improved detection of htmltidy for debian and co.

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

3 files changed, 14 insertions, 5 deletions

diff --git a/configure.ac b/configure.ac
index 0aa1272..a66e53d 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,257 +1,258 @@
 AC_INIT([libopkele], [1.1.99.0], [libopkele-bugs@klever.net])
 AC_CONFIG_SRCDIR([include/opkele/opkele-config.h])
 AC_CONFIG_HEADERS([config.h include/opkele/acconfig.h])
 AM_INIT_AUTOMAKE([dist-bzip2])
 
 AC_PROG_INSTALL
 AC_PROG_CXX
 AC_PROG_CC
 AC_PROG_LIBTOOL
 PKG_PROG_PKG_CONFIG
 
 AC_HEADER_STDC
 
 AC_PATH_PROG([XSLTPROC],[xsltproc],[true])
 
 AC_MSG_CHECKING([for source tree version])
 if headrev=$(cd $srcdir && git rev-parse --verify HEAD 2>/dev/null) ; then
  PACKAGE_SRC_VERSION="$(cd $srcdir && git describe --tags $headrev)"
  test "$PACKAGE_SRC_VERSION" = "$PACKAGE_VERSION" \
   -o "${PACKAGE_SRC_VERSION#${PACKAGE_VERSION}-}" != "$PACKAGE_SRC_VERSION" || PACKAGE_SRC_VERSION="${PACKAGE_VERSION}:${PACKAGE_SRC_VERSION}"
  ( cd $srcdir && git diff-index $headrev | read dirt ) && PACKAGE_SRC_VERSION="${PACKAGE_SRC_VERSION}-dirty"
 else
  PACKAGE_SRC_VERSION="$PACKAGE_VERSION"
 fi
 AC_MSG_RESULT([$PACKAGE_SRC_VERSION])
 AC_SUBST([PACKAGE_SRC_VERSION])
 AC_DEFINE_UNQUOTED([PACKAGE_SRC_VERSION],["$PACKAGE_SRC_VERSION"],[more or less precise source tree version])
 
 tr1_mem_std="false"
 tr1_mem_boost="false"
 AC_CHECK_SHAREDPTR(std::tr1,tr1/memory,[ tr1_mem_std=true ])
 AC_CHECK_SHAREDPTR(boost,boost/shared_ptr.hpp,[ tr1_mem_boost=true ])
 tr1_mem=""
 AC_ARG_WITH([tr1-memory],
  AC_HELP_STRING([--with-tr1-memory=<boost|std>],[select tr1/memory (shared_ptr<>) implementation to use]),
  [ tr1_mem="$withval" ]
 )
 AC_MSG_CHECKING([for tr1/memory implementation to use])
 test -z "$tr1_mem" && $tr1_mem_std && tr1_mem=std
 test -z "$tr1_mem" && $tr1_mem_boost && tr1_mem=boost
 if test -z "$tr1_mem" ; then
  AC_MSG_RESULT([none found])
 else
  AC_MSG_RESULT([$tr1_mem])
 fi
 case "$tr1_mem" in
  std)
   $tr1_mem_std || AC_MSG_ERROR([std implementation requested, but not found])
   OPKELE_TR1_MEM_NS=std::tr1
   OPKELE_TR1_MEM_HEADER=tr1/memory
   ;;
  boost)
   $tr1_mem_boost || AC_MSG_ERROR([boost implementation requested, but not found])
   OPKELE_TR1_MEM_NS=boost
   OPKELE_TR1_MEM_HEADER=boost/shared_ptr.hpp
   ;;
  *)
   AC_MSG_ERROR([no shared_ptr<> implementation found])
   ;;
 esac
 AC_SUBST([OPKELE_TR1_MEM_NS])
 AC_SUBST([OPKELE_TR1_MEM_HEADER])
 
