apr_xml.c

Apache 2.0.63 is the current stable version of the 2.0 series, and is recommended over any previous

    default:
        msg = "There was an unknown error within the XML body.";
        break;
    }

    (void) apr_cpystrn(errbuf, msg, errbufsize);
    return errbuf;
}

APU_DECLARE(apr_status_t) apr_xml_parse_file(apr_pool_t *p,
                                             apr_xml_parser **parser,
                                             apr_xml_doc **ppdoc,
                                             apr_file_t *xmlfd,
                                             apr_size_t buffer_length)
{
    apr_status_t rv;
    char *buffer;
    apr_size_t length;

    *parser = apr_xml_parser_create(p);
    if (*parser == NULL) {
        /* FIXME: returning an error code would be nice,
         * but we dont get one ;( */
        return APR_EGENERAL;
    }
    buffer = apr_palloc(p, buffer_length);
    length = buffer_length;

    rv = apr_file_read(xmlfd, buffer, &length);

    while (rv == APR_SUCCESS) {
        rv = apr_xml_parser_feed(*parser, buffer, length);
        if (rv != APR_SUCCESS) {
            return rv;
        }

        length = buffer_length;
        rv = apr_file_read(xmlfd, buffer, &length);
    }
    if (rv != APR_EOF) {
        return rv;
    }
    rv = apr_xml_parser_done(*parser, ppdoc);
    *parser = NULL;
    return rv;
}

APU_DECLARE(void) apr_text_append(apr_pool_t * p, apr_text_header *hdr,
                                  const char *text)
{
    apr_text *t = apr_palloc(p, sizeof(*t));

    t->text = text;
    t->next = NULL;

    if (hdr->first == NULL) {
	/* no text elements yet */
	hdr->first = hdr->last = t;
    }
    else {
	/* append to the last text element */
	hdr->last->next = t;
	hdr->last = t;
    }
}


/* ---------------------------------------------------------------
**
** XML UTILITY FUNCTIONS
*/

/*
** apr_xml_quote_string: quote an XML string
**
** Replace '<', '>', and '&' with '&lt;', '&gt;', and '&amp;'.
** If quotes is true, then replace '"' with '&quot;'.
**
** quotes is typically set to true for XML strings that will occur within
** double quotes -- attribute values.
*/
APU_DECLARE(const char *) apr_xml_quote_string(apr_pool_t *p, const char *s,
                                               int quotes)
{
    const char *scan;
    apr_size_t len = 0;
    apr_size_t extra = 0;
    char *qstr;
    char *qscan;
    char c;

    for (scan = s; (c = *scan) != '\0'; ++scan, ++len) {
	if (c == '<' || c == '>')
	    extra += 3;		/* &lt; or &gt; */
	else if (c == '&')
	    extra += 4;		/* &amp; */
	else if (quotes && c == '"')
	    extra += 5;		/* &quot; */
    }

    /* nothing to do? */
    if (extra == 0)
	return s;

    qstr = apr_palloc(p, len + extra + 1);
    for (scan = s, qscan = qstr; (c = *scan) != '\0'; ++scan) {
	if (c == '<') {
	    *qscan++ = '&';
	    *qscan++ = 'l';
	    *qscan++ = 't';
	    *qscan++ = ';';
	}
	else if (c == '>') {
	    *qscan++ = '&';
	    *qscan++ = 'g';
	    *qscan++ = 't';
	    *qscan++ = ';';
	}
	else if (c == '&') {
	    *qscan++ = '&';
	    *qscan++ = 'a';
	    *qscan++ = 'm';
	    *qscan++ = 'p';
	    *qscan++ = ';';
	}
	else if (quotes && c == '"') {
	    *qscan++ = '&';
	    *qscan++ = 'q';
	    *qscan++ = 'u';
	    *qscan++ = 'o';
	    *qscan++ = 't';
	    *qscan++ = ';';
	}
	else {
	    *qscan++ = c;
	}
    }

    *qscan = '\0';
    return qstr;
}

/* how many characters for the given integer? */
#define APR_XML_NS_LEN(ns) ((ns) < 10 ? 1 : (ns) < 100 ? 2 : (ns) < 1000 ? 3 : \
                            (ns) < 10000 ? 4 : (ns) < 100000 ? 5 : \
                            (ns) < 1000000 ? 6 : (ns) < 10000000 ? 7 : \
                            (ns) < 100000000 ? 8 : (ns) < 1000000000 ? 9 : 10)

static apr_size_t text_size(const apr_text *t)
{
    apr_size_t size = 0;

    for (; t; t = t->next)
	size += strlen(t->text);
    return size;
}

static apr_size_t elem_size(const apr_xml_elem *elem, int style,
                            apr_array_header_t *namespaces, int *ns_map)
{
    apr_size_t size;

    if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
	const apr_xml_attr *attr;

	size = 0;

	if (style == APR_XML_X2T_FULL_NS_LANG) {
	    int i;

