summaryrefslogtreecommitdiff
path: root/src/utils/encoding.cpp
blob: dc0101c8cdc9261ec499d303a13faecb590030ba (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
#include <utils/encoding.hpp>

#include <utils/scopeguard.hpp>

#include <stdexcept>

#include <assert.h>
#include <string.h>
#include <iconv.h>

#include <config.h>

#include <bitset>

/**
 * The UTF-8-encoded character used as a place holder when a character conversion fails.
 * This is U+FFFD � "replacement character"
 */
static const char* invalid_char = "\xef\xbf\xbd";
static const size_t invalid_char_len = 3;

namespace utils
{
  /**
   * Based on http://en.wikipedia.org/wiki/UTF-8#Description
   */
  bool is_valid_utf8(const char* s)
  {
    if (!s)
      return false;

    const unsigned char* str = reinterpret_cast<const unsigned char*>(s);

    while (*str)
      {
        // 4 bytes:  11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
        if ((str[0] & 0b11111000) == 0b11110000)
          {
            if (!str[1] || !str[2] || !str[3]
                || ((str[1] & 0b11000000) != 0b10000000)
                || ((str[2] & 0b11000000) != 0b10000000)
                || ((str[3] & 0b11000000) != 0b10000000))
              return false;
            str += 4;
          }
        // 3 bytes:  1110xxx 10xxxxxx 10xxxxxx
        else if ((str[0] & 0b11110000) == 0b11100000)
          {
            if (!str[1] || !str[2]
                || ((str[1] & 0b11000000) != 0b10000000)
                || ((str[2] & 0b11000000) != 0b10000000))
              return false;
            str += 3;
          }
        // 2 bytes:  110xxxxx 10xxxxxx
        else if (((str[0]) & 0b11100000) == 0b11000000)
          {
            if (!str[1] ||
                ((str[1] & 0b11000000) != 0b10000000))
              return false;
            str += 2;
          }
        // 1 byte:  0xxxxxxx
        else if ((str[0] & 0b10000000) != 0)
          return false;
        else
          str++;
      }
    return true;
  }

  std::string remove_invalid_xml_chars(const std::string& original)
  {
    // The given string MUST be a valid utf-8 string
    unsigned char* res = new unsigned char[original.size()];
    ScopeGuard sg([&res]() { delete[] res;});

    // pointer where we write valid chars
    unsigned char* r = res;

    const unsigned char* str = reinterpret_cast<const unsigned char*>(original.c_str());
    std::bitset<20> codepoint;

    while (*str)
      {
        // 4 bytes:  11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
        if ((str[0] & 0b11111000) == 0b11110000)
          {
            codepoint  = ((str[0] & 0b00000111) << 18);
            codepoint |= ((str[1] & 0b00111111) << 12);
            codepoint |= ((str[2] & 0b00111111) << 6 );
            codepoint |= ((str[3] & 0b00111111) << 0 );
            if (codepoint.to_ulong() <= 0x10FFFF)
              {
                ::memcpy(r, str, 4);
                r += 4;
              }
            str += 4;
          }
        // 3 bytes:  1110xxx 10xxxxxx 10xxxxxx
        else if ((str[0] & 0b11110000) == 0b11100000)
          {
            codepoint  = ((str[0] & 0b00001111) << 12);
            codepoint |= ((str[1] & 0b00111111) << 6);
            codepoint |= ((str[2] & 0b00111111) << 0 );
            if (codepoint.to_ulong() <= 0xD7FF ||
                (codepoint.to_ulong() >= 0xE000 && codepoint.to_ulong() <= 0xFFFD))
              {
                ::memcpy(r, str, 3);
                r += 3;
              }
            str += 3;
          }
        // 2 bytes:  110xxxxx 10xxxxxx
        else if (((str[0]) & 0b11100000) == 0b11000000)
          {
            // All 2 bytes char are valid, don't even bother calculating
            // the codepoint
            ::memcpy(r, str, 2);
            r += 2;
            str += 2;
          }
        // 1 byte:  0xxxxxxx
        else if ((str[0] & 0b10000000) == 0)
          {
            codepoint = ((str[0] & 0b01111111));
            if (codepoint.to_ulong() == 0x09 ||
                codepoint.to_ulong() == 0x0A ||
                codepoint.to_ulong() == 0x0D ||
                codepoint.to_ulong() >= 0x20)
              {
                ::memcpy(r, str, 1);
                r += 1;
              }
            str += 1;
          }
        else
          throw std::runtime_error("Invalid UTF-8 passed to remove_invalid_xml_chars");
      }
    return std::string(reinterpret_cast<char*>(res), r-res);
  }

  std::string convert_to_utf8(const std::string& str, const char* charset)
  {
    std::string res;

    const iconv_t cd = iconv_open("UTF-8", charset);
    if (cd == (iconv_t)-1)
      throw std::runtime_error("Cannot convert into UTF-8");

    // Make sure cd is always closed when we leave this function
    ScopeGuard sg([&]{ iconv_close(cd); });

    size_t inbytesleft = str.size();

    // iconv will not attempt to modify this buffer, but some plateform
    // require a char** anyway
#ifdef ICONV_SECOND_ARGUMENT_IS_CONST
    const char* inbuf_ptr = str.c_str();
#else
    char* inbuf_ptr = const_cast<char*>(str.c_str());
#endif

    size_t outbytesleft = str.size() * 4;
    char* outbuf = new char[outbytesleft];
    char* outbuf_ptr = outbuf;

    // Make sure outbuf is always deleted when we leave this function
    sg.add_callback([&]{ delete[] outbuf; });

    bool done = false;
    while (done == false)
      {
        size_t error = iconv(cd, &inbuf_ptr, &inbytesleft, &outbuf_ptr, &outbytesleft);
        if ((size_t)-1 == error)
          {
            switch (errno)
              {
              case EILSEQ:
                // Invalid byte found. Insert a placeholder instead of the
                // converted character, jump one byte and continue
                memcpy(outbuf_ptr, invalid_char, invalid_char_len);
                outbuf_ptr += invalid_char_len;
                inbytesleft--;
                inbuf_ptr++;
                break;
              case EINVAL:
                // A multibyte sequence is not terminated, but we can't
                // provide any more data, so we just add a placeholder to
                // indicate that the character is not properly converted,
                // and we stop the conversion
                memcpy(outbuf_ptr, invalid_char, invalid_char_len);
                outbuf_ptr += invalid_char_len;
                outbuf_ptr++;
                done = true;
                break;
              case E2BIG:
                // This should never happen
                done = true;
              default:
                // This should happen even neverer
                done = true;
                break;
              }
          }
        else
          {
            // The conversion finished without any error, stop converting
            done = true;
          }
      }
    // Terminate the converted buffer, and copy that buffer it into the
    // string we return
    *outbuf_ptr = '\0';
    res = outbuf;
    return res;
  }

}