 AC_MSG_CHECKING([for deprecated attribute support])
 AC_COMPILE_IFELSE([
   int __attribute__((deprecated)) deprecated_function();
  ],[
   AC_MSG_RESULT([yes])
   AC_DEFINE([OPKELE_DEPRECATE],[__attribute__((deprecated))],[deprecated function attribute])
  ],[
   AC_MSG_RESULT([no])
   AC_DEFINE([OPKELE_DEPRECATE],,[deprecated function attribute])
  ]
 )
 
 AC_LANG_PUSH([C++])
 AC_MSG_CHECKING([for abi::__cxa_demangle])
 AC_COMPILE_IFELSE([
   #include <typeinfo>
   using namespace std;
   #include <cxxabi.h>
   int main(int c,char **v) {
    int dstat;
    char *demangled = abi::__cxa_demangle(typeid(dstat).name(),0,0,&dstat);
    return 0;
   }
  ],[
   AC_MSG_RESULT([yes])
   AC_DEFINE([HAVE_DEMANGLE],,[defined if abi::__cxa_demangle is available])
  ],[
   AC_MSG_RESULT([no])
  ]
 )
 AC_LANG_POP([C++])
 
 
 
 
 PKG_CHECK_MODULES([OPENSSL],[openssl],,[
  AC_MSG_ERROR([no openssl library found. get one from http://www.openssl.org/])
 ])
 
 WANT_KONFORKA="yes"
 AC_ARG_ENABLE([konforka],
  AC_HELP_STRING([--disable-konforka],[do not use konforka library (default: use if found)]),
  [
   test "${enableval}" = "no" && WANT_KONFORKA="no"
  ]
 )
 if test "${WANT_KONFORKA}" = "yes" ; then
  PKG_CHECK_MODULES([KONFORKA],[konforka],[
   AC_SUBST([KONFORKA_CFLAGS])
   AC_SUBST([KONFORKA_LIBS])
   AC_DEFINE([HAVE_KONFORKA],,[defined in presence of konforka library])
   AC_DEFINE([OPKELE_HAVE_KONFORKA],,[defined in presence of konforka library])
   AC_SUBST([KONFORKA_KONFORKA],[konforka])
  ],[true])
 fi
 
 WANT_DOXYGEN="yes"
 AC_ARG_ENABLE([doxygen],
  AC_HELP_STRING([--disable-doxygen],[do not generate documentation]),
  [
   test "${enableval}" = "no" && WANT_DOXYGEN="no"
  ]
 )
 if test "${WANT_DOXYGEN}" = "yes" ; then
  AC_WITH_DOXYGEN
  AC_WITH_DOT
 else
  AM_CONDITIONAL([HAVE_DOXYGEN],[false])
  AM_CONDITIONAL([HAVE_DOT],[false])
 fi
 
 LIBCURL_CHECK_CONFIG(,,,[
  AC_MSG_ERROR([no required libcurl library. get one from http://curl.haxx.se/])
 ])
 
 AC_CHECK_HEADER([expat.h],[
  AC_CHECK_LIB([expat],[XML_ParserCreate],[
   EXPAT_LIBS=-lexpat
   EXPAT_CFLAGS=
   AC_SUBST([EXPAT_LIBS])
   AC_SUBST([EXPAT_CFLAGS])
  ],[
   AC_MSG_ERROR([no required expat library. get one from http://expat.sourceforge.net/])
  ])
 ],[
  AC_MSG_ERROR([no required expat library. get one from http://expat.sourceforge.net/])
 ])
 
-AC_CHECK_HEADER([tidy.h],[
+AC_CHECK_HEADERS([tidy.h tidy/tidy.h],[
  AC_CHECK_LIB([tidy],[tidyParseBuffer],[
   TIDY_LIBS=-ltidy
   TIDY_CFLAGS=
   AC_SUBST([TIDY_LIBS])
   AC_SUBST([TIDY_CFLAGS])
  ],[
   AC_MSG_ERROR([no required htmltidy library found. get one from http://tidy.sourceforge.net/])
  ])
 ],[
- AC_MSG_ERROR([no required htmltidy library found. get one from http://tidy.sourceforge.net/])
+ test "$ac_header" = "tidy/tidy.h" \
+ && AC_MSG_ERROR([no required htmltidy library found. get one from http://tidy.sourceforge.net/])
 ])
 
