Use internal user list instead of scanning the directory for users

Sarg used to scan the temporary directory for files to process and
extract the user name from the files. But, now, it is a waste of time
to read the directory as a list of the known users is alread in
memory.

This change also keep away stray files left by the crash of a previous
run.

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 13fa3159625c9da030289d8cfe6c1c63afc1a1d9..dd5a1b3d66fae4dc1371494a2461c1c0a8c2ba70 100755 (executable)
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -3,7 +3,7 @@ PROJECT(sarg C)
 SET(sarg_VERSION 2)
 SET(sarg_REVISION "3.2-pre1")
 SET(sarg_BUILD "")
-SET(sarg_BUILDDATE "Dec-21-2010")
+SET(sarg_BUILDDATE "Dec-26-2010")
 
 INCLUDE(AddFileDependencies)
 INCLUDE(CheckIncludeFile)

diff --git a/datafile.c b/datafile.c
index 36af6ae35f2d5cb0680a0821fe297c82c683ecd5..1b14fea69916e1c4d347e6581f9527f9d547b70c 100644 (file)
--- a/datafile.c
+++ b/datafile.c
@@ -43,10 +43,8 @@ void data_file(char *tmp)
        char crc2[50];
        char *str;
        char tmp3[MAXLEN];
-       char user[MAX_USER_LEN];
        char u2[MAX_USER_LEN];
-       DIR *dirp;
-       struct dirent *direntp;
+       userscan uscan;
        long long int nbytes=0;
        long long int nelap=0;
        long long int nacc=0;
@@ -55,7 +53,6 @@ void data_file(char *tmp)
        long long int oucache=0;
        long long int accbytes;
        long long int accelap;
-       int dlen;
        int new_user;
        int same_url;
        int url_len;
@@ -63,7 +60,6 @@ void data_file(char *tmp)
        struct getwordstruct gwarea;
        struct userinfostruct *uinfo;
        longline line;
-       const char logext[]=".log";
 
        ipantes[0]='\0';
        nameantes[0]='\0';
@@ -72,29 +68,12 @@ void data_file(char *tmp)
        oldurl=NULL;
        ourl_size=0;
 
-       if ((dirp = opendir(tmp)) == NULL) {
-               debuga(_("Failed to open directory %s - %s\n"),tmp,strerror(errno));
+       uscan=userinfo_startscan();
+       if (uscan == NULL) {
+               debuga(_("Cannot enumerate the user list\n"));
                exit(EXIT_FAILURE);
        }
-       while ( (direntp = readdir( dirp )) != NULL ) {
-               dlen=strlen(direntp->d_name)-(sizeof(logext)-1);
-               if (dlen<=0) continue;
-               if(strcmp(direntp->d_name+dlen,logext) != 0)
-                       continue;
-
-               if (dlen>0) {
-                       if (dlen>=sizeof(user)) continue;
-                       strncpy(user,direntp->d_name,dlen);
-                       user[dlen]=0;
-               } else {
-                       user[0]='\0';
-               }
-
-               uinfo=userinfo_find_from_file(user);
-               if (!uinfo) {
-                       debuga(_("Ignoring unknown user file %s\n"),user);
-                       continue;
-               }
+       while ( (uinfo = userinfo_advancescan(uscan)) != NULL ) {
                strcpy(u2,uinfo->id);
                if(Ip2Name && uinfo->id_is_ip) {
                        strcpy(ipantes,u2);
@@ -102,9 +81,10 @@ void data_file(char *tmp)
                        strcpy(nameantes,u2);
                }
                user_find(uinfo->label,MAX_USER_LEN, u2);
+               if(debug) debuga(_("Reading user file: %s/%s\n"),tmp,uinfo->filename);
 
-               if (snprintf(tmp3,sizeof(tmp3),"%s/%s",tmp,direntp->d_name)>=sizeof(tmp3)) {
-                       debuga(_("(datafile) directory path too long: %s/%s\n"),tmp,direntp->d_name);
+               if (snprintf(tmp3,sizeof(tmp3),"%s/%s.log",tmp,uinfo->filename)>=sizeof(tmp3)) {
+                       debuga(_("(datafile) directory path too long: %s/%s.log\n"),tmp,uinfo->filename);
                        exit(EXIT_FAILURE);
                }
 
