bashclub/piler
1
0
Fork 0
mirror of https://bitbucket.org/jsuto/piler.git synced 2024-12-25 07:20:11 +01:00
Code Issues Packages Projects Releases Wiki Activity

fixed the date parsing

Browse Source
This commit is contained in:
SJ 2014-02-03 15:38:10 +01:00
parent be0845d1b1
commit fd4184daa9
3 changed files with 106 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/config.h
View File

@ -14,7 +14,7 @@
#define VERSION "0.1.25-master-branch" #define VERSION "0.1.25-master-branch"
#define BUILD 858 #define BUILD 859
#define HOSTID "mailarchiver" #define HOSTID "mailarchiver"

102
src/parser_utils.c
View File

@ -111,7 +111,7 @@ long get_local_timezone_offset(){
unsigned long parse_date_header(char *datestr, struct __config *cfg){ unsigned long parse_date_header(char *datestr, struct __config *cfg){
int n=0; int n=0, len;
long offset=0; long offset=0;
unsigned long ts=0; unsigned long ts=0;
char *p, *q, *r, s[SMALLBUFSIZE]; char *p, *q, *r, s[SMALLBUFSIZE];
@ -120,9 +120,17 @@ unsigned long parse_date_header(char *datestr, struct __config *cfg){
datestr += 5; datestr += 5;
p = datestr; p = datestr;
tm.tm_year = 0;
tm.tm_mon = 0;
tm.tm_mday = 0;
tm.tm_wday = 0;
tm.tm_hour = 0;
tm.tm_min = 0;
tm.tm_sec = 0;
tm.tm_isdst = -1;
for(; *datestr; datestr++){ for(; *datestr; datestr++){
if(isspace(*datestr)) *datestr = ' '; if(isspace(*datestr) || *datestr == '.' || *datestr == ',') *datestr = ' ';
} }
@ -132,14 +140,26 @@ unsigned long parse_date_header(char *datestr, struct __config *cfg){
p = split_str(p, " ", s, sizeof(s)-1); p = split_str(p, " ", s, sizeof(s)-1);
if(strlen(s) > 0){ if(strlen(s) > 0){
n++; n++;
len = strlen(s);
q = strchr(s, ','); if(q) *q='\0'; /*
* A proper Date: header should look like this:
*
* Date: Mon, 3 Feb 2014 13:21:07 +0100
*
*
* However some email applications provide crap, eg.
*
* Sat, 4 Aug 2007 13:36:52 GMT-0700
* Sat, 4 Aug 07 13:36:52 GMT-0700
* 16 Dec 07 20:45:52
* 03 Jun 06 05:59:00 +0100
* 30.06.2005 17:47:42
*
* [wday] mday mon year h:m:s offset
*/
if(strlen(s) <= 2){ tm.tm_mday = atoi(s); continue; } if(n == 1 && len == 3){
if(strlen(s) == 4){ tm.tm_year = atoi(s) - 1900; continue; }
if(strlen(s) == 3){
if(strcmp(s, "Mon") == 0) tm.tm_wday = 1; if(strcmp(s, "Mon") == 0) tm.tm_wday = 1;
else if(strcmp(s, "Tue") == 0) tm.tm_wday = 2; else if(strcmp(s, "Tue") == 0) tm.tm_wday = 2;
else if(strcmp(s, "Wed") == 0) tm.tm_wday = 3; else if(strcmp(s, "Wed") == 0) tm.tm_wday = 3;
@ -147,41 +167,65 @@ unsigned long parse_date_header(char *datestr, struct __config *cfg){
else if(strcmp(s, "Fri") == 0) tm.tm_wday = 5; else if(strcmp(s, "Fri") == 0) tm.tm_wday = 5;
else if(strcmp(s, "Sat") == 0) tm.tm_wday = 6; else if(strcmp(s, "Sat") == 0) tm.tm_wday = 6;
else if(strcmp(s, "Sun") == 0) tm.tm_wday = 0; else if(strcmp(s, "Sun") == 0) tm.tm_wday = 0;
if(strcmp(s, "Jan") == 0) tm.tm_mon = 0;
else if(strcmp(s, "Feb") == 0) tm.tm_mon = 1;
else if(strcmp(s, "Mar") == 0) tm.tm_mon = 2;
else if(strcmp(s, "Apr") == 0) tm.tm_mon = 3;
else if(strcmp(s, "May") == 0) tm.tm_mon = 4;
else if(strcmp(s, "Jun") == 0) tm.tm_mon = 5;
else if(strcmp(s, "Jul") == 0) tm.tm_mon = 6;
else if(strcmp(s, "Aug") == 0) tm.tm_mon = 7;
else if(strcmp(s, "Sep") == 0) tm.tm_mon = 8;
else if(strcmp(s, "Oct") == 0) tm.tm_mon = 9;
else if(strcmp(s, "Nov") == 0) tm.tm_mon = 10;
else if(strcmp(s, "Dec") == 0) tm.tm_mon = 11;
continue;
} }
if(strlen(s) == 8){ if(n == 1 && len <= 2){
n++;
}
if(n == 2 && len <= 2){ tm.tm_mday = atoi(s); continue; }
if(n == 3){
if(len == 3){
if(strcmp(s, "Jan") == 0) tm.tm_mon = 0;
else if(strcmp(s, "Feb") == 0) tm.tm_mon = 1;
else if(strcmp(s, "Mar") == 0) tm.tm_mon = 2;
else if(strcmp(s, "Apr") == 0) tm.tm_mon = 3;
else if(strcmp(s, "May") == 0) tm.tm_mon = 4;
else if(strcmp(s, "Jun") == 0) tm.tm_mon = 5;
else if(strcmp(s, "Jul") == 0) tm.tm_mon = 6;
else if(strcmp(s, "Aug") == 0) tm.tm_mon = 7;
else if(strcmp(s, "Sep") == 0) tm.tm_mon = 8;
else if(strcmp(s, "Oct") == 0) tm.tm_mon = 9;
else if(strcmp(s, "Nov") == 0) tm.tm_mon = 10;
else if(strcmp(s, "Dec") == 0) tm.tm_mon = 11;
continue;
}
if(len == 2){
tm.tm_mon = atoi(s);
continue;
}
}
if(n == 4){
if(len == 4){ tm.tm_year = atoi(s) - 1900; continue; }
if(len == 2){ tm.tm_year = atoi(s); if(tm.tm_year < 70) tm.tm_year += 100; continue; }
}
if(n == 5 && len >= 5){
r = &s[0]; r = &s[0];
q = strchr(r, ':'); if(!q) break; q = strchr(r, ':'); if(!q) break;
*q = '\0'; tm.tm_hour = atoi(r); r = q+1; *q = '\0'; tm.tm_hour = atoi(r); r = q+1;
q = strchr(r, ':'); if(!q) break; q = strchr(r, ':'); if(q) *q = '\0';
*q = '\0'; tm.tm_min = atoi(r); r = q+1; tm.tm_min = atoi(r);
if(len == 8){
r = q+1;
tm.tm_sec = atoi(r);
}
tm.tm_sec = atoi(r);
break; break;
} }
} }
} while(p); } while(p);
tm.tm_isdst = -1;
ts = mktime(&tm); ts = mktime(&tm);
if(p && (*p == '+' || *p == '-')){ if(p && (*p == '+' || *p == '-')){

