Files
TempleOS-Rijn/TempleOS/Demo/WebLogExample/WebLogRep.CPP
T
2018-09-08 21:45:45 +02:00

442 lines
10 KiB
C++

#define DOWNLOAD_FILE1 "TempleOSCD.ISO"
#define DOWNLOAD_FILE1_SIZE 4000000
#define DOWNLOAD_FILE2 "TempleOSCD2.ISO"
#define DOWNLOAD_FILE2_SIZE 4000000
#define MAX_HOURS (24*3)
class LogStruct
{
LogStruct *next,*last;
LogStruct *ip_num_left,*ip_num_right;
U32 ip_num,code;
I64 size;
U8 *file,*link;
CDate datetime;
};
class LinkStruct
{
LinkStruct *left,*right;
U8 *link;
I64 cnt;
};
class BlockedStruct
{
BlockedStruct *next,*last;
U32 ip_num;
};
U0 LogStructDel(LogStruct *templg)
{
Free(templg->file);
Free(templg->link);
Free(templg);
}
U0 PrsSingleLogFile(LogStruct *root,U8 *name)
{
CDoc *doc=DocRead(name,DOCF_PLAIN_TEXT_WITH_TABS|DOCF_DBL_DOLLARS);
CDocEntry *doc_e=doc->root.next;
U8 *src,*src2,*mon_lst=Define("ST_MONTHS");
LogStruct *templg;
CDateStruct ds;
I64 i;
"%Q\n",name;
while (doc_e!=doc) {
if (doc_e->type_u8==DOCT_TEXT) {
templg=CAlloc(sizeof(LogStruct));
try {
src=doc_e->tag;
templg->ip_num.u8[3]=A2I(src,10,&src);
if (*src++!='.') throw;
templg->ip_num.u8[2]=A2I(src,10,&src);
if (*src++!='.') throw;
templg->ip_num.u8[1]=A2I(src,10,&src);
if (*src++!='.') throw;
templg->ip_num.u8[0]=A2I(src,10,&src);
do if (!*src) throw;
while (*src++!='[');
MemSet(&ds,0,sizeof(CDateStruct));
ds.day_of_mon=A2I(src,10,&src);
if (*src++!='/') throw;
src2=src;
do if (!*src2) throw;
while (*src2++!='/');
* --src2=0;
ds.mon=1+LstEntryMatch(src,mon_lst,LEF_IGNORE_CASE);
src=++src2;
ds.year=A2I(src,10,&src);
if (*src++!=':') throw;
ds.hour=A2I(src,10,&src);
if (*src++!=':') throw;
ds.min=A2I(src,10,&src);
if (*src++!=':') throw;
ds.sec=A2I(src,10,&src);
templg->datetime=DateStruct2Date(&ds);
if (*src++!=CH_SPACE) throw;
i=A2I(src,10,&src);
templg->datetime-=(i/100+i%100/60.0)*CDATE_FREQ*60*60;
do if (!*src) throw;
while (*src++!=']');
if (*src++!=CH_SPACE) throw;
if (*src++!='\"') throw;
if (!StrNCmp(src,"GET ",4)) {
src2=src+=4;
do if (!*src2) throw;
while (*src2++!=CH_SPACE);
* --src2=0;
templg->file=StrNew(src);
src=++src2;
do if (!*src) throw;
while (*src++!='\"');
templg->code=A2I(src,10,&src);
if (*src++!=CH_SPACE) throw;
templg->size=A2I(src,10,&src);
if (*src++!=CH_SPACE) throw;
if (*src++!='\"') throw;
src2=src;
do if (!*src2) throw;
while (*src2++!='\"');
* --src2=0;
templg->link=StrNew(src);
src=++src2;
QueIns(templg,root->last);
} else if (!StrNCmp(src,"HEAD ",5)) {
LogStructDel(templg);
} else
throw;
} catch {
Fs->catch_except=TRUE;
"%Q\n",doc_e->tag;
LogStructDel(templg);
}
}
doc_e=doc_e->next;
}
DocDel(doc);
}
LogStruct *PrsLogFiles(U8 *files_find_mask)
{
LogStruct *root=CAlloc(sizeof(LogStruct));
CDirEntry *tempm=FilesFind(files_find_mask),*tempm1=tempm;
QueInit(root);
while (tempm) {
PrsSingleLogFile(root,tempm->full_name);
tempm=tempm->next;
}
CDirLstDel(tempm1);
return root;
}
U0 LogLstDel(LogStruct *root)
{
LogStruct *templg=root->next,*templg1;
while (templg!=root) {
templg1=templg->next;
LogStructDel(templg);
templg=templg1;
}
}
U0 BlockedStructAdd(BlockedStruct *root,U32 ip_num)
{
BlockedStruct *tempb=CAlloc(sizeof(BlockedStruct));
tempb->ip_num=ip_num;
QueIns(tempb,root->last);
}
