Initial paraglob integration.

.gitmodules

@@ -28,3 +28,6 @@
 [submodule "doc"]
 	path = doc
 	url = https://github.com/zeek/zeek-docs
+[submodule "aux/paraglob"]
+	path = aux/paraglob
+	url = https://github.com/zeek/paraglob

CMakeLists.txt

@@ -325,6 +325,10 @@ include_directories(BEFORE ${CAF_INCLUDE_DIR_CORE})
 include_directories(BEFORE ${CAF_INCLUDE_DIR_IO})
 include_directories(BEFORE ${CAF_INCLUDE_DIR_OPENSSL})
 
+add_subdirectory(aux/paraglob)
+set(zeekdeps ${zeekdeps} paraglob)
+include_directories(BEFORE ${CMAKE_CURRENT_SOURCE_DIR}/aux/paraglob)
+
 add_subdirectory(src)
 add_subdirectory(scripts)
 add_subdirectory(man)

aux/paraglob

@@ -0,0 +1 @@
+Subproject commit 757e00b6510d2b0e92510c9c26f9e3279aa442a4

src/OpaqueVal.cc

@@ -869,3 +869,28 @@ void CardinalityVal::Add(const Val* val)
 	c->AddElement(key->Hash());
 	delete key;
 	}
+
+
+ParaglobVal::ParaglobVal(paraglob::Paraglob* p)
+: OpaqueVal(paraglob_type)
+	{
+	this->internal_paraglob = p;
+	}
+
+VectorVal* ParaglobVal::get(StringVal* &pattern)
+	{
+	VectorVal* rval = new VectorVal(internal_type("string_vec")->AsVectorType());
+	std::string string_pattern (pattern->CheckString(), pattern->Len());
+	std::vector<std::string> matches = this->internal_paraglob->get(string_pattern);
+
+	for (unsigned int i = 0; i < matches.size(); i++) {
+		rval->Assign(i, new StringVal(matches.at(i).c_str()));
+	}
+
+	return rval;
+	}
+
+bool ParaglobVal::operator==(const ParaglobVal *other)
+	{
+	return (*(this->internal_paraglob) == *(other->internal_paraglob));
+	}

src/OpaqueVal.h

@@ -8,6 +8,7 @@
 #include "RandTest.h"
 #include "Val.h"
 #include "digest.h"
+#include "src/paraglob.h"
 
 namespace probabilistic {
 	class BloomFilter;
@@ -188,4 +189,14 @@ private:
 	DECLARE_SERIAL(CardinalityVal);
 };
 
+class ParaglobVal : public OpaqueVal {
+public:
+	explicit ParaglobVal(paraglob::Paraglob* p);
+	VectorVal* get(StringVal* &pattern);
+	bool operator==(const ParaglobVal *other);
+
+private:
+	paraglob::Paraglob* internal_paraglob;
+};
+
 #endif

src/Type.h

@@ -639,6 +639,7 @@ extern OpaqueType* topk_type;
 extern OpaqueType* bloomfilter_type;
 extern OpaqueType* x509_opaque_type;
 extern OpaqueType* ocsp_resp_opaque_type;
+extern OpaqueType* paraglob_type;
 
 // Returns the Bro basic (non-parameterized) type with the given type.
 // The reference count of the type is not increased.

src/bro.bif

@@ -789,6 +789,63 @@ function sha256_hash_finish%(handle: opaque of sha256%): string
 	return static_cast<HashVal*>(handle)->Get();
 	%}
 
