[libyaml.git] / src / reader.c


#include "yaml_private.h"

/*
 * Declarations.
 */

static int
yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem,
        size_t offset, int value);

static int
yaml_parser_update_raw_buffer(yaml_parser_t *parser);

static int
yaml_parser_determine_encoding(yaml_parser_t *parser);

YAML_DECLARE(int)
yaml_parser_update_buffer(yaml_parser_t *parser, size_t length);

/*
 * Set the reader error and return 0.
 */

static int
yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem,
        size_t offset, int value)
{
    parser->error = YAML_READER_ERROR;
    parser->problem = problem;
    parser->problem_offset = offset;
    parser->problem_value = value;

    return 0;
}

/*
 * Byte order marks.
 */

#define BOM_UTF8    "\xef\xbb\xbf"
#define BOM_UTF16LE "\xff\xfe"
#define BOM_UTF16BE "\xfe\xff"

/*
 * Determine the input stream encoding by checking the BOM symbol. If no BOM is
 * found, the UTF-8 encoding is assumed. Return 1 on success, 0 on failure.
 */

static int
yaml_parser_determine_encoding(yaml_parser_t *parser)
{
    /* Ensure that we had enough bytes in the raw buffer. */

    while (!parser->eof 
            && parser->raw_buffer.last - parser->raw_buffer.pointer < 3) {
        if (!yaml_parser_update_raw_buffer(parser)) {
            return 0;
        }
    }

    /* Determine the encoding. */

    if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
            && !memcmp(parser->raw_buffer.pointer, BOM_UTF16LE, 2)) {
        parser->encoding = YAML_UTF16LE_ENCODING;
        parser->raw_buffer.pointer += 2;
        parser->offset += 2;
    }
    else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
            && !memcmp(parser->raw_buffer.pointer, BOM_UTF16BE, 2)) {
        parser->encoding = YAML_UTF16BE_ENCODING;
        parser->raw_buffer.pointer += 2;
        parser->offset += 2;
    }
    else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 3
            && !memcmp(parser->raw_buffer.pointer, BOM_UTF8, 3)) {
        parser->encoding = YAML_UTF8_ENCODING;
        parser->raw_buffer.pointer += 3;
        parser->offset += 3;
    }
    else {
        parser->encoding = YAML_UTF8_ENCODING;
    }

    return 1;
}

/*
 * Update the raw buffer.
 */

static int
yaml_parser_update_raw_buffer(yaml_parser_t *parser)
{
    size_t size_read = 0;

    /* Return if the raw buffer is full. */

    if (parser->raw_buffer.start == parser->raw_buffer.pointer
            && parser->raw_buffer.last == parser->raw_buffer.end)
        return 1;

    /* Return on EOF. */

    if (parser->eof) return 1;

    /* Move the remaining bytes in the raw buffer to the beginning. */

    if (parser->raw_buffer.start < parser->raw_buffer.pointer
            && parser->raw_buffer.pointer < parser->raw_buffer.last) {
        memmove(parser->raw_buffer.start, parser->raw_buffer.pointer,
                parser->raw_buffer.last - parser->raw_buffer.pointer);
    }
    parser->raw_buffer.last -=
        parser->raw_buffer.pointer - parser->raw_buffer.start;
    parser->raw_buffer.pointer = parser->raw_buffer.start;

    /* Call the read handler to fill the buffer. */

    if (!parser->read_handler(parser->read_handler_data, parser->raw_buffer.last,
                parser->raw_buffer.end - parser->raw_buffer.last, &size_read)) {
        return yaml_parser_set_reader_error(parser, "Input error",
                parser->offset, -1);
    }
    parser->raw_buffer.last += size_read;
    if (!size_read) {
        parser->eof = 1;
    }

    return 1;
}

/*
 * Ensure that the buffer contains at least `length` characters.
 * Return 1 on success, 0 on failure.
 *
 * The length is supposed to be significantly less that the buffer size.
 */

YAML_DECLARE(int)
yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
{
    assert(parser->read_handler);   /* Read handler must be set. */

