ngx_http_charset_filter_module.c

nginx 反向代理0.7.1版本 用于实现反向代理

            b->shadow = NULL;

            return cl;
        }
    }

    cl = ngx_alloc_chain_link(pool);
    if (cl == NULL) {
        return NULL;
    }

    cl->buf = ngx_create_temp_buf(pool, size);
    if (cl->buf == NULL) {
        return NULL;
    }

    cl->next = NULL;

    cl->buf->temporary = 1;
    cl->buf->tag = (ngx_buf_tag_t) &ngx_http_charset_filter_module;

    return cl;
}


static char *
ngx_http_charset_map_block(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
{
    ngx_http_charset_main_conf_t  *mcf = conf;

    char                         *rv;
    u_char                       *p, *dst2src, **pp;
    ngx_int_t                     src, dst;
    ngx_uint_t                    i, n;
    ngx_str_t                    *value;
    ngx_conf_t                    pvcf;
    ngx_http_charset_t           *charset;
    ngx_http_charset_tables_t    *table;
    ngx_http_charset_conf_ctx_t   ctx;

    value = cf->args->elts;

    src = ngx_http_add_charset(&mcf->charsets, &value[1]);
    if (src == NGX_ERROR) {
        return NGX_CONF_ERROR;
    }

    dst = ngx_http_add_charset(&mcf->charsets, &value[2]);
    if (dst == NGX_ERROR) {
        return NGX_CONF_ERROR;
    }

    if (src == dst) {
        ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                           "\"charset_map\" between the same charsets "
                           "\"%V\" and \"%V\"", &value[1], &value[2]);
        return NGX_CONF_ERROR;
    }

    table = mcf->tables.elts;
    for (i = 0; i < mcf->tables.nelts; i++) {
        if ((src == table->src && dst == table->dst)
             || (src == table->dst && dst == table->src))
        {
            ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                               "duplicate \"charset_map\" between "
                               "\"%V\" and \"%V\"", &value[1], &value[2]);
            return NGX_CONF_ERROR;
        }
    }

    table = ngx_array_push(&mcf->tables);
    if (table == NULL) {
        return NGX_CONF_ERROR;
    }

    table->src = src;
    table->dst = dst;

    if (ngx_strcasecmp(value[2].data, (u_char *) "utf-8") == 0) {
        table->src2dst = ngx_pcalloc(cf->pool, 256 * NGX_UTF_LEN);
        if (table->src2dst == NULL) {
            return NGX_CONF_ERROR;
        }

        table->dst2src = ngx_pcalloc(cf->pool, 256 * sizeof(void *));
        if (table->dst2src == NULL) {
            return NGX_CONF_ERROR;
        }

        dst2src = ngx_pcalloc(cf->pool, 256);
        if (dst2src == NULL) {
            return NGX_CONF_ERROR;
        }

        pp = (u_char **) &table->dst2src[0];
        pp[0] = dst2src;

        for (i = 0; i < 128; i++) {
            p = &table->src2dst[i * NGX_UTF_LEN];
            p[0] = '\1';
            p[1] = (u_char) i;
            dst2src[i] = (u_char) i;
        }

        for (/* void */; i < 256; i++) {
            p = &table->src2dst[i * NGX_UTF_LEN];
            p[0] = '\1';
            p[1] = '?';
        }

    } else {
        table->src2dst = ngx_palloc(cf->pool, 256);
        if (table->src2dst == NULL) {
            return NGX_CONF_ERROR;
        }

        table->dst2src = ngx_palloc(cf->pool, 256);
        if (table->dst2src == NULL) {
            return NGX_CONF_ERROR;
        }

        for (i = 0; i < 128; i++) {
            table->src2dst[i] = (u_char) i;
            table->dst2src[i] = (u_char) i;
        }

        for (/* void */; i < 256; i++) {
            table->src2dst[i] = '?';
            table->dst2src[i] = '?';
        }
    }

    charset = mcf->charsets.elts;

    ctx.table = table;
    ctx.charset = &charset[dst];
    ctx.characters = 0;

    pvcf = *cf;
    cf->ctx = &ctx;
    cf->handler = ngx_http_charset_map;
    cf->handler_conf = conf;

    rv = ngx_conf_parse(cf, NULL);