+
+## Initializes and returns a new paraglob.
+##
+## v: Vector of patterns to initialize the paraglob with.
+##
+## Returns: A new, compiled, paraglob with the patterns in *v*
+##
+## .. zeek:see::paraglob_get paraglob_equals paraglob_add
+function paraglob_init%(v: any%) : opaque of paraglob
+	%{
+	if ( v->Type()->Tag() != TYPE_VECTOR ||
+	     v->Type()->YieldType()->Tag() != TYPE_STRING )
+		{
+		builtin_error("paraglob requires a vector for initialization.");
+		return nullptr;
+		}
+
+	std::vector<std::string> patterns;
+	VectorVal* vv = v->AsVectorVal();
+	for ( unsigned int i = 0; i < vv->Size(); ++i )
+		{
+		const BroString* s = vv->Lookup(i)->AsString();
+		patterns.push_back(std::string(s->CheckString(), s->Len()));
+		}
+
+	return new ParaglobVal(new paraglob::Paraglob(patterns));
+	%}
+
+## Gets all the strings inside the handle associated with an input pattern.
+##
+## handle: A compiled paraglob.
+## pattern: A glob style pattern.
+##
+## Returns: A vector of strings matching the input pattern
+##
+## ## .. zeek:see::paraglob_add paraglob_equals paraglob_init
+function paraglob_get%(handle: opaque of paraglob, pat: string%): string_vec
+	%{
+	return static_cast<ParaglobVal*>(handle)->get(pat);
+	%}
+
+## Compares two paraglobs for equality.
+##
+## p_one: A compiled paraglob.
+## p_two: A compiled paraglob.
+##
+## Returns: True of both paraglobs contain the same patterns, false otherwise.
+##
+## ## .. zeek:see::paraglob_add paraglob_get paraglob_init
+function paraglob_equals%(p_one: opaque of paraglob, p_two: opaque of paraglob%)
+	: bool
+	%{
+	bool eq =
+		(static_cast<ParaglobVal*>(p_one) == static_cast<ParaglobVal*>(p_two));
+	return val_mgr->GetBool(eq);
+	%}
+
 ## Returns 32-bit digest of arbitrary input values using FNV-1a hash algorithm.
 ## See `<https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function>`_.
 ##

src/input/Manager.cc

@@ -224,7 +224,7 @@ ReaderBackend* Manager::CreateBackend(ReaderFrontend* frontend, EnumVal* tag)
 	return backend;
 	}
 
-// Create a new input reader object to be used at whomevers leisure lateron.
+// Create a new input reader object to be used at whomevers leisure later on.
 bool Manager::CreateStream(Stream* info, RecordVal* description)
 	{
 	RecordType* rtype = description->Type()->AsRecordType();
@@ -232,7 +232,7 @@ bool Manager::CreateStream(Stream* info, RecordVal* description)
 		|| same_type(rtype, BifType::Record::Input::EventDescription, 0)
 		|| same_type(rtype, BifType::Record::Input::AnalysisDescription, 0) ) )
 		{
-		reporter->Error("Streamdescription argument not of right type for new input stream");
+		reporter->Error("Stream description argument not of right type for new input stream");
 		return false;
 		}
 
@@ -824,6 +824,7 @@ bool Manager::IsCompatibleType(BroType* t, bool atomic_only)
 	case TYPE_INTERVAL:
 	case TYPE_ENUM:
 	case TYPE_STRING:
+	case TYPE_PATTERN:
 		return true;
 
 	case TYPE_RECORD:
@@ -2074,6 +2075,12 @@ int Manager::GetValueLength(const Value* val) const
 		}
 		break;
 
+	case TYPE_PATTERN:
+		{
+		length += strlen(val->val.pattern_text_val) + 1;
+		break;
+		}
+
 	case TYPE_TABLE:
 		{
 		for ( int i = 0; i < val->val.set_val.size; i++ )
@@ -2193,6 +2200,14 @@ int Manager::CopyValue(char *data, const int startpos, const Value* val) const
 		return length;
 		}
 
+	case TYPE_PATTERN:
+		{
+		// include null-terminator
+		int length = strlen(val->val.pattern_text_val) + 1;
+		memcpy(data + startpos, val->val.pattern_text_val, length);
+		return length;
+		}
+
 	case TYPE_TABLE:
 		{
 		int length = 0;
@@ -2350,6 +2365,13 @@ Val* Manager::ValueToVal(const Stream* i, const Value* val, BroType* request_typ
 		return subnetval;
 		}
 
