/*
* SARG Squid Analysis Report Generator http://sarg.sourceforge.net
- * 1998, 2011
+ * 1998, 2015
*
* SARG donations:
* please look at http://sarg.sourceforge.net/donations.php
#include "include/conf.h"
#include "include/defs.h"
+#ifdef HAVE_PCRE_H
+#include <pcre.h>
+#define USE_PCRE
+#endif
/*!
A host name and the name to report.
const char *Alias;
};
+#ifdef USE_PCRE
+/*!
+A regular expression.
+*/
+struct hostalias_regex
+{
+ //! The next regular expression to test.
+ struct hostalias_regex *Next;
+ //! The regular expression to match against the host name.
+ pcre *Re;
+ //! The replacement name.
+ const char *Alias;
+ //! \c True if this regular expression contains at least one subpattern
+ bool SubPartern;
+};
+#endif
+
//! The first host name.
static struct hostalias_name *FirstAliasName=NULL;
//! The first IPv4 address.
//! The first IPv§ address.
static struct hostalias_ipv6 *FirstAliasIpv6=NULL;
+#ifdef USE_PCRE
+static struct hostalias_regex *FirstAliasRe=NULL;
+#endif
+
/*!
Store a name to alias.
NameEnd=++str;
while (str<next && (unsigned char)*str>' ') {
if (*str=='*') {
- debuga(_("Host name alias \"%s*%s\" contains too many wildcards (*)"),NameBegin,NameEnd);
+ debuga(__FILE__,__LINE__,_("Host name alias \"%s*%s\" contains too many wildcards (*)\n"),NameBegin,NameEnd);
return(-1);
}
str++;
}
NameEndE=str;
if (NameEnd==NameEndE) {
- debuga(_("Host name alias \"%*s\" must not end with a wildcard"),(int)(next-name),name);
+ debuga(__FILE__,__LINE__,_("Host name alias \"%*s\" must not end with a wildcard\n"),(int)(next-name),name);
return(-1);
}
} else {
len=(int)(NameBeginE-NameBegin);
tmp=malloc(len+1);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
memcpy(tmp,NameBegin,len);
tmp=malloc(len+1);
if (!tmp) {
if (NameBegin) free((void*)NameBegin);
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
memcpy(tmp,NameEnd,len);
tmp[len]='\0';
NameEnd=tmp;
}
-
+
// ignore duplicates
prev_alias=NULL;
for (alias=FirstAliasName ; alias ; alias=alias->Next) {
if (!new_alias) {
if (NameBegin) free((void*)NameBegin);
if (NameEnd) free((void*)NameEnd);
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
new_alias->MinLen=0;
len=(int)(ReplaceE-Replace);
tmp=malloc(len+2);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
tmp[0]=ALIAS_PREFIX;
} else {
tmp=malloc(new_alias->MinLen+2);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
tmp[0]=ALIAS_PREFIX;
}
new_alias->Alias=tmp;
}
-
+
new_alias->Next=NULL;
if (prev_alias)
prev_alias->Next=new_alias;
// insert into the list
new_alias=malloc(sizeof(*new_alias));
if (!new_alias) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
memcpy(new_alias->Ip,ipv4,4);
len=(int)(ReplaceE-Replace);
tmp=malloc(len+2);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
tmp[0]=ALIAS_PREFIX;
} else {
tmp=malloc(5*4+1);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
sprintf(tmp,"%c%d.%d.%d.%d/%d",ALIAS_PREFIX,ipv4[0],ipv4[1],ipv4[2],ipv4[3],nbits);
new_alias->Alias=tmp;
}
-
+
if (prev_alias) {
new_alias->Next=prev_alias->Next;
prev_alias->Next=new_alias;
// insert into the list
new_alias=malloc(sizeof(*new_alias));
if (!new_alias) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
memcpy(new_alias->Ip,ipv6,8*sizeof(unsigned short int));
len=ReplaceE-Replace;
tmp=malloc(len+2);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
tmp[0]=ALIAS_PREFIX;
} else {
tmp=malloc(5*8+5);
if (!tmp) {
- debuga(_("Not enough memory to store the host name aliasing directives"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
return(-1);
}
sprintf(tmp,"%c%x:%x:%x:%x:%x:%x:%x:%x/%d",ALIAS_PREFIX,ipv6[0],ipv6[1],ipv6[2],ipv6[3],ipv6[4],ipv6[5],ipv6[6],ipv6[7],nbits);
new_alias->Alias=tmp;
}
-
+
if (prev_alias) {
new_alias->Next=prev_alias->Next;
prev_alias->Next=new_alias;
return(1);
}
+#ifdef USE_PCRE
/*!
