encoding.h

Go to the documentation of this file.

 
#ifndef PRISM_ENCODING_H
#define PRISM_ENCODING_H
 
#include "prism/defines.h"
#include "prism/util/pm_strncasecmp.h"
 
#include <assert.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>
 
typedef struct {
    size_t (*char_width)(const uint8_t *b, ptrdiff_t n);
 
    size_t (*alpha_char)(const uint8_t *b, ptrdiff_t n);
 
    size_t (*alnum_char)(const uint8_t *b, ptrdiff_t n);
 
    bool (*isupper_char)(const uint8_t *b, ptrdiff_t n);
 
    const char *name;
 
    bool multibyte;
} pm_encoding_t;
 
#define PRISM_ENCODING_ALPHABETIC_BIT 1 << 0
 
#define PRISM_ENCODING_ALPHANUMERIC_BIT 1 << 1
 
#define PRISM_ENCODING_UPPERCASE_BIT 1 << 2
 
size_t pm_encoding_utf_8_char_width(const uint8_t *b, ptrdiff_t n);
 
size_t pm_encoding_utf_8_alpha_char(const uint8_t *b, ptrdiff_t n);
 
size_t pm_encoding_utf_8_alnum_char(const uint8_t *b, ptrdiff_t n);
 
bool pm_encoding_utf_8_isupper_char(const uint8_t *b, ptrdiff_t n);
 
extern const uint8_t pm_encoding_unicode_table[256];
 
typedef enum {
    PM_ENCODING_UTF_8 = 0,
    PM_ENCODING_US_ASCII,
    PM_ENCODING_ASCII_8BIT,
    PM_ENCODING_EUC_JP,
    PM_ENCODING_WINDOWS_31J,
 
// We optionally support excluding the full set of encodings to only support the
// minimum necessary to process Ruby code without encoding comments.
#ifndef PRISM_ENCODING_EXCLUDE_FULL
    PM_ENCODING_BIG5,
    PM_ENCODING_BIG5_HKSCS,
    PM_ENCODING_BIG5_UAO,
    PM_ENCODING_CESU_8,
    PM_ENCODING_CP51932,
    PM_ENCODING_CP850,
    PM_ENCODING_CP852,
    PM_ENCODING_CP855,
    PM_ENCODING_CP949,
    PM_ENCODING_CP950,
    PM_ENCODING_CP951,
    PM_ENCODING_EMACS_MULE,
    PM_ENCODING_EUC_JP_MS,
    PM_ENCODING_EUC_JIS_2004,
    PM_ENCODING_EUC_KR,
    PM_ENCODING_EUC_TW,
    PM_ENCODING_GB12345,
    PM_ENCODING_GB18030,
    PM_ENCODING_GB1988,
    PM_ENCODING_GB2312,
    PM_ENCODING_GBK,
    PM_ENCODING_IBM437,
    PM_ENCODING_IBM720,
    PM_ENCODING_IBM737,
    PM_ENCODING_IBM775,
    PM_ENCODING_IBM852,
    PM_ENCODING_IBM855,
    PM_ENCODING_IBM857,
    PM_ENCODING_IBM860,
    PM_ENCODING_IBM861,
    PM_ENCODING_IBM862,
    PM_ENCODING_IBM863,
    PM_ENCODING_IBM864,
    PM_ENCODING_IBM865,
    PM_ENCODING_IBM866,
    PM_ENCODING_IBM869,
    PM_ENCODING_ISO_8859_1,
    PM_ENCODING_ISO_8859_2,
    PM_ENCODING_ISO_8859_3,
    PM_ENCODING_ISO_8859_4,
    PM_ENCODING_ISO_8859_5,
    PM_ENCODING_ISO_8859_6,
    PM_ENCODING_ISO_8859_7,
    PM_ENCODING_ISO_8859_8,
    PM_ENCODING_ISO_8859_9,
    PM_ENCODING_ISO_8859_10,
    PM_ENCODING_ISO_8859_11,
    PM_ENCODING_ISO_8859_13,
    PM_ENCODING_ISO_8859_14,
    PM_ENCODING_ISO_8859_15,
    PM_ENCODING_ISO_8859_16,
    PM_ENCODING_KOI8_R,
    PM_ENCODING_KOI8_U,
    PM_ENCODING_MAC_CENT_EURO,
    PM_ENCODING_MAC_CROATIAN,
    PM_ENCODING_MAC_CYRILLIC,
    PM_ENCODING_MAC_GREEK,
    PM_ENCODING_MAC_ICELAND,
    PM_ENCODING_MAC_JAPANESE,
    PM_ENCODING_MAC_ROMAN,
    PM_ENCODING_MAC_ROMANIA,
    PM_ENCODING_MAC_THAI,
    PM_ENCODING_MAC_TURKISH,
    PM_ENCODING_MAC_UKRAINE,
    PM_ENCODING_SHIFT_JIS,
    PM_ENCODING_SJIS_DOCOMO,
    PM_ENCODING_SJIS_KDDI,
    PM_ENCODING_SJIS_SOFTBANK,
    PM_ENCODING_STATELESS_ISO_2022_JP,
    PM_ENCODING_STATELESS_ISO_2022_JP_KDDI,
    PM_ENCODING_TIS_620,
    PM_ENCODING_UTF8_MAC,
    PM_ENCODING_UTF8_DOCOMO,
    PM_ENCODING_UTF8_KDDI,
    PM_ENCODING_UTF8_SOFTBANK,
    PM_ENCODING_WINDOWS_1250,
    PM_ENCODING_WINDOWS_1251,
    PM_ENCODING_WINDOWS_1252,
    PM_ENCODING_WINDOWS_1253,
    PM_ENCODING_WINDOWS_1254,
    PM_ENCODING_WINDOWS_1255,
    PM_ENCODING_WINDOWS_1256,
    PM_ENCODING_WINDOWS_1257,
    PM_ENCODING_WINDOWS_1258,
    PM_ENCODING_WINDOWS_874,
#endif
 
    PM_ENCODING_MAXIMUM
} pm_encoding_type_t;
 
extern const pm_encoding_t pm_encodings[PM_ENCODING_MAXIMUM];
 
#define PM_ENCODING_UTF_8_ENTRY (&pm_encodings[PM_ENCODING_UTF_8])
 
#define PM_ENCODING_US_ASCII_ENTRY (&pm_encodings[PM_ENCODING_US_ASCII])
 
#define PM_ENCODING_ASCII_8BIT_ENTRY (&pm_encodings[PM_ENCODING_ASCII_8BIT])
 
#define PM_ENCODING_EUC_JP_ENTRY (&pm_encodings[PM_ENCODING_EUC_JP])
 
#define PM_ENCODING_WINDOWS_31J_ENTRY (&pm_encodings[PM_ENCODING_WINDOWS_31J])
 
const pm_encoding_t * pm_encoding_find(const uint8_t *start, const uint8_t *end);
 
#endif