-rw-r--r-- | include/opkele/util.h | 7 | ||||
-rw-r--r-- | lib/util.cc | 19 |
2 files changed, 26 insertions, 0 deletions
diff --git a/include/opkele/util.h b/include/opkele/util.h index 6f3ddf6..719f951 100644 --- a/include/opkele/util.h +++ b/include/opkele/util.h @@ -7,163 +7,170 @@ #include <openssl/bn.h> #include <openssl/dh.h> #include <opkele/types.h> namespace opkele { using std::string; using std::vector; /** * @brief opkele utils namespace */ namespace util { /** * Convenience class encapsulating SSL BIGNUM object for the purpose of * automatical freeing. */ class bignum_t { public: BIGNUM *_bn; bignum_t() : _bn(0) { } bignum_t(BIGNUM *bn) : _bn(bn) { } ~bignum_t() throw() { if(_bn) BN_free(_bn); } bignum_t& operator=(BIGNUM *bn) { if(_bn) BN_free(_bn); _bn = bn; return *this; } operator const BIGNUM*(void) const { return _bn; } operator BIGNUM*(void) { return _bn; } }; /** * Convenience clas encapsulating SSL DH object for the purpose of * automatic freeing. */ class dh_t { public: DH *_dh; dh_t() : _dh(0) { } dh_t(DH *dh) : _dh(dh) { } ~dh_t() throw() { if(_dh) DH_free(_dh); } dh_t& operator=(DH *dh) { if(_dh) DH_free(_dh); _dh = dh; return *this; } operator const DH*(void) const { return _dh; } operator DH*(void) { return _dh; } DH* operator->() { return _dh; } const DH* operator->() const { return _dh; } }; /** * Convert base64-encoded SSL BIGNUM to internal representation. * @param b64 base64-encoded number * @return SSL BIGNUM * @throw failed_conversion in case of error */ BIGNUM *base64_to_bignum(const string& b64); /** * Convert decimal representation to SSL BIGNUM. * @param dec decimal representation * @return resulting BIGNUM * @throw failed_conversion in case of error */ BIGNUM *dec_to_bignum(const string& dec); /** * Convert SSL BIGNUM data to base64 encoded string. * @param bn BIGNUM * @return base64encoded string */ string bignum_to_base64(const BIGNUM *bn); /** * Convert internal time representation to w3c format * @param t internal representation * @return w3c time * @throw failed_conversion in case of error */ string time_to_w3c(time_t t); /** * Convert W3C time representation to internal time_t * @param w w3c representation * @return converted time * @throw failed_conversion in case of error */ time_t w3c_to_time(const string& w); /** * Encode string to the representation suitable for using in URL. * @param str string to encode * @return encoded string * @throw failed_conversion in case of failure */ string url_encode(const string& str); /** + * Make string suitable for using as x(ht)ml attribute. + * @param str string to escape + * @return escaped string + */ + string attr_escape(const string& str); + + /** * Convert number to string * @param l number * @return string representation * @throw failed_conversion in case of failure */ string long_to_string(long l); /** * Convert string to number * @param s string, containing the number * @return the number * @throw failed_conversion in case of failure */ long string_to_long(const string& s); /** * Encode binary data using base64. * @param data pointer to binary data * @param length length of data * @return encoded data */ string encode_base64(const void *data,size_t length); /** * Decode binary data from base64 representation. * @param data base64-encoded data * @param rv container for decoded binary */ void decode_base64(const string& data,vector<unsigned char>& rv); /** * Normalize http(s) URI according to RFC3986, section 6. URI is * expected to have scheme: in front of it. * @param uri URI * @return normalized URI * @throw not_implemented in case of non-httpi(s) URI * @throw bad_input in case of malformed URI */ string rfc_3986_normalize_uri(const string& uri); string& strip_uri_fragment_part(string& uri); string abi_demangle(const char* mn); string base64_signature(const assoc_t& assoc,const basic_openid_message& om); class change_mode_message_proxy : public basic_openid_message { public: const basic_openid_message& x; const string& mode; change_mode_message_proxy(const basic_openid_message& xx,const string& m) : x(xx), mode(m) { } bool has_field(const string& n) const { return x.has_field(n); } const string& get_field(const string& n) const { return (n=="mode")?mode:x.get_field(n); } bool has_ns(const string& uri) const {return x.has_ns(uri); } string get_ns(const string& uri) const { return x.get_ns(uri); } fields_iterator fields_begin() const { return x.fields_begin(); } fields_iterator fields_end() const { return x.fields_end(); } }; } } #endif /* __OPKELE_UTIL_H */ diff --git a/lib/util.cc b/lib/util.cc index 54d6535..b7bc437 100644 --- a/lib/util.cc +++ b/lib/util.cc @@ -68,192 +68,211 @@ namespace opkele { size_t rb = 0; while((rb=BIO_read(b64,tmp,sizeof(tmp)))>0) rv.insert(rv.end(),tmp,&tmp[rb]); BIO_free_all(b64); }catch(...) { if(b64) BIO_free_all(b64); throw; } } /* * big numerics */ BIGNUM *base64_to_bignum(const