	    /*
	    ** The outer element will contain xmlns:ns%d="%s" attributes
	    ** and an xml:lang attribute, if applicable.
	    */

	    for (i = namespaces->nelts; i--;) {
		/* compute size of: ' xmlns:ns%d="%s"' */
		size += (9 + APR_XML_NS_LEN(i) + 2 +
			 strlen(APR_XML_GET_URI_ITEM(namespaces, i)) + 1);
	    }

	    if (elem->lang != NULL) {
		/* compute size of: ' xml:lang="%s"' */
		size += 11 + strlen(elem->lang) + 1;
	    }
	}

	if (elem->ns == APR_XML_NS_NONE) {
	    /* compute size of: <%s> */
	    size += 1 + strlen(elem->name) + 1;
	}
	else {
	    int ns = ns_map ? ns_map[elem->ns] : elem->ns;

	    /* compute size of: <ns%d:%s> */
	    size += 3 + APR_XML_NS_LEN(ns) + 1 + strlen(elem->name) + 1;
	}

	if (APR_XML_ELEM_IS_EMPTY(elem)) {
	    /* insert a closing "/" */
	    size += 1;
	}
	else {
	    /*
	     * two of above plus "/":
	     *     <ns%d:%s> ... </ns%d:%s>
	     * OR  <%s> ... </%s>
	     */
	    size = 2 * size + 1;
	}

	for (attr = elem->attr; attr; attr = attr->next) {
	    if (attr->ns == APR_XML_NS_NONE) {
		/* compute size of: ' %s="%s"' */
		size += 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
	    }
	    else {
		/* compute size of: ' ns%d:%s="%s"' */
                int ns = ns_map ? ns_map[attr->ns] : attr->ns;
                size += 3 + APR_XML_NS_LEN(ns) + 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
	    }
	}

	/*
	** If the element has an xml:lang value that is *different* from
	** its parent, then add the thing in: ' xml:lang="%s"'.
	**
	** NOTE: we take advantage of the pointer equality established by
	** the parsing for "inheriting" the xml:lang values from parents.
	*/
	if (elem->lang != NULL &&
	    (elem->parent == NULL || elem->lang != elem->parent->lang)) {
	    size += 11 + strlen(elem->lang) + 1;
	}
    }
    else if (style == APR_XML_X2T_LANG_INNER) {
	/*
	 * This style prepends the xml:lang value plus a null terminator.
	 * If a lang value is not present, then we insert a null term.
	 */
	size = elem->lang ? strlen(elem->lang) + 1 : 1;
    }
    else
	size = 0;

    size += text_size(elem->first_cdata.first);

    for (elem = elem->first_child; elem; elem = elem->next) {
	/* the size of the child element plus the CDATA that follows it */
	size += (elem_size(elem, APR_XML_X2T_FULL, NULL, ns_map) +
		 text_size(elem->following_cdata.first));
    }

    return size;
}

static char *write_text(char *s, const apr_text *t)
{
    for (; t; t = t->next) {
	apr_size_t len = strlen(t->text);
	memcpy(s, t->text, len);
	s += len;
    }
    return s;
}

static char *write_elem(char *s, const apr_xml_elem *elem, int style,
			apr_array_header_t *namespaces, int *ns_map)
{
    const apr_xml_elem *child;
    apr_size_t len;
    int ns;

    if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
	int empty = APR_XML_ELEM_IS_EMPTY(elem);
	const apr_xml_attr *attr;

	if (elem->ns == APR_XML_NS_NONE) {
	    len = sprintf(s, "<%s", elem->name);
	}
	else {
	    ns = ns_map ? ns_map[elem->ns] : elem->ns;
	    len = sprintf(s, "<ns%d:%s", ns, elem->name);
	}
	s += len;

	for (attr = elem->attr; attr; attr = attr->next) {
	    if (attr->ns == APR_XML_NS_NONE)
		len = sprintf(s, " %s=\"%s\"", attr->name, attr->value);
            else {
                ns = ns_map ? ns_map[attr->ns] : attr->ns;
                len = sprintf(s, " ns%d:%s=\"%s\"", ns, attr->name, attr->value);
            }
	    s += len;
	}

