diff -upN old/Makefile.in new/Makefile.in
--- old/Makefile.in 2009-05-06 00:48:39.000000000 +0530
+++ new/Makefile.in 2009-05-06 00:47:02.000000000 +0530
@@ -43,8 +43,8 @@ webalizer: webalizer.o webalizer.h hasht
linklist.o linklist.h preserve.o preserve.h \
dns_resolv.o dns_resolv.h parser.o parser.h \
output.o output.h graphs.o graphs.h lang.h \
- webalizer_lang.h
- $(CC) ${LDFLAGS} -o webalizer webalizer.o hashtab.o linklist.o preserve.o parser.o output.o dns_resolv.o graphs.o ${LIBS}
+ rssfeed.o rssfeed.h webalizer_lang.h
+ $(CC) ${LDFLAGS} -o webalizer webalizer.o hashtab.o linklist.o preserve.o parser.o output.o dns_resolv.o graphs.o rssfeed.o ${LIBS}
rm -f webazolver
@LN_S@ webalizer webazolver
@@ -75,6 +75,9 @@ dns_resolv.o: dns_resolv.c dns_resolv.h
graphs.o: graphs.c graphs.h webalizer.h lang.h
$(CC) ${CFLAGS} ${DEFS} -c graphs.c
+rssfeed.o: rssfeed.c rssfeed.h webalizer.h output.h hashtab.h
+ $(CC) ${CFLAGS} ${DEFS} -c rssfeed.c
+
wcmgr: wcmgr.o
$(CC) ${LDFLAGS} -o wcmgr wcmgr.o ${WCMGR_LIBS}
diff -upN old/configure new/configure
--- old/configure 2009-05-06 00:48:39.000000000 +0530
+++ new/configure 2009-05-06 00:47:02.000000000 +0530
@@ -706,6 +706,8 @@ with_zlib
with_db
with_dblib
enable_dns
+enable_skiphostid
+enable_rssfeed
enable_bz2
with_bz2
with_bz2lib
@@ -1349,6 +1351,8 @@ Optional Features:
--enable-static Build as static executable [default=no]
--enable-debug Compile with debugging code [default=no]
--enable-dns Enable DNS/GeoDB lookup code [default=yes]
+ --enable-skiphostid Enable skip hostid code [default=no]
+ --enable-rssfeed Enable rssfeed code [default=no]
--enable-bz2 Enable BZip2 decompression code [default=no]
--enable-geoip Enable GeoIP geolocation code [default=no]
@@ -6103,6 +6107,30 @@ fi
fi
fi
+# Check whether --enable-skiphostid was given.
+if test "${enable_skiphostid+set}" = set; then
+ enableval=$enable_skiphostid; SKIP_HOSTID="yes"
+else
+ SKIP_HOSTID="no"
+fi
+
+
+if test "${SKIP_HOSTID}" = "yes"; then
+ OPTS="-DSKIP_HOSTID ${OPTS}"
+fi
+
+# Check whether --enable-rssfeed was given.
+if test "${enable_rssfeed+set}" = set; then
+ enableval=$enable_rssfeed; _RSSFEED="yes"
+else
+ _RSSFEED="no"
+fi
+
+
+if test "${_RSSFEED}" = "yes"; then
+ OPTS="-D_RSSFEED ${OPTS}"
+fi
+
# Check whether --enable-bz2 was given.
if test "${enable_bz2+set}" = set; then
diff -upN old/configure.in new/configure.in
--- old/configure.in 2009-05-06 00:48:39.000000000 +0530
+++ new/configure.in 2009-05-06 00:47:02.000000000 +0530
@@ -54,7 +54,7 @@ if test "$GCC" = "yes"; then
if test "$ac_cv_c_char_unsigned" = "yes"; then
CFLAGS="-fsigned-char ${CFLAGS}"
fi
-
+
AC_ARG_ENABLE(debug,
[ --enable-debug Compile with debugging code [[default=no]]],
CFLAGS="-g ${CFLAGS}")
@@ -191,6 +191,30 @@ if test "${USE_DNS}" = "yes"; then
fi
fi
+dnl ------------
+dnl whether to skip host id in parsing
+dnl ------------
+AC_ARG_ENABLE(skiphostid,
+ [ --enable-skiphostid Enable skip hostid code [[default=no]]],
+ SKIP_HOSTID="yes", SKIP_HOSTID="no")
+
+if test "${SKIP_HOSTID}" = "yes"; then
+ dnl define SKIP_HOSTID
+ OPTS="-DSKIP_HOSTID ${OPTS}"
+fi
+
+dnl ------------
+dnl enable/disable RSS Feed code
+dnl ------------
+AC_ARG_ENABLE(rssfeed,
+ [ --enable-rssfeed Enable rssfeed code [[default=no]]],
+ _RSSFEED="yes", _RSSFEED="no")
+
+if test "${_RSSFEED}" = "yes"; then
+ dnl define _RSSFEED
