Mirror/zeek
1
0
Fork 0
mirror of https://github.com/zeek/zeek.git synced 2025-10-02 14:48:21 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

Move regex matching code to zeek namespaces

Browse source
This commit is contained in:
Tim Wojtulewicz 2020-07-17 15:21:00 -07:00
parent 382812298d
commit c7dc7fc955
26 changed files with 266 additions and 172 deletions
Download patch file Download diff file Expand all files Collapse all files

2
auxil/binpac

@ -1 +1 @@
Subproject commit fbd29e6c8aed54a81402ac60f4b4d859ead35b94
Subproject commit 17fa8f31da6f950d3475113fe5c24000dc3cb6bd

4
src/CCL.cc
View file

@ -8,6 +8,8 @@
#include "RE.h"
#include "DFA.h"
namespace zeek::detail {
CCL::CCL()
{
syms = new int_list;
@ -48,3 +50,5 @@ unsigned int CCL::MemoryAllocation() const
{
return padded_sizeof(*this) + padded_sizeof(*syms) + pad_size(syms->size() * sizeof(int_list::value_type));
}
} // namespace zeek::detail

10
src/CCL.h
View file

@ -2,11 +2,12 @@
#pragma once
#include <vector>
#include "util.h" // for ptr_compat_int
#include <vector>
namespace zeek::detail {
typedef std::vector<ptr_compat_int> int_list;
using int_list = std::vector<ptr_compat_int>;
class CCL {
public:
@ -32,3 +33,8 @@ protected:
int negated;
int index;
};
} // namespace zeek::detail
using int_list [[deprecated("Remove in v4.1. Use zeek::detail::int_list.")]] = zeek::detail::int_list;
using CCL [[deprecated("Remove in v4.1. Use zeek::detail::CCL.")]] = zeek::detail::CCL;

6
src/CompHash.cc
View file

@ -873,12 +873,12 @@ const char* CompositeHash::RecoverOneVal(
case zeek::TYPE_PATTERN:
{
RE_Matcher* re = nullptr;
zeek::RE_Matcher* re = nullptr;
if ( is_singleton )
{
kp1 = kp0;
int divider = strlen(kp0) + 1;
re = new RE_Matcher(kp1, kp1 + divider);
re = new zeek::RE_Matcher(kp1, kp1 + divider);
kp1 += k.Size();
}
else
@ -886,7 +886,7 @@ const char* CompositeHash::RecoverOneVal(
const uint64_t* const len = AlignType<uint64_t>(kp0);
kp1 = reinterpret_cast<const char*>(len+2);
re = new RE_Matcher(kp1, kp1 + len[0]);
re = new zeek::RE_Matcher(kp1, kp1 + len[0]);
kp1 += len[0] + len[1];
}

2
src/Conn.h
View file

@ -25,9 +25,9 @@ class Connection;
class ConnectionTimer;
class NetSessions;
class LoginConn;
class Specific_RE_Matcher;
class EncapsulationStack;
ZEEK_FORWARD_DECLARE_NAMESPACED(Specific_RE_Matcher, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(RuleEndpointState, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(RuleHdrTest, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(Val, zeek);

4
src/DFA.cc
View file

@ -7,6 +7,8 @@
#include "Desc.h"
#include "Hash.h"
namespace zeek::detail {
unsigned int DFA_State::transition_counter = 0;
DFA_State::DFA_State(int arg_state_num, const EquivClass* ec,
@ -469,3 +471,5 @@ int DFA_Machine::Rep(int sym)
return -1;
}
} // namespace zeek::detail

22
src/DFA.h
View file

@ -3,27 +3,26 @@
#pragma once
#include "RE.h" // for typedef AcceptingSet
#include "Obj.h"
#include <map>
#include <string>
#include <assert.h>
#include <sys/types.h> // for u_char
#include "NFA.h"
#include "RE.h" // for typedef AcceptingSet
#include "Obj.h"
namespace zeek::detail {
class DFA_State;
class DFA_Machine;
// Transitions to the uncomputed state indicate that we haven't yet
// computed the state to go to.
#define DFA_UNCOMPUTED_STATE -2
#define DFA_UNCOMPUTED_STATE_PTR ((DFA_State*) DFA_UNCOMPUTED_STATE)
#include "NFA.h"
class DFA_Machine;
class DFA_State;
class DFA_State : public zeek::Obj {
public:
DFA_State(int state_num, const EquivClass* ec,
@ -152,3 +151,10 @@ inline DFA_State* DFA_State::Xtion(int sym, DFA_Machine* machine)
else
return xtions[sym];
}
} // namespace zeek::detail
using DFA_State [[deprecated("Remove in v4.1. Use zeek::detail::DFA_State.")]] = zeek::detail::DFA_State;
using DigestStr [[deprecated("Remove in v4.1. Use zeek::detail::DigestStr.")]] = zeek::detail::DigestStr;
using DFA_State_Cache [[deprecated("Remove in v4.1. Use zeek::detail::DFA_State_Cache.")]] = zeek::detail::DFA_State_Cache;
using DFA_Machine [[deprecated("Remove in v4.1. Use zeek::detail::DFA_Machine.")]] = zeek::detail::DFA_Machine;

