paulo@0: /*
paulo@0:  * $Id: xml.c,v 1.10 2004/04/13 07:25:18 hipnod Exp $
paulo@0:  *
paulo@0:  * Copyright (C) 2001-2003 giFT project (gift.sourceforge.net)
paulo@0:  *
paulo@0:  * This program is free software; you can redistribute it and/or modify it
paulo@0:  * under the terms of the GNU General Public License as published by the
paulo@0:  * Free Software Foundation; either version 2, or (at your option) any
paulo@0:  * later version.
paulo@0:  *
paulo@0:  * This program is distributed in the hope that it will be useful, but
paulo@0:  * WITHOUT ANY WARRANTY; without even the implied warranty of
paulo@0:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
paulo@0:  * General Public License for more details.
paulo@0:  */
paulo@0: 
paulo@0: #include "gt_gnutella.h"
paulo@0: 
paulo@0: #include <zlib.h>
paulo@0: 
paulo@0: #ifdef USE_LIBXML2
paulo@0: #include <libxml/parser.h>         /* xmlParseMemory() */
paulo@0: #include <libxml/xmlerror.h>       /* xmlSetGenericErrorFunc() */
paulo@0: #endif /* USE_LIBXML2 */
paulo@0: 
paulo@0: #include "xml.h"
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: #ifndef USE_LIBXML2
paulo@0: BOOL gt_xml_parse (const char *xml, Dataset **ret)
paulo@0: {
paulo@0: 	return FALSE;
paulo@0: }
paulo@0: 
paulo@0: BOOL gt_xml_parse_indexed (const char *xml, size_t bin_len, Share **shares,
paulo@0:                            size_t shares_len)
paulo@0: {
paulo@0: 	return FALSE;
paulo@0: }
paulo@0: #endif /* !USE_LIBXML2 */
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: /* the rest of this file is conditional on using libxml */
paulo@0: #ifdef USE_LIBXML2
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: #define MAX_XML_BUFSIZE  65536
paulo@0: 
paulo@0: static char      *xml_buf;     /* for decompressing xml */
paulo@0: static size_t     xml_buf_size;
paulo@0: static z_stream   zxml;
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: static void print_nodes (xmlNodePtr node, Dataset **ret)
paulo@0: {
paulo@0: 	while (node != NULL)
paulo@0: 	{
paulo@0: 		/*
paulo@0: 		 * If this node has no children, it is a leaf node,
paulo@0: 		 * so set the metadata from it.
paulo@0: 		 */
paulo@0: 		if (node->xmlChildrenNode)
paulo@0: 			print_nodes (node->xmlChildrenNode, ret);
paulo@0: 		else
paulo@0: 			GT->DBGFN (GT, "name=%s", node->name);
paulo@0: 
paulo@0: 		node = node->next;
paulo@0: 	}
paulo@0: }
paulo@0: 
paulo@0: BOOL gt_xml_parse (const char *xml, Dataset **ret)
paulo@0: {
paulo@0: 	xmlDocPtr doc;
paulo@0: 
paulo@0: 	/* disable for now because it doesn't work anyway: need to share
paulo@0: 	 * code with parse_indexed */
paulo@0: 	if (!XML_DEBUG)
paulo@0: 		return FALSE;
paulo@0: 
paulo@0: 	/* only parse documents starting with '<' */
paulo@0: 	if (!xml || xml[0] != '<')
paulo@0: 		return FALSE;
paulo@0: 
paulo@0: 	if (!(doc = xmlParseMemory (xml, strlen (xml))))
paulo@0: 		return FALSE;
paulo@0: 
paulo@0: 	print_nodes (doc->xmlChildrenNode, ret);
paulo@0: 
paulo@0: 	xmlFreeDoc (doc);
paulo@0: 
paulo@0: 	return TRUE;
paulo@0: }
paulo@0: 
paulo@0: static void add_child (Dataset **children, const char *key, const char *value)
paulo@0: {
paulo@0: 	char *dup = NULL;
paulo@0: 
paulo@0: 	if (!key || !value)
paulo@0: 		return;
paulo@0: 
paulo@0: 	/*
paulo@0: 	 * Hack to map some of the attributes from XML documents found
paulo@0: 	 * on Gnutella to ones peddled by giFT.