+	case TYPE_PATTERN:
+		{
+		RE_Matcher* re = new RE_Matcher(val->val.pattern_text_val);
+		re->Compile();
+		return new PatternVal(re);
+		}
+
 	case TYPE_TABLE:
 		{
 		// all entries have to have the same type...
@@ -2492,6 +2514,13 @@ Val* Manager::ValueToVal(const Stream* i, const Value* val, bool& have_error) co
 		return subnetval;
 		}
 
+	case TYPE_PATTERN:
+		{
+		RE_Matcher* re = new RE_Matcher(val->val.pattern_text_val);
+		re->Compile();
+		return new PatternVal(re);
+		}
+
 	case TYPE_TABLE:
 		{
 		TypeList* set_index;

src/main.cc

@@ -122,6 +122,7 @@ OpaqueType* topk_type = 0;
 OpaqueType* bloomfilter_type = 0;
 OpaqueType* x509_opaque_type = 0;
 OpaqueType* ocsp_resp_opaque_type = 0;
+OpaqueType* paraglob_type = 0;
 
 // Keep copy of command line
 int bro_argc;
@@ -809,6 +810,7 @@ int main(int argc, char** argv)
 	bloomfilter_type = new OpaqueType("bloomfilter");
 	x509_opaque_type = new OpaqueType("x509");
 	ocsp_resp_opaque_type = new OpaqueType("ocsp_resp");
+	paraglob_type = new OpaqueType("paraglob");
 
 	// The leak-checker tends to produce some false
 	// positives (memory which had already been

src/threading/SerialTypes.h

@@ -126,6 +126,7 @@ struct Value {
 		vec_t vector_val;
 		addr_t addr_val;
 		subnet_t subnet_val;
+		const char* pattern_text_val;
 
 		struct {
 			char* data;

src/threading/formatters/Ascii.cc

@@ -325,6 +325,28 @@ threading::Value* Ascii::ParseValue(const string& s, const string& name, TypeTag
 		break;
 		}
 
+	case TYPE_PATTERN:
+		{
+		string cannidate = get_unescaped_string(s);
+		// A string is a cannidate pattern iff it begins and ends with
+		// a '/'. Rather or not the rest of the string is legal will
+		// be determined later when it is given to the RE engine.
+		if ( cannidate.size() >= 2 )
+			{
+				if ( cannidate.front() == cannidate.back() &&
+					cannidate.back() == '/' )
+					{
+					// Remove the '/'s
+					cannidate.erase(0, 1);
+					cannidate.erase(cannidate.size() - 1);
+					val->val.pattern_text_val = copy_string(cannidate.c_str());
+					break;
+					}
+			}
+		GetThread()->Error(GetThread()->Fmt("String '%s' contained no parseable pattern.", cannidate.c_str()));
+		goto parse_error;
+		}
+
 	case TYPE_TABLE:
 	case TYPE_VECTOR:
 		// First - common initialization

testing/btest/Baseline/language.paraglob/out

@@ -0,0 +1,6 @@
+[T, T, T, T, T]
+T
+[*, *og, d?g, d[!wl]g]
+[once]
+[*.gov*, *malware*]
+[*.gov*, *malware*]

testing/btest/Baseline/scripts.base.frameworks.input.bad_patterns/.stderr

@@ -0,0 +1,9 @@
+error: input.log/Input::READER_ASCII: String '/cat/sss' contained no parseable pattern.
+warning: input.log/Input::READER_ASCII: Could not convert line '2	/cat/sss' of input.log to Val. Ignoring line.
+error: input.log/Input::READER_ASCII: String '/foo|bar' contained no parseable pattern.
+warning: input.log/Input::READER_ASCII: Could not convert line '3	/foo|bar' of input.log to Val. Ignoring line.
+error: input.log/Input::READER_ASCII: String 'this is not a pattern' contained no parseable pattern.
+warning: input.log/Input::READER_ASCII: Could not convert line '4	this is not a pattern' of input.log to Val. Ignoring line.
+error: input.log/Input::READER_ASCII: String '/5' contained no parseable pattern.
+warning: input.log/Input::READER_ASCII: Could not convert line '5	/5' of input.log to Val. Ignoring line.
+received termination signal