@@ -210,7 +190,7 @@ void data_file(char *tmp)
                fclose(fp_in);
                longline_destroy(&line);
        }
-       (void)closedir( dirp );
+       userinfo_stopscan(uscan);
        if (oldurl) free(oldurl);
        if (fp_ou) fclose(fp_ou);
 

diff --git a/html.c b/html.c
index 5e8839e11aa17d151fa16c4807f9b1358b1e9377..907f1652ef191be488a941ddfb293e46a7355cb8 100644 (file)
--- a/html.c
+++ b/html.c
@@ -29,10 +29,8 @@
 
 void htmlrel(void)
 {
-       DIR *dirp;
        FILE *fp_in, *fp_ou, *fp_ip, *fp_ip2, *fp_usr;
 
-       struct dirent *direntp;
        long long int nnbytes=0, unbytes=0, tnbytes=0, totbytes=0, totbytes2=0;
        long long int totelap=0, totelap2=0, nnelap=0, unelap=0, tnelap=0;
        long long int incache=0, oucache=0, tnincache=0, tnoucache=0, twork=0;
@@ -42,10 +40,8 @@ void htmlrel(void)
        char *buf;
        char arqin[MAXLEN], arqou[MAXLEN], arqper[MAXLEN], arqip[MAXLEN];
        char *url, tmsg[50], csort[MAXLEN];
-       char user[MAXLEN], duser[MAXLEN];
+       char duser[MAXLEN];
        char user_ip[MAXLEN], olduserip[MAXLEN], tmp2[MAXLEN], tmp3[MAXLEN];
-       char denied_report[255];
-       char *str;
        char warea[MAXLEN];
        char totuser[8];
        char tmp6[MAXLEN];
@@ -54,15 +50,16 @@ void htmlrel(void)
        double perc=0, perc2=0, ouperc=0, inperc=0;
        int count;
        int cstatus;
-       const char txtext[]=".txt";
+       int have_denied_report;
        const char *sort_field;
        const char *sort_order;
-       int dlen;
+       const char *user;
        char siteind[MAX_TRUNCATED_URL];
        struct getwordstruct gwarea;
        longline line,line1;
        struct generalitemstruct item;
        const struct userinfostruct *uinfo;
+       userscan uscan;
 
        if((ReportType & REPORT_TYPE_USERS_SITES) == 0) return;
 
@@ -112,35 +109,13 @@ void htmlrel(void)
 
        sort_labels(&sort_field,&sort_order);
 
-       if ((dirp = opendir(tmp)) == NULL) {
-               debuga(_("Failed to open directory %s - %s\n"),tmp,strerror(errno));
+       uscan=userinfo_startscan();
+       if (uscan == NULL) {
+               debuga(_("Cannot enumerate the user list\n"));
                exit(EXIT_FAILURE);
        }
-       while ( (direntp = readdir( dirp )) != NULL ) {
-               dlen=strlen(direntp->d_name)-(sizeof(txtext)-1);
-               if (dlen<0) continue;
-               if(strcmp(direntp->d_name+dlen,txtext) != 0)
-                       continue;
-
-               count=1;
-               if (dlen>0) {
-                       if (dlen>=sizeof(user)) continue;
-                       strncpy(user,direntp->d_name,dlen);
-                       user[dlen]=0;
-               } else {
-                       user[0]=0;
-               }
-
-               uinfo=userinfo_find_from_file(user);
-               if (!uinfo) {
-                       debuga(_("Unknown user ID %s in directory %s\n"),user,tmp);
-                       exit(EXIT_FAILURE);
-               }
-               str=strrchr(user,'.');
-               if (str)
-                       strcpy(denied_report,str+1);
-               else
-                       strcpy(denied_report,user);
+       while ( (uinfo = userinfo_advancescan(uscan)) != NULL ) {
+               user=uinfo->filename;
 
                if (snprintf(warea,sizeof(warea),"%s/%s",outdirname,user)>=sizeof(warea)) {
                        debuga(_("Destination directory too long: %s/%s\n"),outdirname,user);
@@ -151,25 +126,28 @@ void htmlrel(void)
                report_day(uinfo);
                greport_day(uinfo);
 