paulo@0: 	 */
paulo@0: 	if (!strcasecmp (key, "bitrate"))
paulo@0: 	{
paulo@0: 		dup = stringf_dup ("%s000", value);
paulo@0: 		value = dup;
paulo@0: 	}
paulo@0: 	else if (!strcasecmp (key, "seconds"))
paulo@0: 	{
paulo@0: 		key = "duration";
paulo@0: 	}
paulo@0: 
paulo@0: 	dataset_insertstr (children, key, value);
paulo@0: 	free (dup);
paulo@0: }
paulo@0: 
paulo@0: static Dataset *collect_attributes (xmlNode *node)
paulo@0: {
paulo@0: 	const xmlAttr *attr;
paulo@0: 	Dataset       *children = NULL;
paulo@0: 	BOOL           do_log   = XML_DEBUG;
paulo@0: 
paulo@0: 	for (attr = node->properties; attr != NULL; attr = attr->next)
paulo@0: 	{
paulo@0: 		xmlChar *str;
paulo@0: 
paulo@0: 		/* is there an easier way to get attribute content? */
paulo@0: 		str = xmlGetProp (node, attr->name);
paulo@0: 
paulo@0: 		if (do_log)
paulo@0: 		{
paulo@0: 			GT->dbg (GT, "name=%s content=%s",
paulo@0: 			         (const char *)attr->name, (const char *)str);
paulo@0: 		}
paulo@0: 
paulo@0: 		/* add the key->value pair to the dataset */
paulo@0: 		add_child (&children, (const char *)attr->name,
paulo@0: 		           (const char *)str);
paulo@0: 
paulo@0: 		/* xmlGetProp() allocates memory */
paulo@0: 		free (str);
paulo@0: 	}
paulo@0: 
paulo@0: 	return children;
paulo@0: }
paulo@0: 
paulo@0: static void set_meta_foreach (ds_data_t *key, ds_data_t *value, Share *share)
paulo@0: {
paulo@0: 	char *meta_key = key->data;
paulo@0: 	char *meta_val = value->data;
paulo@0: 
paulo@0: 	share_set_meta (share, meta_key, meta_val);
paulo@0: }
paulo@0: 
paulo@0: static void set_share_meta (Share **shares, size_t shares_len,
paulo@0:                             Dataset *children)
paulo@0: {
paulo@0: 	char      *index_str;
paulo@0: 	size_t     index;
paulo@0: 
paulo@0: 	/*
paulo@0: 	 * Lookup the "index" attribute, and use that to determine
paulo@0: 	 * which Share the XML applies to.
paulo@0: 	 */
paulo@0: 	if (!(index_str = dataset_lookupstr (children, "index")))
paulo@0: 		return;
paulo@0: 
paulo@0: 	index = gift_strtoul (index_str);
paulo@0: 
paulo@0: 	if (index >= shares_len)
paulo@0: 		return;
paulo@0: 
paulo@0: 	if (!shares[index])
paulo@0: 		return;
paulo@0: 
paulo@0: 	/* skip the index attribute */
paulo@0: 	dataset_removestr (children, "index");
paulo@0: 
paulo@0: 	dataset_foreach (children, DS_FOREACH(set_meta_foreach), shares[index]);
paulo@0: }
paulo@0: 
paulo@0: static void set_metadata_from_indexed_xml (Share **shares, size_t shares_len,
paulo@0:                                            xmlDoc *doc)
paulo@0: {
paulo@0: 	xmlNode *node;
paulo@0: 
paulo@0: 	if (!(node = xmlDocGetRootElement (doc)))