+ OPTS="-D_RSSFEED ${OPTS}"
+fi
+
dnl ------------------------------------------
dnl BZip2 code specific tests
dnl ------------------------------------------
diff -upN old/hashtab.c new/hashtab.c
--- old/hashtab.c 2009-05-06 00:48:39.000000000 +0530
+++ new/hashtab.c 2009-05-06 00:47:02.000000000 +0530
@@ -329,9 +329,13 @@ UNODEPTR new_unode(char *str)
/*********************************************/
/* PUT_UNODE - insert/update URL node */
/*********************************************/
-
+#ifdef _RSSFEED
+int put_unode(char *str, int type, u_int64_t count, double xfer,
+ u_int64_t *ctr, u_int64_t entry, u_int64_t exit, UNODEPTR *htab, time_t timestamp)
+#else
int put_unode(char *str, int type, u_int64_t count, double xfer,
u_int64_t *ctr, u_int64_t entry, u_int64_t exit, UNODEPTR *htab)
+#endif
{
UNODEPTR cptr,nptr;
@@ -350,6 +354,9 @@ int put_unode(char *str, int type, u_int
nptr->entry= entry;
nptr->exit = exit;
htab[hash(str)] = nptr;
+#ifdef _RSSFEED
+ nptr->timestamp = timestamp;
+#endif
if (type!=OBJ_GRP) (*ctr)++;
}
}
@@ -365,6 +372,9 @@ int put_unode(char *str, int type, u_int
/* found... bump counter */
cptr->count+=count;
cptr->xfer += xfer;
+#ifdef _RSSFEED
+ cptr->timestamp = timestamp;
+#endif
return 0;
}
}
@@ -379,6 +389,9 @@ int put_unode(char *str, int type, u_int
nptr->next = htab[hash(str)];
nptr->entry= entry;
nptr->exit = exit;
+#ifdef _RSSFEED
+ nptr->timestamp = timestamp;
+#endif
htab[hash(str)]=nptr;
if (type!=OBJ_GRP) (*ctr)++;
}
diff -upN old/hashtab.h new/hashtab.h
--- old/hashtab.h 2009-05-06 00:48:39.000000000 +0530
+++ new/hashtab.h 2009-05-06 00:47:02.000000000 +0530
@@ -41,7 +41,12 @@ struct unode { char *string;
u_int64_t entry; /* entry page counter */
u_int64_t exit; /* exit page counter */
double xfer; /* xfer size in bytes */
- struct unode *next; }; /* pointer to next node */
+#ifdef _RSSFEED
+ time_t timestamp; /* time stamp */
+#endif
+ struct unode *next; /* pointer to next node */
+ };
+
struct rnode { char *string; /* referrer hash table struct */
int flag;
@@ -79,8 +84,13 @@ extern DNODEPTR host_table[MAXHASH];
extern int put_hnode(char *, int, u_int64_t, u_int64_t, double,
u_int64_t *, u_int64_t, u_int64_t, char *, HNODEPTR *);
+#ifdef _RSSFEED
+extern int put_unode(char *, int, u_int64_t, double, u_int64_t *,
+ u_int64_t, u_int64_t, UNODEPTR *, time_t);
+#else
extern int put_unode(char *, int, u_int64_t, double, u_int64_t *,
u_int64_t, u_int64_t, UNODEPTR *);
+#endif
extern int put_inode(char *, int, u_int64_t, u_int64_t, double,
u_int64_t *, u_int64_t, u_int64_t, INODEPTR *);
extern int put_rnode(char *, int, u_int64_t, u_int64_t *, RNODEPTR *);
diff -upN old/output.c new/output.c
--- old/output.c 2009-05-06 00:48:39.000000000 +0530
+++ new/output.c 2009-05-06 00:47:02.000000000 +0530
@@ -65,6 +65,10 @@
#include "graphs.h"
#include "output.h"
+#ifdef _RSSFEED
+#include "rssfeed.h"
+#endif
+
/* internal function prototypes */
void write_html_head(char *, FILE *); /* head of html page */
void write_html_tail(FILE *); /* tail of html page */
@@ -289,6 +293,9 @@ int write_month_html()
if (verbose>1)
printf("%s %s %d\n",msg_gen_rpt, l_month[cur_month-1], cur_year);
+#ifdef _RSSFEED
+ InitRssCounter();
+#endif
/* fill in filenames */
snprintf(html_fname,sizeof(html_fname),"usage_%04d%02d.%s",
cur_year,cur_month,html_ext);
