From a7cb332c30b51b4e27f6e57c6b58dc2d1f73f89c Mon Sep 17 00:00:00 2001 From: Darren Salt Date: Thu, 19 Apr 2007 17:24:08 +0100 Subject: Relaxed XML parsing (ABI addition). Copes with missing close tags. Copes with extra close tags iff they don't match so-far-unclosed elements. Doesn't cope with missing "/>" - that needs more info from the caller. --- src/xine-utils/xmlparser.c | 65 +++++++++++++++++++++++++++++++++++++++------- src/xine-utils/xmlparser.h | 1 + 2 files changed, 57 insertions(+), 9 deletions(-) diff --git a/src/xine-utils/xmlparser.c b/src/xine-utils/xmlparser.c index 47096705a..82e1f0ecd 100644 --- a/src/xine-utils/xmlparser.c +++ b/src/xine-utils/xmlparser.c @@ -156,7 +156,8 @@ void xml_parser_free_tree(xml_node_t *current_node) { #define STATE_NODE 1 #define STATE_COMMENT 7 -static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int rec) { +static int xml_parser_get_node_internal (xml_node_t *current_node, char *root_names[], int rec, int relaxed) +{ char tok[TOKEN_SIZE]; char property_name[TOKEN_SIZE]; char node_name[TOKEN_SIZE]; @@ -164,6 +165,7 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r int res = 0; int parse_res; int bypass_get_token = 0; + int retval = 0; /* used when state==4; non-0 if there are missing */ xml_node_t *subtree = NULL; xml_node_t *current_subtree = NULL; xml_property_t *current_property = NULL; @@ -183,7 +185,7 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r /* do nothing */ break; case (T_EOF): - return 0; /* normal end */ + return retval; /* normal end */ break; case (T_M_START_1): state = STATE_NODE; @@ -252,7 +254,12 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r /* set node propertys */ subtree->props = properties; lprintf("info: rec %d new subtree %s\n", rec, node_name); - parse_res = xml_parser_get_node(subtree, node_name, rec + 1); + root_names[rec + 1] = node_name; + parse_res = xml_parser_get_node_internal(subtree, root_names, rec + 1, relaxed); + if (parse_res < -1) { + /* badly-formed XML (missing close tag) */ + return parse_res + 1 + (parse_res == -2); + } if (parse_res != 0) { return parse_res; } @@ -310,10 +317,26 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r if (xml_parser_mode == XML_PARSER_CASE_INSENSITIVE) { strtoupper(tok); } - if (strcmp(tok, root_name) == 0) { + if (strcmp(tok, root_names[rec]) == 0) { state = 4; - } else { - lprintf("error: xml struct, tok=%s, waited_tok=%s\n", tok, root_name); + } else if (relaxed) { + int r = rec; + while (--r >= 0) + if (strcmp(tok, root_names[r]) == 0) { + lprintf("warning: wanted %s, got %s - assuming missing close tags\n", root_names[rec], tok); + retval = r - rec - 1; /* -1 - (no. of implied close tags) */ + state = 4; + break; + } + /* relaxed parsing, ignoring extra close tag (but we don't handle out-of-order) */ + if (r < 0) { + lprintf("warning: extra close tag %s - ignoring\n", tok); + state = 10; + } + } + else + { + lprintf("error: xml struct, tok=%s, waited_tok=%s\n", tok, root_names[rec]); return -1; } break; @@ -328,7 +351,7 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r case 4: switch (res) { case (T_M_STOP_1): - return 0; + return retval; break; default: lprintf("error: unexpected token \"%s\", state %d\n", tok, state); @@ -435,6 +458,19 @@ static int xml_parser_get_node (xml_node_t *current_node, char *root_name, int r state = 9; break; } + break; + + /* > expected (following unmatched "child) && (!tmp_node->child->next)) { *root_node = tmp_node->child; free_xml_node(tmp_node); @@ -471,6 +514,10 @@ int xml_parser_build_tree(xml_node_t **root_node) { return res; } +int xml_parser_build_tree(xml_node_t **root_node) { + return xml_parser_build_tree_relaxed (root_node, 0); +} + const char *xml_parser_get_property (const xml_node_t *node, const char *name) { xml_property_t *prop; diff --git a/src/xine-utils/xmlparser.h b/src/xine-utils/xmlparser.h index f202ca28d..8d4915abf 100644 --- a/src/xine-utils/xmlparser.h +++ b/src/xine-utils/xmlparser.h @@ -57,6 +57,7 @@ typedef struct xml_node_s { void xml_parser_init(const char * buf, int size, int mode) XINE_PROTECTED; int xml_parser_build_tree(xml_node_t **root_node) XINE_PROTECTED; +int xml_parser_build_tree_relaxed(xml_node_t **root_node, int relaxed) XINE_PROTECTED; void xml_parser_free_tree(xml_node_t *root_node) XINE_PROTECTED; -- cgit v1.2.3