/*
* SARG Squid Analysis Report Generator http://sarg.sourceforge.net
- * 1998, 2010
+ * 1998, 2015
*
* SARG donations:
* please look at http://sarg.sourceforge.net/donations.php
#include "include/conf.h"
#include "include/defs.h"
+#ifdef ENABLE_DOUBLE_CHECK_DATA
+extern struct globalstatstruct globstat;
+#endif
+
void topsites(void)
{
-
- FILE *fp_in, *fp_ou;
-
- char *buf;
- char *url;
- char *ourl=NULL;
- char csort[255];
- char general[MAXLEN];
- char general2[MAXLEN];
- char general3[MAXLEN];
- char per[MAXLEN];
- char sites[MAXLEN];
- char report[MAXLEN];
- char period[100];
- const char *sortf;
- const char *sortt;
- long long int nacc;
- long long int nbytes;
- long long int ntime;
- long long int tnacc=0;
- long long int tnbytes=0;
- long long int tntime=0;
- long long int twork1=0, twork2=0, twork3=0;
- int regs=0;
- int cstatus;
- int url_len;
- int ourl_size=0;
- struct getwordstruct gwarea;
- struct longlinestruct line;
- struct generalitemstruct item;
-
- if(Privacy)
- return;
-
- sprintf(general,"%s/sarg-general",outdirname);
- sprintf(sites,"%s/sarg-sites",outdirname);
- sprintf(general2,"%s/sarg-general2",outdirname);
- sprintf(general3,"%s/sarg-general3",outdirname);
- sprintf(per,"%s/sarg-period",outdirname);
-
- if ((ReportType & REPORT_TYPE_TOPUSERS) == 0)
- sprintf(report,"%s/index.html",outdirname);
- else
- sprintf(report,"%s/topsites.html",outdirname);
-
- if ((fp_in = fopen(per, "r")) == 0) {
- debuga(_("(topsites) Cannot open file: %s\n"),per);
- exit(1);
- }
-
- if (!fgets(period,sizeof(period),fp_in)) {
- debuga(_("(topsites) read error in %s\n"),per);
- exit(1);
- }
- fclose(fp_in);
-
- sprintf(csort,"sort -k 4,4 -o \"%s\" \"%s\"",general2,general);
- cstatus=system(csort);
- if (!WIFEXITED(cstatus) || WEXITSTATUS(cstatus)) {
- debuga(_("sort command return status %d\n"),WEXITSTATUS(cstatus));
- debuga(_("sort command: %s\n"),csort);
- exit(1);
- }
-
- if((fp_in=fopen(general2,"r"))==NULL) {
- debuga(_("(topsites) Cannot open log file %s\n"),general2);
- debuga(_("sort command: %s\n"),csort);
- exit(1);
- }
-
- if((fp_ou=fopen(general3,"w"))==NULL) {
- debuga(_("(topsites) Cannot open log file %s\n"),general3);
- exit(1);
- }
-
- if (longline_prepare(&line)<0) {
- debuga(_("Not enough memory to read file %s\n"),general2);
- exit(1);
- }
-
- while((buf=longline_read(fp_in,&line))!=NULL) {
- ger_read(buf,&item,general2);
- if(item.total) continue;
-
- if(!regs) {
- url_len=strlen(item.url);
- if (!ourl || url_len>=ourl_size) {
- ourl_size=url_len+1;
- ourl=realloc(ourl,ourl_size);
- if (!ourl) {
- debuga(_("Not enough memory to store the url\n"));
