codec_utf16.c

Go to the documentation of this file.

/*
 * This file is part of LibParserUtils.
 * Licensed under the MIT License,
 *                http://www.opensource.org/licenses/mit-license.php
 * Copyright 2007 John-Mark Bell <jmb@netsurf-browser.org>
 */

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

#include <parserutils/charset/mibenum.h>
#include <parserutils/charset/utf16.h>

#include "charset/codecs/codec_impl.h"
#include "utils/endian.h"
#include "utils/utils.h"

typedef struct charset_utf16_codec {
        parserutils_charset_codec base; 
#define INVAL_BUFSIZE (32)
        uint8_t inval_buf[INVAL_BUFSIZE];       
        size_t inval_len;               /*< Byte length of inval_buf **/

#define READ_BUFSIZE (8)
        uint32_t read_buf[READ_BUFSIZE];        
        size_t read_len;                
#define WRITE_BUFSIZE (8)
        uint32_t write_buf[WRITE_BUFSIZE];      
        size_t write_len;               
} charset_utf16_codec;

static bool charset_utf16_codec_handles_charset(const char *charset);
static parserutils_error charset_utf16_codec_create(const char *charset,
                parserutils_charset_codec **codec);
static parserutils_error charset_utf16_codec_destroy(
                parserutils_charset_codec *codec);
static parserutils_error charset_utf16_codec_encode(
                parserutils_charset_codec *codec,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen);
static parserutils_error charset_utf16_codec_decode(
                parserutils_charset_codec *codec,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen);
static parserutils_error charset_utf16_codec_reset(
                parserutils_charset_codec *codec);
static inline parserutils_error charset_utf16_codec_read_char(
                charset_utf16_codec *c,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen);
static inline parserutils_error charset_utf16_codec_output_decoded_char(
                charset_utf16_codec *c,
                uint32_t ucs4, uint8_t **dest, size_t *destlen);

bool charset_utf16_codec_handles_charset(const char *charset)
{
        return parserutils_charset_mibenum_from_name(charset, strlen(charset))
                ==
                parserutils_charset_mibenum_from_name("UTF-16", SLEN("UTF-16"));
}

parserutils_error charset_utf16_codec_create(const char *charset,
                parserutils_charset_codec **codec)
{
        charset_utf16_codec *c;

        UNUSED(charset);

        c = malloc(sizeof(charset_utf16_codec));
        if (c == NULL)
                return PARSERUTILS_NOMEM;

        c->inval_buf[0] = '\0';
        c->inval_len = 0;

        c->read_buf[0] = 0;
        c->read_len = 0;

        c->write_buf[0] = 0;
        c->write_len = 0;

        /* Finally, populate vtable */
        c->base.handler.destroy = charset_utf16_codec_destroy;
        c->base.handler.encode = charset_utf16_codec_encode;
        c->base.handler.decode = charset_utf16_codec_decode;
        c->base.handler.reset = charset_utf16_codec_reset;

        *codec = (parserutils_charset_codec *) c;

        return PARSERUTILS_OK;
}

parserutils_error charset_utf16_codec_destroy (parserutils_charset_codec *codec)
{
        UNUSED(codec);

        return PARSERUTILS_OK;
}

parserutils_error charset_utf16_codec_encode(parserutils_charset_codec *codec,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen)
{
        charset_utf16_codec *c = (charset_utf16_codec *) codec;
        uint32_t ucs4;
        uint32_t *towrite;
        size_t towritelen;
        parserutils_error error;

        /* Process any outstanding characters from the previous call */
        if (c->write_len > 0) {
                uint32_t *pwrite = c->write_buf;
                uint8_t buf[4];
                size_t len;

                while (c->write_len > 0) {
                        error = parserutils_charset_utf16_from_ucs4(
                                        pwrite[0], buf, &len);
                        assert(error == PARSERUTILS_OK);

                        if (*destlen < len) {
                                /* Insufficient output buffer space */
                                for (len = 0; len < c->write_len; len++)
                                        c->write_buf[len] = pwrite[len];

