improved detection of htmltidy for debian and co.

Signed-off-by: Michael Krelin <hacker@klever.net>

1 files changed, 2 insertions, 1 deletions

diff --git a/lib/discovery.cc b/lib/discovery.cc
index 3b90977..d1989ec 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -1,269 +1,270 @@
 #include <list>
 #include <opkele/curl.h>
 #include <opkele/expat.h>
 #include <opkele/uris.h>
 #include <opkele/discovery.h>
 #include <opkele/exception.h>
 #include <opkele/util.h>
-#include <opkele/tidy.h>
 #include <opkele/data.h>
 #include <opkele/debug.h>
 
 #include "config.h"
 
+#include <opkele/tidy.h>
+
 #define XRDS_HEADER "X-XRDS-Location"
 #define CT_HEADER "Content-Type"
 
 namespace opkele {
     using std::list;
     using xrd::XRD_t;
     using xrd::service_t;
 
     /* TODO: the whole discovery thing needs cleanup and optimization due to
      * many changes of concept. */
 
     static const size_t max_html = 16384;
 
     static const struct service_type_t {
         const char *uri;
         const char *forceid;
     } op_service_types[] = {
         { STURI_OPENID20_OP, IDURI_SELECT20 },
         { STURI_OPENID20, 0 },
         { STURI_OPENID11, 0 },
         { STURI_OPENID10, 0 }
     };
     enum {
         st_index_1 = 2, st_index_2 = 1
     };
 
 
     static inline bool is_qelement(const XML_Char *n,const char *qen) {
         return !strcasecmp(n,qen);
     }
     static inline bool is_element(const XML_Char *n,const char *en) {
         if(!strcasecmp(n,en)) return true;
         int nl = strlen(n), enl = strlen(en);
         if( (nl>=(enl+1)) && n[nl-enl-1]=='\t'
                 && !strcasecmp(&n[nl-enl],en) )
             return true;
         return false;
     }
 
     static long element_priority(const XML_Char **a) {
         for(;*a;++a)
             if(!strcasecmp(*(a++),"priority")) {
                 long rv;
                 return (sscanf(*a,"%ld",&rv)==1)?rv:-1;
             }
         return -1;
     }
     /* TODO: ideally all attributes should be
      * retrieved in one run */
     static const char *element_attr(const XML_Char **a, const char *at) {
         for(;*a;++a)
             if(!strcasecmp(*(a++),at)) {
                 return *a;
             }
         return 0;
     }
 
     class idigger_t : public util::curl_t, public util::expat_t {
         public:
             string xri_proxy;
 
             enum {
                 xmode_html = 1, xmode_xrd = 2, xmode_cid = 4,
                 xmode_noredirs = 8
             };
             int xmode;
 
             string xrds_location;
             string http_content_type;
             service_t html_openid1;
             service_t html_openid2;
             string cdata_buf;
             long status_code;
             string status_string;
 
             typedef list<string> pt_stack_t;
             pt_stack_t pt_stack;
             int skipping;
             bool parser_choked;
             string save_html;
 
             XRD_t *xrd;
             service_t *xrd_service;
             string* cdata;
 
             idigger_t()
                 : util::curl_t(easy_init()),
                 util::expat_t(0),
                 xri_proxy(XRI_PROXY_URL) {
                     CURLcode r;
                     (r=misc_sets())
                         || (r=set_write())
                         || (r=set_header())
                         ;
                     if(r)
                         throw exception_curl(OPKELE_CP_ "failed to set curly options",r);
                 }
             ~idigger_t() throw() { }
 
             void yadiscover(endpoint_discovery_iterator oi,const string& yurl,const char **types,bool redirs) {
                 idiscovery_t idis;
                 idis.xri_identity = false;
                 discover_at(idis,yurl,xmode_html|xmode_xrd|(redirs?0:xmode_noredirs));
                 if(!xrds_location.empty()) {
                     idis.clear();
                     discover_at(idis,xrds_location,xmode_xrd);
                 }
                 idis.normalized_id = idis.canonicalized_id = yurl;
                 service_type_t st;
                 for(st.uri=*types;*types;st.uri=*(++types))
                     queue_endpoints(oi,idis,&st);
             }
 