- exit(1);
- }
- }
- strcpy(ourl,item.url);
- regs++;
- }
-
- if(strcmp(item.url,ourl) != 0) {
- fprintf(fp_ou,"%lld\t%lld\t%lld\t%s\n",tnacc,tnbytes,tntime,ourl);
- url_len=strlen(item.url);
- if (url_len>=ourl_size) {
- ourl_size=url_len+1;
- ourl=realloc(ourl,ourl_size);
- if (!ourl) {
- debuga(_("Not enough memory to store the url\n"));
- exit(1);
- }
- }
- strcpy(ourl,item.url);
- tnacc=0;
- tnbytes=0;
- tntime=0;
- }
-
- tnacc+=item.nacc;
- tnbytes+=item.nbytes;
- tntime+=item.nelap;
- }
- fclose(fp_in);
- unlink(general2);
- longline_free(&line);
-
- if (ourl) {
- fprintf(fp_ou,"%lld\t%lld\t%lld\t%s\n",tnacc,tnbytes,tntime,ourl);
- free(ourl);
- }
-
- fclose(fp_ou);
-
- strlow(TopsitesSortField);
- strlow(TopsitesSortType);
-
- if(strcmp(TopsitesSortField,"connect") == 0)
- sortf="-k 1,1 -k 2,2";
- else if(strcmp(TopsitesSortField,"bytes") == 0)
- sortf="-k 2,2 -k 1,1";
- else
- sortf="";
- if(strcmp(TopsitesSortType,"a") == 0)
- sortt="";
- else if(strcmp(TopsitesSortType,"d") == 0)
- sortt="-r";
- else
- sortt="";
-
- sprintf(csort,"sort %s -n %s -o \"%s\" \"%s\"",sortt,sortf,sites,general3);
- cstatus=system(csort);
- if (!WIFEXITED(cstatus) || WEXITSTATUS(cstatus)) {
- debuga(_("sort command return status %d\n"),WEXITSTATUS(cstatus));
- debuga(_("sort command: %s\n"),csort);
- exit(1);
- }
- if((fp_in=fopen(sites,"r"))==NULL) {
- debuga(_("(topsites) Cannot open log file %s\n"),sites);
- debuga(_("sort command: %s\n"),csort);
- exit(1);
- }
-
- unlink(general2);
- unlink(general3);
-
- if((fp_ou=fopen(report,"w"))==NULL) {
- debuga(_("(topsites) Cannot open log file %s\n"),report);
- exit(1);
- }
-
- write_html_header(fp_ou,(IndexTree == INDEX_TREE_DATE) ? 3 : 1,_("Top sites"));
- fputs("<tr><td class=\"header_c\">",fp_ou);
- fprintf(fp_ou,_("Period: %s"),period);
- fputs("</td></tr>\n",fp_ou);
- fputs("<tr><th class=\"header_c\">",fp_ou);
- fprintf(fp_ou,_("Top %d sites"),TopSitesNum);
- fputs("</th></tr>\n",fp_ou);
- close_html_header(fp_ou);
-
- fputs("<div class=\"report\"><table cellpadding=\"1\" cellspacing=\"2\">\n",fp_ou);
- fputs("<tr><td></td></tr>\n",fp_ou);
- fprintf(fp_ou,"<tr><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th></tr>\n",_("NUM"),_("ACCESSED SITE"),_("CONNECT"),_("BYTES"),_("TIME"));
-
- regs=0;
- ntopsites = 0;
-
- if (longline_prepare(&line)<0) {
- debuga(_("Not enough memory to read file %s\n"),sites);
- exit(1);
- }
-
- while(regs<TopSitesNum && (buf=longline_read(fp_in,&line))!=NULL) {
- getword_start(&gwarea,buf);
- if (getword_atoll(&nacc,&gwarea,'\t')<0) {
- debuga(_("Maybe you have a broken record or garbage in your %s file\n"),sites);