                                return PARSERUTILS_NOMEM;
                        }

                        memcpy(*dest, buf, len);

                        *dest += len;
                        *destlen -= len;

                        pwrite++;
                        c->write_len--;
                }
        }

        /* Now process the characters for this call */
        while (*sourcelen > 0) {
                ucs4 = endian_big_to_host(*((uint32_t *) (void *) *source));
                towrite = &ucs4;
                towritelen = 1;

                /* Output current characters */
                while (towritelen > 0) {
                        uint8_t buf[4];
                        size_t len;

                        error = parserutils_charset_utf16_from_ucs4(
                                        towrite[0], buf, &len);
                        assert(error == PARSERUTILS_OK);

                        if (*destlen < len) {
                                /* Insufficient output space */
                                assert(towritelen < WRITE_BUFSIZE);

                                c->write_len = towritelen;

                                /* Copy pending chars to save area, for
                                 * processing next call. */
                                for (len = 0; len < towritelen; len++)
                                        c->write_buf[len] = towrite[len];

                                /* Claim character we've just buffered,
                                 * so it's not reprocessed */
                                *source += 4;
                                *sourcelen -= 4;

                                return PARSERUTILS_NOMEM;
                        }

                        memcpy(*dest, buf, len);

                        *dest += len;
                        *destlen -= len;

                        towrite++;
                        towritelen--;
                }

                *source += 4;
                *sourcelen -= 4;
        }

        (void) error;

        return PARSERUTILS_OK;
}

parserutils_error charset_utf16_codec_decode(parserutils_charset_codec *codec,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen)
{
        charset_utf16_codec *c = (charset_utf16_codec *) codec;
        parserutils_error error;

        if (c->read_len > 0) {
                /* Output left over from last decode */
                uint32_t *pread = c->read_buf;

                while (c->read_len > 0 && *destlen >= c->read_len * 4) {
                        *((uint32_t *) (void *) *dest) =
                                        endian_host_to_big(pread[0]);

                        *dest += 4;
                        *destlen -= 4;

                        pread++;
                        c->read_len--;
                }

                if (*destlen < c->read_len * 4) {
                        /* Ran out of output buffer */
                        size_t i;

                        /* Shuffle remaining output down */
                        for (i = 0; i < c->read_len; i++)
                                c->read_buf[i] = pread[i];

                        return PARSERUTILS_NOMEM;
                }
        }

        if (c->inval_len > 0) {
                /* The last decode ended in an incomplete sequence.
                 * Fill up inval_buf with data from the start of the
                 * new chunk and process it. */
                uint8_t *in = c->inval_buf;
                size_t ol = c->inval_len;
                size_t l = min(INVAL_BUFSIZE - ol - 1, *sourcelen);
                size_t orig_l = l;

                memcpy(c->inval_buf + ol, *source, l);

                l += c->inval_len;

                error = charset_utf16_codec_read_char(c,
                                (const uint8_t **) &in, &l, dest, destlen);
                if (error != PARSERUTILS_OK && error != PARSERUTILS_NOMEM) {
                        return error;
                }

                /* And now, fix up source pointers */
                *source += max((signed) (orig_l - l), 0);
                *sourcelen -= max((signed) (orig_l - l), 0);

                /* Failed to resolve an incomplete character and
                 * ran out of buffer space. No recovery strategy
                 * possible, so explode everywhere. */
                assert((orig_l + ol) - l != 0);

                /* Report memory exhaustion case from above */
                if (error != PARSERUTILS_OK)
                        return error;
        }