4
src/EquivClass.cc
View file

@ -5,6 +5,8 @@
#include "EquivClass.h"
#include "CCL.h"
namespace zeek::detail {
EquivClass::EquivClass(int arg_size)
{
size = arg_size;
@ -188,3 +190,5 @@ int EquivClass::Size() const
{
return padded_sizeof(*this) + pad_size(sizeof(int) * size * (ccl_flags ? 5 : 4));
}
} // namespace zeek::detail

6
src/EquivClass.h
View file

@ -4,7 +4,9 @@
#include <stdio.h>
class CCL;
ZEEK_FORWARD_DECLARE_NAMESPACED(CCL, zeek::detail);
namespace zeek::detail {
class EquivClass {
public:
@ -41,3 +43,5 @@ protected:
int* ccl_flags;
int ec_nil, no_class, no_rep;
};
} // namespace zeek::detail

3
src/EventRegistry.cc
View file

@ -41,7 +41,7 @@ EventHandler* EventRegistry::Lookup(std::string_view name)
return nullptr;
}
EventRegistry::string_list EventRegistry::Match(RE_Matcher* pattern)
EventRegistry::string_list EventRegistry::Match(zeek::RE_Matcher* pattern)
{
string_list names;
@ -120,4 +120,3 @@ void EventRegistry::SetErrorHandler(std::string_view name)
reporter->InternalWarning("unknown event handler '%s' in SetErrorHandler()",
std::string(name).c_str());
}

6
src/EventRegistry.h
View file

@ -2,6 +2,8 @@
#pragma once
#include "zeek-config.h"
#include <map>
#include <memory>
#include <string>
@ -10,7 +12,7 @@
class EventHandler;
class EventHandlerPtr;
class RE_Matcher;
ZEEK_FORWARD_DECLARE_NAMESPACED(RE_Matcher, zeek);
// The registry keeps track of all events that we provide or handle.
class EventRegistry {
@ -34,7 +36,7 @@ public:
// Returns a list of all local handlers that match the given pattern.
// Passes ownership of list.
using string_list = std::vector<std::string>;
string_list Match(RE_Matcher* pattern);
string_list Match(zeek::RE_Matcher* pattern);
// Marks a handler as handling errors. Error handler will not be called
// recursively to avoid infinite loops in case they trigger an error

4
src/NFA.cc
View file

@ -9,6 +9,8 @@
#include <algorithm>
namespace zeek::detail {
static int nfa_state_id = 0;
NFA_State::NFA_State(int arg_sym, EquivClass* ec)
@ -369,3 +371,5 @@ bool NFA_state_cmp_neg(const NFA_State* v1, const NFA_State* v2)
else
return false;
}
} // namespace zeek::detail

24
src/NFA.h
View file

@ -5,14 +5,10 @@
#include "Obj.h"
#include "List.h"
class CCL;
class NFA_State;
class EquivClass;
ZEEK_FORWARD_DECLARE_NAMESPACED(CCL, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(EquivClass, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(Func, zeek);
using NFA_state_list = zeek::PList<NFA_State>;
#define NO_ACCEPT 0
#define NO_UPPER_BOUND -1
@ -24,6 +20,10 @@ using NFA_state_list = zeek::PList<NFA_State>;
#define SYM_EPSILON 259
#define SYM_CCL 260
namespace zeek::detail {
class NFA_State;
using NFA_state_list = zeek::PList<NFA_State>;
class NFA_State : public zeek::Obj {
public:
@ -135,3 +135,15 @@ extern NFA_state_list* epsilon_closure(NFA_state_list* states);
// For sorting NFA states based on their ID fields (decreasing)
extern bool NFA_state_cmp_neg(const NFA_State* v1, const NFA_State* v2);
} // namespace zeek::detail
using NFA_state_list [[deprecated("Remove in v4.1. Use zeek::detail::NFA_state_list.")]] = zeek::detail::NFA_state_list;
using NFA_State [[deprecated("Remove in v4.1. Use zeek::detail::NFA_State.")]] = zeek::detail::NFA_State;
using EpsilonState [[deprecated("Remove in v4.1. Use zeek::detail::EpsilonState.")]] = zeek::detail::EpsilonState;
using NFA_Machine [[deprecated("Remove in v4.1. Use zeek::detail::NFA_Machine.")]] = zeek::detail::NFA_Machine;
// TODO: could at least the first two of these methods be made static methods in NFA_Machine and NFA_State to remove them from the namespace?
constexpr auto make_alternate [[deprecated("Remove in v4.1. Use zeek::detail::make_alternate.")]] = zeek::detail::make_alternate;
constexpr auto epsilon_closure [[deprecated("Remove in v4.1. Use zeek::detail::epsilon_closure.")]] = zeek::detail::epsilon_closure;
constexpr auto NFA_state_cmp_neg [[deprecated("Remove in v4.1. Use zeek::detail::NFA_state_cmp_neg.")]] = zeek::detail::NFA_state_cmp_neg;