- exit(1);
- }
- if (nacc == 0) continue;
- if (getword_atoll(&nbytes,&gwarea,'\t')<0 || getword_atoll(&ntime,&gwarea,'\t')<0) {
- debuga(_("Maybe you have a broken record or garbage in your %s file\n"),sites);
- exit(1);
- }
- if (getword_ptr(buf,&url,&gwarea,'\t')<0) {
- debuga(_("The url is invalid in file %s\n"),sites);
- exit(1);
- }
-
- twork1=nacc;
- twork2=nbytes;
- twork3=ntime;
-
- strcpy(wwork1,fixnum(twork1,1));
- strcpy(wwork2,fixnum(twork2,1));
- strcpy(wwork3,fixtime(twork3));
-
- fprintf(fp_ou,"<tr><td class=\"data\">%d</td><td class=\"data2 link\">",++regs);
-
- if(BlockIt[0] != '\0') {
- fprintf(fp_ou,"<a href=\"%s%s?url=\"",wwwDocumentRoot,BlockIt);
- output_html_url(fp_ou,url);
- fputs("\"><img src=\"../images/sarg-squidguard-block.png\"></a> ",fp_ou);
- }
-
- fputs("<a href=\"http://",fp_ou);
- output_html_url(fp_ou,url);
- fputs("\">",fp_ou);
- output_html_string(fp_ou,url,100);
- fprintf(fp_ou,"</a></td><td class=\"data\">%s</td><td class=\"data\">%s</td><td class=\"data\">%s</td></tr>\n",wwork1,wwork2,wwork3);
- }
- fclose(fp_in);
- longline_free(&line);
-
- fputs("</table></div>\n",fp_ou);
- write_html_trailer(fp_ou);
- fclose(fp_ou);
-
- return;
-
+ FileObject *fp_in;
+ FILE *fp_ou;
+
+ char *buf;
+ char *url;
+ char *ourl=NULL;
+ char csort[4096];
+ char general[MAXLEN];
+ char general2[MAXLEN];
+ char general3[MAXLEN];
+ char sites[MAXLEN];
+ char report[MAXLEN];
+ char ouser[MAX_USER_LEN]="";
+ const char *sortf;
+ const char *sortt;
+ long long int nacc;
+ long long int nbytes;
+ long long int ntime;
+ long long int tnacc=0;
+ long long int tnbytes=0;
+ long long int tntime=0;
+ long long int twork1=0, twork2=0, twork3=0;
+#ifdef ENABLE_DOUBLE_CHECK_DATA
+ long long int ttnacc=0;
+ long long int ttnbytes=0;
+ long long int ttntime=0;
+#endif
+ int nusers=0;
+ int regs=0;
+ int cstatus;
+ int url_len;
+ int ourl_size=0;
+ struct getwordstruct gwarea;
+ longline line;
+ struct generalitemstruct item;
+
+ if(Privacy) {
+ if (debugz>=LogLevel_Process) debugaz(__FILE__,__LINE__,_("Top sites report not produced because privacy option is on\n"));
+ return;
+ }
+ if (debugz>=LogLevel_Process)
+ debuga(__FILE__,__LINE__,_("Creating top sites report...\n"));
+
+ sprintf(general,"%s/sarg-general",outdirname);
+ sprintf(sites,"%s/sarg-sites",outdirname);
+ sprintf(general2,"%s/sarg-general2",outdirname);
+ sprintf(general3,"%s/sarg-general3",outdirname);
+
+ sprintf(report,"%s/topsites.html",outdirname);
+
+ if (snprintf(csort,sizeof(csort),"sort -t \"\t\" -k 4,4 -k 1,1 -o \"%s\" \"%s\"",general2,general)>=sizeof(csort)) {