	/* add the xml:lang value if necessary */
	if (elem->lang != NULL &&
	    (style == APR_XML_X2T_FULL_NS_LANG ||
	     elem->parent == NULL ||
	     elem->lang != elem->parent->lang)) {
	    len = sprintf(s, " xml:lang=\"%s\"", elem->lang);
	    s += len;
	}

	/* add namespace definitions, if required */
	if (style == APR_XML_X2T_FULL_NS_LANG) {
	    int i;

	    for (i = namespaces->nelts; i--;) {
		len = sprintf(s, " xmlns:ns%d=\"%s\"", i,
			      APR_XML_GET_URI_ITEM(namespaces, i));
		s += len;
	    }
	}

	/* no more to do. close it up and go. */
	if (empty) {
	    *s++ = '/';
	    *s++ = '>';
	    return s;
	}

	/* just close it */
	*s++ = '>';
    }
    else if (style == APR_XML_X2T_LANG_INNER) {
	/* prepend the xml:lang value */
	if (elem->lang != NULL) {
	    len = strlen(elem->lang);
	    memcpy(s, elem->lang, len);
	    s += len;
	}
	*s++ = '\0';
    }

    s = write_text(s, elem->first_cdata.first);

    for (child = elem->first_child; child; child = child->next) {
	s = write_elem(s, child, APR_XML_X2T_FULL, NULL, ns_map);
	s = write_text(s, child->following_cdata.first);
    }

    if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
	if (elem->ns == APR_XML_NS_NONE) {
	    len = sprintf(s, "</%s>", elem->name);
	}
	else {
	    ns = ns_map ? ns_map[elem->ns] : elem->ns;
	    len = sprintf(s, "</ns%d:%s>", ns, elem->name);
	}
	s += len;
    }

    return s;
}

APU_DECLARE(void) apr_xml_quote_elem(apr_pool_t *p, apr_xml_elem *elem)
{
    apr_text *scan_txt;
    apr_xml_attr *scan_attr;
    apr_xml_elem *scan_elem;

    /* convert the element's text */
    for (scan_txt = elem->first_cdata.first;
	 scan_txt != NULL;
	 scan_txt = scan_txt->next) {
	scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
    }
    for (scan_txt = elem->following_cdata.first;
	 scan_txt != NULL;
	 scan_txt = scan_txt->next) {
	scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
    }

    /* convert the attribute values */
    for (scan_attr = elem->attr;
	 scan_attr != NULL;
	 scan_attr = scan_attr->next) {
	scan_attr->value = apr_xml_quote_string(p, scan_attr->value, 1);
    }

    /* convert the child elements */
    for (scan_elem = elem->first_child;
	 scan_elem != NULL;
	 scan_elem = scan_elem->next) {
	apr_xml_quote_elem(p, scan_elem);
    }
}

/* convert an element to a text string */
APU_DECLARE(void) apr_xml_to_text(apr_pool_t * p, const apr_xml_elem *elem,
                                  int style, apr_array_header_t *namespaces,
                                  int *ns_map, const char **pbuf,
                                  apr_size_t *psize)
{
    /* get the exact size, plus a null terminator */
    apr_size_t size = elem_size(elem, style, namespaces, ns_map) + 1;
    char *s = apr_palloc(p, size);

    (void) write_elem(s, elem, style, namespaces, ns_map);
    s[size - 1] = '\0';

    *pbuf = s;
    if (psize)
	*psize = size;
}

APU_DECLARE(const char *) apr_xml_empty_elem(apr_pool_t * p,
                                             const apr_xml_elem *elem)
{
    if (elem->ns == APR_XML_NS_NONE) {
	/*
	 * The prefix (xml...) is already within the prop name, or
	 * the element simply has no prefix.
	 */
	return apr_psprintf(p, "<%s/>" DEBUG_CR, elem->name);
    }

    return apr_psprintf(p, "<ns%d:%s/>" DEBUG_CR, elem->ns, elem->name);
}

/* return the URI's (existing) index, or insert it and return a new index */
APU_DECLARE(int) apr_xml_insert_uri(apr_array_header_t *uri_array,
                                    const char *uri)
{
    int i;
    const char **pelt;

    /* never insert an empty URI; this index is always APR_XML_NS_NONE */
    if (*uri == '\0')
        return APR_XML_NS_NONE;  

    for (i = uri_array->nelts; i--;) {
	if (strcmp(uri, APR_XML_GET_URI_ITEM(uri_array, i)) == 0)
	    return i;
    }

    pelt = apr_array_push(uri_array);
    *pelt = uri;		/* assume uri is const or in a pool */
    return uri_array->nelts - 1;
}