From bcb703b9a30d537f55c43e74ef64393dcd2e30e2 Mon Sep 17 00:00:00 2001
From: Arvid Norberg <arvid@libtorrent.org>
Date: Mon, 18 Nov 2013 06:59:47 +0000
Subject: [PATCH] improve duplicate file detection in torrent files

---
 include/libtorrent/file.hpp |  1 +
 src/file.cpp                | 12 ++++++++++++
 src/torrent_info.cpp        | 15 ++++++++++-----
 test/test_file.cpp          |  5 +++++
 test/test_torrent_info.cpp  | 10 +++++++---
 5 files changed, 35 insertions(+), 8 deletions(-)

diff --git a/include/libtorrent/file.hpp b/include/libtorrent/file.hpp
index 75993d934..91086ebdc 100644
--- a/include/libtorrent/file.hpp
+++ b/include/libtorrent/file.hpp
@@ -133,6 +133,7 @@ namespace libtorrent
 	TORRENT_EXTRA_EXPORT std::string split_path(std::string const& f);
 	TORRENT_EXTRA_EXPORT char const* next_path_element(char const* p);
 	TORRENT_EXTRA_EXPORT std::string extension(std::string const& f);
+	TORRENT_EXTRA_EXPORT std::string remove_extension(std::string const& f);
 	TORRENT_EXTRA_EXPORT void replace_extension(std::string& f, std::string const& ext);
 	TORRENT_EXTRA_EXPORT bool is_root_path(std::string const& f);
 
diff --git a/src/file.cpp b/src/file.cpp
index de7f81f3f..60955b18a 100644
--- a/src/file.cpp
+++ b/src/file.cpp
@@ -423,6 +423,18 @@ namespace libtorrent
 		return "";
 	}
 
+	std::string remove_extension(std::string const& f)
+	{
+		char const* slash = strrchr(f.c_str(), '/');
+#ifdef TORRENT_WINDOWS
+		slash = (std::max)((char const*)strrchr(f.c_str(), '\\'), slash);
+#endif
+		char const* ext = strrchr(f.c_str(), '.');
+		// if we don't have an extension, just return f
+		if (ext == 0 || ext == &f[0] || (slash != NULL && ext < slash)) return f;
+		return f.substr(0, ext - &f[0]);
+	}
+
 	void replace_extension(std::string& f, std::string const& ext)
 	{
 		for (int i = f.size() - 1; i >= 0; --i)
diff --git a/src/torrent_info.cpp b/src/torrent_info.cpp
index 299c267d2..a4bfd510d 100644
--- a/src/torrent_info.cpp
+++ b/src/torrent_info.cpp
@@ -402,12 +402,17 @@ namespace libtorrent
 			// as long as this file already exists
 			// increase the counter
 			int cnt = 0;
-			while (!files.insert(e.path).second)
+			if (!files.insert(e.path).second)
 			{
-				++cnt;
-				char suffix[50];
-				snprintf(suffix, sizeof(suffix), "%d%s", cnt, extension(e.path).c_str());
-				replace_extension(e.path, suffix);
+				std::string base = remove_extension(e.path);
+				std::string ext = extension(e.path);
+				do
+				{
+					++cnt;
+					char new_ext[50];
+					snprintf(new_ext, sizeof(new_ext), ".%d%s", cnt, ext.c_str());
+					e.path = base + new_ext;
+				} while (!files.insert(e.path).second);
 			}
 			target.add_file(e, file_hash ? file_hash->string_ptr() + info_ptr_diff : 0);
 
diff --git a/test/test_file.cpp b/test/test_file.cpp
index 8b620a471..f04b0d286 100644
--- a/test/test_file.cpp
+++ b/test/test_file.cpp
@@ -127,6 +127,11 @@ int test_main()
 	TEST_EQUAL(extension("blah.foo."), ".");
 	TEST_EQUAL(extension("blah.foo/bar"), "");
 
+	TEST_EQUAL(remove_extension("blah"), "blah");
+	TEST_EQUAL(remove_extension("blah.exe"), "blah");
+	TEST_EQUAL(remove_extension("blah.foo.bar"), "blah.foo");
+	TEST_EQUAL(remove_extension("blah.foo."), "blah.foo");
+
 	TEST_EQUAL(filename("blah"), "blah");
 	TEST_EQUAL(filename("/blah/foo/bar"), "bar");
 	TEST_EQUAL(filename("/blah/foo/bar/"), "bar");
diff --git a/test/test_torrent_info.cpp b/test/test_torrent_info.cpp
index 2c819572a..178743da9 100644
--- a/test/test_torrent_info.cpp
+++ b/test/test_torrent_info.cpp
@@ -41,7 +41,6 @@ int test_main()
 {
 
 	file_storage fs;
-	int total_size = 100 * 0x4000;
 
 	fs.add_file("test/temporary.txt", 0x4000);
 	fs.add_file("test/A/tmp", 0x4000);
@@ -76,12 +75,17 @@ int test_main()
 	"test/A/tmp",
 	"test/Temporary.1.txt", // duplicate of temporary.txt
 	"test/TeMPorArY.2.txT", // duplicate of temporary.txt
-	"test/a.1", // a file may not have the same name as a directory
+
+	// a file may not have the same name as a directory
+	// however, detecting this is not supported currently.
+	// it is in the next major release
+	"test/a",
+
 	"test/b.exe",
 	"test/B.1.ExE", // duplicate of b.exe
 	"test/B.2.exe", // duplicate of b.exe
 	"test/test/TEMPORARY.TXT", // a file with the same name in a seprate directory is fine
-	"test/A.2", // duplicate of directory a
+	"test/A.1", // duplicate of directory a
 	};
 
 	for (int i = 0; i < ti.num_files(); ++i)