+ debuga(__FILE__,__LINE__,_("Sort command too long when sorting file \"%s\" to \"%s\"\n"),general,general2);
+ exit(EXIT_FAILURE);
+ }
+ cstatus=system(csort);
+ if (!WIFEXITED(cstatus) || WEXITSTATUS(cstatus)) {
+ debuga(__FILE__,__LINE__,_("sort command return status %d\n"),WEXITSTATUS(cstatus));
+ debuga(__FILE__,__LINE__,_("sort command: %s\n"),csort);
+ exit(EXIT_FAILURE);
+ }
+
+ if((fp_in=FileObject_Open(general2))==NULL) {
+ debuga(__FILE__,__LINE__,_("Cannot open file \"%s\": %s\n"),general2,FileObject_GetLastOpenError());
+ debuga(__FILE__,__LINE__,_("sort command: %s\n"),csort);
+ exit(EXIT_FAILURE);
+ }
+
+ if((fp_ou=fopen(general3,"w"))==NULL) {
+ debuga(__FILE__,__LINE__,_("Cannot open file \"%s\": %s\n"),general3,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ if ((line=longline_create())==NULL) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to read file \"%s\"\n"),general2);
+ exit(EXIT_FAILURE);
+ }
+
+ while((buf=longline_read(fp_in,line))!=NULL) {
+ ger_read(buf,&item,general2);
+ if(item.total) continue;
+
+ if(!regs) {
+ url_len=strlen(item.url);
+ if (!ourl || url_len>=ourl_size) {
+ ourl_size=url_len+1;
+ ourl=realloc(ourl,ourl_size);
+ if (!ourl) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the url\n"));
+ exit(EXIT_FAILURE);
+ }
+ }
+ strcpy(ourl,item.url);
+ regs++;
+ }
+
+ if(strcmp(item.url,ourl) != 0) {
+ /*
+ This complicated printf is due to Microsoft's inability to comply with any standard. Msvcrt is unable
+ to print a long long int unless it is exactly 64-bits long.
+ */
+ fprintf(fp_ou,"%"PRIu64"\t%"PRIu64"\t%"PRIu64"\t%d\t%s\n",(uint64_t)tnacc,(uint64_t)tnbytes,(uint64_t)tntime,nusers,ourl);
+ url_len=strlen(item.url);
+ if (url_len>=ourl_size) {
+ ourl_size=url_len+1;
+ ourl=realloc(ourl,ourl_size);
+ if (!ourl) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the url\n"));
+ exit(EXIT_FAILURE);
+ }
+ }
+ strcpy(ourl,item.url);
+ strcpy(ouser,item.user);
+ tnacc=0;
+ tnbytes=0;
+ tntime=0;
+ nusers=1;
+ } else if (strcmp(item.user,ouser)!=0) {
+ strcpy(ouser,item.user);
+ nusers++;
+ }
+
+ tnacc+=item.nacc;
+ tnbytes+=item.nbytes;
+ tntime+=item.nelap;
+#ifdef ENABLE_DOUBLE_CHECK_DATA
+ ttnacc+=item.nacc;
+ ttnbytes+=item.nbytes;
+ ttntime+=item.nelap;
+#endif
+ }
+ if (FileObject_Close(fp_in)) {
+ debuga(__FILE__,__LINE__,_("Read error in \"%s\": %s\n"),general2,FileObject_GetLastCloseError());
+ exit(EXIT_FAILURE);
+ }
+ longline_destroy(&line);
+
+ if (ourl) {
+ /*
+ This complicated printf is due to Microsoft's inability to comply with any standard. Msvcrt is unable
+ to print a long long int unless it is exactly 64-bits long.
