Produce a user agent log if the information is in the extended log

The extended log can contain the user agent name. If it is available, we
produce the user agent log.

diff --git a/include/readlog.h b/include/readlog.h
index a481362292df1dce3eb1290fd7c7e7674214e05d..27a21c3eb10bbe88708f671d6776c1331e4553f9 100644 (file)
--- a/include/readlog.h
+++ b/include/readlog.h
@@ -43,6 +43,8 @@ struct ReadLogStruct
        char *HttpCode;
        //! HTTP method or NULL if the information is not stored in the log.
        char *HttpMethod;
+       //! Useragent string or NULL if it isn't available
+       const char *UserAgent;
 };
 
 /*!

diff --git a/readlog.c b/readlog.c
index d81032e79315a77a4bf787ede68741d38d19ff6d..3805de0cb35f4355509a154933aa6fe797108e72 100644 (file)
--- a/readlog.c
+++ b/readlog.c
@@ -332,6 +332,7 @@ static void ReadOneLogFile(struct ReadLogDataStruct *Filter,const char *arq)
        struct userfilestruct *ufile1;
        struct ReadLogStruct log_entry;
        struct LogLineStruct log_line;
+       FILE *UseragentLog=NULL;
 
        LogLine_Init(&log_line);
        LogLine_File(&log_line,arq);
@@ -714,6 +715,12 @@ static void ReadOneLogFile(struct ReadLogDataStruct *Filter,const char *arq)
                denied_write(&log_entry);
                authfail_write(&log_entry);
                if (download_flag) download_write(&log_entry,download_url);
+               if (log_entry.UserAgent)
+               {
+                       if (!UseragentLog)
+                               UseragentLog=UserAgent_Open();
+                       UserAgent_Write(UseragentLog,log_entry.Ip,log_entry.User,log_entry.UserAgent);
+               }
 
                if (log_line.current_format!=&ReadSargLog) {
                        if (period.start.tm_year==0 || idata<mindate || compare_date(&period.start,&log_entry.EntryTime)>0){
@@ -744,6 +751,7 @@ static void ReadOneLogFile(struct ReadLogDataStruct *Filter,const char *arq)
                debuga(__FILE__,__LINE__,_("Read error in \"%s\": %s\n"),arq,FileObject_GetLastCloseError());
                exit(EXIT_FAILURE);
        }
+       if (UseragentLog) fclose(UseragentLog);
        if (ShowReadStatistics) {
                if (ShowReadPercent)
                        printf(_("SARG: Records in file: %lu, reading: %3.2f%%\n"),recs2, (float) 100 );

diff --git a/readlog_extlog.c b/readlog_extlog.c
index 33491d5b774631263824e1fe5f06f30aea051593..1b8d735ff63bd29e4094207c95427a79b965de57 100644 (file)
--- a/readlog_extlog.c
+++ b/readlog_extlog.c
@@ -50,6 +50,7 @@ enum ext_col_id {
        EXTCOL_Path,
        EXTCOL_Query,
        EXTCOL_Status,
+       EXTCOL_UserAgent,
        EXTCOL_Last //last entry of the list !
 };
 
@@ -161,6 +162,8 @@ static bool ExtLog_Fields(const char *columns)
                        if (strncasecmp(columns,"cs-uri-query",len)==0) col_id=EXTCOL_Query;
                } else if (len==13) {
                        if (strncasecmp(columns,"cs-uri-scheme",len)==0) col_id=EXTCOL_Scheme;
+               } else if (len==14) {
+                       if (strncasecmp(columns,"cs(User-Agent)",len)==0) col_id=EXTCOL_UserAgent;
                }
                if (col_id!=EXTCOL_Last) {
                        ExtCols[col_id]=col;
@@ -227,23 +230,22 @@ Scan through the string of a column.
 static char *ExtLog_GetString(char *Line,int col,char **End)
 {
        bool quote;
-       bool dequote;
 
        //skip opening double quote
        quote=(*Line=='\"');
        if (quote) ++Line;
 