-               if (snprintf(arqin,sizeof(arqin),"%s/%s",tmp,direntp->d_name)>=sizeof(arqin)) {
-                       debuga(_("Input file name too long: %s/%s\n"),tmp,direntp->d_name);
+               if (snprintf(arqin,sizeof(arqin),"%s/%s.txt",tmp,uinfo->filename)>=sizeof(arqin)) {
+                       debuga(_("Input file name too long: %s/%s.txt\n"),tmp,uinfo->filename);
+                       exit(EXIT_FAILURE);
+               }
+               if ((fp_in = fopen(arqin, "r")) == 0){
+                       if (uinfo->no_report) continue;
+                       debuga(_("(html3) Cannot open file %s\n"),arqin);
                        exit(EXIT_FAILURE);
                }
+
                if (snprintf(arqou,sizeof(arqou),"%s/%s/%s.html",outdirname,uinfo->filename,uinfo->filename)>=sizeof(arqou)) {
                        debuga(_("Output file name too long: %s/%s/%s.html\n"),outdirname,uinfo->filename,uinfo->filename);
                        exit(EXIT_FAILURE);
                }
-               if (snprintf(duser,sizeof(duser),"%s/%s/denied_%s.html",tmp,direntp->d_name,denied_report)>=sizeof(duser)) {
-                       debuga(_("File name too long: %s/%s/denied_%s.html\n"),tmp,direntp->d_name,denied_report);
+               if (snprintf(duser,sizeof(duser),"%s/%s/denied_%s.html",tmp,uinfo->filename,uinfo->filename)>=sizeof(duser)) {
+                       debuga(_("File name too long: %s/%s/denied_%s.html\n"),tmp,uinfo->filename,uinfo->filename);
                        exit(EXIT_FAILURE);
                }
                if(access(duser, R_OK) != 0)
-                       denied_report[0]='\0';
-
-               if ((fp_in = fopen(arqin, "r")) == 0){
-                       debuga(_("(html3) Cannot open file %s\n"),arqin);
-                       exit(EXIT_FAILURE);
-               }
+                       have_denied_report=0;
+               else
+                       have_denied_report=1;
 
                if ((line=longline_create())==NULL) {
                        debuga(_("Not enough memory to read file %s\n"),arqin);
@@ -241,8 +219,8 @@ void htmlrel(void)
                /*! \bug This line lays before the header of the table and is therefore invalid
                HTML. It will also break the sorttable script. Beside, I suspect it serves no purpose.
                */
-               if(denied_report[0]!='\0') {
-                       fprintf(fp_ou,"<tr><td class=\"header_l\" colspan=\"11\"><a href=\"denied_%s.html\">%s</a> %s</td></tr>\n",denied_report,_("SmartFilter"),_("Report"));
+               if(have_denied_report) {
+                       fprintf(fp_ou,"<tr><td class=\"header_l\" colspan=\"11\"><a href=\"denied_%s.html\">%s</a> %s</td></tr>\n",uinfo->filename,_("SmartFilter"),_("Report"));
                }
 
                fputs("<thead><tr><th class=\"sorttable_nosort\"></th><th class=\"header_l",fp_ou);
@@ -624,7 +602,7 @@ void htmlrel(void)
                htaccess(uinfo);
        }
 
-       (void)closedir(dirp);
+       userinfo_stopscan(uscan);
        greport_cleanup();
 
        return;

diff --git a/include/conf.h b/include/conf.h
index 7bb87dff1090b54e8b80e5fcb4c22a0e6482128d..2d5546c5e1e6b34b6e706f1b37e2ecc4dbe19fad 100755 (executable)
--- a/include/conf.h
+++ b/include/conf.h
@@ -472,6 +472,3 @@ typedef struct
 { int list[ 24 ];
   int len;
 } numlist;
-
-DIR *dirp;
-struct dirent *direntp;

diff --git a/include/defs.h b/include/defs.h
index 86b837059e88fd56c0e703f7f5608e92d88c5a57..f1de44a54c6610851159350fb3bc1ff746cf2778 100755 (executable)
--- a/include/defs.h
+++ b/include/defs.h
@@ -37,18 +37,25 @@ struct generalitemstruct
    long long oucache;
 };
 
