marc8.c

harvest是一个下载html网页得机器人

  {"\x6F", 0x00F2},
  {"\xAC", 0x1EDC},
  {"\xBC", 0x1EDD},
  {"\xE5\x4F", 0x1E50},
  {"\xE5\x6F", 0x1E51},
  {"\x55", 0x00D9},
  {"\x75", 0x00F9},
  {"\xAD", 0x1EEA},
  {"\xBD", 0x1EEB},
  {"\xE8\x55", 0x01DB},
  {"\xE8\x75", 0x01DC},
  {"\x57", 0x1E80},
  {"\x77", 0x1E81},
  {"\x59", 0x1EF2},
  {"\x79", 0x1EF3},
  {0, 0}
};
struct yaz_iconv_trie marc8page3 = {
  marc8page3_flat, 0
};
/* PAGE 2 */
struct yaz_iconv_trie_flat marc8page2_flat[] = {
  {"\xF2\x41", 0x1EB6},
  {"\xF2\x61", 0x1EB7},
  {"\x41", 0x0102},
  {"\x61", 0x0103},
  {"\x45", 0x0114},
  {"\x65", 0x0115},
  {"\xF0\x45", 0x1E1C},
  {"\xF0\x65", 0x1E1D},
  {"\x47", 0x011E},
  {"\x67", 0x011F},
  {"\x49", 0x012C},
  {"\x69", 0x012D},
  {"\x4F", 0x014E},
  {"\x6F", 0x014F},
  {"\x55", 0x016C},
  {"\x75", 0x016D},
  {0, 0}
};
struct yaz_iconv_trie marc8page2 = {
  marc8page2_flat, 0
};
/* PAGE 1 */
struct yaz_iconv_trie_dir marc8page1_dir[256] = {
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x00C1} /* 41 */,
  {0, 0},
  {0, 0x0106} /* 43 */,
  {0, 0},
  {0, 0x00C9} /* 45 */,
  {0, 0},
  {0, 0x01F4} /* 47 */,
  {0, 0},
  {0, 0x00CD} /* 49 */,
  {0, 0},
  {0, 0x1E30} /* 4B */,
  {0, 0x0139} /* 4C */,
  {0, 0x1E3E} /* 4D */,
  {0, 0x0143} /* 4E */,
  {0, 0x00D3} /* 4F */,
  {0, 0x1E54} /* 50 */,
  {0, 0},
  {0, 0x0154} /* 52 */,
  {0, 0x015A} /* 53 */,
  {0, 0},
  {0, 0x00DA} /* 55 */,
  {0, 0},
  {0, 0x1E82} /* 57 */,
  {0, 0},
  {0, 0x00DD} /* 59 */,
  {0, 0x0179} /* 5A */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x00E1} /* 61 */,
  {0, 0},
  {0, 0x0107} /* 63 */,
  {0, 0},
  {0, 0x00E9} /* 65 */,
  {0, 0},
  {0, 0x01F5} /* 67 */,
  {0, 0},
  {0, 0x00ED} /* 69 */,
  {0, 0},
  {0, 0x1E31} /* 6B */,
  {0, 0x013A} /* 6C */,
  {0, 0x1E3F} /* 6D */,
  {0, 0x0144} /* 6E */,
  {0, 0x00F3} /* 6F */,
  {0, 0x1E55} /* 70 */,
  {0, 0},
  {0, 0x0155} /* 72 */,
  {0, 0x015B} /* 73 */,
  {0, 0},
  {0, 0x00FA} /* 75 */,
  {0, 0},
  {0, 0x1E83} /* 77 */,
  {0, 0},
  {0, 0x00FD} /* 79 */,
  {0, 0x017A} /* 7A */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x01FE} /* A2 */,
  {0, 0},
  {0, 0},
  {0, 0x01FC} /* A5 */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x1EDA} /* AC */,
  {0, 0x1EE8} /* AD */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x01FF} /* B2 */,
  {0, 0},
  {0, 0},
  {0, 0x01FD} /* B5 */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x1EDB} /* BC */,
  {0, 0x1EE9} /* BD */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page31, 0} /* E4 */,
  {&marc8page29, 0} /* E5 */,
  {&marc8page26, 0} /* E6 */,
  {0, 0},
  {&marc8page30, 0} /* E8 */,
  {0, 0},
  {&marc8page27, 0} /* EA */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page28, 0} /* F0 */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0}
};
struct yaz_iconv_trie marc8page1 = {
  0, marc8page1_dir
};
/* PAGE 0 */
struct yaz_iconv_trie_dir marc8page0_dir[256] = {
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page11, 0} /* 1B */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page20, 0} /* 4C */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page21, 0} /* 6C */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0x0141} /* A1 */,
  {0, 0x00D8} /* A2 */,
  {0, 0x0110} /* A3 */,
  {0, 0x00DE} /* A4 */,
  {0, 0x00C6} /* A5 */,
  {0, 0x0152} /* A6 */,
  {0, 0x02B9} /* A7 */,
  {0, 0x00B7} /* A8 */,
  {0, 0x266D} /* A9 */,
  {0, 0x00AE} /* AA */,
  {0, 0x00B1} /* AB */,
  {0, 0x01A0} /* AC */,
  {0, 0x01AF} /* AD */,
  {0, 0x02BE} /* AE */,
  {0, 0},
  {0, 0x02BF} /* B0 */,
  {0, 0x0142} /* B1 */,
  {0, 0x00F8} /* B2 */,
  {0, 0x0111} /* B3 */,
  {0, 0x00FE} /* B4 */,
  {0, 0x00E6} /* B5 */,
  {0, 0x0153} /* B6 */,
  {0, 0x02BA} /* B7 */,
  {0, 0x0131} /* B8 */,
  {0, 0x00A3} /* B9 */,
  {0, 0x00F0} /* BA */,
  {0, 0},
  {0, 0x01A1} /* BC */,
  {0, 0x01B0} /* BD */,
  {0, 0},
  {0, 0},
  {0, 0x00B0} /* C0 */,
  {0, 0x2113} /* C1 */,
  {0, 0x2117} /* C2 */,
  {0, 0x00A9} /* C3 */,
  {0, 0x266F} /* C4 */,
  {0, 0x00BF} /* C5 */,
  {0, 0x00A1} /* C6 */,
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {0, 0},
  {&marc8page4, 0x0309} /* E0 */,
  {&marc8page3, 0x0300} /* E1 */,
  {&marc8page1, 0x0301} /* E2 */,
  {&marc8page7, 0x0302} /* E3 */,
  {&marc8page5, 0x0303} /* E4 */,
  {&marc8page8, 0x0304} /* E5 */,
  {&marc8page2, 0x0306} /* E6 */,
  {&marc8page10, 0x0307} /* E7 */,
  {&marc8page14, 0x0308} /* E8 */,
  {&marc8page6, 0x030C} /* E9 */,
  {&marc8page13, 0x030A} /* EA */,
  {0, 0xFE20} /* EB */,
  {0, 0xFE21} /* EC */,
  {&marc8page22, 0x0315} /* ED */,
  {&marc8page23, 0x030B} /* EE */,
  {0, 0x0310} /* EF */,
  {&marc8page16, 0x0327} /* F0 */,
  {&marc8page12, 0x0328} /* F1 */,
  {&marc8page9, 0x0323} /* F2 */,
  {&marc8page25, 0x0324} /* F3 */,
  {0, 0x0325} /* F4 */,
  {&marc8page18, 0x0333} /* F5 */,
  {&marc8page15, 0x0332} /* F6 */,
  {&marc8page17, 0x0326} /* F7 */,
  {&marc8page24, 0x031C} /* F8 */,
  {&marc8page19, 0x032E} /* F9 */,
  {0, 0xFE22} /* FA */,
  {0, 0xFE23} /* FB */,
  {0, 0},
  {0, 0},
  {0, 0x0313} /* FE */,
  {0, 0}
};
struct yaz_iconv_trie marc8page0 = {
  0, marc8page0_dir
};

