/src/libyaml/reader.c

  1
  2#include "yaml_private.h"
  3
  4/*
  5 * Declarations.
  6 */
  7
  8static int
  9yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem,
 10        size_t offset, int value);
 11
 12static int
 13yaml_parser_update_raw_buffer(yaml_parser_t *parser);
 14
 15static int
 16yaml_parser_determine_encoding(yaml_parser_t *parser);
 17
 18YAML_DECLARE(int)
 19yaml_parser_update_buffer(yaml_parser_t *parser, size_t length);
 20
 21/*
 22 * Set the reader error and return 0.
 23 */
 24
 25static int
 26yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem,
 27        size_t offset, int value)
 28{
 29    parser->error = YAML_READER_ERROR;
 30    parser->problem = problem;
 31    parser->problem_offset = offset;
 32    parser->problem_value = value;
 33
 34    return 0;
 35}
 36
 37/*
 38 * Byte order marks.
 39 */
 40
 41#define BOM_UTF8    "\xef\xbb\xbf"
 42#define BOM_UTF16LE "\xff\xfe"
 43#define BOM_UTF16BE "\xfe\xff"
 44
 45/*
 46 * Determine the input stream encoding by checking the BOM symbol. If no BOM is
 47 * found, the UTF-8 encoding is assumed. Return 1 on success, 0 on failure.
 48 */
 49
 50static int
 51yaml_parser_determine_encoding(yaml_parser_t *parser)
 52{
 53    /* Ensure that we had enough bytes in the raw buffer. */
 54
 55    while (!parser->eof 
 56            && parser->raw_buffer.last - parser->raw_buffer.pointer < 3) {
 57        if (!yaml_parser_update_raw_buffer(parser)) {
 58            return 0;
 59        }
 60    }
 61
 62    /* Determine the encoding. */
 63
 64    if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
 65            && !memcmp(parser->raw_buffer.pointer, BOM_UTF16LE, 2)) {
 66        parser->encoding = YAML_UTF16LE_ENCODING;
 67        parser->raw_buffer.pointer += 2;
 68        parser->offset += 2;
 69    }
 70    else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
 71            && !memcmp(parser->raw_buffer.pointer, BOM_UTF16BE, 2)) {
 72        parser->encoding = YAML_UTF16BE_ENCODING;
 73        parser->raw_buffer.pointer += 2;
 74        parser->offset += 2;
 75    }
 76    else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 3
 77            && !memcmp(parser->raw_buffer.pointer, BOM_UTF8, 3)) {
 78        parser->encoding = YAML_UTF8_ENCODING;
 79        parser->raw_buffer.pointer += 3;
 80        parser->offset += 3;
 81    }
 82    else {
 83        parser->encoding = YAML_UTF8_ENCODING;
 84    }
 85
 86    return 1;
 87}
 88
 89/*
 90 * Update the raw buffer.
 91 */
 92
 93static int
 94yaml_parser_update_raw_buffer(yaml_parser_t *parser)
 95{
 96    size_t size_read = 0;
 97
 98    /* Return if the raw buffer is full. */
 99
100    if (parser->raw_buffer.start == parser->raw_buffer.pointer
101            && parser->raw_buffer.last == parser->raw_buffer.end)
102        return 1;
103
104    /* Return on EOF. */
105
106    if (parser->eof) return 1;
107
108    /* Move the remaining bytes in the raw buffer to the beginning. */
109
110    if (parser->raw_buffer.start < parser->raw_buffer.pointer
111            && parser->raw_buffer.pointer < parser->raw_buffer.last) {
112        memmove(parser->raw_buffer.start, parser->raw_buffer.pointer,
113                parser->raw_buffer.last - parser->raw_buffer.pointer);
114    }
115    parser->raw_buffer.last -=
116        parser->raw_buffer.pointer - parser->raw_buffer.start;
117    parser->raw_buffer.pointer = parser->raw_buffer.start;
118
119    /* Call the read handler to fill the buffer. */
120
121    if (!parser->read_handler(parser->read_handler_data, parser->raw_buffer.last,
122                parser->raw_buffer.end - parser->raw_buffer.last, &size_read)) {
123        return yaml_parser_set_reader_error(parser, "input error",
124                parser->offset, -1);
125    }
126    parser->raw_buffer.last += size_read;
127    if (!size_read) {
128        parser->eof = 1;
129    }
130
131    return 1;
132}
133
134/*
135 * Ensure that the buffer contains at least `length` characters.