110
src/RE.cc
View file

@ -12,16 +12,22 @@
#include "Reporter.h"
#include "ZeekString.h"
CCL* curr_ccl = nullptr;
Specific_RE_Matcher* rem;
NFA_Machine* nfa = nullptr;
int case_insensitive = 0;
zeek::detail::CCL* zeek::detail::curr_ccl = nullptr;
zeek::detail::CCL*& curr_ccl = zeek::detail::curr_ccl;
zeek::detail::Specific_RE_Matcher* zeek::detail::rem = nullptr;
zeek::detail::Specific_RE_Matcher*& rem = zeek::detail::rem;
zeek::detail::NFA_Machine* zeek::detail::nfa = nullptr;
zeek::detail::NFA_Machine*& nfa = zeek::detail::nfa;
int zeek::detail::case_insensitive = 0;
int& case_insensitive = zeek::detail::case_insensitive;
extern int RE_parse(void);
extern void RE_set_input(const char* str);
extern void RE_done_with_scan();
namespace zeek {
namespace detail {
Specific_RE_Matcher::Specific_RE_Matcher(match_type arg_mt, int arg_multiline)
: equiv_class(NUM_SYM)
{
@ -440,51 +446,6 @@ unsigned int Specific_RE_Matcher::MemoryAllocation() const
+ accepted->size() * padded_sizeof(AcceptingSet::key_type);
}
RE_Matcher::RE_Matcher()
{
re_anywhere = new Specific_RE_Matcher(MATCH_ANYWHERE);
re_exact = new Specific_RE_Matcher(MATCH_EXACTLY);
}
RE_Matcher::RE_Matcher(const char* pat)
{
re_anywhere = new Specific_RE_Matcher(MATCH_ANYWHERE);
re_exact = new Specific_RE_Matcher(MATCH_EXACTLY);
AddPat(pat);
}
RE_Matcher::RE_Matcher(const char* exact_pat, const char* anywhere_pat)
{
re_anywhere = new Specific_RE_Matcher(MATCH_ANYWHERE);
re_anywhere->SetPat(anywhere_pat);
re_exact = new Specific_RE_Matcher(MATCH_EXACTLY);
re_exact->SetPat(exact_pat);
}
RE_Matcher::~RE_Matcher()
{
delete re_anywhere;
delete re_exact;
}
void RE_Matcher::AddPat(const char* new_pat)
{
re_anywhere->AddPat(new_pat);
re_exact->AddPat(new_pat);
}
void RE_Matcher::MakeCaseInsensitive()
{
re_anywhere->MakeCaseInsensitive();
re_exact->MakeCaseInsensitive();
}
bool RE_Matcher::Compile(bool lazy)
{
return re_anywhere->Compile(lazy) && re_exact->Compile(lazy);
}
static RE_Matcher* matcher_merge(const RE_Matcher* re1, const RE_Matcher* re2,
const char* merge_op)
{
@ -513,3 +474,52 @@ RE_Matcher* RE_Matcher_disjunction(const RE_Matcher* re1, const RE_Matcher* re2)
{
return matcher_merge(re1, re2, "|");
}
} // namespace detail
RE_Matcher::RE_Matcher()
{
re_anywhere = new detail::Specific_RE_Matcher(zeek::detail::MATCH_ANYWHERE);
re_exact = new detail::Specific_RE_Matcher(zeek::detail::MATCH_EXACTLY);
}
RE_Matcher::RE_Matcher(const char* pat)
{
re_anywhere = new detail::Specific_RE_Matcher(zeek::detail::MATCH_ANYWHERE);
re_exact = new detail::Specific_RE_Matcher(zeek::detail::MATCH_EXACTLY);
AddPat(pat);
}
RE_Matcher::RE_Matcher(const char* exact_pat, const char* anywhere_pat)
{
re_anywhere = new detail::Specific_RE_Matcher(zeek::detail::MATCH_ANYWHERE);
re_anywhere->SetPat(anywhere_pat);
re_exact = new detail::Specific_RE_Matcher(zeek::detail::MATCH_EXACTLY);
re_exact->SetPat(exact_pat);
}
RE_Matcher::~RE_Matcher()
{
delete re_anywhere;
delete re_exact;
}
void RE_Matcher::AddPat(const char* new_pat)
{
re_anywhere->AddPat(new_pat);
re_exact->AddPat(new_pat);
}
void RE_Matcher::MakeCaseInsensitive()
{
re_anywhere->MakeCaseInsensitive();
re_exact->MakeCaseInsensitive();
}
bool RE_Matcher::Compile(bool lazy)
{
return re_anywhere->Compile(lazy) && re_exact->Compile(lazy);
}
} // namespace zeek