        static unsigned long lookup(struct yaz_iconv_trie *t, unsigned char *inp,
                                    size_t inbytesleft, size_t *no_read)
        {
            if (!t || inbytesleft < 1)
            return 0;
            if (t->dir)
            {
                size_t ch = inp[0] & 0xff;
                unsigned long code =
                lookup(t->dir[ch].ptr, inp+1, inbytesleft-1, no_read);
                if (code)
                {
                    (*no_read)++;
                    return code;
                }
                if (t->dir[ch].to)
                {
                    code = t->dir[ch].to;
                    *no_read = 1;
                    return code;
                }
            }
            else
            {
                struct yaz_iconv_trie_flat *flat = t->flat;
                while (flat->from)
                {
                    size_t len = strlen(flat->from);
                    if (len <= inbytesleft)
                    {
                        if (memcmp(flat->from, inp, len) == 0)
                        {
                            *no_read = len;
                            return flat->to;
                        }
                    }
                    flat++;
                }
            }
            return 0;
        }
        
    
unsigned long yaz_marc8_conv
            (unsigned char *inp, size_t inbytesleft, size_t *no_read)
        {
            unsigned long code;
            
            code = lookup(&marc8page0, inp, inbytesleft, no_read);
            if (!code)
            {
                *no_read = 1;
                code = *inp;
            }
            return code;
        }