    /* If the EOF flag is set and the raw buffer is empty, do nothing. */

    if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last)
        return 1;

    /* Return if the buffer contains enough characters. */

    if (parser->unread >= length)
        return 1;

    /* Determine the input encoding if it is not known yet. */

    if (!parser->encoding) {
        if (!yaml_parser_determine_encoding(parser))
            return 0;
    }

    /* Move the unread characters to the beginning of the buffer. */

    if (parser->buffer.start < parser->buffer.pointer
            && parser->buffer.pointer < parser->buffer.last) {
        size_t size = parser->buffer.last - parser->buffer.pointer;
        memmove(parser->buffer.start, parser->buffer.pointer, size);
        parser->buffer.pointer = parser->buffer.start;
        parser->buffer.last = parser->buffer.start + size;
    }
    else if (parser->buffer.pointer == parser->buffer.last) {
        parser->buffer.pointer = parser->buffer.start;
        parser->buffer.last = parser->buffer.start;
    }

    /* Fill the buffer until it has enough characters. */

    while (parser->unread < length)
    {
        /* Fill the raw buffer. */

        if (!yaml_parser_update_raw_buffer(parser)) return 0;

        /* Decode the raw buffer. */

        while (parser->raw_buffer.pointer != parser->raw_buffer.last)
        {
            unsigned int value = 0, value2 = 0;
            int incomplete = 0;
            unsigned char octet;
            unsigned int width = 0;
            int k, low, high;
            int raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer;

            /* Decode the next character. */

            switch (parser->encoding)
            {
                case YAML_UTF8_ENCODING:

                    /*
                     * Decode a UTF-8 character.  Check RFC 3629
                     * (http://www.ietf.org/rfc/rfc3629.txt) for more details.
                     *
                     * The following table (taken from the RFC) is used for
                     * decoding.
                     *
                     *    Char. number range |        UTF-8 octet sequence
                     *      (hexadecimal)    |              (binary)
                     *   --------------------+------------------------------------
                     *   0000 0000-0000 007F | 0xxxxxxx
                     *   0000 0080-0000 07FF | 110xxxxx 10xxxxxx
                     *   0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
                     *   0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
                     *
                     * Additionally, the characters in the range 0xD800-0xDFFF
                     * are prohibited as they are reserved for use with UTF-16
                     * surrogate pairs.
                     */

                    /* Determine the length of the UTF-8 sequence. */

                    octet = parser->raw_buffer.pointer[0];
                    width = (octet & 0x80) == 0x00 ? 1 :
                            (octet & 0xE0) == 0xC0 ? 2 :
                            (octet & 0xF0) == 0xE0 ? 3 :
                            (octet & 0xF8) == 0xF0 ? 4 : 0;

                    /* Check if the leading octet is valid. */

                    if (!width)
                        return yaml_parser_set_reader_error(parser,
                                "Invalid leading UTF-8 octet",
                                parser->offset, octet);

                    /* Check if the raw buffer contains an incomplete character. */

                    if (width > raw_unread) {
                        if (parser->eof) {
                            return yaml_parser_set_reader_error(parser,
                                    "Incomplete UTF-8 octet sequence",
                                    parser->offset, -1);
                        }
                        incomplete = 1;
                        break;
                    }

                    /* Decode the leading octet. */

                    value = (octet & 0x80) == 0x00 ? octet & 0x7F :
                            (octet & 0xE0) == 0xC0 ? octet & 0x1F :
                            (octet & 0xF0) == 0xE0 ? octet & 0x0F :
                            (octet & 0xF8) == 0xF0 ? octet & 0x07 : 0;

                    /* Check and decode the trailing octets. */

                    for (k = 1; k < width; k ++)
                    {
                        octet = parser->raw_buffer.pointer[k];

                        /* Check if the octet is valid. */

                        if ((octet & 0xC0) != 0x80)
                            return yaml_parser_set_reader_error(parser,
                                    "Invalid trailing UTF-8 octet",
                                    parser->offset+k, octet);