@@ -327,6 +334,10 @@ int write_month_html()
/* first, open the file */
if ( (out_fp=open_out_file(html_fname))==NULL ) return 1;
+#ifdef _RSSFEED
+ InitRssFiles(l_month[cur_month-1]);
+#endif
+
snprintf(buffer,sizeof(buffer),"%s %d",l_month[cur_month-1],cur_year);
write_html_head(buffer, out_fp);
month_links();
@@ -464,6 +475,9 @@ int write_month_html()
write_html_tail(out_fp); /* finish up the HTML document */
fclose(out_fp); /* close the file */
+#ifdef _RSSFEED
+ CloseTopUrlRss(); /* close all rss feed file */
+#endif
return (0); /* done... */
}
@@ -1031,6 +1045,11 @@ void top_urls_table(int flag)
case OBJ_GRP: u_grp++; break;
case OBJ_HIDE: u_hid++; break;
}
+
+#ifdef _RSSFEED
+ if (IsRssTopCountReached())
+ AddAsTopUrl((*pointer)->string, (*pointer)->timestamp, (*pointer)->flag);
+#endif
pointer++;
}
@@ -2907,4 +2926,3 @@ FILE *open_out_file(char *filename)
}
return out_fp;
}
-
diff -upN old/parser.c new/parser.c
--- old/parser.c 2009-05-06 00:48:39.000000000 +0530
+++ new/parser.c 2009-05-06 00:47:02.000000000 +0530
@@ -335,6 +335,12 @@ int parse_record_clf(char *buffer)
/* done with CLF record */
if (cp1>=eob) return 1;
+#ifdef SKIP_HOSTID
+ cp1++;
+ while ( (*cp1 != '\0') && (cp1 < eob) ) cp1++; /* skip host id */
+ if (cp1 < eob) cp1++;
+#endif
+
while ( (*cp1 != '\0') && (*cp1 != '\n') && (cp1 < eob) ) cp1++;
if (cp1 < eob) cp1++;
/* get referrer if present */
diff -upN old/preserve.c new/preserve.c
--- old/preserve.c 2009-05-06 00:48:39.000000000 +0530
+++ new/preserve.c 2009-05-06 00:47:02.000000000 +0530
@@ -658,8 +658,13 @@ int restore_state()
&t_unode.entry, &t_unode.exit);
/* Good record, insert into hash table */
+#ifdef _RSSFEED
+ if (put_unode(tmp_buf,t_unode.flag,t_unode.count,
+ t_unode.xfer,&ul_bogus,t_unode.entry,t_unode.exit,um_htab, 0))
+#else
if (put_unode(tmp_buf,t_unode.flag,t_unode.count,
t_unode.xfer,&ul_bogus,t_unode.entry,t_unode.exit,um_htab))
+#endif
{
if (verbose)
/* Error adding URL node, skipping ... */
diff -upN old/rssfeed.c new/rssfeed.c
--- old/rssfeed.c 1970-01-01 05:30:00.000000000 +0530
+++ new/rssfeed.c 2009-05-06 00:47:02.000000000 +0530
@@ -0,0 +1,548 @@
+#include "rssfeed.h"
+#define MAX_RSSFEED_COUNT 50
+#define MAX_BLOCK 10000
+
+char *rssIncludeUrlExts[MAX_RSSFEED_COUNT];
+char *rssExcludeUrls[MAX_RSSFEED_COUNT];
+char *rssFileName[MAX_RSSFEED_COUNT];
+char *rssHeaderFile[MAX_RSSFEED_COUNT];
+char *rssUrlRootFolders[MAX_RSSFEED_COUNT];
+char *rssFooter[MAX_RSSFEED_COUNT];
+time_t rssStartDate[MAX_RSSFEED_COUNT];
+time_t rssEndDate[MAX_RSSFEED_COUNT];
+
+int rssFilePopulated[MAX_RSSFEED_COUNT];
+
+int rssTopUrlCount[MAX_RSSFEED_COUNT];
+int rssTopUrlCounter[MAX_RSSFEED_COUNT];
+
+char *rssUrlDownloadTimeout=NULL;
+char *rssDateRange = NULL;
+
+
+int nRssFileName=0;
+int nRssHeaderFile=0;
+int nRsstopUrlCount=0;
+int nRssIncludeUrlExts=0;
+int nRssUrlRootFolders=0;
+int nRssExcludeUrls=0;
+int rssFeedCount=0;
+int nRssDateRange=0;
+int nRssFooter=0;
+
+FILE **rssFile_fp = NULL;
+
+void CloseTopUrlRss()
+{
+ int nRss=0;
+ for(; nRss < rssFeedCount; nRss++)
+ {
+ fprintf(rssFile_fp[nRss], rssFooter[nRss]);
+ fclose(rssFile_fp[nRss]);
+ }
+}
+long GetFileLength ( FILE * fp)
+{
+ int pos = ftell(fp);
+ fseek ( fp, 0L, SEEK_END);
+ long len = ftell (fp);
+ fseek ( fp, pos, SEEK_SET);
+ return (len);
+}
+void InitRssCounter()
+{
+ int nRss=0;
