user-agent support

README.C-LANG

@@ -28,6 +28,10 @@ parsleyPtr parsley_compile(char* parsley, char* incl)
 	Notes: This is *NOT* thread-safe. (Usage of the parselet via parsley_parse_* *IS* 
 	thread-safe, however.)
 
+void parsley_set_user_agent(char *);
+
+	Sets the user-agent used by parsley's internal http library.
+
 void parsley_free(parsleyPtr);
 
 	Frees the parsleyPtr's memory.

parsley.c

@@ -316,11 +316,8 @@ collate(xmlNodePtr xml) {
     free(empty);
     free(multi);
     free(sortable);
-    // xml->children = NULL;
-    // xmlFreeNode(xml);
 
     collate(groups);
-		// TODO: done? figure out what the recursion target needs to be and reimplement
   } else {
     xmlNodePtr child = xml->children;
     while(child != NULL){
@@ -330,6 +327,11 @@ collate(xmlNodePtr xml) {
   }
 }
 
+void
+parsley_set_user_agent(char const * agent) {
+  _parsley_set_user_agent(agent);
+}
+
 static void
 visit(parsedParsleyPtr ptr, xmlNodePtr xml, char* err);
 

parsley.h

@@ -75,6 +75,7 @@ parsedParsleyPtr parsley_parse_file(parsleyPtr parsley, char* file, int flags);
 parsedParsleyPtr parsley_parse_string(parsleyPtr parsley, char* string, size_t size, char* base_uri, int flags);
 parsedParsleyPtr parsley_parse_doc(parsleyPtr, xmlDocPtr, int);
 
+void parsley_set_user_agent(char const *agent);
 static contextPtr parsley_parsing_context;
 
 #endif

parsley_main.c

@@ -21,6 +21,7 @@ struct arguments
 	int flags;
 	int output_xml;
 	char *parsley;
+	char *user_agent;
   char *input_file;
   char *output_file;
 };
@@ -42,6 +43,7 @@ static struct argp_option options[] = {
 	{"output",   				'o', "FILE", 0, 	"Output to FILE instead of standard output" },
   {"include",  				'i', "FILE", 0, 	"Include the contents of FILE in the compiled XSLT" },
 	{"no-prune",        'n', 0, 0, 	"Don't prune empty subtrees" },
+	{"user-agent",      'U', "USER_AGENT", 0, 	"Value of HTTP User-Agent header" },
 	{"no-net",          'z', 0, 0, 	"Disable ftp and http access for parselets" },
 	{"no-filesystem",   'Z', 0, 0, 	"Disable filesystem access for parselets" },
   { 0 }
@@ -58,6 +60,8 @@ static error_t parse_opt (int key, char *arg, struct argp_state *state)
     case 'x':
       arguments->flags &= ~PARSLEY_OPTIONS_HTML;
 			break;
+		case 'U':
+      parsley_set_user_agent(arg);
     case 'n':
 			arguments->flags &= ~PARSLEY_OPTIONS_PRUNE;
 			break;

util.c

@@ -22,13 +22,25 @@ FILE* parsley_fopen(char* name, char* mode) {
 	return fo;	
 }
 
-static int parsley_io_mode = ~0;
+static int parsley_io_mode = 0;
+static char *parsley_user_agent_header = NULL;
 
 int
 parsley_io_get_mode() {
   return parsley_io_mode;
 }
 
+void
+_parsley_set_user_agent(char * agent) {
+  if(parsley_user_agent_header != NULL) free(parsley_user_agent_header);
+  asprintf(&parsley_user_agent_header, "User-Agent: %s\n", agent);
+}
+
+static void *
+xmlUserAgentIOHTTPOpen(const char * file_name) {
+  return(xmlNanoHTTPMethod(file_name, NULL, NULL, NULL, parsley_user_agent_header, 0));
+}
+
 void
 parsley_io_set_mode(int mode) {
   if(mode == parsley_io_mode) return;
@@ -47,7 +59,7 @@ parsley_io_set_mode(int mode) {
   }
   if(parsley_io_mode & PARSLEY_OPTIONS_ALLOW_NET) {
     #ifdef LIBXML_HTTP_ENABLED
-        xmlRegisterInputCallbacks(xmlIOHTTPMatch, xmlIOHTTPOpen,
+        xmlRegisterInputCallbacks(xmlIOHTTPMatch, xmlUserAgentIOHTTPOpen,
     	                      xmlIOHTTPRead, xmlIOHTTPClose);
     #endif /* LIBXML_HTTP_ENABLED */
 

util.h

@@ -27,5 +27,6 @@ char* parsley_key_tag(char*);
 pxpathPtr parsley_key_filter(char*);
 int parsley_io_get_mode();
 void parsley_io_set_mode(int mode);
+void _parsley_set_user_agent(char *agent);
 
 #endif