                        /* Decode the octet. */

                        value = (value << 6) + (octet & 0x3F);
                    }

                    /* Check the length of the sequence against the value. */

                    if (!((width == 1) ||
                            (width == 2 && value >= 0x80) ||
                            (width == 3 && value >= 0x800) ||
                            (width == 4 && value >= 0x10000)))
                        return yaml_parser_set_reader_error(parser,
                                "Invalid length of a UTF-8 sequence",
                                parser->offset, -1);

                    /* Check the range of the value. */

                    if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF)
                        return yaml_parser_set_reader_error(parser,
                                "Invalid Unicode character",
                                parser->offset, value);

                    break;
                
                case YAML_UTF16LE_ENCODING:
                case YAML_UTF16BE_ENCODING:

                    low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1);
                    high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0);

                    /*
                     * The UTF-16 encoding is not as simple as one might
                     * naively think.  Check RFC 2781
                     * (http://www.ietf.org/rfc/rfc2781.txt).
                     *
                     * Normally, two subsequent bytes describe a Unicode
                     * character.  However a special technique (called a
                     * surrogate pair) is used for specifying character
                     * values larger than 0xFFFF.
                     *
                     * A surrogate pair consists of two pseudo-characters:
                     *      high surrogate area (0xD800-0xDBFF)
                     *      low surrogate area (0xDC00-0xDFFF)
                     *
                     * The following formulas are used for decoding
                     * and encoding characters using surrogate pairs:
                     * 
                     *  U  = U' + 0x10000   (0x01 00 00 <= U <= 0x10 FF FF)
                     *  U' = yyyyyyyyyyxxxxxxxxxx   (0 <= U' <= 0x0F FF FF)
                     *  W1 = 110110yyyyyyyyyy
                     *  W2 = 110111xxxxxxxxxx
                     *
                     * where U is the character value, W1 is the high surrogate
                     * area, W2 is the low surrogate area.
                     */

                    /* Check for incomplete UTF-16 character. */

                    if (raw_unread < 2) {
                        if (parser->eof) {
                            return yaml_parser_set_reader_error(parser,
                                    "Incomplete UTF-16 character",
                                    parser->offset, -1);
                        }
                        incomplete = 1;
                        break;
                    }

                    /* Get the character. */

                    value = parser->raw_buffer.pointer[low]
                        + (parser->raw_buffer.pointer[high] << 8);

                    /* Check for unexpected low surrogate area. */

                    if ((value & 0xFC00) == 0xDC00)
                        return yaml_parser_set_reader_error(parser,
                                "Unexpected low surrogate area",
                                parser->offset, value);

                    /* Check for a high surrogate area. */

                    if ((value & 0xFC00) == 0xD800) {

                        width = 4;

                        /* Check for incomplete surrogate pair. */

                        if (raw_unread < 4) {
                            if (parser->eof) {
                                return yaml_parser_set_reader_error(parser,
                                        "Incomplete UTF-16 surrogate pair",
                                        parser->offset, -1);
                            }
                            incomplete = 1;
                            break;
                        }

                        /* Get the next character. */

                        value2 = parser->raw_buffer.pointer[low+2]
                            + (parser->raw_buffer.pointer[high+2] << 8);

                        /* Check for a low surrogate area. */

                        if ((value2 & 0xFC00) != 0xDC00)
                            return yaml_parser_set_reader_error(parser,
                                    "Expected low surrogate area",
                                    parser->offset+2, value2);

                        /* Generate the value of the surrogate pair. */

                        value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF);
                    }

                    else {
                        width = 2;
                    }

                    break;

                default:
                    assert(1);      /* Impossible. */
            }

            /* Check if the raw buffer contains enough bytes to form a character. */

            if (incomplete) break;

            /*
             * Check if the character is in the allowed range:
             *      #x9 | #xA | #xD | [#x20-#x7E]               (8 bit)
             *      | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD]    (16 bit)
             *      | [#x10000-#x10FFFF]                        (32 bit)
             */

            if (! (value == 0x09 || value == 0x0A || value == 0x0D
                        || (value >= 0x20 && value <= 0x7E)
                        || (value == 0x85) || (value >= 0xA0 && value <= 0xD7FF)
                        || (value >= 0xE000 && value <= 0xFFFD)
                        || (value >= 0x10000 && value <= 0x10FFFF)))
                return yaml_parser_set_reader_error(parser,
                        "Control characters are not allowed",
                        parser->offset, value);