35
test/parser.c
View File

@ -138,12 +138,38 @@ int test_htmls(){
return count; return count;
} }
int test_dates(){
int count=0;
unsigned long ts;
char datestr[SMALLBUFSIZE];
struct __config cfg;
cfg.tweak_sent_time_offset = 0;
snprintf(datestr, sizeof(datestr)-2, "Date: Mon, 3 Feb 2014 13:16:09 +0100");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
snprintf(datestr, sizeof(datestr)-2, "Date: Sat, 4 Aug 07 13:36:52 GMT-0700");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
snprintf(datestr, sizeof(datestr)-2, "Date: 23 Sep 09 07:03 -0800");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
snprintf(datestr, sizeof(datestr)-2, "Date: 16 Dec 07 20:45:52");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
snprintf(datestr, sizeof(datestr)-2, "Date: 30.06.2005 17:47:42");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
snprintf(datestr, sizeof(datestr)-2, "Date: 03 Jun 06 05:59:00 +0100");
ts = parse_date_header(datestr, &cfg); printf("%s => %ld\n", datestr, ts);
return count;
}
int main(int argc, char **argv){ int main(int argc, char **argv){
int n; int n;
//struct __config cfg;
//cfg = read_config(CONFIG_FILE);
n = test_urls(); n = test_urls();
printf("testing fixURL(), errors: %d\n", n); printf("testing fixURL(), errors: %d\n", n);
@ -156,5 +182,8 @@ int main(int argc, char **argv){
n = test_htmls(); n = test_htmls();
printf("testing markHTML(), errors: %d\n", n); printf("testing markHTML(), errors: %d\n", n);
n = test_dates();
printf("testing parse_date_header(), errors: %d\n", n);
return 0; return 0;
} }