75
src/RE.h
View file

@ -17,30 +17,35 @@ typedef int (*cce_func)(int);
namespace zeek { class String; }
using BroString [[deprecated("Remove in v4.1. Use zeek::String instead.")]] = zeek::String;
class CCL;
class NFA_Machine;
class DFA_Machine;
class Specific_RE_Matcher;
class RE_Matcher;
class DFA_State;
ZEEK_FORWARD_DECLARE_NAMESPACED(NFA_Machine, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(DFA_Machine, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(DFA_State, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(Specific_RE_Matcher, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(RE_Matcher, zeek);
ZEEK_FORWARD_DECLARE_NAMESPACED(CCL, zeek::detail);
// This method is automatically generated by flex and shouldn't be namespaced
extern int re_lex(void);
namespace zeek {
namespace detail {
extern int case_insensitive;
extern CCL* curr_ccl;
extern NFA_Machine* nfa;
extern zeek::detail::NFA_Machine* nfa;
extern Specific_RE_Matcher* rem;
extern const char* RE_parse_input;
extern int re_lex(void);
extern int clower(int);
extern void synerr(const char str[]);
typedef int AcceptIdx;
typedef std::set<AcceptIdx> AcceptingSet;
typedef uint64_t MatchPos;
typedef std::map<AcceptIdx, MatchPos> AcceptingMatchSet;
typedef zeek::name_list string_list;
using AcceptIdx = int;
using AcceptingSet = std::set<AcceptIdx>;
using MatchPos = uint64_t;
using AcceptingMatchSet = std::map<AcceptIdx, MatchPos>;
using string_list = zeek::name_list;
typedef enum { MATCH_ANYWHERE, MATCH_EXACTLY, } match_type;
enum match_type { MATCH_ANYWHERE, MATCH_EXACTLY };
// A "specific" RE matcher will match one type of pattern: either
// MATCH_ANYWHERE or MATCH_EXACTLY.
@ -106,7 +111,7 @@ public:
const char* PatternText() const { return pattern_text; }
DFA_Machine* DFA() const { return dfa; }
zeek::detail::DFA_Machine* DFA() const { return dfa; }
void Dump(FILE* f);
@ -133,7 +138,7 @@ protected:
zeek::PList<CCL> ccl_list;
EquivClass equiv_class;
int* ecs;
DFA_Machine* dfa;
zeek::detail::DFA_Machine* dfa;
CCL* any_ccl;
AcceptingSet* accepted;
};
@ -168,14 +173,19 @@ public:
void AddMatches(const AcceptingSet& as, MatchPos position);
protected:
DFA_Machine* dfa;
zeek::detail::DFA_Machine* dfa;
int* ecs;
AcceptingMatchSet accepted_matches;
DFA_State* current_state;
zeek::detail::DFA_State* current_state;
int current_pos;
};
extern RE_Matcher* RE_Matcher_conjunction(const RE_Matcher* re1, const RE_Matcher* re2);
extern RE_Matcher* RE_Matcher_disjunction(const RE_Matcher* re1, const RE_Matcher* re2);
} // namespace detail
class RE_Matcher final {
public:
RE_Matcher();
@ -225,9 +235,30 @@ public:
}
protected:
Specific_RE_Matcher* re_anywhere;
Specific_RE_Matcher* re_exact;
detail::Specific_RE_Matcher* re_anywhere;
detail::Specific_RE_Matcher* re_exact;
};
extern RE_Matcher* RE_Matcher_conjunction(const RE_Matcher* re1, const RE_Matcher* re2);
extern RE_Matcher* RE_Matcher_disjunction(const RE_Matcher* re1, const RE_Matcher* re2);
} // namespace zeek
constexpr auto clower [[deprecated("Remove in v4.1. Use zeek::detail::clower.")]] = zeek::detail::clower;
constexpr auto synerr [[deprecated("Remove in v4.1. Use zeek::detail::synerr.")]] = zeek::detail::synerr;
using AcceptIdx [[deprecated("Remove in v4.1. Use zeek::detail::AcceptIdx.")]] = zeek::detail::AcceptIdx;
using AcceptingSet [[deprecated("Remove in v4.1. Use zeek::detail::AcceptingSet.")]] = zeek::detail::AcceptingSet;
using MatchPos [[deprecated("Remove in v4.1. Use zeek::detail::MatchPos.")]] = zeek::detail::MatchPos;
using AcceptingMatchSet [[deprecated("Remove in v4.1. Use zeek::detail::AcceptingMatchSet.")]] = zeek::detail::AcceptingMatchSet;
using string_list [[deprecated("Remove in v4.1. Use zeek::detail::string_list.")]] = zeek::detail::string_list;
constexpr auto MATCH_ANYWHERE [[deprecated("Remove in v4.1. Use zeek::detail::MATCH_ANYWHERE.")]] = zeek::detail::MATCH_ANYWHERE;
constexpr auto MATCH_EXACTLY [[deprecated("Remove in v4.1. Use zeek::detail::MATCH_EXACTLY.")]] = zeek::detail::MATCH_EXACTLY;
using Specific_RE_Matcher [[deprecated("Remove in v4.1. Use zeek::detail::Specific_RE_Matcher.")]] = zeek::detail::Specific_RE_Matcher;
using RE_Match_State [[deprecated("Remove in v4.1. Use zeek::detail::RE_Match_State.")]] = zeek::detail::RE_Match_State;
using RE_Matcher [[deprecated("Remove in v4.1. Use zeek::RE_Matcher.")]] = zeek::RE_Matcher;
extern int& case_insensitive [[deprecated("Remove in v4.1. Use zeek::detail::case_insensitive")]];
extern zeek::detail::CCL*& curr_ccl [[deprecated("Remove in v4.1. Use zeek::detail::curr_ccl")]];;
extern zeek::detail::NFA_Machine*& nfa [[deprecated("Remove in v4.1. Use zeek::detail::nfa")]];;
extern zeek::detail::Specific_RE_Matcher*& rem [[deprecated("Remove in v4.1. Use zeek::detail::rem")]];;
extern const char*& RE_parse_input [[deprecated("Remove in v4.1. Use zeek::detail::RE_parse_input")]];;