+ for(; nRss < rssFeedCount; nRss++)
+ {
+ rssTopUrlCounter[nRss] = 0;
+ }
+}
+int ParseTitle(char *buffer, int *titlePos, int *titleSize)
+{
+ char *temp=NULL, *temp1=NULL;
+
+ temp = strstr(buffer, "");
+ if (temp)
+ temp1 = strstr(temp, "");
+ else
+ return 0;
+
+ if (temp && temp1)
+ {
+ temp = temp + strlen("");
+ *titleSize = temp1 - temp;
+ *titlePos = temp - buffer;
+ }
+ else
+ return 0;
+
+ return 1;
+}
+
+int ParseDescription(char *buffer, int *descPos, int *descSize)
+{
+ char *temp=NULL;
+ char *temp1=NULL;
+
+ if (!buffer)
+ return 0;
+
+ char *buf = buffer;
+
+ while (temp = strstr(buffer, "<meta "))
+ {
+ temp += strlen("<meta ");
+ while (isspace(*temp)) temp ++;
+
+ if (strstr(temp, "name") == temp)
+ {
+ temp += strlen("name");
+ while (isspace(*temp)) temp ++;
+ if (*temp == '=')
+ {
+ temp ++;
+ while (isspace(*temp)) temp ++;
+ if (*temp == '"' || *temp == '\'')
+ temp ++;
+
+ if (strstr(temp, "description") == temp)
+ {
+ temp = strstr(temp, "content");
+ temp += strlen("content");
+ while (isspace(*temp)) temp ++;
+ if (*temp == '=')
+ {
+ temp ++;
+
+ while (isspace(*temp)) temp ++;
+
+ if (*temp == '"' || *temp == '\'')
+ temp ++;
+
+ char *temp1 = strstr(temp, "\"");
+
+ if (temp1)
+ {
+ *descSize = temp1 - temp;
+ *descPos = temp - buf;
+ return 1;
+ }
+ else
+ return 0;
+ }
+ else
+ return 0;
+ }
+ }
+ else
+ return 0;
+
+ }
+
+ buffer = temp;
+ }
+
+ return 0;
+
+}
+
+int ParseTitleAndDesc(char *fileName, char *url, char **title, char **desc)
+{
+ char *buffer = malloc(MAX_BLOCK + 1);
+ if (buffer == NULL)
+ return 0;
+
+ buffer[MAX_BLOCK] = 0;
+
+ FILE *fstream = fopen(fileName, "r");
+ if (fstream == NULL)
+ return 0;
+
+ if (GetFileLength(fstream) == 0)
+ {
+ if (verbose)
+ {
+ printf("Failed to download the url: %s%s\n", hname, url);
+ }
+ }
+ else
+ {
+ int loop = 0;
+ while(fread (buffer + loop*MAX_BLOCK, sizeof(char), MAX_BLOCK, fstream))
+ {
+ char *strHead = strstr(buffer, "</head>");
+
+ if (strHead != NULL)
+ {
+ break;
+ }
+ else
+ {
+ loop ++;
+ buffer = realloc(buffer, MAX_BLOCK*(loop+1));
+ buffer[MAX_BLOCK*(loop+1)] = 0;
+ }
+ }
+
+ char *buf = strdup(buffer);
+ int i=0;
+ while(buf[i] = tolower(buf[i])) i++;
+
+ int titlePos = 0;
+ int descPos = 0;
+
+ int titleSize=0, descSize=0;
+ if (ParseTitle(buf, &titlePos, &titleSize))
+ {
+ *title = malloc(titleSize + 1);
+ strncpy(*title, buffer + titlePos, titleSize);
+ (*title)[titleSize] = 0;
+ }
+
+ if (ParseDescription(buf, &descPos, &descSize))
+ {
+ *desc = malloc(descSize + 1);
+ strncpy(*desc, buffer + descPos, descSize);
+ (*desc)[descSize] = 0;
+ }
+ }
+
+ fclose(fstream);
+
+ return 1;
+}
+
+int IsUrlExtIncluded(int nRss, char *url)
+{
+ extern char *rssIncludeUrlExts[];
+ if (!rssIncludeUrlExts[nRss])
+ return 1;
+
+ char noext[]="noext";
+ char *temp = strrchr(url, '/');
+
+ if (temp)
+ temp = strchr (temp, '.');
+
+ if (!temp)
+ temp = noext;
+
+ return (NULL != strstr(rssIncludeUrlExts[nRss], temp));
+}
+
+int IsUrlExcluded(int nRss, char *url)
+{
+ int ret = 0;
+ extern char *rssExcludeUrls[];
+
+ if (rssExcludeUrls[nRss] && *rssExcludeUrls[nRss])
+ {
+ char *rssExcludedUrls = strdup(rssExcludeUrls[nRss]);
+ char *token = strtok(rssExcludedUrls, ";");
+ while (token)
+ {
+ if (strstr(url, token))
+ {
+ ret = 1;
+ break;
+ }
+ token = strtok( NULL, ";");
+ }
+
+ if (rssExcludedUrls)
+ {
