Move ScannedFile class and associated globals into ScannedFile.h and out of Net.h and scan.l

src/CMakeLists.txt

@@ -255,6 +255,7 @@ set(MAIN_SRCS
     NFA.cc
     Net.cc
     NetVar.cc
+    Notifier.cc
     Obj.cc
     OpaqueVal.cc
     Options.cc
@@ -270,12 +271,12 @@ set(MAIN_SRCS
     RuleAction.cc
     RuleCondition.cc
     RuleMatcher.cc
-    SmithWaterman.cc
+    ScannedFile.cc
     Scope.cc
     ScriptCoverageManager.cc
     SerializationFormat.cc
     Sessions.cc
-    Notifier.cc
+    SmithWaterman.cc
     Stats.cc
     Stmt.cc
     Tag.cc

src/Net.cc

@@ -397,9 +397,6 @@ void net_delete()
 		delete zeek::detail::ip_anonymizer[i];
 	}
 
-std::list<ScannedFile> files_scanned;
-std::vector<std::string> sig_files;
-
 } // namespace detail
 
 bool reading_live = false;
@@ -461,6 +458,3 @@ bool& is_parsing = zeek::net::is_parsing;
 const zeek::Packet*& current_pkt = zeek::net::current_pkt;
 int& current_dispatched = zeek::net::current_dispatched;
 double& current_timestamp = zeek::net::current_timestamp;
-
-std::list<zeek::net::detail::ScannedFile>& files_scanned = zeek::net::detail::files_scanned;
-std::vector<std::string>& sig_files = zeek::net::detail::sig_files;

src/Net.h

@@ -11,6 +11,8 @@
 #include <string>
 #include <optional>
 
+#include "ScannedFile.h"
+
 ZEEK_FORWARD_DECLARE_NAMESPACED(IOSource, zeek, iosource);
 ZEEK_FORWARD_DECLARE_NAMESPACED(PktSrc, zeek, iosource);
 ZEEK_FORWARD_DECLARE_NAMESPACED(PktDumper, zeek, iosource);
@@ -44,30 +46,6 @@ extern zeek::iosource::PktDumper* pkt_dumper;	// where to save packets
 // on future timers).
 extern bool have_pending_timers;
 
-
-// Script file we have already scanned (or are in the process of scanning).
-// They are identified by normalized realpath.
-struct ScannedFile {
-	int include_level;
-	bool skipped;		// This ScannedFile was @unload'd.
-	bool prefixes_checked;	// If loading prefixes for this file has been tried.
-	std::string name;
-	std::string canonical_path; // normalized, absolute path via realpath()
-
-	ScannedFile(int arg_include_level,
-	            std::string arg_name, bool arg_skipped = false,
-	            bool arg_prefixes_checked = false);
-
-	/**
-	 * Compares the canonical path of this file against every canonical path
-	 * in files_scanned and returns whether there's any match.
-	 */
-	bool AlreadyScanned() const;
-};
-
-extern std::list<ScannedFile> files_scanned;
-extern std::vector<std::string> sig_files;
-
 } // namespace detail
 
 // Functions to temporarily suspend processing of live input (network packets
@@ -144,7 +122,3 @@ extern bool& is_parsing [[deprecated("Remove in v4.1. Use zeek::net::is_parsing.
 extern const zeek::Packet*& current_pkt [[deprecated("Remove in v4.1. Use zeek::net::current_pkt.")]];
 extern int& current_dispatched [[deprecated("Remove in v4.1. Use zeek::net::current_dispatched.")]];
 extern double& current_timestamp [[deprecated("Remove in v4.1. Use zeek::net::current_timestamp.")]];
-
-using ScannedFile [[deprecated("Remove in v4.1. Use zeek::net::detail::ScannedFile.")]] = zeek::net::detail::ScannedFile;
-extern std::list<zeek::net::detail::ScannedFile>& files_scanned [[deprecated("Remove in v4.1. Use zeek::net::detail::files_scanned.")]];
-extern std::vector<std::string>& sig_files [[deprecated("Remove in v4.1. Use zeek::net::detail::sig_files.")]];