+ */
+ fprintf(fp_ou,"%"PRIu64"\t%"PRIu64"\t%"PRIu64"\t%d\t%s\n",(uint64_t)tnacc,(uint64_t)tnbytes,(uint64_t)tntime,nusers,ourl);
+ free(ourl);
+ }
+
+ if (fclose(fp_ou)==EOF) {
+ debuga(__FILE__,__LINE__,_("Write error in \"%s\": %s\n"),general3,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+#ifdef ENABLE_DOUBLE_CHECK_DATA
+ if (ttnacc!=globstat.nacc || ttnbytes!=globstat.nbytes || ttntime!=globstat.elap) {
+ debuga(__FILE__,__LINE__,_("Total statistics mismatch when reading \"%s\" to produce the top sites\n"),general2);
+ exit(EXIT_FAILURE);
+ }
+#endif
+
+ if (!KeepTempLog && unlink(general2)) {
+ debuga(__FILE__,__LINE__,_("Cannot delete \"%s\": %s\n"),general2,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ if((TopsitesSort & TOPSITE_SORT_CONNECT) != 0) {
+ sortf="-k 1,1 -k 2,2";
+ } else if((TopsitesSort & TOPSITE_SORT_BYTES) != 0) {
+ sortf="-k 2,2 -k 1,1";
+ } else if((TopsitesSort & TOPSITE_SORT_TIME) != 0) {
+ sortf="-k 3,3";
+ } else if((TopsitesSort & TOPSITE_SORT_USER) != 0) {
+ sortf="-k 4,4 -k 1,1 -k 2,2";
+ } else {
+ sortf="-k 2,2 -k 1,1"; //default is BYTES
+ }
+ if((TopsitesSort & TOPSITE_SORT_REVERSE) != 0) {
+ sortt="-r";
+ } else {
+ sortt="";
+ }
+
+ if (snprintf(csort,sizeof(csort),"sort -t \"\t\" %s -n %s -o \"%s\" \"%s\"",sortt,sortf,sites,general3)>=sizeof(csort)) {
+ debuga(__FILE__,__LINE__,_("Sort command too long when sorting file \"%s\" to \"%s\"\n"),general3,sites);
+ exit(EXIT_FAILURE);
+ }
+ cstatus=system(csort);
+ if (!WIFEXITED(cstatus) || WEXITSTATUS(cstatus)) {
+ debuga(__FILE__,__LINE__,_("sort command return status %d\n"),WEXITSTATUS(cstatus));
+ debuga(__FILE__,__LINE__,_("sort command: %s\n"),csort);
+ exit(EXIT_FAILURE);
+ }
+ if((fp_in=FileObject_Open(sites))==NULL) {
+ debuga(__FILE__,__LINE__,_("Cannot open file \"%s\": %s\n"),sites,FileObject_GetLastOpenError());
+ debuga(__FILE__,__LINE__,_("sort command: %s\n"),csort);
+ exit(EXIT_FAILURE);
+ }
+
+ if (!KeepTempLog && unlink(general3)) {
+ debuga(__FILE__,__LINE__,_("Cannot delete \"%s\": %s\n"),general3,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ if((fp_ou=fopen(report,"w"))==NULL) {
+ debuga(__FILE__,__LINE__,_("Cannot open file \"%s\": %s\n"),report,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ write_html_header(fp_ou,(IndexTree == INDEX_TREE_DATE) ? 3 : 1,_("Top sites"),HTML_JS_SORTTABLE);
+ fputs("<tr><td class=\"header_c\">",fp_ou);
+ fprintf(fp_ou,_("Period: %s"),period.html);
+ fputs("</td></tr>\n",fp_ou);
+ fputs("<tr><th class=\"header_c\">",fp_ou);
+ fprintf(fp_ou,_("Top %d sites"),TopSitesNum);
+ fputs("</th></tr>\n",fp_ou);
+ close_html_header(fp_ou);
+
+ fputs("<div class=\"report\"><table cellpadding=\"1\" cellspacing=\"2\"",fp_ou);
+ if (SortTableJs[0]) fputs(" class=\"sortable\"",fp_ou);
+ fputs(">\n",fp_ou);
+ fprintf(fp_ou,"<thead><tr><th class=\"header_l\">%s</th><th class=\"header_l",
+ /* TRANSLATORS: This is a column header showing the position of the entry in the sorted list. */
+ _("NUM"));
+ if (SortTableJs[0]) fputs(" sorttable_alpha",fp_ou);
