_8_LEN_4_MATCH

#define UTF_8_LEN_4_MATCH   0xf0

Definition at line 94 of file zip_utf-8.c.

Function Documentation

_zip_cp437_to_utf8()

zip_uint8_t* _zip_cp437_to_utf8	(	const zip_uint8_t *const	_cp437buf,
		zip_uint32_t	len,
		zip_uint32_t *	utf8_lenp,
		zip_error_t *	error
	)

Definition at line 198 of file zip_utf-8.c.

                                                                                                                      {
    zip_uint8_t *cp437buf = (zip_uint8_t *)_cp437buf;
    zip_uint8_t *utf8buf;
    zip_uint32_t buflen, i, offset;
 
    if (len == 0) {
        if (utf8_lenp)
            *utf8_lenp = 0;
        return NULL;
    }
 
    buflen = 1;
    for (i = 0; i < len; i++)
        buflen += _zip_unicode_to_utf8_len(_cp437_to_unicode[cp437buf[i]]);
 
    if ((utf8buf = (zip_uint8_t *)malloc(buflen)) == NULL) {
        zip_error_set(error, ZIP_ER_MEMORY, 0);
        return NULL;
    }
 
    offset = 0;
    for (i = 0; i < len; i++)
        offset += _zip_unicode_to_utf8(_cp437_to_unicode[cp437buf[i]], utf8buf + offset);
 
    utf8buf[buflen - 1] = 0;
    if (utf8_lenp)
        *utf8_lenp = buflen - 1;
    return utf8buf;
}

References _cp437_to_unicode, _zip_unicode_to_utf8(), _zip_unicode_to_utf8_len(), buflen, error(), i, len, malloc(), NULL, ZIP_ER_MEMORY, and zip_error_set().

Referenced by _zip_string_get().

_zip_guess_encoding()

zip_encoding_type_t _zip_guess_encoding	(	zip_string_t *	str,
		zip_encoding_type_t	expected_encoding
	)

Definition at line 100 of file zip_utf-8.c.

                                                                              {
    zip_encoding_type_t enc;
    const zip_uint8_t *name;
    zip_uint32_t i, j, ulen;
 
    if (str == NULL)
        return ZIP_ENCODING_ASCII;
 
    name = str->raw;
 
    if (str->encoding != ZIP_ENCODING_UNKNOWN)
        enc = str->encoding;
    else {
        enc = ZIP_ENCODING_ASCII;
        for (i = 0; i < str->length; i++) {
            if ((name[i] > 31 && name[i] < 128) || name[i] == '\r' || name[i] == '\n' || name[i] == '\t')
                continue;
 
            enc = ZIP_ENCODING_UTF8_GUESSED;
            if ((name[i] & UTF_8_LEN_2_MASK) == UTF_8_LEN_2_MATCH)
                ulen = 1;
            else if ((name[i] & UTF_8_LEN_3_MASK) == UTF_8_LEN_3_MATCH)
                ulen = 2;
            else if ((name[i] & UTF_8_LEN_4_MASK) == UTF_8_LEN_4_MATCH)
                ulen = 3;
            else {
                enc = ZIP_ENCODING_CP437;
                break;
            }
 
            if (i + ulen >= str->length) {
                enc = ZIP_ENCODING_CP437;
                break;
            }
 
            for (j = 1; j <= ulen; j++) {
                if ((name[i + j] & UTF_8_CONTINUE_MASK) != UTF_8_CONTINUE_MATCH) {
                    enc = ZIP_ENCODING_CP437;
                    goto done;
                }
            }
            i += ulen;
        }
    }
 
done:
    str->encoding = enc;
 
    if (expected_encoding != ZIP_ENCODING_UNKNOWN) {
        if (expected_encoding == ZIP_ENCODING_UTF8_KNOWN && enc == ZIP_ENCODING_UTF8_GUESSED)
            str->encoding = enc = ZIP_ENCODING_UTF8_KNOWN;
 
        if (expected_encoding != enc && enc != ZIP_ENCODING_ASCII)
            return ZIP_ENCODING_ERROR;
    }
 
    return enc;
}

References done, i, name, NULL, cmd_descs_generate::str, UTF_8_CONTINUE_MASK, UTF_8_CONTINUE_MATCH, UTF_8_LEN_2_MASK, UTF_8_LEN_2_MATCH, UTF_8_LEN_3_MASK, UTF_8_LEN_3_MATCH, UTF_8_LEN_4_MASK, UTF_8_LEN_4_MATCH, ZIP_ENCODING_ASCII, ZIP_ENCODING_CP437, ZIP_ENCODING_ERROR, ZIP_ENCODING_UNKNOWN, ZIP_ENCODING_UTF8_GUESSED, and ZIP_ENCODING_UTF8_KNOWN.

Referenced by _zip_dirent_read(), _zip_dirent_write(), _zip_set_name(), _zip_string_get(), _zip_string_new(), zip_file_set_comment(), and zip_set_archive_comment().

_zip_unicode_to_utf8()

static zip_uint32_t _zip_unicode_to_utf8 ( zip_uint32_t  codepoint, zip_uint8_t *  buf  )

static

Definition at line 173 of file zip_utf-8.c.

                                                               {
    if (codepoint < 0x0080) {
        buf[0] = codepoint & 0xff;
        return 1;
    }
    if (codepoint < 0x0800) {
        buf[0] = (zip_uint8_t)(UTF_8_LEN_2_MATCH | ((codepoint >> 6) & 0x1f));
        buf[1] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | (codepoint & 0x3f));
        return 2;
    }
    if (codepoint < 0x10000) {
        buf[0] = (zip_uint8_t)(UTF_8_LEN_3_MATCH | ((codepoint >> 12) & 0x0f));
        buf[1] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | ((codepoint >> 6) & 0x3f));
        buf[2] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | (codepoint & 0x3f));
        return 3;
    }
    buf[0] = (zip_uint8_t)(UTF_8_LEN_4_MATCH | ((codepoint >> 18) & 0x07));
    buf[1] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | ((codepoint >> 12) & 0x3f));
    buf[2] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | ((codepoint >> 6) & 0x3f));
    buf[3] = (zip_uint8_t)(UTF_8_CONTINUE_MATCH | (codepoint & 0x3f));
    return 4;
}

References UTF_8_CONTINUE_MATCH, UTF_8_LEN_2_MATCH, UTF_8_LEN_3_MATCH, and UTF_8_LEN_4_MATCH.

Referenced by _zip_cp437_to_utf8().

_zip_unicode_to_utf8_len()

static zip_uint32_t _zip_unicode_to_utf8_len ( zip_uint32_t  codepoint )

static

Definition at line 161 of file zip_utf-8.c.

                                                 {
    if (codepoint < 0x0080)
        return 1;
    if (codepoint < 0x0800)
        return 2;
    if (codepoint < 0x10000)
        return 3;
    return 4;
}

Referenced by _zip_cp437_to_utf8().

Variable Documentation

_cp437_to_unicode

const zip_uint16_t _cp437_to_unicode[256]

static

Definition at line 40 of file zip_utf-8.c.

Referenced by _zip_cp437_to_utf8().