path: root/jni/ruby/ext/psych/yaml/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)
{
    int first = 1;

    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 necessary. */

        if (!first || parser->raw_buffer.pointer == parser->raw_buffer.last) {
            if (!yaml_parser_update_raw_buffer(parser)) return 0;
        }
        first = 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 low, high;
            size_t k;
            size_t 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;
        }

    }

    if (parser->offset >= PTRDIFF_MAX)
        return yaml_parser_set_reader_error(parser, "input is too long",
                PTRDIFF_MAX, -1);

    return 1;
}