        /* Finally, the "normal" case; process all outstanding characters */
        while (*sourcelen > 0) {
                error = charset_utf16_codec_read_char(c,
                                source, sourcelen, dest, destlen);
                if (error != PARSERUTILS_OK) {
                        return error;
                }
        }

        return PARSERUTILS_OK;
}

parserutils_error charset_utf16_codec_reset(parserutils_charset_codec *codec)
{
        charset_utf16_codec *c = (charset_utf16_codec *) codec;

        c->inval_buf[0] = '\0';
        c->inval_len = 0;

        c->read_buf[0] = 0;
        c->read_len = 0;

        c->write_buf[0] = 0;
        c->write_len = 0;

        return PARSERUTILS_OK;
}


parserutils_error charset_utf16_codec_read_char(charset_utf16_codec *c,
                const uint8_t **source, size_t *sourcelen,
                uint8_t **dest, size_t *destlen)
{
        uint32_t ucs4;
        size_t sucs4;
        parserutils_error error;

        /* Convert a single character */
        error = parserutils_charset_utf16_to_ucs4(*source, *sourcelen,
                        &ucs4, &sucs4);
        if (error == PARSERUTILS_OK) {
                /* Read a character */
                error = charset_utf16_codec_output_decoded_char(c,
                                ucs4, dest, destlen);
                if (error == PARSERUTILS_OK || error == PARSERUTILS_NOMEM) {
                        /* output succeeded; update source pointers */
                        *source += sucs4;
                        *sourcelen -= sucs4;
                }

                /* Clear inval buffer */
                c->inval_buf[0] = '\0';
                c->inval_len = 0;

                return error;
        } else if (error == PARSERUTILS_NEEDDATA) {
                /* Incomplete input sequence */
                assert(*sourcelen < INVAL_BUFSIZE);

                memmove(c->inval_buf, *source, *sourcelen);
                c->inval_buf[*sourcelen] = '\0';
                c->inval_len = *sourcelen;

                *source += *sourcelen;
                *sourcelen = 0;

                return PARSERUTILS_OK;
        } else if (error == PARSERUTILS_INVALID) {
                /* Illegal input sequence */
                uint32_t nextchar;

                /* Clear inval buffer */
                c->inval_buf[0] = '\0';
                c->inval_len = 0;

                /* Strict errormode; simply flag invalid character */
                if (c->base.errormode ==
                                PARSERUTILS_CHARSET_CODEC_ERROR_STRICT) {
                        return PARSERUTILS_INVALID;
                }

                /* Find next valid UTF-16 sequence.
                 * We're processing client-provided data, so let's
                 * be paranoid about its validity. */
                error = parserutils_charset_utf16_next_paranoid(
                                *source, *sourcelen, 0, &nextchar);
                if (error != PARSERUTILS_OK) {
                        if (error == PARSERUTILS_NEEDDATA) {
                                /* Need more data to be sure */
                                assert(*sourcelen < INVAL_BUFSIZE);

                                memmove(c->inval_buf, *source, *sourcelen);
                                c->inval_buf[*sourcelen] = '\0';
                                c->inval_len = *sourcelen;

                                *source += *sourcelen;
                                *sourcelen = 0;

                                nextchar = 0;
                        } else {
                                return error;
                        }
                }

                /* output U+FFFD and continue processing. */
                error = charset_utf16_codec_output_decoded_char(c,
                                0xFFFD, dest, destlen);
                if (error == PARSERUTILS_OK || error == PARSERUTILS_NOMEM) {
                        /* output succeeded; update source pointers */
                        *source += nextchar;
                        *sourcelen -= nextchar;
                }

                return error;
        }

        return PARSERUTILS_OK;
}

parserutils_error charset_utf16_codec_output_decoded_char(charset_utf16_codec *c,
                uint32_t ucs4, uint8_t **dest, size_t *destlen)
{
        if (*destlen < 4) {
                /* Run out of output buffer */
                c->read_len = 1;
                c->read_buf[0] = ucs4;

                return PARSERUTILS_NOMEM;
        }

        *((uint32_t *) (void *) *dest) = endian_host_to_big(ucs4);
        *dest += 4;
        *destlen -= 4;

        return PARSERUTILS_OK;
}


const parserutils_charset_handler charset_utf16_codec_handler = {
        charset_utf16_codec_handles_charset,
        charset_utf16_codec_create
};