testing/btest/Baseline/scripts.base.frameworks.input.patterns/out

@@ -0,0 +1,9 @@
+T
+F
+T
+{
+[2] = [p=/^?(cat)$?/],
+[4] = [p=/^?(^oob)$?/],
+[1] = [p=/^?(dog)$?/],
+[3] = [p=/^?(foo|bar)$?/]
+}

testing/btest/language/paraglob.zeek

@@ -0,0 +1,34 @@
+# @TEST-EXEC: bro -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+event zeek_init ()
+{
+  local v1 = vector("*", "d?g", "*og", "d?", "d[!wl]g");
+  local v2 = vector("once", "!o*", "once");
+  local v3 = vector("https://*.google.com/*", "*malware*", "*.gov*");
+
+  local p1 = paraglob_init(v1);
+  local p2: opaque of paraglob = paraglob_init(v2);
+  local p3 = paraglob_init(v3);
+  local p_eq = paraglob_init(v1);
+
+  # paraglob_init should not modify v1
+  print (v1 == vector("*", "d?g", "*og", "d?", "d[!wl]g"));
+  # p_eq and p1 should be the same paraglobs
+  print paraglob_equals(p1, p_eq);
+
+  print paraglob_get(p1, "dog");
+  print paraglob_get(p2, "once");
+  print paraglob_get(p3, "www.strange-malware-domain.gov");
+
+  # This looks like a lot, but really should complete quickly.
+  # Paraglob should stop addition of duplicate patterns.
+  local i = 1000000;
+  while (i > 0) {
+    i = i - 1;
+    v3 += v3[1];
+  }
+
+  local large_glob: opaque of paraglob = paraglob_init(v3);
+  print paraglob_get(large_glob, "www.strange-malware-domain.gov");
+}

testing/btest/scripts/base/frameworks/input/bad_patterns.zeek

@@ -0,0 +1,38 @@
+# @TEST-EXEC: zeek -b %INPUT
+# @TEST-EXEC: btest-diff .stderr
+
+@TEST-START-FILE input.log
+#separator \x09
+#fields	i	p
+#types	count	pattern
+1	/d/og/
+2	/cat/sss
+3	/foo|bar
+4	this is not a pattern
+5	/5
+@TEST-END-FILE
+
+redef exit_only_after_terminate = T;
+
+module A;
+
+type Idx: record {
+	i: int;
+};
+
+type Val: record {
+	p: pattern;
+};
+
+event kill_me()
+	{
+	terminate();
+	}
+
+global pats: table[int] of Val = table();
+
+event zeek_init()
+	{
+	Input::add_table([$source="input.log", $name="pats", $idx=Idx, $val=Val, $destination=pats]);
+	schedule 10msec { kill_me() };
+	}

testing/btest/scripts/base/frameworks/input/patterns.zeek

@@ -0,0 +1,47 @@
+# @TEST-EXEC: btest-bg-run zeek zeek -b %INPUT
+# @TEST-EXEC: btest-bg-wait 10
+
+
+redef exit_only_after_terminate = T;
+
+@TEST-START-FILE input.log
+#separator \x09
+#fields	i	p
+#types	count	pattern
+1	/dog/
+2	/cat/
+3	/foo|bar/
+4	/^oob/
+@TEST-END-FILE
+
+global outfile: file;
+
+module A;
+
+type Idx: record {
+	i: int;
+};
+
+type Val: record {
+	p: pattern;
+};
+
+global pats: table[int] of Val = table();
+
+event zeek_init()
+	{
+	outfile = open("../out");
+	# first read in the old stuff into the table...
+	Input::add_table([$source="../input.log", $name="pats", $idx=Idx, $val=Val, $destination=pats]);
+	}
+
+event Input::end_of_data(name: string, source:string)
+	{
+	print outfile, (pats[3]$p in "foobar"); # T
+	print outfile, (pats[4]$p in "foobar"); # F
+	print outfile, (pats[3]$p == "foo"); # T
+	print outfile, pats;
+	Input::remove("pats");
+	close(outfile);
+	terminate();
+	}