made util::url_encode refrain from encoding unreserved chars

as per rfc 3986

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

1 files changed, 33 insertions, 6 deletions

diff --git a/lib/util.cc b/lib/util.cc
index 29e6738..a6e08e2 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -1,61 +1,60 @@
 #include <errno.h>
 #include <cassert>
 #include <cctype>
 #include <cstring>
 #include <vector>
 #include <string>
 #include <stack>
 #include <algorithm>
 #include <openssl/bio.h>
 #include <openssl/evp.h>
 #include <openssl/sha.h>
 #include <openssl/hmac.h>
-#include <curl/curl.h>
 #include <opkele/util.h>
 #include <opkele/exception.h>
 #include <opkele/data.h>
 #include <opkele/debug.h>
 
 #include <config.h>
 #ifdef HAVE_DEMANGLE
 # include <cxxabi.h>
 #endif
 
 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();
@@ -111,102 +110,130 @@ namespace opkele {
 
         /*
          * 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;
         }
 
         /*
          *
          */
 
+        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) {
-            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);
+            string rv;
+            for_each(str.begin(),str.end(),
+                    __url_encoder(rv));
             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 '/'