-       dequote=false;
        while (*Line) {
                if (quote) {
                        if (*Line=='\"') {
-                               if (Line[1]!='\"') {
-                                       if (End) *End=(dequote) ? NULL : Line;
-                                       Line++;//skip the closing quote
+                               if (Line[1]=='\"') {
+                                       Line++;//skip the first quote here, the second is skipped by the other Line++
+                               } else {
+                                       if (End) *End=Line;
+                                       Line++;//skip closing quote
                                        quote=false;
                                        break;
                                }
-                               dequote=true;
                        }
                } else {
                        if (*Line==ExtColSep[col]) {
@@ -367,12 +369,16 @@ static void ExtLog_FixString(char *string,char *end_ptr)
        char *dest;
 
        if (!string) return;//string not parsed
-       if (end_ptr) { //end is known and no quotes are in the string
+       if (*string!='\"' && end_ptr) { //no quotes to remove from the string
                *end_ptr='\0';
                return;
        }
-       // remove the quotes and end at the first unremoveable quote
+
+       // remove first quote
        dest=string;
+       if (string[1]!='\"') string++;
+
+       // remove the quotes and end at the first unremoveable quote
        while (*string)
        {
                if (*string=='\"') {
@@ -502,6 +508,7 @@ static enum ReadLogReturnCodeEnum ExtLog_ReadEntry(char *Line,struct ReadLogStru
        char *UrlPort=NULL,*UrlPortEnd;
        char *UrlPath=NULL,*UrlPathEnd;
        char *UrlQuery=NULL,*UrlQueryEnd;
+       char *UserAgent=NULL,*UserAgentEnd;
 
        // is it a directive
        if (*Line=='#') {
@@ -581,6 +588,11 @@ static enum ReadLogReturnCodeEnum ExtLog_ReadEntry(char *Line,struct ReadLogStru
                                Line=ExtLog_GetString(Line,col,&HttpCodeEnd);
                                if (!Line) return(RLRC_Unknown);
                                break;
+                       case EXTCOL_UserAgent:
+                               UserAgent=Line;
+                               Line=ExtLog_GetString(Line,col,&UserAgentEnd);
+                               if (!Line) return(RLRC_Unknown);
+                               break;
                        case EXTCOL_Last://ignored column
                                Line=ExtLog_GetString(Line,col,NULL);
                                if (!Line) return(RLRC_Unknown);
@@ -614,6 +626,8 @@ static enum ReadLogReturnCodeEnum ExtLog_ReadEntry(char *Line,struct ReadLogStru
                ExtLog_FixString(UrlQuery,UrlQueryEnd);
                Entry->Url=ExtLog_ConcatUrl(UrlScheme,UrlHost,UrlPort,UrlPath,UrlQuery);
        }
+       ExtLog_FixString(UserAgent,UserAgentEnd);
+       Entry->UserAgent=ExtLog_FixEmptyString(UserAgent);
 
        return(RLRC_NoError);
 }

diff --git a/topuser.c b/topuser.c
index 762478f04f79458fc40d246d77d955fd02c3f1a3..06cca7dcdfc091c7c274f3d6179f48011199dafe 100644 (file)
--- a/topuser.c
+++ b/topuser.c
@@ -46,7 +46,6 @@ struct SortInfoStruct
 
 extern struct globalstatstruct globstat;
 extern bool smartfilter;
-extern FileListObject UserAgentLog;
 
 /*!
 Save the total number of users. The number is written in sarg-users and set
@@ -139,7 +138,7 @@ static void TopUser_HtmlReport(const char *ListFile,struct TopUserStatistics *St
                if (is_denied()) fprintf(fp_top3,"<tr><td class=\"link\" colspan=\"0\"><a href=\"denied.html\">%s</a></td></tr>\n",_("Denied accesses"));
                if (is_authfail()) fprintf(fp_top3,"<tr><td class=\"link\" colspan=\"0\"><a href=\"authfail.html\">%s</a></td></tr>\n",_("Authentication Failures"));
                if(smartfilter) fprintf(fp_top3,"<tr><td class=\"link\" colspan=\"0\"><a href=\"smartfilter.html\">%s</a></td></tr>\n",_("SmartFilter"));
-               if (!FileList_IsEmpty(UserAgentLog) && useragent_count) fprintf(fp_top3,"<tr><td class=\"link\" colspan=\"0\"><a href=\"useragent.html\">%s</a></td></tr>\n",_("Useragent"));
+               if (useragent_count) fprintf(fp_top3,"<tr><td class=\"link\" colspan=\"0\"><a href=\"useragent.html\">%s</a></td></tr>\n",_("Useragent"));
                fputs("<tr><td></td></tr>\n</table></div>\n",fp_top3);
        }