+ free(rssExcludedUrls);
+ rssExcludedUrls = NULL;
+ }
+ }
+
+ return ret;
+}
+
+int IsUrlUnderRssRoot(int nRss, char *url)
+{
+ extern char *rssUrlRootFolders[];
+
+ if (rssUrlRootFolders[nRss])
+ {
+ char *rssRootFolder = strdup(rssUrlRootFolders[nRss]);
+ char *token = strtok(rssRootFolder, ";");
+ while (token)
+ {
+ if (strstr(url, token) == url)
+ {
+ return 1;
+ }
+ token = strtok( NULL, ";");
+ }
+
+ free (rssRootFolder);
+ rssRootFolder = NULL;
+ }
+
+ return 0;
+}
+int IsWithinTheDateRange(int nRss, time_t timestamp)
+{
+ if(rssStartDate[nRss] == 0 || rssEndDate[nRss] == 0 || (rssStartDate[nRss] <= timestamp && rssEndDate[nRss] >= timestamp))
+ return 1;
+ else
+ return 0;
+}
+int IsRssTopCountReached()
+{
+ int nRss=0;
+ int bRssFeedCountNotReached = 0;
+ for(nRss=0; nRss < rssFeedCount; nRss++)
+ {
+ if (rssTopUrlCounter[nRss] < rssTopUrlCount[nRss])
+ {
+ bRssFeedCountNotReached = 1;
+ break;
+ }
+ }
+ return bRssFeedCountNotReached;
+}
+void AddAsTopUrl(char *url, time_t timestamp, int flag)
+{
+ int *rssAddUrlFlag = (int *)malloc(sizeof(int)*rssFeedCount);
+ int parseUrl=0;
+
+ int nRss=0;
+ for(nRss=0; nRss < rssFeedCount; nRss++)
+ {
+ rssAddUrlFlag[nRss] = (rssTopUrlCounter[nRss] < rssTopUrlCount[nRss] &&
+ IsUrlExtIncluded(nRss, url) &&
+ !IsUrlExcluded(nRss, url) &&
+ IsUrlUnderRssRoot(nRss, url) &&
+ IsWithinTheDateRange(nRss, timestamp));
+
+ if (rssAddUrlFlag[nRss])
+ parseUrl = 1;
+ }
+
+ char *title=NULL;
+ char *desc=NULL;
+ if (parseUrl)
+ {
+ char fullyQualifiedUrl[2048];
+ strcpy(fullyQualifiedUrl, hname);
+ strncat(fullyQualifiedUrl, url, sizeof(fullyQualifiedUrl) - strlen(fullyQualifiedUrl));
+
+ if (flag != OBJ_HIDE)
+ {
+ char wgetCmd[2048];
+ char tempFile[] = ".temp.txt";
+
+ strcpy(wgetCmd, "wget -q --save-header -O ");
+ strcat(wgetCmd, tempFile);
+ strcat(wgetCmd, " --tries=1 --timeout=");
+
+ strcat(wgetCmd, rssUrlDownloadTimeout ? rssUrlDownloadTimeout : "0");
+
+ if (use_https)
+ strcat(wgetCmd, " https://");
+ else
+ strcat(wgetCmd, " http://");
+
+ strncat(wgetCmd, fullyQualifiedUrl, sizeof(wgetCmd) - strlen(wgetCmd));
+
+ if (verbose)
+ {
+ fprintf(stderr, "Downloading page: %s\n", fullyQualifiedUrl);
+ }
+
+ system(wgetCmd);
+
+ if (!ParseTitleAndDesc(tempFile, url, &title, &desc))
+ {
+ if (verbose)
+ fprintf(stderr, "failed to create Rss feed for top urls");
+ }
+ }
+
+ for(nRss=0; nRss < rssFeedCount; nRss++)
+ {
+ if (rssAddUrlFlag[nRss])
+ {
+ rssTopUrlCounter[nRss] ++;
+ rssFilePopulated[nRss] = 1;
+ fprintf(rssFile_fp[nRss], "<item>\n\t<title>%s\n\thttp://%s%s\n\t%s\n\thttp://%s%s\n", title?title:fullyQualifiedUrl, hname, url, desc?desc:"", hname, url);
+ }
+ }
+ }
+
+ if (rssAddUrlFlag)
+ {
+ free (rssAddUrlFlag);
+ rssAddUrlFlag = NULL;
+ }
+}
+
+FILE *OpenAndInitializeRssFile(char *rssOutFilename, char *rssHeaderFile)
+{
+ FILE *rss_fp=NULL;
+
+ if ( (rss_fp=fopen(rssOutFilename,"w")) == NULL)
+ {
+ if (verbose)
+ fprintf(stderr,"%s %s!\n",msg_no_open,rssOutFilename);
+
+ return NULL;
+ }
+
+ FILE *rssTemplate_fp = fopen(rssHeaderFile,"r");
+ if (rssTemplate_fp == NULL)
+ {
+ fclose(rss_fp);
+ rss_fp = NULL;
+ if (verbose)
+ fprintf(stderr,"%s %s!\n",msg_no_open,rssHeaderFile);
+ }
+ else
+ {
+ long templateFileLen = GetFileLength(rssTemplate_fp);
+ if (templateFileLen > 0)
+ {
+ char *buffer = malloc(templateFileLen + 1);