            /* Move the raw pointers. */

            parser->raw_buffer.pointer += width;
            parser->offset += width;

            /* Finally put the character into the buffer. */

            /* 0000 0000-0000 007F -> 0xxxxxxx */
            if (value <= 0x7F) {
                *(parser->buffer.last++) = value;
            }
            /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
            else if (value <= 0x7FF) {
                *(parser->buffer.last++) = 0xC0 + (value >> 6);
                *(parser->buffer.last++) = 0x80 + (value & 0x3F);
            }
            /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
            else if (value <= 0xFFFF) {
                *(parser->buffer.last++) = 0xE0 + (value >> 12);
                *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
                *(parser->buffer.last++) = 0x80 + (value & 0x3F);
            }
            /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
            else {
                *(parser->buffer.last++) = 0xF0 + (value >> 18);
                *(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F);
                *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
                *(parser->buffer.last++) = 0x80 + (value & 0x3F);
            }

            parser->unread ++;
        }

        /* On EOF, put NUL into the buffer and return. */

        if (parser->eof) {
            *(parser->buffer.last++) = '\0';
            parser->unread ++;
            return 1;
        }

    }

    return 1;
}
Commit	Line	Data
95b98ba9	1
625fcfe9	2	#include "yaml_private.h"
6eb1ded4	3
625fcfe9 KS	4	/*
	5	* Declarations.
	6	*/
	7
	8	static int
	9	yaml_parser_set_reader_error(yaml_parser_t parser, const char problem,
	10	size_t offset, int value);
6eb1ded4	11
625fcfe9 KS	12	static int
	13	yaml_parser_update_raw_buffer(yaml_parser_t *parser);
	14
	15	static int
	16	yaml_parser_determine_encoding(yaml_parser_t *parser);
	17
	18	YAML_DECLARE(int)
	19	yaml_parser_update_buffer(yaml_parser_t *parser, size_t length);
6eb1ded4	20
6eb1ded4	21	/*
3d790dfd	22	* Set the reader error and return 0.
6eb1ded4 KS	23	*/
6eb1ded4 KS	24
f642fd11	25	static int
2fa16060 KS	26	yaml_parser_set_reader_error(yaml_parser_t parser, const char problem,
2fa16060 KS	27	size_t offset, int value)
3d790dfd KS	28	{
	29	parser->error = YAML_READER_ERROR;
	30	parser->problem = problem;
2fa16060 KS	31	parser->problem_offset = offset;
2fa16060 KS	32	parser->problem_value = value;
6eb1ded4	33
3d790dfd KS	34	return 0;
3d790dfd KS	35	}
6eb1ded4	36
f642fd11	37	/*
625fcfe9	38	* Byte order marks.
f642fd11 KS	39	*/
f642fd11 KS	40
625fcfe9 KS	41	#define BOM_UTF8 "\xef\xbb\xbf"
	42	#define BOM_UTF16LE "\xff\xfe"
	43	#define BOM_UTF16BE "\xfe\xff"
f642fd11 KS	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
f642fd11 KS	50	static int
	51	yaml_parser_determine_encoding(yaml_parser_t *parser)
	52	{
	53	/* Ensure that we had enough bytes in the raw buffer. */
	54
625fcfe9 KS	55	while (!parser->eof
625fcfe9 KS	56	&& parser->raw_buffer.last - parser->raw_buffer.pointer < 3) {
f642fd11 KS	57	if (!yaml_parser_update_raw_buffer(parser)) {
