added URI normalization procedure to opkele::util

as specified in RFC3896, section 6

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

3 files changed, 208 insertions, 0 deletions

diff --git a/include/opkele/util.h b/include/opkele/util.h
index edc1859..085c9e6 100644
--- a/include/opkele/util.h
+++ b/include/opkele/util.h
@@ -1,133 +1,143 @@
 #ifndef __OPKELE_UTIL_H
 #define __OPKELE_UTIL_H
 
 #include <time.h>
 #include <string>
 #include <vector>
 #include <openssl/bn.h>
 #include <openssl/dh.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);
 
 	/**
 	 * 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);
     }
 
 }
 
 #endif /* __OPKELE_UTIL_H */
diff --git a/lib/util.cc b/lib/util.cc
index 26be66a..eacf6d7 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -1,164 +1,295 @@
 #include <errno.h>
 #include <cassert>
+#include <cctype>
 #include <cstring>
 #include <vector>
 #include <string>
+#include <stack>
 #include <openssl/bio.h>
 #include <openssl/evp.h>
 #include <curl/curl.h>
 #include "opkele/util.h"
 #include "opkele/exception.h"
 
 namespace opkele {
     using namespace std;
 
     namespace util {
 
 	/*
 	 * base64
 	 */
 	string encode_base64(const void *data,size_t length) {
 	    BIO *b64 = 0, *bmem = 0;
 	    try {
 		b64 = BIO_new(BIO_f_base64());
 		if(!b64)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_new() base64 encoder");
 		BIO_set_flags(b64,BIO_FLAGS_BASE64_NO_NL);
 		bmem = BIO_new(BIO_s_mem());
 		BIO_set_flags(b64,BIO_CLOSE);
 		if(!bmem)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_new() memory buffer");
 		BIO_push(b64,bmem);
 		if(((size_t)BIO_write(b64,data,length))!=length)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_write()");
 		if(BIO_flush(b64)!=1)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_flush()");
 		char *rvd;
 		long rvl = BIO_get_mem_data(bmem,&rvd);
 		string rv(rvd,rvl);
 		BIO_free_all(b64);
 		return rv;
 	    }catch(...) {
 		if(b64) BIO_free_all(b64);
 		throw;
 	    }
 	}
 
 	void decode_base64(const string& data,vector<unsigned char>& rv) {
 	    BIO *b64 = 0, *bmem = 0;
 	    rv.clear();
 	    try {
 		bmem = BIO_new_mem_buf((void*)data.data(),data.size());
 		if(!bmem)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_new_mem_buf()");
 		b64 = BIO_new(BIO_f_base64());
 		if(!b64)
 		    throw exception_openssl(OPKELE_CP_ "failed to BIO_new() base64 decoder");
 		BIO_set_flags(b64,BIO_FLAGS_BASE64_NO_NL);
 		BIO_push(b64,bmem);
 		unsigned char tmp[512];
 		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) {
 	    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 )
 		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;
 	}
 