4
src/RuleMatcher.h
View file

@ -29,9 +29,9 @@ extern const char* current_rule_file;
class BroFile;
class IntSet;
class RE_Match_State;
class Specific_RE_Matcher;
ZEEK_FORWARD_DECLARE_NAMESPACED(RE_Match_State, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(Specific_RE_Matcher, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(RuleMatcher, zeek::detail);
ZEEK_FORWARD_DECLARE_NAMESPACED(IP_Hdr, zeek);
ZEEK_FORWARD_DECLARE_NAMESPACED(IPPrefix, zeek);

2
src/Val.h
View file

@ -44,7 +44,7 @@ using BroFunc [[deprecated("Remove in v4.1. Use zeek::detail::ScriptFunc instead
class BroFile;
class PrefixTable;
class StateAccess;
class RE_Matcher;
ZEEK_FORWARD_DECLARE_NAMESPACED(RE_Matcher, zeek);
class CompositeHash;
class HashKey;

20
src/analyzer/protocol/login/Login.cc
View file

@ -17,15 +17,15 @@
using namespace analyzer::login;
static RE_Matcher* re_skip_authentication = nullptr;
static RE_Matcher* re_direct_login_prompts;
static RE_Matcher* re_login_prompts;
static RE_Matcher* re_login_non_failure_msgs;
static RE_Matcher* re_login_failure_msgs;
static RE_Matcher* re_login_success_msgs;
static RE_Matcher* re_login_timeouts;
static zeek::RE_Matcher* re_skip_authentication = nullptr;
static zeek::RE_Matcher* re_direct_login_prompts;
static zeek::RE_Matcher* re_login_prompts;
static zeek::RE_Matcher* re_login_non_failure_msgs;
static zeek::RE_Matcher* re_login_failure_msgs;
static zeek::RE_Matcher* re_login_success_msgs;
static zeek::RE_Matcher* re_login_timeouts;
static RE_Matcher* init_RE(zeek::ListVal* l);
static zeek::RE_Matcher* init_RE(zeek::ListVal* l);
Login_Analyzer::Login_Analyzer(const char* name, Connection* conn)
: tcp::TCP_ApplicationAnalyzer(name, conn), user_text()
@ -625,9 +625,9 @@ void Login_Analyzer::FlushEmptyTypeahead()
delete [] PopUserText();
}
RE_Matcher* init_RE(zeek::ListVal* l)
zeek::RE_Matcher* init_RE(zeek::ListVal* l)
{
RE_Matcher* re = l->BuildRE();
zeek::RE_Matcher* re = l->BuildRE();
if ( re )
re->Compile();

6
src/broker/Data.cc
View file

@ -421,7 +421,7 @@ struct val_converter {
if ( ! exact_text || ! anywhere_text )
return nullptr;
RE_Matcher* re = new RE_Matcher(exact_text->c_str(),
auto* re = new zeek::RE_Matcher(exact_text->c_str(),
anywhere_text->c_str());
if ( ! re->Compile() )
@ -745,7 +745,7 @@ struct type_checker {
if ( ! exact_text || ! anywhere_text )
return false;
RE_Matcher* re = new RE_Matcher(exact_text->c_str(),
auto* re = new zeek::RE_Matcher(exact_text->c_str(),
anywhere_text->c_str());
auto compiled = re->Compile();
delete re;
@ -986,7 +986,7 @@ broker::expected<broker::data> bro_broker::val_to_data(const zeek::Val* v)
}
case zeek::TYPE_PATTERN:
{
const RE_Matcher* p = v->AsPattern();
const zeek::RE_Matcher* p = v->AsPattern();
broker::vector rval = {p->PatternText(), p->AnywherePatternText()};
return {std::move(rval)};
}