	58	return 0;
	59	}
	60	}
	61
	62	/* Determine the encoding. */
	63
625fcfe9 KS	64	if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
625fcfe9 KS	65	&& !memcmp(parser->raw_buffer.pointer, BOM_UTF16LE, 2)) {
f642fd11	66	parser->encoding = YAML_UTF16LE_ENCODING;
625fcfe9	67	parser->raw_buffer.pointer += 2;
f642fd11 KS	68	parser->offset += 2;
f642fd11 KS	69	}
625fcfe9 KS	70	else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
625fcfe9 KS	71	&& !memcmp(parser->raw_buffer.pointer, BOM_UTF16BE, 2)) {
f642fd11	72	parser->encoding = YAML_UTF16BE_ENCODING;
625fcfe9	73	parser->raw_buffer.pointer += 2;
f642fd11 KS	74	parser->offset += 2;
f642fd11 KS	75	}
625fcfe9 KS	76	else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 3
625fcfe9 KS	77	&& !memcmp(parser->raw_buffer.pointer, BOM_UTF8, 3)) {
f642fd11	78	parser->encoding = YAML_UTF8_ENCODING;
625fcfe9	79	parser->raw_buffer.pointer += 3;
f642fd11 KS	80	parser->offset += 3;
	81	}
	82	else {
	83	parser->encoding = YAML_UTF8_ENCODING;
	84	}
	85
	86	return 1;
	87	}
95b98ba9 KS	88
95b98ba9 KS	89	/*
625fcfe9 KS	90	* Update the raw buffer.
	91	*/
	92
	93	static int
	94	yaml_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.
95b98ba9	136	* Return 1 on success, 0 on failure.
6eb1ded4 KS	137	*
6eb1ded4 KS	138	* The length is supposed to be significantly less that the buffer size.
95b98ba9 KS	139	*/
95b98ba9 KS	140
f642fd11	141	YAML_DECLARE(int)
6eb1ded4	142	yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
95b98ba9	143	{
625fcfe9 KS	144	assert(parser->read_handler); /* Read handler must be set. */
625fcfe9 KS	145
2fa16060	146	/* If the EOF flag is set and the raw buffer is empty, do nothing. */
95b98ba9	147
625fcfe9	148	if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last)
95b98ba9 KS	149	return 1;
95b98ba9 KS	150
6eb1ded4 KS	151	/* Return if the buffer contains enough characters. */
	152
	153	if (parser->unread >= length)
	154	return 1;
	155
	156	/* Determine the input encoding if it is not known yet. */
95b98ba9	157
6eb1ded4 KS	158	if (!parser->encoding) {
6eb1ded4 KS	159	if (!yaml_parser_determine_encoding(parser))
95b98ba9	160	return 0;
95b98ba9 KS	161	}
95b98ba9 KS	162
6eb1ded4 KS	163	/* Move the unread characters to the beginning of the buffer. */
6eb1ded4 KS	164
625fcfe9 KS	165	if (parser->buffer.start < parser->buffer.pointer
	166	&& parser->buffer.pointer < parser->buffer.last) {
	167	size_t size = parser->buffer.last - parser->buffer.pointer;
	168	memmove(parser->buffer.start, parser->buffer.pointer, size);
	169	parser->buffer.pointer = parser->buffer.start;
	170	parser->buffer.last = parser->buffer.start + size;
6eb1ded4	171	}
625fcfe9 KS	172	else if (parser->buffer.pointer == parser->buffer.last) {
	173	parser->buffer.pointer = parser->buffer.start;
	174	parser->buffer.last = parser->buffer.start;
6eb1ded4 KS	175	}
	176
	177	/* Fill the buffer until it has enough characters. */
	178
	179	while (parser->unread < length)
	180	{
	181	/* Fill the raw buffer. */
	182
	183	if (!yaml_parser_update_raw_buffer(parser)) return 0;
	184
	185	/* Decode the raw buffer. */
	186