+ fread (buffer, sizeof(char), templateFileLen, rssTemplate_fp);
+ buffer[templateFileLen] = 0;
+
+ fprintf(rss_fp, buffer);
+
+ free (buffer);
+ buffer = NULL;
+ }
+ else
+ {
+ fclose(rss_fp);
+ rss_fp = NULL;
+ if (verbose)
+ {
+ fprintf(stderr,"Invalid file: %s!\n", rssHeaderFile);
+ }
+ }
+ }
+
+ return rss_fp;
+}
+void ParseRssDateRange (char *rssEndDateAndDuration, time_t *startDate, time_t *endDate)
+{
+ int duration = 0;
+
+ if(!strncasecmp(rssEndDateAndDuration, "none", strlen("none")))
+ {
+ *startDate = *endDate = 0;
+ return;
+ }
+
+ if (!strncasecmp(rssEndDateAndDuration, "today", strlen("today")))
+ {
+ int endDateOffset = 0;
+ char *temp=strstr(rssEndDateAndDuration,",");
+ if (temp)
+ {
+ *temp = ' ';
+ temp = strstr(rssEndDateAndDuration,"-");
+ if (temp)
+ {
+ *temp = ' ';
+ sscanf(rssEndDateAndDuration, "%*s%d%d", &endDateOffset, &duration);
+ }
+ else
+ {
+ sscanf(rssEndDateAndDuration, "%*s%d", &duration);
+ }
+ }
+
+ *endDate = time(NULL);
+ *endDate -= endDateOffset*86400; //24*60*60
+ }
+ else
+ {
+ int year=0, month=0, day=0;
+ char *temp = NULL;
+ temp = rssEndDateAndDuration;
+ while(temp = strstr(temp, "/"))
+ {
+ *temp = ' ';
+ }
+
+ if (temp = strstr(rssEndDateAndDuration, ","))
+ *temp = ' ';
+
+ sscanf(rssEndDateAndDuration, "%d %d %d %d", &month, &day, &year, &duration);
+
+ struct tm timeinfo;
+ memset(&timeinfo, 0, sizeof(timeinfo));
+ timeinfo.tm_year = year - 1900;
+ timeinfo.tm_mon = month - 1;
+ timeinfo.tm_mday = day;
+ timeinfo.tm_hour = 23;
+ timeinfo.tm_min = 59;
+ timeinfo.tm_sec = 59;
+
+ *endDate = mktime(&timeinfo);
+ }
+
+ *endDate += 86400;
+ *endDate -= (*endDate % 86400);
+ *endDate -= 1;
+
+ *startDate = *endDate - (duration-1)*86400; //24*60*60
+ *startDate -= (*startDate % 86400);
+
+}
+int InitRssFiles(char *month)
+{
+ // open RSS file and write the header content
+ rssFile_fp = (FILE **)malloc(sizeof(FILE *)*rssFeedCount);
+ int nRss = 0;
+ for(; nRss < rssFeedCount; nRss++)
+ {
+ char rssTopUrlFileNameForMonth[260];
+ strcpy(rssTopUrlFileNameForMonth, rssFileName[nRss]);
+
+ if (rssFilePopulated[nRss])
+ {
+ strcat(rssTopUrlFileNameForMonth, ".");
+ strcat(rssTopUrlFileNameForMonth, month);
+ }
+
+ if ( (rssFile_fp[nRss]=OpenAndInitializeRssFile(rssTopUrlFileNameForMonth, rssHeaderFile[nRss]))==NULL )
+ {
+ if (verbose)
+ fprintf(stderr, "Failed to create file %s", rssFileName[nRss]);
+ return 1;
+ }
+ }
+ return 0;
+}
+int ValidateRssFeedInputs()
+{
+ if (nRssFileName != nRssHeaderFile ||
+ nRssHeaderFile != nRsstopUrlCount ||
+ nRsstopUrlCount != nRssIncludeUrlExts ||
+ nRssIncludeUrlExts != nRssExcludeUrls ||
+ nRssIncludeUrlExts != nRssUrlRootFolders ||
+ nRssUrlRootFolders != nRssDateRange ||
+ nRssDateRange != nRssFooter)
+ {
+ if (verbose)
+ {
+ fprintf(stderr,"One of the required field to generate RSS feed is missing in config file. The following gives the count of different RSS settings found in config file:\n");
+ fprintf(stderr,"RssFileName=%d\n", nRssFileName);
+ fprintf(stderr,"RssHeaderFile=%d\n", nRssHeaderFile);
+ fprintf(stderr,"RsstopUrlCount=%d\n", nRsstopUrlCount);
+ fprintf(stderr,"RssIncludeUrlExts=%d\n", nRssIncludeUrlExts);
+ fprintf(stderr,"RssExcludeUrls=%d\n", nRssExcludeUrls);
+ fprintf(stderr,"RssUrlRootFolders=%d\n", nRssUrlRootFolders);
+ fprintf(stderr,"RssEndDateAndDuration=%d\n", nRssDateRange);