136 * Return 1 on success, 0 on failure.
137 *
138 * The length is supposed to be significantly less that the buffer size.
139 */
140
141YAML_DECLARE(int)
142yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
143{
144    int first = 1;
145
146    assert(parser->read_handler);   /* Read handler must be set. */
147
148    /* If the EOF flag is set and the raw buffer is empty, do nothing. */
149
150    if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last)
151        return 1;
152
153    /* Return if the buffer contains enough characters. */
154
155    if (parser->unread >= length)
156        return 1;
157
158    /* Determine the input encoding if it is not known yet. */
159
160    if (!parser->encoding) {
161        if (!yaml_parser_determine_encoding(parser))
162            return 0;
163    }
164
165    /* Move the unread characters to the beginning of the buffer. */
166
167    if (parser->buffer.start < parser->buffer.pointer
168            && parser->buffer.pointer < parser->buffer.last) {
169        size_t size = parser->buffer.last - parser->buffer.pointer;
170        memmove(parser->buffer.start, parser->buffer.pointer, size);
171        parser->buffer.pointer = parser->buffer.start;
172        parser->buffer.last = parser->buffer.start + size;
173    }
174    else if (parser->buffer.pointer == parser->buffer.last) {
175        parser->buffer.pointer = parser->buffer.start;
176        parser->buffer.last = parser->buffer.start;
177    }
178
179    /* Fill the buffer until it has enough characters. */
180
181    while (parser->unread < length)
182    {
183        /* Fill the raw buffer if necessary. */
184
185        if (!first || parser->raw_buffer.pointer == parser->raw_buffer.last) {
186            if (!yaml_parser_update_raw_buffer(parser)) return 0;
187        }
188        first = 0;
189
190        /* Decode the raw buffer. */
191
192        while (parser->raw_buffer.pointer != parser->raw_buffer.last)
193        {
194            unsigned int value = 0, value2 = 0;
195            unsigned char octet;
196            unsigned int width = 0;
197            int low, high;
198            size_t raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer;
199
200            /* Decode the next character. */
201
202            switch (parser->encoding)
203            {
204                case YAML_UTF8_ENCODING:
205                    octet = parser->raw_buffer.pointer[0];
206                    /*
207                      We can only disallow characters without the high
208                      bit set. Characters with the high bit set are required
209                      for invalid UTF-8 strings to be encoded, as we
210                      cannot rely on any backslash sequences working.
211                    */
212                    if (! (octet == 0x09 || octet == 0x0A || octet == 0x0D
213                           || (octet >= 0x20 && octet != 0x7F) ))
214                        return yaml_parser_set_reader_error(parser,
215                                "Control characters are not allowed",
216                                parser->offset, value);
217
218                    parser->raw_buffer.pointer++;
219                    parser->offset++;
220                    *(parser->buffer.last++) = octet;
221                    break;
222                
223                case YAML_UTF16LE_ENCODING:
224                case YAML_UTF16BE_ENCODING:
225
226                    low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1);
227                    high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0);
228
229                    /*
230                     * The UTF-16 encoding is not as simple as one might
231                     * naively think.  Check RFC 2781
232                     * (http://www.ietf.org/rfc/rfc2781.txt).
233                     *
234                     * Normally, two subsequent bytes describe a Unicode
235                     * character.  However a special technique (called a
236                     * surrogate pair) is used for specifying character
237                     * values larger than 0xFFFF.