+/*! \brief What is known about a user.
+*/
 struct userinfostruct
 {
-   //! The ID of the user as found in the input file.
-   char id[MAX_USER_LEN];
-   //! \c True if the ID is in fact the IP address from which the user connected.
-   bool id_is_ip;
-   //! The name of the user to display in the report.
-   char label[MAX_USER_LEN];
-   //! The mangled name to use in file names of that user.
-   char filename[MAX_USER_FNAME_LEN];
+       //! The ID of the user as found in the input file.
+       char id[MAX_USER_LEN];
+       //! \c True if the ID is in fact the IP address from which the user connected.
+       bool id_is_ip;
+       //! \c True if the user doesn't have a report file.
+       bool no_report;
+       //! The name of the user to display in the report.
+       char label[MAX_USER_LEN];
+       //! The mangled name to use in file names of that user.
+       char filename[MAX_USER_FNAME_LEN];
 };
 
+//! Scan through the known users.
+typedef struct userscanstruct *userscan;
+
 // auth.c
 void htaccess(const struct userinfostruct *uinfo);
 
@@ -180,6 +187,9 @@ void useragent(void);
 void userinfo_free(void);
 /*@shared@*/struct userinfostruct *userinfo_find_from_file(const char *filename);
 /*@shared@*/struct userinfostruct *userinfo_find_from_id(const char *id);
+userscan userinfo_startscan(void);
+void userinfo_stopscan(userscan uscan);
+struct userinfostruct *userinfo_advancescan(userscan uscan);
 
 // usertab.c
 void init_usertab(const char *UserTabFile);

diff --git a/include/info.h b/include/info.h
index 1b0c7d2630dd26923426c65f0b3245ff730eb79d..f82617428d6a19742da067345f2c7b9497e9f492 100755 (executable)
--- a/include/info.h
+++ b/include/info.h
@@ -1,3 +1,3 @@
-#define VERSION PACKAGE_VERSION" Dec-21-2010"
+#define VERSION PACKAGE_VERSION" Dec-26-2010"
 #define PGM PACKAGE_NAME
 #define URL "http://sarg.sourceforge.net"

diff --git a/report.c b/report.c
index 4e8e57b22ee9373c01dd5e7d2f938e0fb8c8a5d6..8b90ed3b3fef5b7035d3f5928c430523eb51b026 100644 (file)
--- a/report.c
+++ b/report.c
@@ -47,7 +47,7 @@ void gerarel(void)
        char oldaccdia[11], oldacchora[9], oldaccip[MAXLEN];
        char wdirname[MAXLEN];
        char *oldurl=NULL;
-       char oldmsg[50], acccode[MAXLEN/2 - 1], oldacccode[MAXLEN/2 - 1], user[MAX_USER_LEN];
+       char oldmsg[50], acccode[MAXLEN/2 - 1], oldacccode[MAXLEN/2 - 1];
        char ipantes[MAXLEN], nameantes[MAXLEN];
        char accsmart[MAXLEN];
        char crc2[MAXLEN/2 -1];
@@ -66,12 +66,10 @@ void gerarel(void)
        long long int oucache=0;
        long long int accbytes, accelap;
        char *str;
-       DIR *dirp;
-       struct dirent *direntp;
-       const char logext[]=".log";
+       userscan uscan;
        const char *sort_field;
        const char *sort_order;
-       int dlen;
+       const char *user;
        int url_len;
        int ourl_size=0;
        int ourltt_size=0;
@@ -104,40 +102,22 @@ void gerarel(void)
        puinfo=NULL;
        fp_tt=NULL;
 