+ fprintf(stderr,"RssFooter=%d\n", nRssFooter);
+ }
+ return 1;
+ }
+
+ rssFeedCount = nRssFileName;
+ return 0;
+}
diff -upN old/rssfeed.h new/rssfeed.h
--- old/rssfeed.h 1970-01-01 05:30:00.000000000 +0530
+++ new/rssfeed.h 2009-05-06 00:47:02.000000000 +0530
@@ -0,0 +1,47 @@
+#include
+#include
+#include
+
+#define MAXHASH 2048 /* Size of our hash tables */
+#include "hashtab.h"
+
+extern char *rssIncludeUrlExts[];
+extern char *rssExcludeUrls[];
+extern char *rssFileName[];
+extern char *rssHeaderFile[];
+extern char *rssUrlRootFolders[];
+extern char *rssFooter[];
+extern time_t rssStartDate[];
+extern time_t rssEndDate[];
+
+extern int rssFilePopulated[];
+
+extern int rssTopUrlCount[];
+extern int rssTopUrlCounter[];
+
+extern char *rssUrlDownloadTimeout;
+extern char *rssDateRange;
+
+
+extern int nRssFileName;
+extern int nRssHeaderFile;
+extern int nRsstopUrlCount;
+extern int nRssIncludeUrlExts;
+extern int nRssUrlRootFolders;
+extern int nRssExcludeUrls;
+extern int nRssDateRange;
+extern int nRssFooter;
+extern int rssFeedCount;
+
+extern FILE **rssFile_fp;
+
+// from webalizer.h
+extern int verbose ; /* 2=verbose,1=err, 0=none */
+extern char *hname ; /* hostname for reports */
+extern int use_https ; /* use 'https://' on URLs */
+
+// from lang.h
+extern char *msg_no_open ;
+
+void ParseRssDateRange (char *, time_t *, time_t *);
+FILE *OpenAndInitializeRssFile(char *, char *);
diff -upN old/webalizer.c new/webalizer.c
--- old/webalizer.c 2009-05-06 00:48:39.000000000 +0530
+++ new/webalizer.c 2009-05-06 00:47:02.000000000 +0530
@@ -91,6 +91,9 @@ int bz2_rewind(void **, char *, char *);
#include "dns_resolv.h"
#endif
+#ifdef _RSSFEED
+#include "rssfeed.h"
+#endif
/* internal function prototypes */
void clear_month(); /* clear monthly stuff */
@@ -106,7 +109,6 @@ void agent_mangle(char *);
char *our_gzgets(void *, char *, int); /* our gzgets */
int stricmp(char *, char *); /* case ins. compare */
int isipaddr(char *); /* is IP address test */
-
/*********************************************/
/* GLOBAL VARIABLES */
/*********************************************/
@@ -753,6 +755,19 @@ int main(int argc, char *argv[])
continue; /* and ignore this record */
}
+#ifdef _RSSFEED
+ // Calculate time
+ struct tm timeinfo;
+ memset(&timeinfo, 0, sizeof(timeinfo));
+ timeinfo.tm_year = rec_year - 1900;
+ timeinfo.tm_mon = rec_month - 1;
+ timeinfo.tm_mday = rec_day;
+ timeinfo.tm_hour = rec_hour;
+ timeinfo.tm_min = rec_min;
+ timeinfo.tm_sec = rec_sec;
+
+ log_rec.timestamp = mktime(&timeinfo);
+#endif
/*********************************************/
/* GOOD RECORD, CHECK INCREMENTAL/TIMESTAMPS */
/*********************************************/
@@ -1201,8 +1216,13 @@ int main(int argc, char *argv[])
(log_rec.resp_code==RC_PARTIALCONTENT))
{
/* URL hash table */
+#ifdef _RSSFEED
+ if (put_unode(log_rec.url,OBJ_REG,(u_int64_t)1,
+ log_rec.xfer_size,&t_url,(u_int64_t)0,(u_int64_t)0,um_htab,log_rec.timestamp))
+#else
if (put_unode(log_rec.url,OBJ_REG,(u_int64_t)1,
log_rec.xfer_size,&t_url,(u_int64_t)0,(u_int64_t)0,um_htab))
+#endif
{
if (verbose)
/* Error adding URL node, skipping ... */
@@ -1296,8 +1316,13 @@ int main(int argc, char *argv[])
/* URL Grouping */
if ( (cp1=isinglist(group_urls,log_rec.url))!=NULL)
{
+#ifdef _RSSFEED
+ if (put_unode(cp1,OBJ_GRP,(u_int64_t)1,log_rec.xfer_size,