238                     *
239                     * A surrogate pair consists of two pseudo-characters:
240                     *      high surrogate area (0xD800-0xDBFF)
241                     *      low surrogate area (0xDC00-0xDFFF)
242                     *
243                     * The following formulas are used for decoding
244                     * and encoding characters using surrogate pairs:
245                     * 
246                     *  U  = U' + 0x10000   (0x01 00 00 <= U <= 0x10 FF FF)
247                     *  U' = yyyyyyyyyyxxxxxxxxxx   (0 <= U' <= 0x0F FF FF)
248                     *  W1 = 110110yyyyyyyyyy
249                     *  W2 = 110111xxxxxxxxxx
250                     *
251                     * where U is the character value, W1 is the high surrogate
252                     * area, W2 is the low surrogate area.
253                     */
254
255                    /* Check for incomplete UTF-16 character. */
256
257                    if (raw_unread < 2) {
258                        if (parser->eof) {
259                            return yaml_parser_set_reader_error(parser,
260                                    "incomplete UTF-16 character",
261                                    parser->offset, -1);
262                        }
263                        break;
264                    }
265
266                    /* Get the character. */
267
268                    value = parser->raw_buffer.pointer[low]
269                        + (parser->raw_buffer.pointer[high] << 8);
270                    width = 2;
271
272                    /* Check for a high surrogate area. */
273
274                    if ((value & 0xFC00) == 0xD800) {
275
276                        /* Check for incomplete surrogate pair. */
277
278                        if (raw_unread < 4) {
279                            if (parser->eof) { /* trailing high surrogate */
280                                width = 2;
281                            } else {
282                                break; /* Can't tell until we have more raw characters */
283                            }
284                        } else {
285
286                            /* Get the next character. */
287
288                            value2 = parser->raw_buffer.pointer[low+2]
289                                + (parser->raw_buffer.pointer[high+2] << 8);
290
291                            /* Check for a low surrogate area. */
292                            if ((value2 & 0xFC00) == 0xDC00) {
293                                width = 4;
294                                /* Generate the value of the surrogate pair. */
295                                value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF);
296                            }
297                        }
298                    }
299
300                    /* Check if the raw buffer contains enough bytes to form a character. */
301
302                    /*
303                     * Check if the character is in the allowed range:
304                     *      #x9 | #xA | #xD | [#x20-#x7E]               (8 bit)
305                     *      | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD]    (16 bit)
306                     *      | [#x10000-#x10FFFF]                        (32 bit)
307                     */
308                    /* Modified to allow all 16-bit values as \uNNNN may not
309                     work for some parsers for these values. */
310                    if (! (value == 0x09 || value == 0x0A || value == 0x0D
311                           || (value >= 0x20 && value <= 0x7E)
312                           || (value == 0x85) || value >= 0xA0))
313                        return yaml_parser_set_reader_error(parser,
314                                "Control characters are not allowed",
315                                parser->offset, value);
316
317                    /* Move the raw pointers. */
318
319                    parser->raw_buffer.pointer += width;
320                    parser->offset += width;
321
322                    /* Finally put the character into the buffer. */
323
324                    /* 0000 0000-0000 007F -> 0xxxxxxx */
325                    if (value <= 0x7F) {
326                        *(parser->buffer.last++) = value;
327                    }
328                    /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
329                    else if (value <= 0x7FF) {
330                        *(parser->buffer.last++) = 0xC0 + (value >> 6);
331                        *(parser->buffer.last++) = 0x80 + (value & 0x3F);
332                    }
333                    /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
334                    else if (value <= 0xFFFF) {
335                        *(parser->buffer.last++) = 0xE0 + (value >> 12);
336                        *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
337                        *(parser->buffer.last++) = 0x80 + (value & 0x3F);
338                    }
339                    /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
340                    else {
341                        *(parser->buffer.last++) = 0xF0 + (value >> 18);
342                        *(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F);
343                        *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
344                        *(parser->buffer.last++) = 0x80 + (value & 0x3F);
345                    }
346                    break;
347
348                default:
349                    assert(1);      /* Impossible. */
350            }
351
352            parser->unread ++;
353        }
354
355        /* On EOF, put NUL into the buffer and return. */
356
357        if (parser->eof) {
358            *(parser->buffer.last++) = '\0';
359            parser->unread ++;
360            return 1;
361        }
362
363    }
364
365    return 1;
366}
367