    *cf = pvcf;

    if (ctx.characters) {
        n = ctx.charset->length;
        ctx.charset->length /= ctx.characters;

        if (((n * 10) / ctx.characters) % 10 > 4) {
            ctx.charset->length++;
        }
    }

    return rv;
}


static char *
ngx_http_charset_map(ngx_conf_t *cf, ngx_command_t *dummy, void *conf)
{
    u_char                       *p, *dst2src, **pp;
    uint32_t                      n;
    ngx_int_t                     src, dst;
    ngx_str_t                    *value;
    ngx_uint_t                    i;
    ngx_http_charset_tables_t    *table;
    ngx_http_charset_conf_ctx_t  *ctx;

    if (cf->args->nelts != 2) {
        ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, "invalid parameters number");
        return NGX_CONF_ERROR;
    }

    value = cf->args->elts;

    src = ngx_hextoi(value[0].data, value[0].len);
    if (src == NGX_ERROR || src > 255) {
        ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                           "invalid value \"%V\"", &value[0]);
        return NGX_CONF_ERROR;
    }

    ctx = cf->ctx;
    table = ctx->table;

    if (ctx->charset->utf8) {
        p = &table->src2dst[src * NGX_UTF_LEN];

        *p++ = (u_char) (value[1].len / 2);

        for (i = 0; i < value[1].len; i += 2) {
            dst = ngx_hextoi(&value[1].data[i], 2);
            if (dst == NGX_ERROR || dst > 255) {
                ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                                   "invalid value \"%V\"", &value[1]);
                return NGX_CONF_ERROR;
            }

            *p++ = (u_char) dst;
        }

        i /= 2;

        ctx->charset->length += i;
        ctx->characters++;

        p = &table->src2dst[src * NGX_UTF_LEN] + 1;

        n = ngx_utf_decode(&p, i);

        if (n > 0xffff) {
            ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                               "invalid value \"%V\"", &value[1]);
            return NGX_CONF_ERROR;
        }

        pp = (u_char **) &table->dst2src[0];

        dst2src = pp[n >> 8];

        if (dst2src == NULL) {
            dst2src = ngx_pcalloc(cf->pool, 256);
            if (dst2src == NULL) {
                return NGX_CONF_ERROR;
            }

            pp[n >> 8] = dst2src;
        }

        dst2src[n & 0xff] = (u_char) src;

    } else {
        dst = ngx_hextoi(value[1].data, value[1].len);
        if (dst == NGX_ERROR || dst > 255) {
            ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                               "invalid value \"%V\"", &value[1]);
            return NGX_CONF_ERROR;
        }

        table->src2dst[src] = (u_char) dst;
        table->dst2src[dst] = (u_char) src;
    }

    return NGX_CONF_OK;
}


static char *
ngx_http_set_charset_slot(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
{
    char  *p = conf;

    ngx_int_t                     *cp;
    ngx_str_t                     *value, var;
    ngx_http_charset_main_conf_t  *mcf;

    cp = (ngx_int_t *) (p + cmd->offset);

    if (*cp != NGX_CONF_UNSET) {
        return "is duplicate";
    }

    value = cf->args->elts;

    if (cmd->offset == offsetof(ngx_http_charset_loc_conf_t, charset)
        && ngx_strcmp(value[1].data, "off") == 0)
    {
        *cp = NGX_HTTP_NO_CHARSET;
        return NGX_CONF_OK;
    }


    if (value[1].data[0] == '$') {
        var.len = value[1].len - 1;
        var.data = value[1].data + 1;

        *cp = ngx_http_get_variable_index(cf, &var);

        if (*cp == NGX_ERROR) {
            return NGX_CONF_ERROR;
        }

        *cp += NGX_HTTP_CHARSET_VAR;

        return NGX_CONF_OK;
    }

    mcf = ngx_http_conf_get_module_main_conf(cf,
                                             ngx_http_charset_filter_module);

    *cp = ngx_http_add_charset(&mcf->charsets, &value[1]);
    if (*cp == NGX_ERROR) {
        return NGX_CONF_ERROR;
    }

    return NGX_CONF_OK;
}


static ngx_int_t
ngx_http_add_charset(ngx_array_t *charsets, ngx_str_t *name)
{
    ngx_uint_t           i;
    ngx_http_charset_t  *c;

    c = charsets->elts;
    for (i = 0; i < charsets->nelts; i++) {
        if (name->len != c[i].name.len) {
            continue;
        }

        if (ngx_strcasecmp(name->data, c[i].name.data) == 0) {
            break;
        }
    }

    if (i < charsets->nelts) {
        return i;
    }

    c = ngx_array_push(charsets);
    if (c == NULL) {
        return NGX_ERROR;
    }

    c->tables = NULL;
    c->name = *name;
    c->length = 0;

    if (ngx_strcasecmp(name->data, (u_char *) "utf-8") == 0) {
        c->utf8 = 1;