 if test -n "$PCRE_LIBS" -a -n "$PCRE_CFLAGS" ; then
  AC_SUBST([PCRE_CFLAGS])
  AC_SUBST([PCRE_LIBS])
  :
 else
  PKG_CHECK_MODULES([PCRE],[libpcre],,[
   AC_MSG_ERROR([no libpcre found, go get it at http://www.pcre.org/])
  ])
 fi
 
 PKG_CHECK_MODULES([SQLITE3],[sqlite3],[have_sqlite3=true],[have_sqlite3=false])
 AM_CONDITIONAL([HAVE_SQLITE3],[$have_sqlite3])
 PKG_CHECK_MODULES([KINGATE],[kingate-plaincgi],[have_kingate=true],[have_kingate=false])
 AM_CONDITIONAL([HAVE_KINGATE],[$have_kingate])
 PKG_CHECK_MODULES([UUID],[uuid],[have_uuid=true],[have_uuid=false])
 AM_CONDITIONAL([HAVE_UUID],[$have_uuid])
 if $have_uuid ; then
  AC_DEFINE([HAVE_LIBUUID],,[defined in presence of libuuid])
  AC_SUBST([UUID_UUID],[uuid])
 fi
 
 curl_ssl_verify_host="true"
 AC_ARG_ENABLE([ssl-verify-host],
  AC_HELP_STRING([--disable-ssl-verify-host],[disable cURL cert/host relationships verification]),
  [ test "${enableval}" = "no" && curl_ssl_verify_host="false" ]
 )
 ${curl_ssl_verify_host} || AC_DEFINE([DISABLE_CURL_SSL_VERIFYHOST],,[defined if cURL is not to verify cert/host])
 
 curl_ssl_verify_peer="true"
 AC_ARG_ENABLE([ssl-verify-peer],
  AC_HELP_STRING([--disable-ssl-verify-peer],[disable cURL cert validity verification]),
  [ test "${enableval}" = "no" && curl_ssl_verify_peer="false" ]
 )
 ${curl_ssl_verify_peer} || AC_DEFINE([DISABLE_CURL_SSL_VERIFYPEER],,[defined if cURL is not to verify cert validity])
 
 postels_law=true
 AC_ARG_ENABLE([postels-law],
  AC_HELP_STRING([--disable-postels-law],[Be strict, do not adhere to Postel's Law ("be conservative in what you do, be liberal in what you accept from others", RFC 793)]),
  [ test "${enableval}" = "no" && postels_law=false ]
 )
 $postels_law && AC_DEFINE([POSTELS_LAW],,[defined if we want to adhere to Postel's Law])
 
 AC_DEFINE_UNQUOTED([OPKELE_SRC_DIR],["$PWD"],[source directory])
 
 nitpick=false
 AC_ARG_ENABLE([nitpicking],
  AC_HELP_STRING([--enable-nitpicking],[make compiler somewhat overly fastidious about the code it deals with]),
  [ test "$enableval" = "no" || nitpick=true ]
 )
 if $nitpick ; then
  CPP_NITPICK="-pedantic -Wall -Wextra -Wundef -Wshadow \
   -Wunsafe-loop-optimizations -Wconversion -Wmissing-format-attribute \
   -Wredundant-decls -ansi"
  # -Wlogical-op -Wmissing-noreturn 
  C_NITPICK="$CPP_NITPICK"
  CXX_NITPICK="$C_NITPICK"
 