paulo@0: 		return;
paulo@0: 
paulo@0: 	for (node = node->xmlChildrenNode; node != NULL; node = node->next)
paulo@0: 	{
paulo@0: 		Dataset *children;
paulo@0: 
paulo@0: 		children = collect_attributes (node);
paulo@0: 
paulo@0: 		set_share_meta (shares, shares_len, children);
paulo@0: 		dataset_clear (children);
paulo@0: 	}
paulo@0: }
paulo@0: 
paulo@0: static int try_inflate_xml (const char *xml, size_t bin_len)
paulo@0: {
paulo@0: 	int ret;
paulo@0: 
paulo@0: 	/* set zlib allocation data */
paulo@0: 	zxml.zalloc    = Z_NULL;
paulo@0: 	zxml.zfree     = Z_NULL;
paulo@0: 	zxml.opaque    = Z_NULL;
paulo@0: 
paulo@0: 	/* set the input parameters */
paulo@0: 	zxml.next_in   = (char *)xml;
paulo@0: 	zxml.avail_in  = bin_len;
paulo@0: 
paulo@0: 	/* set the output parameters */
paulo@0: 	zxml.next_out  = xml_buf;
paulo@0: 	zxml.avail_out = xml_buf_size - 1;
paulo@0: 
paulo@0: 	if ((ret = inflateInit (&zxml)) != Z_OK)
paulo@0: 		return ret;
paulo@0: 
paulo@0: 	ret = inflate (&zxml, Z_FINISH);
paulo@0: 	inflateEnd (&zxml);
paulo@0: 
paulo@0: 	return ret;
paulo@0: }
paulo@0: 
paulo@0: static const char *inflate_xml (const char *xml, size_t bin_len)
paulo@0: {
paulo@0: 	size_t xml_len;
paulo@0: 	int    ret;
paulo@0: 
paulo@0: 	ret = try_inflate_xml (xml, bin_len);
paulo@0: 
paulo@0: 	if (ret == Z_BUF_ERROR && xml_buf_size < MAX_XML_BUFSIZE)
paulo@0: 	{
paulo@0: 		size_t newsize = xml_buf_size * 2;
paulo@0: 		char  *newbuf;
paulo@0: 
paulo@0: 		if (!(newbuf = realloc (xml_buf, newsize)))
paulo@0: 			return NULL;
paulo@0: 
paulo@0: 		xml_buf      = newbuf;
paulo@0: 		xml_buf_size = newsize;
paulo@0: 
paulo@0: 		/* retry with bigger buffer */
paulo@0: 		return inflate_xml (xml, bin_len);
paulo@0: 	}
paulo@0: 
paulo@0: 	if (ret != Z_STREAM_END)
paulo@0: 		return NULL;
paulo@0: 
paulo@0: 	/* null terminate (the now hopefully plaintext) XML */
paulo@0: 	xml_len = (xml_buf_size - 1) - zxml.avail_out;
paulo@0: 	xml_buf[xml_len] = 0;
paulo@0: 
paulo@0: 	if (XML_DEBUG)
paulo@0: 		GT->dbg (GT, "inflated xml: %s", xml_buf);
paulo@0: 
paulo@0: 	return xml_buf;
paulo@0: }
paulo@0: 
paulo@0: BOOL gt_xml_parse_indexed (const char *xml, size_t bin_len, Share **shares,
paulo@0:                            size_t shares_len)
paulo@0: {
paulo@0: 	xmlDoc     *doc;
paulo@0: 	size_t      xml_len;
paulo@0: 	const char *next;
paulo@0: 	const char *ptr;
paulo@0: 
paulo@0: 	if (!xml || bin_len <= 4)
paulo@0: 		return FALSE;
paulo@0: 
paulo@0: 	/*
paulo@0: 	 * Look for the encoding type, currently possible
paulo@0: 	 * encoding values are: "{}" meaning plain text, "{plaintext}",
paulo@0: 	 * and "{deflate}".