-       if ((dirp = opendir(tmp)) == NULL) {
-               debuga(_("Failed to open directory %s - %s\n"),tmp,strerror(errno));
+       uscan=userinfo_startscan();
+       if (uscan == NULL) {
+               debuga(_("Cannot enumerate the user list\n"));
                exit(EXIT_FAILURE);
        }
-       while ((direntp = readdir( dirp )) != NULL ) {
-               dlen=strlen(direntp->d_name)-(sizeof(logext)-1);
-               if (dlen<0) continue;
-               if((strcmp(direntp->d_name+dlen,logext) != 0) ||
-                  (strncmp(direntp->d_name,"download.log",12) == 0) ||
-                  (strncmp(direntp->d_name,"denied.log",10) == 0) ||
-                  (strncmp(direntp->d_name,"authfail.log.unsort",19) == 0))
-                       continue;
-               if (snprintf(tmp3,sizeof(tmp3),"%s/%s",tmp,direntp->d_name)>=sizeof(tmp3)) {
-                       debuga(_("(report) directory entry too long: %s/%s\n"),tmp,direntp->d_name);
+       while ((uinfo = userinfo_advancescan(uscan)) != NULL ) {
+               if (snprintf(tmp3,sizeof(tmp3),"%s/%s.log",tmp,uinfo->filename)>=sizeof(tmp3)) {
+                       debuga(_("(report) directory entry too long: %s/%s.log\n"),tmp,uinfo->filename);
                        exit(EXIT_FAILURE);
                }
                if((fp_in=MY_FOPEN(tmp3,"r"))==NULL){
                        debuga(_("(report) Cannot open file %s\n"),tmp);
                        exit(EXIT_FAILURE);
                }
+               user=uinfo->filename;
 
-               if (dlen>0) {
-                       if (dlen>=sizeof(user)) continue;
-                       strncpy(user,direntp->d_name,dlen);
-                       user[dlen]=0;
-               } else {
-                       user[0]='\0';
-               }
-
-               uinfo=userinfo_find_from_file(user);
-               if (!uinfo) {
-                       debuga(_("Ignoring unknown user file %s\n"),user);
-                       continue;
-               }
                strcpy(u2,uinfo->id);
                if(Ip2Name && uinfo->id_is_ip) {
                        strcpy(ipantes,u2);
@@ -345,7 +325,7 @@ void gerarel(void)
                if (oldurltt) free(oldurltt);
                unlink(tmp3);
        }
-       closedir(dirp);
+       userinfo_stopscan(uscan);
 
        if (oldurl) {
                if(strstr(oldacccode,"DENIED") != 0)

diff --git a/sort.c b/sort.c
index 3fad666a5a76006a2ac2ad32adcc2679457cf9cd..0b229a525433bfa981bb9eab27ae8f9eea334b84 100644 (file)
--- a/sort.c
+++ b/sort.c
@@ -29,13 +29,11 @@
 
 void tmpsort(void)
 {
-       DIR *dirp;
-       struct dirent *direntp;
+       userscan uscan;
+       struct userinfostruct *uinfo;
        int cstatus;
-       const char tmpext[]=".utmp";
-       int dlen;
        char csort[MAXLEN];
-       char arqou[MAXLEN], arqin[MAXLEN], wnome[MAXLEN];
+       char arqou[MAXLEN], arqin[MAXLEN];
        const char *field1="2,2";
        const char *field2="1,1";
        const char *field3="3,3";
@@ -63,35 +61,25 @@ void tmpsort(void)
        else
                order="-r";
 
-       if ((dirp = opendir(tmp)) == NULL) {
-               debuga(_("Failed to open directory %s - %s\n"),tmp,strerror(errno));
+       uscan=userinfo_startscan();
+       if (uscan == NULL) {
+               debuga(_("Cannot enumerate the user list\n"));
                exit(EXIT_FAILURE);
        }
-       while ((direntp = readdir( dirp )) != NULL ){
-               dlen=strlen(direntp->d_name)-(sizeof(tmpext)-1);
-               if (dlen<0) continue;
-               if(strcmp(direntp->d_name+dlen,tmpext) != 0)
-                       continue;
-
-               if (dlen>0) {
-                       if (dlen>=sizeof(wnome)) continue;
-                       strncpy(wnome,direntp->d_name,dlen);
-                       wnome[dlen]='\0';
-               } else {
-                       wnome[0]='\0';
+       while ((uinfo = userinfo_advancescan(uscan)) != NULL ){
+               if (snprintf(arqin,sizeof(arqin),"%s/%s.utmp",tmp,uinfo->filename)>=sizeof(arqin)) {
+                       debuga(_("file name too long: %s/%s.utmp\n"),tmp,uinfo->filename);
+                       exit(EXIT_FAILURE);
+               }
+               if (snprintf(arqou,sizeof(arqou),"%s/%s.txt",tmp,uinfo->filename)>=sizeof(arqou)) {
+                       debuga(_("file name too long: %s/%s.utmp\n"),tmp,uinfo->filename);
+                       exit(EXIT_FAILURE);
                }
-
-               strcpy(arqou,tmp);
-               strcat(arqou,"/");
-               strcpy(arqin,arqou);
-               strcat(arqou,wnome);
-               strcat(arqin,direntp->d_name);
 
                if(debug) {
                        debuga(_("Sorting file: %s\n"),arqou);
                }
 