Bool IsBlocked(BlockedStruct *root,U32 ip_num)
{
BlockedStruct *tempb=root->next;
while (tempb!=root) {
if (tempb->ip_num==ip_num)
return TRUE;
tempb=tempb->next;
}
return FALSE;
}
U0 BlockIPNums(LogStruct *root)
{
BlockedStruct blocked_root;
LogStruct *templg=root->next,*templg1;
QueInit(&blocked_root);
BlockedStructAdd(&blocked_root,68<<24+224<<16+130<<8+96); //Me 68.224.130.96
BlockedStructAdd(&blocked_root,68<<24+96<<16+100<<8+126); //Me 68.96.100.126
//pass 1: collect robot lst
while (templg!=root) {
if (StrIStr(templg->file,"ROBOT") &&
!IsBlocked(&blocked_root,templg->ip_num))
BlockedStructAdd(&blocked_root,templg->ip_num);
templg=templg->next;
}
//pass 2: removed blocked ip_nums
templg=root->next;
while (templg!=root) {
templg1=templg->next;
if (IsBlocked(&blocked_root,templg->ip_num)) {
QueRem(templg);
LogStructDel(templg);
}
templg=templg1;
}
QueDel(&blocked_root);
}
Bool IsDownLoad(LogStruct *templg)
{
if (
StrStr(templg->file,DOWNLOAD_FILE1) && templg->size>=DOWNLOAD_FILE1_SIZE ||
StrStr(templg->file,DOWNLOAD_FILE2) && templg->size>=DOWNLOAD_FILE2_SIZE )
return TRUE;
else
return FALSE;
}
Bool IsIndex(LogStruct *templg)
{
if (!StrCmp(templg->file,"/index.html") ||
!StrNCmp(templg->file,"/v",2) ||
!StrCmp(templg->file,"/"))
return TRUE;
else
return FALSE;
}
Bool IPNumTreeAdd(LogStruct **_root,LogStruct *templg)
{
LogStruct *root;
if (UnusedStk<0x200) {
PutErr("Stk Overflow");
ThrowBreak;
}
if (root=*_root) {
if (templg->ip_num==root->ip_num)
return TRUE;
else if (templg->ip_num<root->ip_num)
return IPNumTreeAdd(&root->ip_num_left,templg);
else
return IPNumTreeAdd(&root->ip_num_right,templg);
} else {
templg->ip_num_left=NULL;
templg->ip_num_right=NULL;
*_root=templg;
return FALSE;
}
}
U0 LinkTreeAdd(LinkStruct **_root,LogStruct *templg)
{
I64 i;
LinkStruct *root,*templk;
if (UnusedStk<0x200) {
PutErr("Stk Overflow");
ThrowBreak;
}
if (root=*_root) {
if (!(i=StrCmp(templg->link,root->link)))
root->cnt++;
else if (i<0)
LinkTreeAdd(&root->left,templg);
else
LinkTreeAdd(&root->right,templg);
} else {
templk=CAlloc(sizeof(LinkStruct));
templk->link=templg->link;
templk->cnt=1;
*_root=templk;
}
}
U0 LinkTreeDel(LinkStruct *root)
{
if (root) {
LinkTreeDel(root->left);
LinkTreeDel(root->right);
Free(root);
}
}
U0 LinkTreeTraverse(LinkStruct *root)
{
if (root) {
LinkTreeTraverse(root->left);
"%3d:%Q\n",root->cnt,root->link;
LinkTreeTraverse(root->right);
}
}
U0 DownLoadRep(LogStruct *root,CDate start,CDate end)
{
I64 i,j,cnt,dups,
hours_start,hours_end,*hour_cnts,*dup_cnts,
days_start,days_end,*day_cnts,*day_dup_cnts;
LogStruct *templg=root->next,*dup_root=NULL;
LinkStruct *link_root=NULL;
i=start*24;hours_start=i.u32[1];
i=end *24;hours_end =i.u32[1];
days_start=(start+local_time_offset)>>32;
days_end =(end+local_time_offset)>>32;
hour_cnts=CAlloc((hours_end-hours_start+1)*sizeof(I64));
dup_cnts =CAlloc((hours_end-hours_start+1)*sizeof(I64));
day_cnts =CAlloc((days_end-days_start+1)*sizeof(I64));
day_dup_cnts=CAlloc((days_end-days_start+1)*sizeof(I64));
dups=cnt=0;
while (templg!=root) {
if (start<=templg->datetime<=end && IsDownLoad(templg)) {
i=templg->datetime*24;
hour_cnts[i.u32[1]-hours_start]++;
