From 4c1c5bf00f9784ba5e86b03af1b700fcc78e0936 Mon Sep 17 00:00:00 2001
From: Armin Friedl <dev@friedl.net>
Date: Fri, 19 Feb 2021 06:07:32 +0100
Subject: [PATCH] Compression

---
 src/Archiver.cpp             | 37 +++++++++++++++++++++++++-----------
 src/Xwim.cpp                 |  9 ++++-----
 src/archiver/LibArchiver.cpp | 35 +++++++++++++++++++++++++++++++++-
 3 files changed, 64 insertions(+), 17 deletions(-)

diff --git a/src/Archiver.cpp b/src/Archiver.cpp
index b3e3a91..df32a25 100644
--- a/src/Archiver.cpp
+++ b/src/Archiver.cpp
@@ -1,7 +1,9 @@
 #include "Archiver.hpp"
 
+#include <spdlog/common.h>
 #include <spdlog/spdlog.h>
 
+#include <filesystem>
 #include <map>
 #include <memory>
 
@@ -15,16 +17,19 @@ namespace fs = std::filesystem;
 fs::path archive_extension(const fs::path& path) {
   // TODO: creates lots of paths, refactor
   fs::path ext;
+  fs::path tmp_ext;
   fs::path tmp_path = path;
   while (tmp_path.has_extension()) {
-    fs::path tmp_ext = tmp_path.extension() += ext;
+    tmp_ext = tmp_path.extension() += tmp_ext;
     auto search = extensions_format.find(tmp_ext);
 
-    // (Combined) extension not known, return last known extension
-    if (search == extensions_format.end()) return ext;
+    if (search != extensions_format.end()) {
+      // (Combined) extension known. Remember as `ext` and keep
+      // looking for even longer extensions.
+      ext = tmp_ext;
+    }  // else: (Combined) extension not known, keep `ext` as-is but try longer
+       // extensions
 
-    // Continue extending extension
-    ext = tmp_ext;
     tmp_path = tmp_path.stem();
   }
 
@@ -34,20 +39,28 @@ fs::path archive_extension(const fs::path& path) {
 // Strip longest known extension from path
 fs::path strip_archive_extension(const fs::path& path) {
   // TODO: creates lots of paths, refactor
-  fs::path ext;
+  int longest_ext = 0;
+  fs::path tmp_ext;
   fs::path tmp_path = path;
+  fs::path stem_path = path;
+
   while (tmp_path.has_extension()) {
-    fs::path tmp_ext = tmp_path.extension() += ext;
+    tmp_ext = tmp_path.extension() += tmp_ext;
     auto search = extensions_format.find(tmp_ext);
 
-    // (Combined) extension not known, return stripped path
-    if (search == extensions_format.end()) return tmp_path;
+    if (search != extensions_format.end()) {
+      // (Combined) extension known. Remember as `longest_ext` and keep
+      // looking for even longer extensions.
+      longest_ext++;
+    }  // else: (Combined) extension not known, keep `longest_ext` as-is but try longer
+       // extensions
 
-    // Continue stripping path
-    ext = tmp_ext;
     tmp_path = tmp_path.stem();
   }
 
+  tmp_path = path;
+  for(int i=0; i<longest_ext; i++) tmp_path = tmp_path.stem();
+
   return tmp_path;
 }
 
@@ -63,7 +76,9 @@ bool can_extract(const fs::path& path) {
 }
 
 Format parse_format(const fs::path& path) {
+  spdlog::debug("Looking for path {}", path);
   fs::path ext = archive_extension(path);
+  spdlog::debug("Looking for ext {}", ext);
   auto search = extensions_format.find(ext);
   if (search == extensions_format.end()) {
     throw XwimError{"No known archiver for {}", path};
diff --git a/src/Xwim.cpp b/src/Xwim.cpp
index 052a9cc..d6d7ef4 100644
--- a/src/Xwim.cpp
+++ b/src/Xwim.cpp
@@ -83,11 +83,10 @@ void Xwim::infer_compression_output() {
   if (ins.size() == 1) {
     // archive name is just the name of the input with default archive
     // extension
-
-    fs::path p = *ins.begin();
-    while (p.has_extension()) p = p.stem();
-    p += default_extension;
-    out = p;
+    fs::path archive_stem = xwim::strip_archive_extension(*ins.begin());
+    fs::path path = (*ins.begin()).stem();
+    path += default_extension;
+    out = path;
   } else {
     // We cannot guess the name of the output archive
 
diff --git a/src/archiver/LibArchiver.cpp b/src/archiver/LibArchiver.cpp
index 1198965..765f33c 100644
--- a/src/archiver/LibArchiver.cpp
+++ b/src/archiver/LibArchiver.cpp
@@ -1,7 +1,9 @@
 #include <archive.h>
 #include <archive_entry.h>
+#include <fcntl.h>
 #include <fmt/core.h>
 #include <spdlog/spdlog.h>
+#include <sys/stat.h>
 
 #include <filesystem>
 #include <iostream>
@@ -16,7 +18,38 @@ namespace fs = std::filesystem;
 
 static int copy_data(shared_ptr<archive> reader, shared_ptr<archive> writer);
 
-void LibArchiver::compress(set<fs::path> ins, fs::path archive_out) { return; }
+void LibArchiver::compress(set<fs::path> ins, fs::path archive_out) {
+  spdlog::debug("Compressing to {}", archive_out);
+
+  // cannot use unique_ptr here since unique_ptr requires a
+  // complete type. `archive` is forward declared only.
+  shared_ptr<archive> writer;
+  writer = shared_ptr<archive>(archive_write_new(), archive_write_free);
+  archive_write_add_filter_gzip(writer.get());
+  archive_write_set_format_pax_restricted(writer.get());
+  archive_write_open_filename(writer.get(), archive_out.c_str());
+
+  archive_entry *entry = archive_entry_new();
+  char buff[8192];
+
+  for(auto path: ins) {
+    archive_entry_set_pathname(entry, path.c_str());
+    archive_entry_set_size(entry, fs::file_size(path));
+    archive_entry_set_filetype(entry, AE_IFREG);
+    archive_entry_set_perm(entry, 0644);
+    archive_write_header(writer.get(), entry);
+
+    int fd = open(path.c_str(), O_RDONLY);
+    int len = read(fd, buff, sizeof(buff));
+    while (len > 0) {
+      archive_write_data(writer.get(), buff, len);
+      len = read(fd, buff, sizeof(buff));
+    }
+
+    close(fd);
+    archive_entry_clear(entry);
+  }
+}
 
 void LibArchiver::extract(fs::path archive_in, fs::path out) {
   spdlog::debug("Extracting archive {} to {}", archive_in, out);