-               strcat(arqou,".txt");
                sprintf(csort,"sort -n -T \"%s\" %s -k %s -k %s -k %s -o \"%s\" \"%s\"",tmp,order,field1,field2,field3,arqou,arqin);
                cstatus=system(csort);
                if (!WIFEXITED(cstatus) || WEXITSTATUS(cstatus)) {
@@ -100,10 +88,9 @@ void tmpsort(void)
                        exit(EXIT_FAILURE);
                }
                unlink(arqin);
-
        }
 
-       (void)closedir( dirp );
+       userinfo_stopscan(uscan);
        return;
 }
 

diff --git a/squidguard_log.c b/squidguard_log.c
index 8d88512eb292817b61088cfbf242aeeaec9d8428..4c0d7a53757a0555bc6c2ce1d61ce5969141b728 100644 (file)
--- a/squidguard_log.c
+++ b/squidguard_log.c
@@ -251,6 +251,7 @@ static void read_log(const char *wentp, FILE *fp_ou,int dfrom,int duntil)
                if (!uinfo) {
                        uinfo=userinfo_create(user);
                        uinfo->id_is_ip=id_is_ip;
+                       uinfo->no_report=true;
                        if(Ip2Name && id_is_ip) ip2name(user,sizeof(user));
                        user_find(uinfo->label,MAX_USER_LEN, user);
                }

diff --git a/userinfo.c b/userinfo.c
index 4b315fa1fab9db4e8a378de7bdf089f146303cb6..6d82aa1df4d4b2c99598da8005a38017cd728032 100644 (file)
--- a/userinfo.c
+++ b/userinfo.c
@@ -44,6 +44,16 @@ struct usergroupstruct
        int nusers;
 };
 
+/*! \brief Hold pointer to scan through the user list.
+*/
+struct userscanstruct
+{
+       //! The group containing the user.
+       struct usergroupstruct *group;
+       //! The index of the user in the group.
+       int index;
+};
+
 //! The first group of users.
 static struct usergroupstruct *first_user_group=NULL;
 
@@ -148,3 +158,57 @@ struct userinfostruct *userinfo_find_from_id(const char *id)
        }
        return(NULL);
 }
+
+/*!
+Start the scanning of the user list.
+
+\return The object to pass to subsequent scanning functions or NULL
+if it failed. The object must be freed with a call to userinfo_stop().
+*/
+userscan userinfo_startscan(void)
+{
+       userscan uscan;
+
+       uscan=malloc(sizeof(*uscan));
+       if (!uscan) return(NULL);
+       uscan->group=first_user_group;
+       uscan->index=0;
+       return(uscan);
+}
+
+/*!
+Free the memory allocated by userinfo_start().
+
+\param uscan The object created by userinfo_start().
+*/
+void userinfo_stopscan(userscan uscan)
+{
+       free(uscan);
+}
+
+/*!
+Get the user pointed to by the object and advance the object
+to the next user.
+
+\param uscan The object created by userinfo_start().
+
+\return The user in the list or NULL if the end of the list
+is reached.
+*/
+struct userinfostruct *userinfo_advancescan(userscan uscan)
+{
+       struct userinfostruct *uinfo;
+
+       if (!uscan) return(NULL);
+       if (!uscan->group) return(NULL);
+       if (uscan->index<0 || uscan->index>=uscan->group->nusers) return(NULL);
+
+       uinfo=uscan->group->list+uscan->index;
+
+       ++uscan->index;
+       if (uscan->index>=uscan->group->nusers) {
+               uscan->group=uscan->group->next;
+               uscan->index=0;
+       }
+       return(uinfo);
+}