625fcfe9	187	while (parser->raw_buffer.pointer != parser->raw_buffer.last)
6eb1ded4	188	{
54815ffd	189	unsigned int value = 0, value2 = 0;
6eb1ded4	190	int incomplete = 0;
2fa16060	191	unsigned char octet;
54815ffd	192	unsigned int width = 0;
3d790dfd	193	int k, low, high;
625fcfe9	194	int raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer;
6eb1ded4 KS	195
	196	/* Decode the next character. */
	197
	198	switch (parser->encoding)
	199	{
	200	case YAML_UTF8_ENCODING:
	201
3d790dfd KS	202	/*
	203	* Decode a UTF-8 character. Check RFC 3629
	204	* (http://www.ietf.org/rfc/rfc3629.txt) for more details.
	205	*
	206	* The following table (taken from the RFC) is used for
	207	* decoding.
	208	*
	209	* Char. number range \| UTF-8 octet sequence
	210	* (hexadecimal) \| (binary)
	211	* --------------------+------------------------------------
	212	* 0000 0000-0000 007F \| 0xxxxxxx
	213	* 0000 0080-0000 07FF \| 110xxxxx 10xxxxxx
	214	* 0000 0800-0000 FFFF \| 1110xxxx 10xxxxxx 10xxxxxx
	215	* 0001 0000-0010 FFFF \| 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
	216	*
	217	* Additionally, the characters in the range 0xD800-0xDFFF
	218	* are prohibited as they are reserved for use with UTF-16
	219	* surrogate pairs.
	220	*/
	221
	222	/* Determine the length of the UTF-8 sequence. */
	223
625fcfe9	224	octet = parser->raw_buffer.pointer[0];
2fa16060 KS	225	width = (octet & 0x80) == 0x00 ? 1 :
	226	(octet & 0xE0) == 0xC0 ? 2 :
	227	(octet & 0xF0) == 0xE0 ? 3 :
	228	(octet & 0xF8) == 0xF0 ? 4 : 0;
3d790dfd KS	229
	230	/* Check if the leading octet is valid. */
	231
2fa16060	232	if (!width)
3d790dfd	233	return yaml_parser_set_reader_error(parser,
2fa16060 KS	234	"Invalid leading UTF-8 octet",
2fa16060 KS	235	parser->offset, octet);
6eb1ded4 KS	236
	237	/* Check if the raw buffer contains an incomplete character. */
	238
625fcfe9	239	if (width > raw_unread) {
6eb1ded4	240	if (parser->eof) {
3d790dfd	241	return yaml_parser_set_reader_error(parser,
2fa16060 KS	242	"Incomplete UTF-8 octet sequence",
2fa16060 KS	243	parser->offset, -1);
6eb1ded4 KS	244	}
6eb1ded4 KS	245	incomplete = 1;
3d790dfd	246	break;
6eb1ded4 KS	247	}
6eb1ded4 KS	248
3d790dfd	249	/* Decode the leading octet. */
6eb1ded4	250
2fa16060 KS	251	value = (octet & 0x80) == 0x00 ? octet & 0x7F :
	252	(octet & 0xE0) == 0xC0 ? octet & 0x1F :
	253	(octet & 0xF0) == 0xE0 ? octet & 0x0F :
	254	(octet & 0xF8) == 0xF0 ? octet & 0x07 : 0;
3d790dfd KS	255
	256	/* Check and decode the trailing octets. */
	257
2fa16060	258	for (k = 1; k < width; k ++)
3d790dfd	259	{
625fcfe9	260	octet = parser->raw_buffer.pointer[k];
3d790dfd KS	261
	262	/* Check if the octet is valid. */
	263
2fa16060	264	if ((octet & 0xC0) != 0x80)
3d790dfd	265	return yaml_parser_set_reader_error(parser,
2fa16060 KS	266	"Invalid trailing UTF-8 octet",
2fa16060 KS	267	parser->offset+k, octet);
3d790dfd KS	268
	269	/* Decode the octet. */
	270
2fa16060	271	value = (value << 6) + (octet & 0x3F);
6eb1ded4 KS	272	}
6eb1ded4 KS	273
3d790dfd KS	274	/* Check the length of the sequence against the value. */
3d790dfd KS	275
2fa16060 KS	276	if (!((width == 1) \|\|