paulo@0: 	 */
paulo@0: 
paulo@0: 	if (!strncmp (xml, "{}", 2))
paulo@0: 	{
paulo@0: 		xml += 2;
paulo@0: 	}
paulo@0: 	else if (bin_len >= sizeof("{plaintext}") - 1 &&
paulo@0: 	         !strncasecmp (xml, "{plaintext}", sizeof("{plaintext}") - 1))
paulo@0: 	{
paulo@0: 		xml += sizeof("{plaintext}") - 1;
paulo@0: 	}
paulo@0: 	else if (bin_len >= sizeof("{deflate}") - 1 &&
paulo@0: 	         !strncasecmp (xml, "{deflate}", sizeof("{deflate}") - 1))
paulo@0: 	{
paulo@0: 		/* the len passed here should be bin_len - 1, but some servents (MRPH)
paulo@0: 		 * don't terminate the XML */
paulo@0: 		xml = inflate_xml (xml + sizeof("{deflate}") - 1, bin_len);
paulo@0: 
paulo@0: 		if (XML_DEBUG)
paulo@0: 			assert (xml != NULL);    /* assume valid input */
paulo@0: 
paulo@0: 		if (!xml)
paulo@0: 			return FALSE;
paulo@0: 	}
paulo@0: 
paulo@0: 	xml_len = strlen (xml);
paulo@0: 
paulo@0: 	/*
paulo@0: 	 * The XML block is a sequence of XML documents, separated by the <?xml
paulo@0: 	 * version="1.0"> document prefix.  Parse each one separately.
paulo@0: 	 */
paulo@0: 	for (ptr = xml; ptr != NULL; ptr = next)
paulo@0: 	{
paulo@0: 		size_t chunk_len;
paulo@0: 
paulo@0: 		if (ptr[0] != '<')
paulo@0: 			return FALSE;
paulo@0: 
paulo@0: 		next = strstr (ptr + 1, "<?xml");
paulo@0: 
paulo@0: 		chunk_len = xml_len;
paulo@0: 		if (next)
paulo@0: 			chunk_len = next - ptr;
paulo@0: 
paulo@0: 		if (!(doc = xmlParseMemory (ptr, chunk_len)))
paulo@0: 			return FALSE;
paulo@0: 
paulo@0: 		xml_len -= chunk_len;
paulo@0: 
paulo@0: 		set_metadata_from_indexed_xml (shares, shares_len, doc);
paulo@0: 		xmlFreeDoc (doc);
paulo@0: 	}
paulo@0: 
paulo@0: 	return TRUE;
paulo@0: }
paulo@0: 
paulo@0: /* gets called when there are parsing errors */
paulo@0: static void error_handler_func (void *udata, const char *msg, ...)
paulo@0: {
paulo@0: 	char     buf[1024];
paulo@0: 	va_list  args;
paulo@0: 
paulo@0: 	/* this is here until i figure out why i get a message about
paulo@0: 	 * namespace errors (but it still seems to work...) */
paulo@0: 	if (!XML_DEBUG)
paulo@0: 		return;
paulo@0: 
paulo@0: 	va_start (args, msg);
paulo@0: 	vsnprintf (buf, sizeof (buf) - 1, msg, args);
paulo@0: 	va_end (args);
paulo@0: 
paulo@0: 	GT->DBGFN (GT, "xml parse error: %s", buf);
paulo@0: }
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: #endif /* USE_LIBXML2 */
paulo@0: 
paulo@0: /*****************************************************************************/
paulo@0: 
paulo@0: void gt_xml_init (void)
paulo@0: {
paulo@0: #ifdef USE_LIBXML2
paulo@0: 	/* so libxml doesn't print messages on stderr */
paulo@0: 	xmlSetGenericErrorFunc (NULL, error_handler_func);
paulo@0: 
paulo@0: 	xml_buf = malloc (32);
paulo@0: 	assert (xml_buf != NULL);
paulo@0: 	xml_buf_size = 32;
paulo@0: 
paulo@0: 	memset (&zxml, 0, sizeof (zxml));
paulo@0: #endif /* USE_LIBXML2 */
paulo@0: }
paulo@0: 
paulo@0: void gt_xml_cleanup (void)
paulo@0: {
paulo@0: #ifdef USE_LIBXML2
paulo@0: 	free (xml_buf);
paulo@0: 	xml_buf      = NULL;
paulo@0: 	xml_buf_size = 0;
paulo@0: #endif /* USE_LIBXML2 */
paulo@0: }