src/ScannedFile.cc

@@ -0,0 +1,56 @@
+#include "ScannedFile.h"
+
+#include <sys/errno.h>
+#include <limits.h> // for PATH_MAX
+
+#include "DebugLogger.h"
+#include "Reporter.h"
+
+namespace zeek::detail {
+
+std::list<ScannedFile> files_scanned;
+std::vector<std::string> sig_files;
+
+ScannedFile::ScannedFile(int arg_include_level,
+                         std::string arg_name,
+                         bool arg_skipped,
+                         bool arg_prefixes_checked)
+	: include_level(arg_include_level),
+	  skipped(arg_skipped),
+	  prefixes_checked(arg_prefixes_checked),
+	  name(std::move(arg_name))
+	{
+	if ( name == canonical_stdin_path )
+		canonical_path = canonical_stdin_path;
+	else
+		{
+		char buf[PATH_MAX];
+		auto res = realpath(name.data(), buf);
+
+		if ( ! res )
+			zeek::reporter->FatalError("failed to get realpath() of %s: %s",
+			                           name.data(), strerror(errno));
+
+		canonical_path = res;
+		}
+	}
+
+bool ScannedFile::AlreadyScanned() const
+	{
+	auto rval = false;
+
+	for ( const auto& it : files_scanned )
+		if ( it.canonical_path == canonical_path )
+			{
+			rval = true;
+			break;
+			}
+
+	DBG_LOG(zeek::DBG_SCRIPTS, "AlreadyScanned result (%d) %s", rval, canonical_path.data());
+	return rval;
+	}
+
+} // namespace zeek::detail
+
+std::list<zeek::detail::ScannedFile>& files_scanned = zeek::detail::files_scanned;
+std::vector<std::string>& sig_files = zeek::detail::sig_files;

src/ScannedFile.h

@@ -0,0 +1,44 @@
+// See the file "COPYING" in the main distribution directory for copyright.
+
+#pragma once
+
+#include <string>
+#include <list>
+#include <vector>
+
+namespace zeek::detail {
+
+// Script file we have already scanned (or are in the process of scanning).
+// They are identified by normalized realpath.
+class ScannedFile {
+
+public:
+
+	ScannedFile(int arg_include_level,
+	            std::string arg_name, bool arg_skipped = false,
+	            bool arg_prefixes_checked = false);
+
+	/**
+	 * Compares the canonical path of this file against every canonical path
+	 * in files_scanned and returns whether there's any match.
+	 */
+	bool AlreadyScanned() const;
+
+	int include_level;
+	bool skipped;		// This ScannedFile was @unload'd.
+	bool prefixes_checked;	// If loading prefixes for this file has been tried.
+	std::string name;
+	std::string canonical_path; // normalized, absolute path via realpath()
+
+	static auto constexpr canonical_stdin_path = "<stdin>";
+
+};
+
+extern std::list<ScannedFile> files_scanned;
+extern std::vector<std::string> sig_files;
+
+} // namespace zeek::detail
+
+using ScannedFile [[deprecated("Remove in v4.1. Use zeek::detail::ScannedFile.")]] = zeek::detail::ScannedFile;
+extern std::list<zeek::detail::ScannedFile>& files_scanned [[deprecated("Remove in v4.1. Use zeek::detail::files_scanned.")]];
+extern std::vector<std::string>& sig_files [[deprecated("Remove in v4.1. Use zeek::detail::sig_files.")]];

src/scan.l

@@ -34,6 +34,7 @@
 #include "Net.h"
 #include "Traverse.h"
 #include "module_util.h"
+#include "ScannedFile.h"
 
 #include "analyzer/Analyzer.h"
 #include "zeekygen/Manager.h"
@@ -347,7 +348,7 @@ when	return TOK_WHEN;
 			zeek::reporter->Error("failed to find file associated with @load-sigs %s",
 			                      file);
 		else
-			zeek::net::detail::sig_files.push_back(zeek::util::copy_string(path.c_str()));
+			zeek::detail::sig_files.push_back(zeek::util::copy_string(path.c_str()));
 		break;
 
 	case 0:
@@ -404,8 +405,8 @@ when	return TOK_WHEN;
 	else
 		{
 		// All we have to do is pretend we've already scanned it.
-		zeek::net::detail::ScannedFile sf(file_stack.length(), std::move(path), true);
-		zeek::net::detail::files_scanned.push_back(std::move(sf));
+		zeek::detail::ScannedFile sf(file_stack.length(), std::move(path), true);
+		zeek::detail::files_scanned.push_back(std::move(sf));
 		}
 	}
 
@@ -571,8 +572,6 @@ YYLTYPE zeek::detail::GetCurrentLocation()
 	return currloc;
 	}
 
-static auto constexpr canonical_stdin_path = "<stdin>";
-
 static int load_files(const char* orig_file)
 	{
 	std::string file_path = find_relative_script_file(orig_file);
@@ -599,7 +598,7 @@ static int load_files(const char* orig_file)
 	if ( zeek::util::streq(orig_file, "-") )
 		{
 		f = stdin;
-		file_path = canonical_stdin_path;
+		file_path = zeek::detail::ScannedFile::canonical_stdin_path;
 
 		if ( zeek::detail::g_policy_debug )
 			{
@@ -622,7 +621,7 @@ static int load_files(const char* orig_file)
 			zeek::reporter->FatalError("can't open %s", file_path.c_str());
 		}
 