	277	(width == 2 && value >= 0x80) \|\|
	278	(width == 3 && value >= 0x800) \|\|
	279	(width == 4 && value >= 0x10000)))
3d790dfd	280	return yaml_parser_set_reader_error(parser,
2fa16060 KS	281	"Invalid length of a UTF-8 sequence",
2fa16060 KS	282	parser->offset, -1);
3d790dfd KS	283
	284	/* Check the range of the value. */
	285
	286	if ((value >= 0xD800 && value <= 0xDFFF) \|\| value > 0x10FFFF)
	287	return yaml_parser_set_reader_error(parser,
2fa16060 KS	288	"Invalid Unicode character",
2fa16060 KS	289	parser->offset, value);
3d790dfd	290
6eb1ded4 KS	291	break;
	292
	293	case YAML_UTF16LE_ENCODING:
3d790dfd	294	case YAML_UTF16BE_ENCODING:
6eb1ded4	295
3d790dfd KS	296	low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1);
	297	high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0);
	298
	299	/*
	300	* The UTF-16 encoding is not as simple as one might
	301	* naively think. Check RFC 2781
	302	* (http://www.ietf.org/rfc/rfc2781.txt).
	303	*
	304	* Normally, two subsequent bytes describe a Unicode
	305	* character. However a special technique (called a
	306	* surrogate pair) is used for specifying character
	307	* values larger than 0xFFFF.
	308	*
	309	* A surrogate pair consists of two pseudo-characters:
	310	* high surrogate area (0xD800-0xDBFF)
	311	* low surrogate area (0xDC00-0xDFFF)
	312	*
	313	* The following formulas are used for decoding
	314	* and encoding characters using surrogate pairs:
	315	*
	316	* U = U' + 0x10000 (0x01 00 00 <= U <= 0x10 FF FF)
	317	* U' = yyyyyyyyyyxxxxxxxxxx (0 <= U' <= 0x0F FF FF)
	318	* W1 = 110110yyyyyyyyyy
	319	* W2 = 110111xxxxxxxxxx
	320	*
	321	* where U is the character value, W1 is the high surrogate
	322	* area, W2 is the low surrogate area.
	323	*/
	324
	325	/* Check for incomplete UTF-16 character. */
6eb1ded4	326
625fcfe9	327	if (raw_unread < 2) {
6eb1ded4	328	if (parser->eof) {
3d790dfd	329	return yaml_parser_set_reader_error(parser,
2fa16060 KS	330	"Incomplete UTF-16 character",
2fa16060 KS	331	parser->offset, -1);
6eb1ded4 KS	332	}
6eb1ded4 KS	333	incomplete = 1;
3d790dfd	334	break;
6eb1ded4 KS	335	}
6eb1ded4 KS	336
3d790dfd	337	/* Get the character. */
6eb1ded4	338
625fcfe9 KS	339	value = parser->raw_buffer.pointer[low]
625fcfe9 KS	340	+ (parser->raw_buffer.pointer[high] << 8);
6eb1ded4	341
3d790dfd	342	/* Check for unexpected low surrogate area. */
6eb1ded4	343
3d790dfd KS	344	if ((value & 0xFC00) == 0xDC00)
3d790dfd KS	345	return yaml_parser_set_reader_error(parser,
2fa16060 KS	346	"Unexpected low surrogate area",
2fa16060 KS	347	parser->offset, value);
6eb1ded4	348
3d790dfd	349	/* Check for a high surrogate area. */
6eb1ded4	350
3d790dfd KS	351	if ((value & 0xFC00) == 0xD800) {
3d790dfd KS	352
2fa16060 KS	353	width = 4;
2fa16060 KS	354
3d790dfd KS	355	/* Check for incomplete surrogate pair. */
3d790dfd KS	356
625fcfe9	357	if (raw_unread < 4) {
3d790dfd KS	358	if (parser->eof) {
3d790dfd KS	359	return yaml_parser_set_reader_error(parser,
2fa16060 KS	360	"Incomplete UTF-16 surrogate pair",
2fa16060 KS	361	parser->offset, -1);
3d790dfd KS	362	}
	363	incomplete = 1;
	364	break;
6eb1ded4	365	}
6eb1ded4	366