 	/*
 	 *
 	 */
 
 	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 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 hext 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) {
+	     string rv;
+	     string::size_type colon = uri.find(':');
+	     if(colon==string::npos)
+		 throw bad_input(OPKELE_CP_ "No scheme specified in URI");
+	     transform(
+		     uri.begin(), uri.begin()+colon+1,
+		     back_inserter(rv), ::tolower );
+	     bool s;
+	     if(rv=="http:")
+		 s = false;
+	     else if(rv=="https:")
+		 s = true;
+	     else
+		 throw not_implemented(OPKELE_CP_ "Only http(s) URIs can be normalized here");
+	     string::size_type ul = uri.length();
+	     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");
+	     rv += "//";
+	     string::size_type interesting = uri.find_first_of(":/#?",colon+3);
+	     if(interesting==string::npos) {
+		 transform(
+			 uri.begin()+colon+3,uri.end(),
+			 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[6];
+		     snprintf(tmp,sizeof(tmp),"%d",port);
+		     rv += ':'; 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;
+		 }else{
+		     pseg.append(uri,n,unsafe-n);
+		     n = unsafe;
+		 }
+		 char c = uri[n++];
+		 if(c=='%') {
+		     if((n+1)>=ul)
+			 throw bad_input(OPKELE_CP_ "Unexpected end of URI encountered while parsing percent-encoded character");
+		     tmp[0] = uri[n++];
+		     tmp[1] = uri[n++];
+		     if(!( isxdigit(tmp[0]) && isxdigit(tmp[1]) ))
+			 throw bad_input(OPKELE_CP_ "Invalid percent-encoded character in URI being normalized");
+		     int cc = strtol(tmp,0,16);
+		     if( isalpha(cc) || isdigit(cc) || strchr("._~-",cc) )
+			 pseg += cc;
+		     else{
+			 pseg += '%';
+			 pseg += toupper(tmp[0]); pseg += toupper(tmp[1]);
+		     }
+		 }else if(qf) {
+		     rv += pseg; rv += c;
+		     pseg.clear();
+		 }else if(n>=ul || strchr("?/#",c)) {
+		     if(pseg.empty() || pseg==".") {
+		     }else if(pseg=="..") {
+			 if(psegs.size()>1) {
+			     rv.resize(psegs.top()); psegs.pop();
+			 }
+		     }else{
+			 psegs.push(rv.length());
+			 if(c!='/') {
+			     pseg += c;
+			     qf = true;
+			 }
+			 rv += '/'; rv += pseg;
+		     }
+		     if(c=='/' && (n>=ul || strchr("?#",uri[n])) ) {
+			 rv += '/';
+			 if(n<ul)
+			     qf = true;
+		     }else if(strchr("?#",c)) {
+			 if(psegs.size()==1 && psegs.top()==rv.length())
+			     rv += '/';
+			 if(pseg.empty())
+			     rv += c;
+			 qf = true;
+		     }
+		     pseg.clear();
+		 }else{
+		     pseg += c;
+		 }
+	     }
+	     if(!pseg.empty()) {
+		 rv += '/'; rv += pseg;
+	     }
+	     return rv;
+	 }
+
     }
 
 }
diff --git a/test/test.cc b/test/test.cc
index f92284c..1a012b5 100644
--- a/test/test.cc
+++ b/test/test.cc
@@ -1,72 +1,139 @@
 #include <iostream>
 #include <stdexcept>
 using namespace std;
 #include <opkele/exception.h>
 #include <opkele/consumer.h>
+#include <opkele/util.h>
 
 #include "config.h"
 
 class failed_test : public opkele::exception {
     public:
 	failed_test(OPKELE_E_PARS)
 	    : exception(OPKELE_E_CONS) { }
 };
 
 class dummy_consumer_t : public opkele::consumer_t {
     public:
 	virtual opkele::assoc_t store_assoc(const string& /* server */,const string& /* handle */,const opkele::secret_t& /* secret */,int /* expires_in */) {
 	    throw opkele::not_implemented(OPKELE_CP_ "Not implemented");
 	}
 	virtual opkele::assoc_t retrieve_assoc(const string& /* server */ ,const string& /* handle */) {
 	    throw opkele::not_implemented(OPKELE_CP_ "Not implemented");
 	}
 	virtual void invalidate_assoc(const string& /* server */,const string& /* handle */) {
 	    throw opkele::not_implemented(OPKELE_CP_ "Not implemented");
 	}
 };
 
 
 void test_retrieve_links(const string& f,bool success,const string& s="",const string& d="") {
     dummy_consumer_t dc;
     string server, delegate;
     try {
 	dc.retrieve_links("file://" OPKELE_SRC_DIR "/test/html/"+f,server,delegate);
 	if(!success)
 	    throw failed_test(OPKELE_CP_ "Retrieved links when it shouldn't");
 	if(server!=s)
 	    throw failed_test(OPKELE_CP_ "retrieve_links test failed, expected server '"+s+"', got '"+server+"'");
 	if(delegate!=d)
 	    throw failed_test(OPKELE_CP_ "retrieve_links test failed, expected delegate '"+d+"', got '"+delegate+"'");
     }catch(opkele::bad_input& obi) {
 	if(success)
 	    throw failed_test(OPKELE_CP_ "Test '"+f+"' failed due to 'bad_input'["+obi.what()+"]");
     }catch(opkele::failed_assertion& ofa) {
 	if(success)
 	    throw failed_test(OPKELE_CP_ "Test '"+f+"' failed due to 'failed_assertion'["+ofa.what()+"]");
     }
 }
 