+ fprintf(fp_ou,"\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th><th class=\"header_l\">%s</th></tr></thead>\n",
+ /* TRANSLATORS: This is a column header showing the URL of the visited sites. */
+ _("ACCESSED SITE"),
+ /* TRANSLATORS: This is a column header showing the number of connections to a visited site. */
+ _("CONNECT"),
+ /* TRANSLATORS: This is a column header showing the number of transfered bytes. */
+ _("BYTES"),
+ /* TRANSLATORS: This is a column header showing the time spent by the proxy processing the requests. */
+ pgettext("duration","TIME"),
+ /* TRANSLATORS: This is a column header showing the number of users who visited a sites. */
+ _("USERS"));
+
+ regs=0;
+ ntopsites = 0;
+
+ if ((line=longline_create())==NULL) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to read file \"%s\"\n"),sites);
+ exit(EXIT_FAILURE);
+ }
+
+ while(regs<TopSitesNum && (buf=longline_read(fp_in,line))!=NULL) {
+ getword_start(&gwarea,buf);
+ if (getword_atoll(&nacc,&gwarea,'\t')<0) {
+ debuga(__FILE__,__LINE__,_("Invalid record in file \"%s\"\n"),sites);
+ exit(EXIT_FAILURE);
+ }
+ if (nacc == 0) continue;
+ if (getword_atoll(&nbytes,&gwarea,'\t')<0 || getword_atoll(&ntime,&gwarea,'\t')<0) {
+ debuga(__FILE__,__LINE__,_("Invalid record in file \"%s\"\n"),sites);
+ exit(EXIT_FAILURE);
+ }
+ if (getword_atoi(&nusers,&gwarea,'\t')<0) {
+ debuga(__FILE__,__LINE__,_("Invalid number of users in file \"%s\"\n"),sites);
+ exit(EXIT_FAILURE);
+ }
+ if (getword_ptr(buf,&url,&gwarea,'\t')<0) {
+ debuga(__FILE__,__LINE__,_("Invalid url in file \"%s\"\n"),sites);
+ exit(EXIT_FAILURE);
+ }
+
+ twork1=nacc;
+ twork2=nbytes;
+ twork3=ntime;
+
+ fprintf(fp_ou,"<tr><td class=\"data\">%d</td><td class=\"data2\">",++regs);
+
+ if(BlockIt[0] != '\0' && url[0]!=ALIAS_PREFIX) {
+ fprintf(fp_ou,"<a href=\"%s%s?url=\"",wwwDocumentRoot,BlockIt);
+ output_html_url(fp_ou,url);
+ fputs("\"><img src=\"../images/sarg-squidguard-block.png\"></a> ",fp_ou);
+ }
+
+ output_html_link(fp_ou,url,100);
+ fputs("</td><td class=\"data\"",fp_ou);
+ if (SortTableJs[0]) fprintf(fp_ou," sorttable_customkey=\"%"PRId64"\"",(uint64_t)twork1);
+ fprintf(fp_ou,">%s</td>",fixnum(twork1,1));
+ fputs("<td class=\"data\"",fp_ou);
+ if (SortTableJs[0]) fprintf(fp_ou," sorttable_customkey=\"%"PRId64"\"",(uint64_t)twork2);
+ fprintf(fp_ou,">%s</td>",fixnum(twork2,1));
+ fputs("<td class=\"data\"",fp_ou);
+ if (SortTableJs[0]) fprintf(fp_ou," sorttable_customkey=\"%"PRId64"\"",(uint64_t)twork3);
+ fprintf(fp_ou,">%s</td>",fixtime(twork3));
+ fputs("<td class=\"data\"",fp_ou);
+ if (SortTableJs[0]) fprintf(fp_ou," sorttable_customkey=\"%d\"",nusers);
+ fprintf(fp_ou,">%s</td></tr>\n",fixnum(nusers,1));
+ }
+ if (FileObject_Close(fp_in)) {
+ debuga(__FILE__,__LINE__,_("Read error in \"%s\": %s\n"),sites,FileObject_GetLastCloseError());
+ exit(EXIT_FAILURE);
+ }
+ longline_destroy(&line);
+
+ fputs("</table></div>\n",fp_ou);
+ write_html_trailer(fp_ou);
+ if (fclose(fp_ou)==EOF) {
+ debuga(__FILE__,__LINE__,_("Write error in \"%s\": %s\n"),report,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ return;
}
projects / thirdparty / sarg.git / blobdiff