    } else {
        c->utf8 = 0;
    }

    return i;
}


static void *
ngx_http_charset_create_main_conf(ngx_conf_t *cf)
{
    ngx_http_charset_main_conf_t  *mcf;

    mcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_main_conf_t));
    if (mcf == NULL) {
        return NGX_CONF_ERROR;
    }

    if (ngx_array_init(&mcf->charsets, cf->pool, 2, sizeof(ngx_http_charset_t))
        == NGX_ERROR)
    {
        return NGX_CONF_ERROR;
    }

    if (ngx_array_init(&mcf->tables, cf->pool, 1,
                       sizeof(ngx_http_charset_tables_t)) == NGX_ERROR)
    {
        return NGX_CONF_ERROR;
    }

    if (ngx_array_init(&mcf->recodes, cf->pool, 2,
                       sizeof(ngx_http_charset_recode_t)) == NGX_ERROR)
    {
        return NGX_CONF_ERROR;
    }

    return mcf;
}


static void *
ngx_http_charset_create_loc_conf(ngx_conf_t *cf)
{
    ngx_http_charset_loc_conf_t  *lcf;

    lcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_loc_conf_t));
    if (lcf == NULL) {
        return NGX_CONF_ERROR;
    }

    lcf->charset = NGX_CONF_UNSET;
    lcf->source_charset = NGX_CONF_UNSET;
    lcf->override_charset = NGX_CONF_UNSET;

    return lcf;
}


static char *
ngx_http_charset_merge_loc_conf(ngx_conf_t *cf, void *parent, void *child)
{
    ngx_http_charset_loc_conf_t *prev = parent;
    ngx_http_charset_loc_conf_t *conf = child;

    ngx_uint_t                     i;
    ngx_http_charset_recode_t     *recode;
    ngx_http_charset_main_conf_t  *mcf;

    ngx_conf_merge_value(conf->override_charset, prev->override_charset, 0);
    ngx_conf_merge_value(conf->charset, prev->charset, NGX_HTTP_NO_CHARSET);

    if (conf->source_charset == NGX_CONF_UNSET) {
        conf->source_charset = prev->source_charset;
    }

    if (conf->charset == NGX_HTTP_NO_CHARSET
        || conf->source_charset == NGX_CONF_UNSET
        || conf->charset == conf->source_charset)
    {
        return NGX_CONF_OK;
    }

    if (conf->source_charset >= NGX_HTTP_CHARSET_VAR
        || conf->charset >= NGX_HTTP_CHARSET_VAR)
    {
        return NGX_CONF_OK;
    }

    mcf = ngx_http_conf_get_module_main_conf(cf,
                                             ngx_http_charset_filter_module);
    recode = mcf->recodes.elts;
    for (i = 0; i < mcf->recodes.nelts; i++) {
        if (conf->source_charset == recode[i].src
            && conf->charset == recode[i].dst)
        {
            return NGX_CONF_OK;
        }
    }

    recode = ngx_array_push(&mcf->recodes);
    if (recode == NULL) {
        return NGX_CONF_ERROR;
    }

    recode->src = conf->source_charset;
    recode->dst = conf->charset;

    return NGX_CONF_OK;
}


static ngx_int_t
ngx_http_charset_postconfiguration(ngx_conf_t *cf)
{
    u_char                       **src, **dst;
    ngx_int_t                      c;
    ngx_uint_t                     i, t;
    ngx_http_charset_t            *charset;
    ngx_http_charset_recode_t     *recode;
    ngx_http_charset_tables_t     *tables;
    ngx_http_charset_main_conf_t  *mcf;

    mcf = ngx_http_conf_get_module_main_conf(cf,
                                             ngx_http_charset_filter_module);

    recode = mcf->recodes.elts;
    tables = mcf->tables.elts;
    charset = mcf->charsets.elts;

    for (i = 0; i < mcf->recodes.nelts; i++) {

        c = recode[i].src;

        for (t = 0; t < mcf->tables.nelts; t++) {

            if (c == tables[t].src && recode[i].dst == tables[t].dst) {
                goto next;
            }

            if (c == tables[t].dst && recode[i].dst == tables[t].src) {
                goto next;
            }
        }

        ngx_log_error(NGX_LOG_EMERG, cf->log, 0,
                   "no \"charset_map\" between the charsets \"%V\" and \"%V\"",
                   &charset[c].name, &charset[recode[i].dst].name);
        return NGX_ERROR;

    next:
        continue;
    }


    for (t = 0; t < mcf->tables.nelts; t++) {

        src = charset[tables[t].src].tables;

        if (src == NULL) {
            src = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
            if (src == NULL) {
                return NGX_ERROR;
            }

            charset[tables[t].src].tables = src;
        }

        dst = charset[tables[t].dst].tables;

        if (dst == NULL) {
            dst = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
            if (dst == NULL) {
                return NGX_ERROR;
            }

            charset[tables[t].dst].tables = dst;
        }

        src[tables[t].dst] = tables[t].src2dst;
        dst[tables[t].src] = tables[t].dst2src;
    }

    ngx_http_next_header_filter = ngx_http_top_header_filter;
    ngx_http_top_header_filter = ngx_http_charset_header_filter;

    ngx_http_next_body_filter = ngx_http_top_body_filter;
    ngx_http_top_body_filter = ngx_http_charset_body_filter;

    return NGX_OK;
}