aded util::url_decode()

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

2 files changed, 37 insertions, 1 deletions

diff --git a/include/opkele/util.h b/include/opkele/util.h
index fd974a1..0130bff 100644
--- a/include/opkele/util.h
+++ b/include/opkele/util.h
@@ -1,89 +1,97 @@
 #ifndef __OPKELE_UTIL_H
 #define __OPKELE_UTIL_H
 
 #include <time.h>
 #include <string>
 #include <vector>
 #include <opkele/types.h>
 
 namespace opkele {
     using std::string;
     using std::vector;
 
     /**
      * @brief opkele utils namespace
      */
     namespace util {
 
 	/**
 	 * 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.
+	 * 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);
 
 	/**
+	 * Decode url-encoded string back to normal
+	 * @param str url-encoded string
+	 * @return decoded string
+	 * @throw failed_conversion in case of failure
+	 */
+	string url_decode(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 normalize_identifier(const string& usi,bool strip_fragment);
 
 	/**
diff --git a/lib/util.cc b/lib/util.cc
index a6e08e2..3e7f3aa 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -147,96 +147,124 @@ namespace opkele {
 	    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;
 	}
 
 	/*
 	 *
 	 */
 
 	static inline bool isrfc3986unreserved(int c) {
 	    if(c<'-') return false;
 	    if(c<='.') return true;
 	    if(c<'0') return false; if(c<='9') return true;
 	    if(c<'A') return false; if(c<='Z') return true;
 	    if(c<'_') return false;
 	    if(c=='_') return true;
 	    if(c<'a') return false; if(c<='z') return true;
 	    if(c=='~') return true;
 	    return false;
 	}
 
 	struct __url_encoder : public unary_function<char,void> {
 	    public:
 		string& rv;
 
 		__url_encoder(string& r) : rv(r) { }
 
 		result_type operator()(argument_type c) {
 		    if(isrfc3986unreserved(c))
 			rv += c;
 		    else{
 			char tmp[4];
 			snprintf(tmp,sizeof(tmp),"%%%02X",
 				(c&0xff));
 			rv += tmp;
 		    }
 		}
 	};
 
 	string url_encode(const string& str) {
 	    string rv;
 	    for_each(str.begin(),str.end(),
 		    __url_encoder(rv));
 	    return rv;
 	}
 
+	string url_decode(const string& str) {
+	    string rv;
+	    back_insert_iterator<string> ii(rv);
+	    for(string::const_iterator i=str.begin(),ie=str.end();
+		    i!=ie;++i) {
+		switch(*i) {
+		    case '+':
+			*(ii++) = ' '; break;
+		    case '%':
+			++i;
+			static char tmp[3] = {0,0,0};
+			if(i==ie)
+			    throw failed_conversion(OPKELE_CP_ "trailing percent in the url-encoded string");
+			tmp[0] == *(i++);
+			if(i==ie)
+			    throw failed_conversion(OPKELE_CP_ "not enough hexadecimals after the percent sign in url-encoded string");
+			tmp[1] == *i;
+			if(!(isxdigit(tmp[0]) && isxdigit(tmp[1])))
+			    throw failed_conversion(OPKELE_CP_ "non-hex follows percent in url-encoded string");
+			*(ii++) = strtol(tmp,0,16);
+			break;
+		    default:
+			*(ii++) = *i; break;
+		}
+	    }
+	    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) {
 	     string rv;