 void test_retrieve_links() {
     test_retrieve_links("empty.html",false);
     test_retrieve_links("in-body.html",false);
     test_retrieve_links("head-in-body.html",false);
     test_retrieve_links("hkn.html",true,"http://www.klever.net/openid.server","http://hacker.klever.net/");
     test_retrieve_links("hkn-server.html",true,"http://www.klever.net/openid.server");
     test_retrieve_links("hkn-delegate.html",false);
     test_retrieve_links("unclosed-head.html",true,"http://www.klever.net/openid.server","http://hacker.klever.net/");
     test_retrieve_links("spaced-links.html",true,"http://www.klever.net/openid.server","http://hacker.klever.net/");
     test_retrieve_links("spaced-link-attrs.html",true,"http://www.klever.net/openid.server","http://hacker.klever.net/");
     test_retrieve_links("2rels.html",true,"http://www.klever.net/openid.server","http://www.klever.net/openid.server");
 }
 
+void test_rfc_3986_normalize_uri(const string &ouri,bool success,const string& nuri="") {
+    try {
+	string n = opkele::util::rfc_3986_normalize_uri(ouri);
+	if(!success)
+	    throw failed_test(OPKELE_CP_ "Normalized URI when it shouldn't");
+	if(n!=nuri)
+	    throw failed_test(OPKELE_CP_ "rfc_3986_test_failed for '"+ouri+"' failed, expected '"+nuri+"', got '"+n+"'");
+    }catch(opkele::bad_input& obi) {
+	if(success)
+	    throw failed_test(OPKELE_CP_ "Test '"+ouri+"' failed due to 'bad_input'["+obi.what()+"]");
+    }catch(opkele::not_implemented& oni) {
+	if(success)
+	    throw failed_test(OPKELE_CP_ "Test '"+ouri+"' failed due to 'not_implemented'["+oni.what()+"]");
+    }
+}
+
+void test_rfc_3986_normalize_uri() {
+    test_rfc_3986_normalize_uri(
+	    "invalid", false );
+    test_rfc_3986_normalize_uri(
+	    "ftp://hacker.klever.net/", false );
+    test_rfc_3986_normalize_uri(
+	    "http://", false );
+    test_rfc_3986_normalize_uri(
+	    "http:/hacker.klever.net/", false );
+    test_rfc_3986_normalize_uri(
+	    "hTTp://hacker.klever.net#uh?oh", true, "http://hacker.klever.net/#uh?oh" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net?uh#oh", true, "http://hacker.klever.net/?uh#oh" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net:80/", true, "http://hacker.klever.net/" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net:80?uh", true, "http://hacker.klever.net/?uh" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net:80#uh", true, "http://hacker.klever.net/#uh" );
+    test_rfc_3986_normalize_uri(
+	    "https://hacker.klever.net:443", true, "https://hacker.klever.net/" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net:?oh", true, "http://hacker.klever.net/?oh" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah%2E", true, "http://hacker.klever.net/ah." );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%2E/", true, "http://hacker.klever.net/ah/" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%2b/", true, "http://hacker.klever.net/ah/%2B/" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/./oh?eh", true, "http://hacker.klever.net/ah/oh?eh" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/../oh?", true, "http://hacker.klever.net/oh?" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah//oh?", true, "http://hacker.klever.net/ah/oh?" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/?", true, "http://hacker.klever.net/ah/?" );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%", false );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%a", false );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%zx", false );
+    test_rfc_3986_normalize_uri(
+	    "http://hacker.klever.net/ah/%5x", false );
+    test_rfc_3986_normalize_uri(
+	    "Http://Hacker.Klever.Net:", true, "http://hacker.klever.net/" );
+}
+
 int main() {
     try {
+	test_rfc_3986_normalize_uri();
 	test_retrieve_links();
     }catch(failed_test& ft) {
 	cerr << "Test failed: " << ft.what() << endl;
     }catch(exception& e) {
 	cerr << "oops: " << e.what() << endl;
 	_exit(1);
     }
     _exit(0);
 }