+ &ul_bogus,(u_int64_t)0,(u_int64_t)0,um_htab,log_rec.timestamp))
+#else
if (put_unode(cp1,OBJ_GRP,(u_int64_t)1,log_rec.xfer_size,
&ul_bogus,(u_int64_t)0,(u_int64_t)0,um_htab))
+#endif
{
if (verbose)
/* Error adding URL node, skipping ... */
@@ -1629,6 +1654,18 @@ void get_config(char *fname)
"CountryFlags", /* show country flags? (0-no) 118 */
"FlagDir", /* directory w/flag images 119 */
"SearchCaseI" /* srch str case insensitive 120 */
+#ifdef _RSSFEED
+ ,
+ "RssFileName", /* rss feed filename 121 */
+ "RssHeaderFile", /* rss template filename 122 */
+ "RssTopUrlCount", /* count of top urls for rss 123 */
+ "RssIncludeUrlExts", /* extenstion to be included 124*/
+ "RssExcludeUrls", /* urls to be excluded 125*/
+ "RssUrlRootFolders", /* root folders rss urls 126*/
+ "RssEndDateAndDuration",/* access date range for rss 127*/
+ "RssFooter", /* RSS feed footer 128*/
+ "RssUrlDownloadTimeout" /* url download timeout to fetch title & desc 129*/
+#endif
};
FILE *fp;
@@ -1863,8 +1900,30 @@ void get_config(char *fname)
case 119: use_flags=1; flag_dir=save_opt(value); break; /* FlagDir */
case 120: searchcasei=
(tolower(value[0])=='n')?0:1; break; /* SearchCaseI */
+#ifdef _RSSFEED
+ case 121: rssFileName[nRssFileName ++]=save_opt(value); break;
+ case 122: rssHeaderFile[nRssHeaderFile ++]=save_opt(value); break;
+ case 123: rssTopUrlCount[nRsstopUrlCount ++]=atoi(value); break;
+ case 124: rssIncludeUrlExts[nRssIncludeUrlExts ++]=save_opt(value); break;
+ case 125: rssExcludeUrls[nRssExcludeUrls ++]=save_opt(value); break;
+ case 126: rssUrlRootFolders[nRssUrlRootFolders ++]=save_opt(value); break;
+ case 127: rssDateRange = save_opt(value);
+ if (rssDateRange)
+ {
+ ParseRssDateRange(rssDateRange, &rssStartDate[nRssDateRange], &rssEndDate[nRssDateRange]);
+ nRssDateRange ++;
+ free(rssDateRange);rssDateRange = NULL;
+ }
+ break;
+ case 128: rssFooter[nRssFooter ++] = save_opt(value);break;
+ case 129: rssUrlDownloadTimeout = save_opt(value); break;
+#endif
}
}
+#ifdef _RSSFEED
+ ValidateRssFeedInputs();
+#endif
+
fclose(fp);
}
@@ -1999,7 +2058,6 @@ char *cur_time()
return timestamp;
}
-
/*********************************************/
/* ISPAGE - determine if an HTML page or not */
/*********************************************/
diff -upN old/webalizer.h new/webalizer.h
--- old/webalizer.h 2009-05-06 00:48:39.000000000 +0530
+++ new/webalizer.h 2009-05-06 00:47:02.000000000 +0530
@@ -17,8 +17,8 @@
#define MAXHOST 128 /* Max hostname buffer size */
#define MAXURL 4096 /* Max HTTP request/URL field size */
#define MAXURLH 512 /* Max URL field size in htab */
-#define MAXREF 1024 /* Max referrer field size */
-#define MAXREFH 256 /* Max referrer field size in htab */
+#define MAXREF 4096 /* Max referrer field size */
+#define MAXREFH 512 /* Max referrer field size in htab */
#define MAXAGENT 128 /* Max user agent field size */
#define MAXCTRY 48 /* Max country name size */
#define MAXSRCH 256 /* Max size of search string buffer */
@@ -151,6 +151,9 @@ typedef struct country_code *CLISTPTR;
/* log record structure */
struct log_struct { char hostname[MAXHOST]; /* hostname */
char datetime[29]; /* raw timestamp */
+#ifdef _RSSFEED
+ time_t timestamp; /* timestamp */
+#endif
char url[MAXURL]; /* raw request field */
int resp_code; /* response code */
u_int64_t xfer_size; /* xfer size in bytes */