             string discover(endpoint_discovery_iterator& oi,const string& identity) {
                 string rv;
                 idiscovery_t idis;
                 string::size_type fsc = identity.find_first_not_of(data::_whitespace_chars);
                 if(fsc==string::npos)
                     throw bad_input(OPKELE_CP_ "whitespace-only identity");
                 string::size_type lsc = identity.find_last_not_of(data::_whitespace_chars);
                 assert(lsc!=string::npos);
                 if(!strncasecmp(identity.c_str()+fsc,"xri://",sizeof("xri://")-1))
                     fsc += sizeof("xri://")-1;
                 if((fsc+1)>=lsc)
                     throw bad_input(OPKELE_CP_ "not a character of importance in identity");
                 string id(identity,fsc,lsc-fsc+1);
                 idis.clear();
                 if(strchr(data::_iname_leaders,id[0])) {
                     /* TODO: further normalize xri identity? Like folding case
                      * or whatever... */
                     rv = id;
                     set<string> cids;
                     for(const struct service_type_t *st=op_service_types;
                             st<&op_service_types[sizeof(op_service_types)/sizeof(*op_service_types)];++st) {
                         idis.clear();
                         discover_at( idis,
                                 xri_proxy + util::url_encode(id)+
                                 "?_xrd_t="+util::url_encode(st->uri)+
                                 "&_xrd_r=application/xrd%2Bxml"
                                 ";sep=true;refs=true",
                                 xmode_xrd );
                         if(status_code==241) continue;
                         if(status_code!=100)
                             throw failed_xri_resolution(OPKELE_CP_
                                     "XRI resolution failed with '"+status_string+"' message"
                                     ", while looking for SEP with type '"+st->uri+"'", status_code);
                         if(idis.xrd.canonical_ids.empty())
                             throw opkele::failed_discovery(OPKELE_CP_ "No CanonicalID for XRI identity found");
                         string cid = idis.xrd.canonical_ids.begin()->second;
                         if(cids.find(cid)==cids.end()) {
                             cids.insert(cid);
                             idis.clear();
                             discover_at( idis,
                                     xri_proxy + util::url_encode(id)+
                                     "?_xrd_t="+util::url_encode(st->uri)+
                                     "&_xrd_r=application/xrd%2Bxml"
                                     ";sep=true;refs=true",
                                     xmode_xrd );
                             if(status_code==241) continue;
                             if(status_code!=100)
                                 throw failed_xri_resolution(OPKELE_CP_
                                         "XRI resolution failed with '"+status_string+"' message"
                                         ", while looking for SEP with type '"+st->uri+"'"
                                         " on canonical id", status_code);
                         }
                         idis.canonicalized_id = cid;
                         idis.normalized_id = rv; idis.xri_identity = true;
                         queue_endpoints(oi,idis,st);
                     }
                 }else{
                     idis.xri_identity = false;
                     if(id.find("://")==string::npos)
                         id.insert(0,"http://");
                     string::size_type fp = id.find('#');
                     if(fp!=string::npos) {
                         string::size_type qp = id.find('?');
                         if(qp==string::npos || qp<fp)
                             id.erase(fp);
                         else if(qp>fp)
                             id.erase(fp,qp-fp);
                     }
                     rv = idis.normalized_id = util::rfc_3986_normalize_uri(id);
                     discover_at(idis,id,xmode_html|xmode_xrd);
                     const char * eu = 0;
                     CURLcode r = easy_getinfo(CURLINFO_EFFECTIVE_URL,&eu);
                     if(r)
                         throw exception_curl(OPKELE_CP_ "failed to get CURLINFO_EFFECTIVE_URL",r);
                     string cid = util::strip_uri_fragment_part( idis.canonicalized_id = util::rfc_3986_normalize_uri(eu) );
                     if(xrds_location.empty()) {
                         if(idis.xrd.empty())
                             html2xrd(oi,idis);
                         else{
                             for(const service_type_t *st=op_service_types;
                                     st<&op_service_types[sizeof(op_service_types)/sizeof(*op_service_types)];++st)
                                 queue_endpoints(oi,idis,st);
                         }
                     }else{
                         idis.clear();
                         idis.canonicalized_id = cid;
                         discover_at(idis,xrds_location,xmode_xrd);
                         if(idis.xrd.empty())
                             html2xrd(oi,idis);
                         else{
                             for(const service_type_t *st=op_service_types;
                                     st<&op_service_types[sizeof(op_service_types)/sizeof(*op_service_types)];++st)
                                 queue_endpoints(oi,idis,st);
                         }
                     }
                 }
                 return rv;
             }
 
             void discover_at(idiscovery_t& idis,const string& url,int xm) {
                 CURLcode r = easy_setopt(CURLOPT_MAXREDIRS, (xm&xmode_noredirs)?0:5);
                 if(r)
                     throw exception_curl(OPKELE_CP_ "failed to set curly maxredirs option");
                 if( (r=easy_setopt(CURLOPT_URL,url.c_str())) )
                     throw exception_curl(OPKELE_CP_ "failed to set curly urlie",r);
 
                 http_content_type.clear();
                 xmode = xm;
                 prepare_to_parse();
                 if(xmode&xmode_html) {
                     xrds_location.clear();
                     save_html.clear();
                     save_html.reserve(max_html);
                 }
                 xrd = &idis.xrd;
 
                 r = easy_perform();
                 if(r && r!=CURLE_WRITE_ERROR)
                     throw exception_curl(OPKELE_CP_ "failed to perform curly request",r);
 
                 if(!parser_choked) {
                     parse(0,0,true);
                 }else if(xmode&xmode_html){
                     /* TODO: do not bother if we've seen xml */
                     try {
                         util::tidy_doc_t td = util::tidy_doc_t::create();
                         if(!td)
                             throw exception_tidy(OPKELE_CP_ "failed to create htmltidy document");
 #ifndef NDEBUG
                         td.opt_set(TidyQuiet,false);
                         td.opt_set(TidyShowWarnings,false);
 #endif /* NDEBUG */
                         td.opt_set(TidyForceOutput,true);
                         td.opt_set(TidyXhtmlOut,true);
                         td.opt_set(TidyDoctypeMode,TidyDoctypeOmit);
                         td.opt_set(TidyMark,false);
                         td.opt_set(TidyNumEntities,true);
                         if(td.parse_string(save_html)<=0)
                             throw exception_tidy(OPKELE_CP_ "tidy failed to parse document");
                         if(td.clean_and_repair()<=0)
                             throw exception_tidy(OPKELE_CP_ "tidy failed to clean and repair");
                         util::tidy_buf_t tide;
                         if(td.save_buffer(tide)<=0)