2
src/input/Manager.cc
View file

@ -2269,7 +2269,7 @@ zeek::Val* Manager::ValueToVal(const Stream* i, const Value* val, zeek::Type* re
case zeek::TYPE_PATTERN:
{
RE_Matcher* re = new RE_Matcher(val->val.pattern_text_val);
auto* re = new zeek::RE_Matcher(val->val.pattern_text_val);
re->Compile();
return new zeek::PatternVal(re);
}

4
src/parse.y
View file

@ -236,7 +236,7 @@ static bool expr_is_table_type_name(const zeek::detail::Expr* expr)
id_list* id_l;
zeek::detail::InitClass ic;
zeek::Val* val;
RE_Matcher* re;
zeek::RE_Matcher* re;
zeek::detail::Expr* expr;
zeek::detail::EventExpr* event_expr;
zeek::detail::Stmt* stmt;
@ -715,7 +715,7 @@ expr:
{
zeek::detail::set_location(@3);
RE_Matcher* re = new RE_Matcher($3);
auto* re = new zeek::RE_Matcher($3);
delete [] $3;
if ( $4 )

71
src/re-parse.y
View file

@ -12,8 +12,11 @@
int csize = 256;
int syntax_error = 0;
int cupper(int sym);
int clower(int sym);
namespace zeek::detail {
int cupper(int sym);
int clower(int sym);
}
void yyerror(const char msg[]);
%}
@ -22,8 +25,8 @@ void yyerror(const char msg[]);
%union {
int int_val;
cce_func cce_val;
CCL* ccl_val;
NFA_Machine* mach_val;
zeek::detail::CCL* ccl_val;
zeek::detail::NFA_Machine* mach_val;
}
%type <int_val> TOK_CHAR TOK_NUMBER
@ -33,17 +36,17 @@ void yyerror(const char msg[]);
%%
flexrule : re
{ $1->AddAccept(1); nfa = $1; }
{ $1->AddAccept(1); zeek::detail::nfa = $1; }
| error
{ return 1; }
;
re : re '|' series
{ $$ = make_alternate($1, $3); }
{ $$ = zeek::detail::make_alternate($1, $3); }
| series
|
{ $$ = new NFA_Machine(new EpsilonState()); }
|
{ $$ = new zeek::detail::NFA_Machine(new zeek::detail::EpsilonState()); }
;
series : series singleton
@ -63,14 +66,14 @@ singleton : singleton '*'
| singleton '{' TOK_NUMBER ',' TOK_NUMBER '}'
{
if ( $3 > $5 || $3 < 0 )
synerr("bad iteration values");
zeek::detail::synerr("bad iteration values");
else
{
if ( $3 == 0 )
{
if ( $5 == 0 )
{
$$ = new NFA_Machine(new EpsilonState());
$$ = new zeek::detail::NFA_Machine(new zeek::detail::EpsilonState());
Unref($1);
}
else
@ -87,7 +90,7 @@ singleton : singleton '*'
| singleton '{' TOK_NUMBER ',' '}'
{
if ( $3 < 0 )
synerr("iteration value must be positive");
zeek::detail::synerr("iteration value must be positive");
else if ( $3 == 0 )
$1->MakeClosure();
else
@ -97,11 +100,11 @@ singleton : singleton '*'
| singleton '{' TOK_NUMBER '}'
{
if ( $3 < 0 )
synerr("iteration value must be positive");
zeek::detail::synerr("iteration value must be positive");
else if ( $3 == 0 )
{
Unref($1);
$$ = new NFA_Machine(new EpsilonState());
$$ = new zeek::detail::NFA_Machine(new zeek::detail::EpsilonState());
}
else
$1->LinkCopies($3-1);
@ -109,18 +112,18 @@ singleton : singleton '*'
| '.'
{
$$ = new NFA_Machine(new NFA_State(rem->AnyCCL()));
$$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State(zeek::detail::rem->AnyCCL()));
}
| full_ccl
{
$1->Sort();
rem->EC()->CCL_Use($1);
$$ = new NFA_Machine(new NFA_State($1));