-Read the file containing the host names to alias in the report.
+Store a regular expression to match the alias.
-\param Filename The name of the file.
+\retval 1 Alias added.
+\retval 0 Ignore the line.
+\retval -1 Error.
*/
-void read_hostalias(const char *Filename)
+static int Alias_StoreRegexp(char *buf)
+{
+ char Delimiter;
+ char *End;
+ struct hostalias_regex *alias;
+ struct hostalias_regex *new_alias;
+ struct hostalias_regex **prev_alias;
+ const char *PcreError;
+ int ErrorOffset;
+ char *Replace;
+ int len;
+ char *tmp;
+ int i;
+
+ // find the pattern
+ Delimiter=*buf++;
+ for (End=buf ; *End && *End!=Delimiter ; End++) {
+ if (*End=='\\') {
+ if (End[1]=='\0') {
+ debuga(__FILE__,__LINE__,_("Invalid NUL character found in regular expression\n"));
+ return(-1);
+ }
+ End++; //ignore the escaped character
+ }
+ }
+ if (*End!=Delimiter) {
+ debuga(__FILE__,__LINE__,_("Unterminated regular expression\n"));
+ return(-1);
+ }
+ *End++='\0';
+
+ // find the alias
+ for (Replace=End ; *Replace==' ' || *Replace=='\t' ; Replace++);
+ for (End=Replace ; *End && (unsigned char)*End>' ' ; End++);
+ *End='\0';
+
+ // store it
+ new_alias=malloc(sizeof(*new_alias));
+ if (!new_alias) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
+ return(-1);
+ }
+ new_alias->Next=NULL;
+ new_alias->Re=pcre_compile(buf,0,&PcreError,&ErrorOffset,NULL);
+ if (new_alias->Re==NULL) {
+ debuga(__FILE__,__LINE__,_("Failed to compile the regular expression \"%s\": %s\n"),buf,PcreError);
+ free(new_alias);
+ return(-1);
+ }
+ len=strlen(Replace);
+ tmp=malloc(len+2);
+ if (!tmp) {
+ debuga(__FILE__,__LINE__,_("Not enough memory to store the host name aliasing directives\n"));
+ pcre_free(new_alias->Re);
+ return(-1);
+ }
+ tmp[0]=ALIAS_PREFIX;
+ memcpy(tmp+1,Replace,len);
+ tmp[len+1]='\0';
+ new_alias->Alias=tmp;
+
+ new_alias->SubPartern=false;
+ for (i=1 ; tmp[i] ; i++)
+ // both the sed \1 and the perl $1 replacement operators are accepted
+ if ((tmp[i]=='\\' || tmp[i]=='$') && isdigit(tmp[i+1])) {
+ new_alias->SubPartern=true;
+ break;
+ }
+
+ // chain it
+ prev_alias=&FirstAliasRe;
+ for (alias=FirstAliasRe ; alias ; alias=alias->Next)
+ prev_alias=&alias->Next;
+ *prev_alias=new_alias;
+
+ return(1);
+}
+#endif
+
+/*!
+Store an alias in the corresponding list.
+
+\param buf The string to parse and store.
+
+\retval 0 No error.
+\retval -1 Error in file.
+\retval -2 Unknown string type to store.
+*/
+static int Alias_Store(char *buf)
{
- FILE *fi;
- longline line;
- char *buf;
int type;
const char *name;
unsigned char ipv4[4];
unsigned short int ipv6[8];
int nbits;
const char *next;
+ int Error=-2;
+
+ if (strncasecmp(buf,"re:",3)==0) {
+#ifdef USE_PCRE
+ if (Alias_StoreRegexp(buf+3)<0)
+ return(-1);
+ return(0);
+#else
+ debuga(__FILE__,__LINE__,_("PCRE not compiled in therefore the regular expressions are not available in the host alias file\n"));
+ return(-1);
+#endif
+ }
+ type=extract_address_mask(buf,&name,ipv4,ipv6,&nbits,&next);
+ if (type<0) {
+ return(-1);
+ }
+
+ if (type==1) {
+ Error=Alias_StoreName(name,next);
+ } else if (type==2) {
+ Error=Alias_StoreIpv4(ipv4,nbits,next);
+ } else if (type==3) {
+ Error=Alias_StoreIpv6(ipv6,nbits,next);
+ }
+ if (Error<0) return(-1);
+ return(0);
+}
+
+/*!