day_cnts[(templg->datetime+local_time_offset)>>32-days_start]++;
cnt++;
if (IPNumTreeAdd(&dup_root,templg)) {
day_dup_cnts[(templg->datetime+local_time_offset)>>32-days_start]++;
dup_cnts[i.u32[1]-hours_start]++;
dups++;
}
LinkTreeAdd(&link_root,templg);
}
templg=templg->next;
}
"\n\nDownloads of /TempleOSCD.ISO\n";
for (i=start;i<=end;i+=1<<32)
"%D:%5d%5d\n",i,day_dup_cnts[(i+local_time_offset)>>32-days_start],
day_cnts[(i+local_time_offset)>>32-days_start];
"\n\nDownloads of /TempleOSCD.ISO\n"
"'-' is a dup. '+' is not a dup.\n";
if (hours_end-hours_start>=MAX_HOURS)
i=hours_end-MAX_HOURS+1;
else
i=hours_start;
for (;i<=hours_end;i++) {
"%D %h20T: ",i<<32/24,i<<32/24;
for (j=0;j<dup_cnts[i-hours_start];j++)
'-';
for (;j<hour_cnts[i-hours_start];j++)
'+';
NewLine;
}
"Total:%d Dups:%d\n",cnt,dups;
"\n\nDownloads of /TempleOSCD.ISO\n";
LinkTreeTraverse(link_root);
NewLine;
LinkTreeDel(link_root);
Free(hour_cnts);
Free(dup_cnts);
Free(day_cnts);
Free(day_dup_cnts);
}
U0 IndexRep(LogStruct *root,CDate start,CDate end)
{
I64 i,j,cnt,dups,
hours_start,hours_end,*hour_cnts,*dup_cnts,
days_start,days_end,*day_cnts,*day_dup_cnts;
LogStruct *templg=root->next,*dup_root=NULL;
LinkStruct *link_root=NULL;
i=start*24;hours_start=i.u32[1];
i=end *24;hours_end =i.u32[1];
days_start=(start+local_time_offset)>>32;
days_end =(end+local_time_offset)>>32;
hour_cnts=CAlloc((hours_end-hours_start+1)*sizeof(I64));
dup_cnts =CAlloc((hours_end-hours_start+1)*sizeof(I64));
day_cnts =CAlloc((days_end-days_start+1)*sizeof(I64));
day_dup_cnts=CAlloc((days_end-days_start+1)*sizeof(I64));
dups=cnt=0;
while (templg!=root) {
if (start<=templg->datetime<=end && IsIndex(templg)) {
i=templg->datetime*24;
hour_cnts[i.u32[1]-hours_start]++;
day_cnts[(templg->datetime+local_time_offset)>>32-days_start]++;
cnt++;
if (IPNumTreeAdd(&dup_root,templg)) {
day_dup_cnts[(templg->datetime+local_time_offset)>>32-days_start]++;
dup_cnts[i.u32[1]-hours_start]++;
dups++;
}
LinkTreeAdd(&link_root,templg);
}
templg=templg->next;
}
"\n\nHits on /index.html\n"
"'-' is a dup. '+' is not a dup.\n";
for (i=start;i<=end;i+=1<<32)
"%D:%5d%5d\n",i,day_dup_cnts[(i+local_time_offset)>>32-days_start],
day_cnts[(i+local_time_offset)>>32-days_start];
"\n\nHits on /index.html\n";
if (hours_end-hours_start>=MAX_HOURS)
i=hours_end-MAX_HOURS+1;
else
i=hours_start;
for (;i<=hours_end;i++) {
"%D %h20T: ",i<<32/24,i<<32/24;
for (j=0;j<dup_cnts[i-hours_start];j++)
'-';
for (;j<hour_cnts[i-hours_start];j++)
'+';
NewLine;
}
"Total:%d Dups:%d\n",cnt,dups;
"\n\nHits on /index.html\n";
LinkTreeTraverse(link_root);
NewLine;
LinkTreeDel(link_root);
Free(hour_cnts);
Free(dup_cnts);
Free(day_cnts);
Free(day_dup_cnts);
}
U0 WebLogRep(U8 *mask,U8 *output_filename)
{
LogStruct *root;
CDate start,end;
DocMax;
root=PrsLogFiles(mask);
start=PmtCDate("Start(%D):",root->next->datetime);
end =PmtCDate("End (%D):",root->last->datetime);
BlockIPNums(root);
DocClear;
"$$WW,0$$";
IndexRep(root,start,end);
DownLoadRep(root,start,end);
StrCpy(DocPut->filename.name,output_filename);
DocWrite(DocPut,TRUE);
"$$WW,1$$";
LogLstDel(root);
}
WebLogRep("::/TempleOS/Demo/WebLogExample/*.LOG*","::/TempleOS/Demo/WebLogExample/WebLogRep.TXT.Z");