zeek::detail::rem->EC()->CCL_Use($1);
$$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State($1));
}
| TOK_CCL
{ $$ = new NFA_Machine(new NFA_State($1)); }
{ $$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State($1)); }
| '"' string '"'
{ $$ = $2; }
@ -129,7 +132,7 @@ singleton : singleton '*'
{ $$ = $2; }
| TOK_CASE_INSENSITIVE re ')'
{ $$ = $2; case_insensitive = 0; }
{ $$ = $2; zeek::detail::case_insensitive = 0; }
| TOK_CHAR
{
@ -138,22 +141,22 @@ singleton : singleton '*'
if ( sym < 0 || ( sym >= NUM_SYM && sym != SYM_EPSILON ) )
{
reporter->Error("bad symbol %d (compiling pattern /%s/)", sym,
RE_parse_input);
zeek::detail::RE_parse_input);
return 1;
}
$$ = new NFA_Machine(new NFA_State(sym, rem->EC()));
$$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State(sym, zeek::detail::rem->EC()));
}
| '^'
{
$$ = new NFA_Machine(new NFA_State(SYM_BOL, rem->EC()));
$$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State(SYM_BOL, zeek::detail::rem->EC()));
$$->MarkBOL();
}
| '$'
{
$$ = new NFA_Machine(new NFA_State(SYM_EOL, rem->EC()));
$$ = new zeek::detail::NFA_Machine(new zeek::detail::NFA_State(SYM_EOL, zeek::detail::rem->EC()));
$$->MarkEOL();
}
;
@ -171,9 +174,9 @@ full_ccl : '[' ccl ']'
ccl : ccl TOK_CHAR '-' TOK_CHAR
{
if ( $2 > $4 )
synerr("negative range in character class");
zeek::detail::synerr("negative range in character class");
else if ( case_insensitive &&
else if ( zeek::detail::case_insensitive &&
(isalpha($2) || isalpha($4)) )
{
if ( isalpha($2) && isalpha($4) &&
@ -190,7 +193,7 @@ ccl : ccl TOK_CHAR '-' TOK_CHAR
}
else
synerr("ambiguous case-insensitive character class");
zeek::detail::synerr("ambiguous case-insensitive character class");
}
else
@ -202,10 +205,10 @@ ccl : ccl TOK_CHAR '-' TOK_CHAR
| ccl TOK_CHAR
{
if ( case_insensitive && isalpha($2) )
if ( zeek::detail::case_insensitive && isalpha($2) )
{
$1->Add(clower($2));
$1->Add(cupper($2));
$1->Add(zeek::detail::clower($2));
$1->Add(zeek::detail::cupper($2));
}
else
$1->Add($2);
@ -214,14 +217,14 @@ ccl : ccl TOK_CHAR '-' TOK_CHAR
| ccl ccl_expr
|
{ $$ = curr_ccl; }
{ $$ = zeek::detail::curr_ccl; }
;
ccl_expr: TOK_CCE
{
for ( int c = 0; c < csize; ++c )
if ( isascii(c) && $1(c) )
curr_ccl->Add(c);
zeek::detail::curr_ccl->Add(c);
}
;
@ -231,14 +234,16 @@ string : string TOK_CHAR
// leave this alone; that provides a way
// of "escaping" out of insensitivity
// if needed.
$1->AppendState(new NFA_State($2, rem->EC()));
$1->AppendState(new zeek::detail::NFA_State($2, zeek::detail::rem->EC()));
}
|
{ $$ = new NFA_Machine(new EpsilonState()); }
{ $$ = new zeek::detail::NFA_Machine(new zeek::detail::EpsilonState()); }
;
%%
namespace zeek::detail {
int cupper(int sym)
{
return (isascii(sym) && islower(sym)) ? toupper(sym) : sym;
@ -255,6 +260,8 @@ void synerr(const char str[])
reporter->Error("%s (compiling pattern /%s/)", str, RE_parse_input);
}
} // namespace zeek::detail
void yyerror(const char msg[])
{
}

