From e59968444ed79df4b4bffa3421f688967b79465e Mon Sep 17 00:00:00 2001
From: Armin Friedl <dev@friedl.net>
Date: Sun, 23 Feb 2020 15:20:47 +0100
Subject: [PATCH] [feat] Add known fileformats

Reliably stripping (multiple) archive extensions (e.g. .tar.gz) is not supported
by std::filesystem::path. Dots in the regular root folder name can cause issues.

fileformats.hpp adds well-known format extensions which are used for stripping
the extension.

Note that this is not used by libarchive or xwim to determine the
filters/formats for extraction. This is done by libarchive's `bidding`.
---
 meson.build         |  2 +-
 src/archive.cpp     |  5 ++---
 src/archive.hpp     |  5 +----
 src/fileformats.hpp | 38 ++++++++++++++++++++++++++++++++++++++
 src/main.cpp        |  6 ++----
 src/meson.build     |  4 +++-
 src/spec.hpp        |  5 +----
 7 files changed, 48 insertions(+), 17 deletions(-)
 create mode 100644 src/fileformats.hpp

diff --git a/meson.build b/meson.build
index 5ec2ae2..aa1a0d1 100644
--- a/meson.build
+++ b/meson.build
@@ -1,5 +1,5 @@
 project('xwim', 'cpp',
-        version: '0.1',
+        version: '0.2',
         default_options: ['cpp_std=c++17'])
 
 subdir('src')
diff --git a/src/archive.cpp b/src/archive.cpp
index 50026da..7ff9bc6 100644
--- a/src/archive.cpp
+++ b/src/archive.cpp
@@ -14,6 +14,7 @@ namespace logger = spdlog;
 #include "archive_sys.hpp"
 #include "archive.hpp"
 #include "spec.hpp"
+#include "fileformats.hpp"
 
 namespace xwim {
 
@@ -22,9 +23,7 @@ static void _spec_is_root_filename(ArchiveSpec* spec,
                                    std::filesystem::path* filepath) {
   auto entry_path = entry.path();
   auto norm_stem = filepath->filename();
-
-  while (norm_stem.has_extension())
-    norm_stem = norm_stem.stem();
+  norm_stem = xwim::stem(norm_stem);
 
   if (*entry_path.begin() != norm_stem) {
     logger::debug("Archive root does not match archive name");
diff --git a/src/archive.hpp b/src/archive.hpp
index bbf36b9..0e3b494 100644
--- a/src/archive.hpp
+++ b/src/archive.hpp
@@ -1,5 +1,4 @@
-#ifndef ARCHIVE_H
-#define ARCHIVE_H
+#pragma once
 
 #include <archive.h>
 #include <fmt/format.h>
@@ -42,5 +41,3 @@ class ArchiveException : public std::exception {
 };
 
 }  // namespace xwim
-
-#endif
diff --git a/src/fileformats.hpp b/src/fileformats.hpp
new file mode 100644
index 0000000..4dc3562
--- /dev/null
+++ b/src/fileformats.hpp
@@ -0,0 +1,38 @@
+#pragma once
+
+#include <spdlog/spdlog.h>
+namespace logger = spdlog;
+
+/** Common archive formats understood by xwim
+ *
+ * The underlying libarchive backend retrieves format information by a process
+ * called `bidding`. Hence, this information is mainly used to strip extensions.
+ *
+ * Stripping extensions via `std::filesystem::path` does not work reliably since
+ * it gets easily confused by dots in the regular file name.
+ */
+
+#include <filesystem>
+#include <set>
+#include <string>
+
+namespace xwim {
+
+  const std::set<std::string> fileformats{
+    ".7z",   ".7zip", ".jar", ".tgz",    ".bz2",     ".bzip2",  ".gz",
+    ".gzip", ".rar",  ".tar", ".tar.gz", ".tar.bz2", ".tar.xz", ".zip"};
+
+inline std::filesystem::path stem(std::filesystem::path& path) {
+  std::filesystem::path p_stem {path};
+  logger::trace("Stemming {}", p_stem.string());
+
+  while( fileformats.find(p_stem.extension().string()) != fileformats.end() ) {
+    p_stem = p_stem.stem();
+    logger::trace("Stemmed to {}", p_stem.string());
+  }
+
+  logger::trace("Finished stemming {}", p_stem.string());
+  return p_stem;
+}
+
+}  // namespace xwim
diff --git a/src/main.cpp b/src/main.cpp
index c34037a..a5e64c9 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -9,6 +9,7 @@ namespace logger = spdlog;
 
 #include "archive.hpp"
 #include "spec.hpp"
+#include "fileformats.hpp"
 
 int main(int argc, char** argv) {
   logger::set_level(logger::level::trace);
@@ -25,10 +26,7 @@ int main(int argc, char** argv) {
     if (!archive_spec.has_single_root || !archive_spec.is_root_filename) {
       extract_spec.make_dir = true;
 
-      std::filesystem::path stem = filepath.stem();
-
-      while (stem.has_extension())
-        stem = stem.stem();
+      std::filesystem::path stem = xwim::stem(filepath);
 
       extract_spec.dirname = stem;
     }
diff --git a/src/meson.build b/src/meson.build
index caba555..bfebd28 100644
--- a/src/meson.build
+++ b/src/meson.build
@@ -1,9 +1,11 @@
 src = ['main.cpp',
        'archive.cpp',
        'archive_sys.cpp']
+
 inc = ['archive.hpp',
        'spec.hpp',
-       'archive_sys.hpp']
+       'archive_sys.hpp',
+       'fileformats.hpp']
 
 libs = [dependency('libarchive', required: true),
         dependency('fmt', required: true),
diff --git a/src/spec.hpp b/src/spec.hpp
index c0bf840..e9ab8d2 100644
--- a/src/spec.hpp
+++ b/src/spec.hpp
@@ -1,5 +1,4 @@
-#ifndef SPEC_H
-#define SPEC_H
+#pragma once
 
 #include <archive.h>
 #include <fmt/core.h>
@@ -55,5 +54,3 @@ struct fmt::formatter<xwim::ExtractSpec> {
                      spec.make_dir, spec.dirname.string(), spec.extract_subarchive);
   }
 };
-
-#endif