-	zeek::net::detail::ScannedFile sf(file_stack.length(), file_path);
+	zeek::detail::ScannedFile sf(file_stack.length(), file_path);
 
 	if ( sf.AlreadyScanned() )
 		{
@@ -632,7 +631,7 @@ static int load_files(const char* orig_file)
 		return 0;
 		}
 
-	zeek::net::detail::files_scanned.push_back(std::move(sf));
+	zeek::detail::files_scanned.push_back(std::move(sf));
 
 	if ( zeek::detail::g_policy_debug && ! file_path.empty() )
 		{
@@ -889,13 +888,12 @@ int yywrap()
 	// file name is discarded. If the prefix is non-empty, it gets placed
 	// in front of the flattened path, separated with another '.'
 	bool found_prefixed_files = false;
-	for ( auto it = zeek::net::detail::files_scanned.begin();
-	      it != zeek::net::detail::files_scanned.end(); ++it )
+	for ( auto& scanned_file : zeek::detail::files_scanned )
 		{
-		if ( it->skipped || it->prefixes_checked )
+		if ( scanned_file.skipped || scanned_file.prefixes_checked )
 			continue;
 
-		it->prefixes_checked = true;
+		scanned_file.prefixes_checked = true;
 		// Prefixes are pushed onto a stack, so iterate backwards.
 		for ( int i = zeek::detail::zeek_script_prefixes.size() - 1; i >= 0; --i )
 			{
@@ -903,7 +901,7 @@ int yywrap()
 			if ( ! zeek::detail::zeek_script_prefixes[i][0] )
 				continue;
 
-			std::string canon = zeek::util::detail::without_zeekpath_component(it->name);
+			std::string canon = zeek::util::detail::without_zeekpath_component(scanned_file.name);
 			std::string flat = zeek::util::detail::flatten_script_name(canon, zeek::detail::zeek_script_prefixes[i]);
 			std::string path = find_relative_script_file(flat);
 
@@ -914,7 +912,7 @@ int yywrap()
 				}
 
 			//printf("====== prefix search ======\n");
-			//printf("File  : %s\n", it->name.c_str());
+			//printf("File  : %s\n", scanned_file.name.c_str());
 			//printf("Canon : %s\n", canon.c_str());
 			//printf("Flat  : %s\n", flat.c_str());
 			//printf("Found : %s\n", path.empty() ? "F" : "T");
@@ -1027,42 +1025,3 @@ FileInfo::~FileInfo()
 	if ( restore_module != "" )
 		zeek::detail::current_module = restore_module;
 	}
-
-zeek::net::detail::ScannedFile::ScannedFile(int arg_include_level,
-                                            std::string arg_name,
-                                            bool arg_skipped,
-                                            bool arg_prefixes_checked)
-	: include_level(arg_include_level),
-	  skipped(arg_skipped),
-	  prefixes_checked(arg_prefixes_checked),
-	  name(std::move(arg_name))
-	{
-	if ( name == canonical_stdin_path )
-		canonical_path = canonical_stdin_path;
-	else
-		{
-		char buf[PATH_MAX];
-		auto res = realpath(name.data(), buf);
-
-		if ( ! res )
-			zeek::reporter->FatalError("failed to get realpath() of %s: %s",
-			                           name.data(), strerror(errno));
-
-		canonical_path = res;
-		}
-	}
-
-bool zeek::net::detail::ScannedFile::AlreadyScanned() const
-	{
-	auto rval = false;
-
-	for ( const auto& it : zeek::net::detail::files_scanned )
-		if ( it.canonical_path == canonical_path )
-			{
-			rval = true;
-			break;
-			}
-
-	DBG_LOG(zeek::DBG_SCRIPTS, "AlreadyScanned result (%d) %s", rval, canonical_path.data());
-	return rval;
-	}

src/zeek-setup.cc

@@ -46,6 +46,7 @@ extern "C" {
 #include "Trigger.h"
 #include "Hash.h"
 #include "Func.h"
+#include "ScannedFile.h"
 
 #include "supervisor/Supervisor.h"
 #include "threading/Manager.h"
@@ -712,7 +713,7 @@ zeek::detail::SetupResult setup(int argc, char** argv,
 		all_signature_files.emplace_back(std::move(sf));
 
 	// Append signature files defined in @load-sigs
-	for ( const auto& sf : zeek::net::detail::sig_files )
+	for ( const auto& sf : zeek::detail::sig_files )
 		all_signature_files.emplace_back(sf);
 
 	if ( ! all_signature_files.empty() )
@@ -850,7 +851,7 @@ zeek::detail::SetupResult setup(int argc, char** argv,
 	if ( zeek_script_loaded )
 		{
 		// Queue events reporting loaded scripts.
-		for ( const auto& file : zeek::net::detail::files_scanned )
+		for ( const auto& file : zeek::detail::files_scanned )
 			{
 			if ( file.skipped )
 				continue;