string& b64) { vector<unsigned char> bin; decode_base64(b64,bin); BIGNUM *rv = BN_bin2bn(&(bin.front()),bin.size(),0); if(!rv) throw failed_conversion(OPKELE_CP_ "failed to BN_bin2bn()"); return rv; } BIGNUM *dec_to_bignum(const string& dec) { BIGNUM *rv = 0; if(!BN_dec2bn(&rv,dec.c_str())) throw failed_conversion(OPKELE_CP_ "failed to BN_dec2bn()"); return rv; } string bignum_to_base64(const BIGNUM *bn) { vector<unsigned char> bin(BN_num_bytes(bn)+1); unsigned char *binptr = &(bin.front())+1; int l = BN_bn2bin(bn,binptr); if(l && (*binptr)&0x80){ (*(--binptr)) = 0; ++l; } return encode_base64(binptr,l); } /* * w3c times */ string time_to_w3c(time_t t) { struct tm tm_t; if(!gmtime_r(&t,&tm_t)) throw failed_conversion(OPKELE_CP_ "failed to BN_dec2bn()"); char rv[25]; if(!strftime(rv,sizeof(rv)-1,"%Y-%m-%dT%H:%M:%SZ",&tm_t)) throw failed_conversion(OPKELE_CP_ "failed to strftime()"); return rv; } time_t w3c_to_time(const string& w) { int fraction; struct tm tm_t; memset(&tm_t,0,sizeof(tm_t)); if( ( sscanf( w.c_str(), "%04d-%02d-%02dT%02d:%02d:%02dZ", &tm_t.tm_year,&tm_t.tm_mon,&tm_t.tm_mday, &tm_t.tm_hour,&tm_t.tm_min,&tm_t.tm_sec ) != 6 ) && ( sscanf( w.c_str(), "%04d-%02d-%02dT%02d:%02d:%02d.%03dZ", &tm_t.tm_year,&tm_t.tm_mon,&tm_t.tm_mday, &tm_t.tm_hour,&tm_t.tm_min,&tm_t.tm_sec, &fraction ) != 7 ) ) throw failed_conversion(OPKELE_CP_ "failed to sscanf()"); tm_t.tm_mon--; tm_t.tm_year-=1900; time_t rv = mktime(&tm_t); if(rv==(time_t)-1) throw failed_conversion(OPKELE_CP_ "failed to mktime()"); return rv-timezone; } /* * */ string url_encode(const string& str) { char * t = curl_escape(str.c_str(),str.length()); if(!t) throw failed_conversion(OPKELE_CP_ "failed to curl_escape()"); string rv(t); curl_free(t); return rv; } + string attr_escape(const string& str) { + static const char *unsafechars = "<>&\n\"'"; + string rv; + string::size_type p=0; + while(true) { + string::size_type us = str.find_first_of(unsafechars,p); + if(us==string::npos) { + if(p!=str.length()) + rv.append(str,p,str.length()-p); + return rv; + } + rv.append(str,p,us-p); + rv += "&#"; + rv += long_to_string((long)str[us]); + rv += ';'; + p = us+1; + } + } + string long_to_string(long l) { char rv[32]; int r=snprintf(rv,sizeof(rv),"%ld",l); if(r<0 || r>=(int)sizeof(rv)) throw failed_conversion(OPKELE_CP_ "failed to snprintf()"); return rv; } long string_to_long(const string& s) { char *endptr = 0; long rv = strtol(s.c_str(),&endptr,10); if((!endptr) || endptr==s.c_str()) throw failed_conversion(OPKELE_CP_ "failed to strtol()"); return rv; } /* * Normalize URL according to the rules, described in rfc 3986, section 6 * * - uppercase hex triplets (e.g. %ab -> %AB) * - lowercase scheme and host * - decode %-encoded characters, specified as unreserved in rfc 3986, section 2.3, * that is - [:alpha:][:digit:]._~- * - remove dot segments * - remove empty and default ports * - if there's no path component, add '/' */ string rfc_3986_normalize_uri(const string& uri) { static const char *whitespace = " \t\r\n"; string rv; string::size_type ns = uri.find_first_not_of(whitespace); if(ns==string::npos) throw bad_input(OPKELE_CP_ "Can't normalize empty URI"); string::size_type colon = uri.find(':',ns); if(colon==string::npos) throw bad_input(OPKELE_CP_ "No scheme specified in URI"); transform( uri.begin()+ns, uri.begin()+colon+1, back_inserter(rv), ::tolower ); bool s; string::size_type ul = uri.find_last_not_of(whitespace)+1; if(ul <= (colon+3)) throw bad_input(OPKELE_CP_ "Unexpected end of URI being normalized encountered"); if(uri[colon+1]!='/' || uri[colon+2]!='/') throw bad_input(OPKELE_CP_ "Unexpected input in URI being normalized after scheme component"); if(rv=="http:") s = false; else if(rv=="https:") s = true; else{ /* TODO: support more schemes. e.g. xri. How do we normalize * xri? */ rv.append(uri,colon+1,ul-colon-1); return rv; } rv += "//"; string::size_type interesting = uri.find_first_of(":/#?",colon+3); if(interesting==string::npos) { transform( uri.begin()+colon+3,uri.begin()+ul, back_inserter(rv), ::tolower ); rv += '/'; return rv; } transform( uri.begin()+colon+3,uri.begin()+interesting, back_inserter(rv), ::tolower ); bool qf = false; char ic = uri[interesting]; if(ic==':') { string::size_type ni = uri.find_first_of("/#?%",interesting+1); const char *nptr = uri.data()+interesting+1; char *eptr = 0; long port = strtol(nptr,&eptr,10); if( (port>0) && (port<65535) && port!=(s?443:80) ) { char tmp[8]; snprintf(tmp,sizeof(tmp),":%ld",port); rv += tmp; } if(ni==string::npos) { rv += '/'; return rv; } interesting = ni; }else if(ic!='/') { rv += '/'; rv += ic; qf = true; ++interesting; } string::size_type n = interesting; char tmp[3] = { 0,0,0 }; stack<string::size_type> psegs; psegs.push(rv.length()); string pseg; for(;n<ul;) { string::size_type unsafe = uri.find_first_of(qf?"%":"%/?#",n); if(unsafe==string::npos) { pseg.append(uri,n,ul-n-1); n = ul-1; |