+Read the file containing the host names to alias in the report.
+
+\param Filename The name of the file.
+*/
+void read_hostalias(const char *Filename)
+{
+ FILE *fi;
+ longline line;
+ char *buf;
- if (debug) debuga(_("Reading host alias file \"%s\"\n"),Filename);
+ if (debug) debuga(__FILE__,__LINE__,_("Reading host alias file \"%s\"\n"),Filename);
fi=fopen(Filename,"rt");
if (!fi) {
- debuga(_("Cannot read host name alias file \"%s\" - %s\n"),Filename,strerror(errno));
+ debuga(__FILE__,__LINE__,_("Cannot read host name alias file \"%s\": %s\n"),Filename,strerror(errno));
exit(EXIT_FAILURE);
}
-
+
if ((line=longline_create())==NULL) {
- debuga(_("Not enough memory to read the host name aliases\n"));
+ debuga(__FILE__,__LINE__,_("Not enough memory to read file \"%s\"\n"),Filename);
exit(EXIT_FAILURE);
}
while ((buf=longline_read(fi,line)) != NULL) {
- type=extract_address_mask(buf,&name,ipv4,ipv6,&nbits,&next);
- if (type<0) {
- debuga(_("While reading \"%s\""),Filename);
+ if (Alias_Store(buf)<0) {
+ debuga(__FILE__,__LINE__,_("While reading \"%s\"\n"),Filename);
exit(EXIT_FAILURE);
}
-
- if (type==1) {
- Alias_StoreName(name,next);
- } else if (type==2) {
- Alias_StoreIpv4(ipv4,nbits,next);
- } else if (type==3) {
- Alias_StoreIpv6(ipv6,nbits,next);
- }
}
-
+
longline_destroy(&line);
- fclose(fi);
-
+ if (fclose(fi)==EOF) {
+ debuga(__FILE__,__LINE__,_("Read error in \"%s\": %s\n"),Filename,strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
if (debug) {
struct hostalias_name *alias1;
struct hostalias_ipv4 *alias4;
struct hostalias_ipv6 *alias6;
- debuga(_("List of host names to alias:\n"));
+ debuga(__FILE__,__LINE__,_("List of host names to alias:\n"));
for (alias1=FirstAliasName ; alias1 ; alias1=alias1->Next) {
if (alias1->HostName_Prefix && alias1->HostName_Suffix)
- debuga(_(" %s*%s => %s\n"),alias1->HostName_Prefix,alias1->HostName_Suffix,alias1->Alias);
+ debuga(__FILE__,__LINE__,_(" %s*%s => %s\n"),alias1->HostName_Prefix,alias1->HostName_Suffix,alias1->Alias);
else if (alias1->HostName_Prefix)
- debuga(_(" %s => %s\n"),alias1->HostName_Prefix,alias1->Alias);
+ debuga(__FILE__,__LINE__,_(" %s => %s\n"),alias1->HostName_Prefix,alias1->Alias);
else
- debuga(_(" *%s => %s\n"),alias1->HostName_Suffix,alias1->Alias);
+ debuga(__FILE__,__LINE__,_(" *%s => %s\n"),alias1->HostName_Suffix,alias1->Alias);
}
for (alias4=FirstAliasIpv4 ; alias4 ; alias4=alias4->Next) {
- debuga(_(" %d.%d.%d.%d/%d => %s\n"),alias4->Ip[0],alias4->Ip[1],alias4->Ip[2],alias4->Ip[3],alias4->NBits,alias4->Alias);
+ debuga(__FILE__,__LINE__,_(" %d.%d.%d.%d/%d => %s\n"),alias4->Ip[0],alias4->Ip[1],alias4->Ip[2],alias4->Ip[3],alias4->NBits,alias4->Alias);
}
for (alias6=FirstAliasIpv6 ; alias6 ; alias6=alias6->Next) {
- debuga(_(" %x:%x:%x:%x:%x:%x:%x:%x/%d => %s\n"),alias6->Ip[0],alias6->Ip[1],alias6->Ip[2],alias6->Ip[3],
+ debuga(__FILE__,__LINE__,_(" %x:%x:%x:%x:%x:%x:%x:%x/%d => %s\n"),alias6->Ip[0],alias6->Ip[1],alias6->Ip[2],alias6->Ip[3],
alias6->Ip[4],alias6->Ip[5],alias6->Ip[6],alias6->Ip[7],alias6->NBits,alias6->Alias);
}
}
*/
void free_hostalias(void)
{
- struct hostalias_name *alias1;
