Improve a bit the processing of the URL

Detect more strictly the scheme prefixing the URL to avoid any false
positive.

The caller decides if the full URL must be kept or truncated.

diff --git a/include/defs.h b/include/defs.h
index 440948b5b5e08933fa65d7c162dd841087a424aa..7c9e508f8deaa2bf37271a7742e9d32851fe72a2 100755 (executable)
--- a/include/defs.h
+++ b/include/defs.h
@@ -193,7 +193,7 @@ void day_totalize(const char *tmp, const struct userinfostruct *uinfo);
 // url.c
 void read_hostalias(const char *Filename);
 void free_hostalias(void);
-const char *process_url(char *url);
+const char *process_url(char *url,bool full_url);
 void url_hostname(const char *url,char *hostname,int hostsize);
 
 // usage.c

diff --git a/log.c b/log.c
index 59aca0dfad8516def2334e062984d6ad3a5f918f..a0eedf1b1a6e3e90daaca30f9a4691854c679230 100644 (file)
--- a/log.c
+++ b/log.c
@@ -1338,7 +1338,7 @@ int main(int argc,char *argv[])
                        } else
                                download_flag=false;
 
-                       url=process_url(full_url);
+                       url=process_url(full_url,LongUrl);
                        if (!url || url[0] == '\0') continue;
 
                        if(addr[0] != '\0'){

diff --git a/url.c b/url.c
index 74324206872ddb13c8c9e2856adcc87e0c22dbe7..a4ee3658715b1b08532b3fca5b3056c8b9c2d616 100644 (file)
--- a/url.c
+++ b/url.c
@@ -227,6 +227,7 @@ void free_hostalias(void)
                free((void *)alias->Alias);
                free(alias);
        }
+       FirstAlias=NULL;
 }
 
 /*!
@@ -266,21 +267,23 @@ const char *alias_url(const char *url)
 Get the part of the URL necessary to generate the report.
 
 \param url The URL as extracted from the report.
+\param full_url \c True to keep the whole URL. If \c false,
+the URL is truncated to only keep the host name and port number.
 */
-const char *process_url(char *url)
+const char *process_url(char *url,bool full_url)
 {
        char *str;
        const char *start;
 
        // remove any scheme:// at the beginning of the URL (see rfc2396 section 3.1)
        for (str=url ; *str && (isalnum(*str) || *str=='+' || *str=='-' || *str=='.') ; str++);
-       if (*str==':' && str[1]=='/') {
-               url=str+1;
+       if (*str==':' && str[1]=='/' && str[2]=='/') {
+               url=str+3;
                while (*url=='/') url++;
        }
 
        start=url;
-       if (!LongUrl) {
+       if (!full_url) {
                for (str=url ; *str && *str!='/' ; str++);
                if (*str=='/') *str='\0';
                if (FirstAlias)