3d790dfd	367	/* Get the next character. */
6eb1ded4	368
54815ffd	369	value2 = parser->raw_buffer.pointer[low+2]
625fcfe9	370	+ (parser->raw_buffer.pointer[high+2] << 8);
3d790dfd KS	371
	372	/* Check for a low surrogate area. */
	373
	374	if ((value2 & 0xFC00) != 0xDC00)
	375	return yaml_parser_set_reader_error(parser,
2fa16060 KS	376	"Expected low surrogate area",
2fa16060 KS	377	parser->offset+2, value2);
3d790dfd KS	378
	379	/* Generate the value of the surrogate pair. */
	380
	381	value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF);
3d790dfd KS	382	}
	383
	384	else {
2fa16060	385	width = 2;
3d790dfd	386	}
6eb1ded4 KS	387
6eb1ded4 KS	388	break;
54815ffd KS	389
	390	default:
	391	assert(1); /* Impossible. */
6eb1ded4 KS	392	}
6eb1ded4 KS	393
3d790dfd KS	394	/* Check if the raw buffer contains enough bytes to form a character. */
	395
	396	if (incomplete) break;
	397
6eb1ded4 KS	398	/*
	399	* Check if the character is in the allowed range:
	400	* #x9 \| #xA \| #xD \| [#x20-#x7E] (8 bit)
	401	* \| #x85 \| [#xA0-#xD7FF] \| [#xE000-#xFFFD] (16 bit)
	402	* \| [#x10000-#x10FFFF] (32 bit)
	403	*/
	404
3d790dfd KS	405	if (! (value == 0x09 \|\| value == 0x0A \|\| value == 0x0D
	406	\|\| (value >= 0x20 && value <= 0x7E)
	407	\|\| (value == 0x85) \|\| (value >= 0xA0 && value <= 0xD7FF)
	408	\|\| (value >= 0xE000 && value <= 0xFFFD)
	409	\|\| (value >= 0x10000 && value <= 0x10FFFF)))
	410	return yaml_parser_set_reader_error(parser,
2fa16060 KS	411	"Control characters are not allowed",
	412	parser->offset, value);
	413
	414	/* Move the raw pointers. */
	415
625fcfe9	416	parser->raw_buffer.pointer += width;
2fa16060	417	parser->offset += width;
6eb1ded4 KS	418
	419	/* Finally put the character into the buffer. */
	420
	421	/* 0000 0000-0000 007F -> 0xxxxxxx */
3d790dfd	422	if (value <= 0x7F) {
625fcfe9	423	*(parser->buffer.last++) = value;
6eb1ded4 KS	424	}
6eb1ded4 KS	425	/* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
3d790dfd	426	else if (value <= 0x7FF) {
625fcfe9 KS	427	*(parser->buffer.last++) = 0xC0 + (value >> 6);
625fcfe9 KS	428	*(parser->buffer.last++) = 0x80 + (value & 0x3F);
6eb1ded4 KS	429	}
6eb1ded4 KS	430	/* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
3d790dfd	431	else if (value <= 0xFFFF) {
625fcfe9 KS	432	*(parser->buffer.last++) = 0xE0 + (value >> 12);
	433	*(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
	434	*(parser->buffer.last++) = 0x80 + (value & 0x3F);
6eb1ded4 KS	435	}
	436	/* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
	437	else {
625fcfe9 KS	438	*(parser->buffer.last++) = 0xF0 + (value >> 18);
	439	*(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F);
	440	*(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
	441	*(parser->buffer.last++) = 0x80 + (value & 0x3F);
6eb1ded4	442	}
2fa16060 KS	443
	444	parser->unread ++;
	445	}
	446
	447	/* On EOF, put NUL into the buffer and return. */
	448
	449	if (parser->eof) {
625fcfe9	450	*(parser->buffer.last++) = '\0';
2fa16060 KS	451	parser->unread ++;
2fa16060 KS	452	return 1;
95b98ba9	453	}
2fa16060	454
95b98ba9 KS	455	}
95b98ba9 KS	456
3d790dfd	457	return 1;
6eb1ded4	458	}
95b98ba9	459