39
src/re-scan.l
View file

@ -14,7 +14,8 @@
#include "re-parse.h"
const char* RE_parse_input = 0;
const char* zeek::detail::RE_parse_input = nullptr;
const char*& RE_parse_input = zeek::detail::RE_parse_input;
#define RET_CCE(func) \
BEGIN(SC_CCL); \
@ -60,18 +61,18 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
"$" return '$';
"["({FIRST_CCL_CHAR}|{CCL_EXPR})({CCL_CHAR}|{CCL_EXPR})* {
curr_ccl = rem->LookupCCL(yytext);
if ( curr_ccl )
zeek::detail::curr_ccl = zeek::detail::rem->LookupCCL(yytext);
if ( zeek::detail::curr_ccl )
{
if ( yyinput() != ']' )
synerr("bad character class");
yylval.ccl_val = curr_ccl;
zeek::detail::synerr("bad character class");
yylval.ccl_val = zeek::detail::curr_ccl;
return TOK_CCL;
}
else
{
curr_ccl = new CCL();
rem->InsertCCL(yytext, curr_ccl);
zeek::detail::curr_ccl = new zeek::detail::CCL();
zeek::detail::rem->InsertCCL(yytext, zeek::detail::curr_ccl);
// Push back everything but the leading bracket
// so the ccl can be rescanned.
@ -86,11 +87,11 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
char* nmstr = copy_string(yytext+1);
nmstr[yyleng - 2] = '\0'; // chop trailing brace
std::string namedef = rem->LookupDef(nmstr);
std::string namedef = zeek::detail::rem->LookupDef(nmstr);
delete nmstr;
if ( namedef.empty() )
synerr("undefined definition");
zeek::detail::synerr("undefined definition");
else
{ // push back name surrounded by ()'s
int len = namedef.size();
@ -115,10 +116,10 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
}
}
"(?i:" case_insensitive = 1; return TOK_CASE_INSENSITIVE;
"(?i:" zeek::detail::case_insensitive = 1; return TOK_CASE_INSENSITIVE;
[a-zA-Z] {
if ( case_insensitive )
if ( zeek::detail::case_insensitive )
{
char c = yytext[0]; // unput trashes yytext!
// Push back the character inside a CCL,
@ -140,7 +141,7 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
}
<SC_QUOTE>{
[^"\n]$ synerr("missing quote"); return '"';
[^"\n]$ zeek::detail::synerr("missing quote"); return '"';
[^"\n] yylval.int_val = yytext[0]; return TOK_CHAR;
\" BEGIN(INITIAL); return '"';
}
@ -156,7 +157,7 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
[^\]\n] yylval.int_val = yytext[0]; return TOK_CHAR;
"]" BEGIN(INITIAL); return ']';
[^\]]$ {
synerr("bad character class");
zeek::detail::synerr("bad character class");
BEGIN(INITIAL);
return ']';
}
@ -177,19 +178,19 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
"[:lower:]" {
BEGIN(SC_CCL);
yylval.cce_val =
case_insensitive ? my_isalpha : my_islower;
zeek::detail::case_insensitive ? my_isalpha : my_islower;
return TOK_CCE;
}
"[:upper:]" {
BEGIN(SC_CCL);
yylval.cce_val =
case_insensitive ? my_isalpha : my_isupper;
zeek::detail::case_insensitive ? my_isalpha : my_isupper;
return TOK_CCE;
}
{CCL_EXPR} {
synerr("bad character class expression");
zeek::detail::synerr("bad character class expression");
BEGIN(SC_CCL);
yylval.cce_val = my_isalnum;
return TOK_CCE;
@ -203,7 +204,7 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
"}" BEGIN(INITIAL); return '}';
. {
synerr("bad character inside {}'s");
zeek::detail::synerr("bad character inside {}'s");
BEGIN(INITIAL);
return '}';
}
@ -219,7 +220,7 @@ CCL_EXPR ("[:"[[:alpha:]]+":]")
return TOK_CHAR;
}
<*>.|\n synerr("bad character");
<*>.|\n zeek::detail::synerr("bad character");
%%
@ -227,7 +228,7 @@ YY_BUFFER_STATE RE_buf;
void RE_set_input(const char* str)
{
RE_parse_input = str;
zeek::detail::RE_parse_input = str;
RE_buf = yy_scan_string(str);
}

2
src/scan.l
View file

@ -48,7 +48,7 @@ extern zeek::EnumType* cur_enum_type;
// Track the @if... depth.
ptr_compat_int current_depth = 0;
int_list if_stack;
zeek::detail::int_list if_stack;
int line_number = 1;
const char* filename = 0; // Absolute path of file currently being parsed.

4
src/strings.bif
View file

@ -200,7 +200,7 @@ static int match_prefix(int s_len, const char* s, int t_len, const char* t)
}
static zeek::VectorValPtr do_split_string(zeek::StringVal* str_val,
RE_Matcher* re, int incl_sep,
zeek::RE_Matcher* re, int incl_sep,
int max_num_sep)
{
// string_vec is used early in the version script - do not use the NetVar.
@ -257,7 +257,7 @@ static zeek::VectorValPtr do_split_string(zeek::StringVal* str_val,
return rval;
}
zeek::Val* do_split(zeek::StringVal* str_val, RE_Matcher* re, int incl_sep, int max_num_sep)
zeek::Val* do_split(zeek::StringVal* str_val, zeek::RE_Matcher* re, int incl_sep, int max_num_sep)
{
auto* a = new zeek::TableVal(zeek::id::string_array);
const u_char* s = str_val->Bytes();

2
src/threading/SerialTypes.cc
View file

@ -517,7 +517,7 @@ zeek::Val* Value::ValueToVal(const std::string& source, const Value* val, bool&
case zeek::TYPE_PATTERN:
{
RE_Matcher* re = new RE_Matcher(val->val.pattern_text_val);
auto* re = new zeek::RE_Matcher(val->val.pattern_text_val);
re->Compile();
return new zeek::PatternVal(re);
}