- struct hostalias_name *next1;
- struct hostalias_ipv4 *alias4;
- struct hostalias_ipv4 *next4;
- struct hostalias_ipv6 *alias6;
- struct hostalias_ipv6 *next6;
-
- for (alias1=FirstAliasName ; alias1 ; alias1=next1) {
- next1=alias1->Next;
- if (alias1->HostName_Prefix) free((void *)alias1->HostName_Prefix);
- if (alias1->HostName_Suffix) free((void *)alias1->HostName_Suffix);
- free((void *)alias1->Alias);
- free(alias1);
- }
- FirstAliasName=NULL;
- for (alias4=FirstAliasIpv4 ; alias4 ; alias4=next4) {
- next4=alias4->Next;
- free((void *)alias4->Alias);
- free(alias4);
- }
- FirstAliasIpv4=NULL;
- for (alias6=FirstAliasIpv6 ; alias6 ; alias6=next6) {
- next6=alias6->Next;
- free((void *)alias6->Alias);
- free(alias6);
- }
- FirstAliasIpv6=NULL;
+ {
+ struct hostalias_name *alias1;
+ struct hostalias_name *next1;
+
+ for (alias1=FirstAliasName ; alias1 ; alias1=next1) {
+ next1=alias1->Next;
+ if (alias1->HostName_Prefix) free((void *)alias1->HostName_Prefix);
+ if (alias1->HostName_Suffix) free((void *)alias1->HostName_Suffix);
+ free((void *)alias1->Alias);
+ free(alias1);
+ }
+ FirstAliasName=NULL;
+ }
+ {
+ struct hostalias_ipv4 *alias4;
+ struct hostalias_ipv4 *next4;
+
+ for (alias4=FirstAliasIpv4 ; alias4 ; alias4=next4) {
+ next4=alias4->Next;
+ free((void *)alias4->Alias);
+ free(alias4);
+ }
+ FirstAliasIpv4=NULL;
+ }
+ {
+ struct hostalias_ipv6 *alias6;
+ struct hostalias_ipv6 *next6;
+
+ for (alias6=FirstAliasIpv6 ; alias6 ; alias6=next6) {
+ next6=alias6->Next;
+ free((void *)alias6->Alias);
+ free(alias6);
+ }
+ FirstAliasIpv6=NULL;
+ }
+#ifdef USE_PCRE
+ {
+ struct hostalias_regex *alias;
+ struct hostalias_regex *next;
+
+ for (alias=FirstAliasRe ; alias ; alias=next) {
+ next=alias->Next;
+ pcre_free(alias->Re);
+ free((void *)alias->Alias);
+ free(alias);
+ }
+ FirstAliasRe=NULL;
+ }
+#endif
}
/*!
\return The pointer to the host name or its alias.
*/
-const char *alias_url_name(const char *url,const char *next)
+static const char *alias_url_name(const char *url,const char *next)
{
struct hostalias_name *alias;
int len;
\return The pointer to the host name or its alias.
*/
-const char *alias_url_ipv4(const char *url,unsigned char *ipv4)
+static const char *alias_url_ipv4(const char *url,unsigned char *ipv4)
{
struct hostalias_ipv4 *alias;
int len;
for (alias=FirstAliasIpv4 ; alias ; alias=alias->Next) {
- len=alias->NBits;
+ len=alias->NBits;
if ((len<8 || memcmp(ipv4,alias->Ip,len/8)==0) && ((len%8)==0 || (ipv4[len/8] ^ alias->Ip[len/8]) & (0xFFU<<(8-len%8)))==0) {
return(alias->Alias);
}
\return The pointer to the host name or its alias.
*/
-const char *alias_url_ipv6(const char *url,unsigned short int *ipv6)
+static const char *alias_url_ipv6(const char *url,unsigned short int *ipv6)
{
struct hostalias_ipv6 *alias;
int len;
return(url);
}
+#ifdef USE_PCRE
+/*!
+Replace the host name by its alias if it is in our list.
+
+\param url_ptr A pointer to the host name to match. It is replaced
+by a pointer to the alias if a match is found.
+
+\return \c True if a match is found or \c false if it failed.
+
+\warning The function is not thread safe as it may return a static
+internal buffer.