  CPPFLAGS="$CPPFLAGS $CPP_NITPICK"
  CFLAGS="$CFLAGS $C_NITPICK"
  CXXFLAGS="$CXXFLAGS $CXX_NITPICK"
 fi
 
 ndebug=true
 AC_ARG_ENABLE([debug],
  AC_HELP_STRING([--enable-debug],[enable debugging code]),
  [ test "$enableval" = "no" || ndebug=false ]
 )
 if $ndebug ; then
  CPPFLAGS_DEBUG="-DNDEBUG"
 else
  CPPFLAGS_DEBUG=""
 fi
 AC_SUBST([CPPFLAGS_DEBUG])
 
 xri_proxy_url="https://xri.net/"
 AC_MSG_CHECKING([for XRI resolver proxy])
 AC_ARG_ENABLE([xri-proxy],
  AC_HELP_STRING([--with-xri-proxy=url],[set xri proxy for use when resolving xri identities, default is https://xri.net/]),
  [ xri_proxy_url="$withval" ]
 )
 AC_MSG_RESULT([$xri_proxy_url])
 AC_DEFINE_UNQUOTED([XRI_PROXY_URL],["$xri_proxy_url"],[XRI proxy resolver URL])
 
 AC_CONFIG_FILES([
  Makefile
   libopkele.pc
   Doxyfile
  include/Makefile
   include/opkele/tr1-mem.h
  lib/Makefile
  test/Makefile
 ])
 AC_OUTPUT
diff --git a/include/opkele/tidy.h b/include/opkele/tidy.h
index 888e7d4..059656d 100644
--- a/include/opkele/tidy.h
+++ b/include/opkele/tidy.h
@@ -1,73 +1,80 @@
 #ifndef __OPKELE_TIDY_H
 #define __OPKELE_TIDY_H
 
 #include <cassert>
-#include <tidy.h>
-#include <buffio.h>
+#ifdef HAVE_TIDY_H
+# include <tidy.h>
+# include <buffio.h>
+#elif HAVE_TIDY_TIDY_H
+# include <tidy/tidy.h>
+# include <tidy/buffio.h>
+#else
+# error "Don't know where to look for htmltidy headers"
+#endif
 
 namespace opkele {
     namespace util {
 
         class tidy_buf_t {
             public:
                 TidyBuffer _x;
 
                 tidy_buf_t() { tidyBufInit(&_x); }
                 virtual ~tidy_buf_t() throw() {
                     tidyBufFree(&_x); }
 
                 inline operator const TidyBuffer&(void) const { return _x; }
                 inline operator TidyBuffer&(void) { return _x; }
 
                 inline operator const char*(void) const { return (const char*)_x.bp; }
                 inline operator char*(void) { return (char*)_x.bp; }
 
                 inline const char *c_str() const {
                     return (const char*)_x.bp; }
                 inline size_t size() const {
                     return _x.size; }
         };
 
         class tidy_doc_t {
             public:
                 TidyDoc _x;
 
                 tidy_doc_t() : _x(0) { }
                 tidy_doc_t(TidyDoc x) : _x(x) { }
                 virtual ~tidy_doc_t() throw() {
                     if(_x) tidyRelease(_x); }
 
                 tidy_doc_t& operator=(TidyDoc x) {
                     if(_x) tidyRelease(_x);
                     _x = x;
                     return *this;
                 }
 
                 operator const TidyDoc(void) const { return _x; }
                 operator TidyDoc(void) { return _x; }
 
                 inline bool opt_set(TidyOptionId o,bool v) {
                     assert(_x);
                     return tidyOptSetBool(_x,o,v?yes:no); }
                 inline bool opt_set(TidyOptionId o,int v) {
                     assert(_x);
                     return tidyOptSetInt(_x,o,v); }
 
                 inline int parse_string(const string& s) {
                     assert(_x);
                     return tidyParseString(_x,s.c_str()); }
                 inline int clean_and_repair() {
                     assert(_x);
                     return tidyCleanAndRepair(_x); }
                 inline int save_buffer(TidyBuffer& ob) {
                     assert(_x);
                     return tidySaveBuffer(_x,&ob); }
 
                 static inline TidyDoc create() {
                     return tidyCreate(); }
         };
 
     }
 }
 
 #endif /* __OPKELE_TIDY_H */
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)