+*/
+static bool alias_url_regex(const char **url_ptr)
+{
+ struct hostalias_regex *alias;
+ int nmatches;
+ const char *url;
+ int url_len;
+ int ovector[30];//size must be a multiple of 3
+ static char Replacement[1024];
+ const char *str;
+ int i;
+ int sub;
+ int repl_idx;
+
+ url=*url_ptr;
+ url_len=strlen(url);
+ for (alias=FirstAliasRe ; alias ; alias=alias->Next) {
+ nmatches=pcre_exec(alias->Re,NULL,url,url_len,0,0,ovector,sizeof(ovector)/sizeof(ovector[0]));
+ if (nmatches>=0) {
+ if (nmatches==0) nmatches=(int)(sizeof(ovector)/sizeof(ovector[0]))/3*2; //only 2/3 of the vector is used by pcre_exec
+ if (nmatches==1 || !alias->SubPartern) { //no subpattern to replace
+ *url_ptr=alias->Alias;
+ } else {
+ repl_idx=0;
+ str=alias->Alias;
+ for (i=0 ; str[i] ; i++) {
+ // both the sed \1 and the perl $1 replacement operators are accepted
+ if ((str[i]=='\\' || str[i]=='$') && isdigit(str[i+1])) {
+ sub=str[++i]-'0';
+ if (sub>=1 && sub<=nmatches) {
+ /*
+ * ovector[sub] is the start position of the match.
+ * ovector[sub+1] is the end position of the match.
+ */
+ sub<<=1;
+ if (repl_idx+ovector[sub+1]-ovector[sub]>=sizeof(Replacement)-1) break;
+ memcpy(Replacement+repl_idx,url+ovector[sub],ovector[sub+1]-ovector[sub]);
+ repl_idx+=ovector[sub+1]-ovector[sub];
+ continue;
+ }
+ }
+ if (repl_idx>=sizeof(Replacement)-1) break;
+ Replacement[repl_idx++]=str[i];
+ }
+ Replacement[repl_idx]='\0';
+ *url_ptr=Replacement;
+ }
+ return(true);
+ }
+ }
+ return(false);
+}
+#endif
+
+/*!
+Find the beginning of the URL beyond the scheme://
+
+\param url The url possibly containing a scheme.
+
+\return The beginning of the url beyond the scheme.
+*/
+const char *skip_scheme(const char *url)
+{
+ const char *str;
+
+ /*
+ Skip any scheme:// at the beginning of the URL (see rfc2396 section 3.1).
+ The underscore is not part of the standard but is found in the squid logs as cache_object://.
+ */
+ for (str=url ; *str && (isalnum(*str) || *str=='+' || *str=='-' || *str=='.' || *str=='_') ; str++);
+ if (str[0]==':' && str[1]=='/' && str[2]=='/') {
+ url=str+3;
+ while (*url=='/') url++;
+ }
+ return(url);
+}
+
/*!
Get the part of the URL necessary to generate the report.
\param full_url \c True to keep the whole URL. If \c false,
the URL is truncated to only keep the host name and port number.
*/
-const char *process_url(char *url,bool full_url)
+const char *process_url(const char *url,bool full_url)
{
- char *str;
+ static char short_url[1024];
+ int i;
const char *start;
int type;
- const char *address;
unsigned char ipv4[4];
unsigned short int ipv6[8];
const char *next;
- /*
- Remove any scheme:// at the beginning of the URL (see rfc2396 section 3.1).
- The underscore is not part of the standard but is found in the squid logs as cache_object://.
- */
- for (str=url ; *str && (isalnum(*str) || *str=='+' || *str=='-' || *str=='.' || *str=='_') ; str++);
- if (*str==':' && str[1]=='/' && str[2]=='/') {
- url=str+3;
- while (*url=='/') url++;
- }
-
- start=url;
+ start=skip_scheme(url);
if (!full_url) {
- for (str=url ; *str && *str!='/' && *str!='?' ; str++);
- *str='\0';
- type=extract_address_mask(url,&address,ipv4,ipv6,NULL,&next);
+ for (i=0 ; i<sizeof(short_url)-1 && start[i] && start[i]!='/' && start[i]!='?' ; i++)
+ short_url[i]=start[i];
+ short_url[i]='\0';
+ start=short_url;
+#ifdef USE_PCRE
+ if (FirstAliasRe) {
+ if (alias_url_regex(&start)) return(start);
+ }
+#endif
+ type=extract_address_mask(start,NULL,ipv4,ipv6,NULL,&next);
if (type==1) {
if (FirstAliasName)
start=alias_url_name(start,next);
projects / thirdparty / sarg.git / blobdiff