Merge branch 'libarchive' of https://github.com/yorickvP/nix

5 files changed, 122 insertions, 52 deletions

diff --git a/src/libutil/local.mk b/src/libutil/local.mk
index 35c1f6c13..16c1fa03f 100644
--- a/src/libutil/local.mk
+++ b/src/libutil/local.mk
@@ -6,6 +6,6 @@ libutil_DIR := $(d)
 
 libutil_SOURCES := $(wildcard $(d)/*.cc)
 
-libutil_LDFLAGS = $(LIBLZMA_LIBS) -lbz2 -pthread $(OPENSSL_LIBS) $(LIBBROTLI_LIBS) $(BOOST_LDFLAGS) -lboost_context
+libutil_LDFLAGS = $(LIBLZMA_LIBS) -lbz2 -pthread $(OPENSSL_LIBS) $(LIBBROTLI_LIBS) $(LIBARCHIVE_LIBS) $(BOOST_LDFLAGS) -lboost_context
 
 libutil_LIBS = libnixrust
diff --git a/src/libutil/rust-ffi.cc b/src/libutil/rust-ffi.cc
index 8b8b7b75d..6f36b3192 100644
--- a/src/libutil/rust-ffi.cc
+++ b/src/libutil/rust-ffi.cc
@@ -19,15 +19,4 @@ std::ostream & operator << (std::ostream & str, const String & s)
     return str;
 }
 
-size_t Source::sourceWrapper(void * _this, rust::Slice<uint8_t> data)
-{
-    try {
-        // FIXME: how to propagate exceptions?
-        auto n = ((nix::Source *) _this)->read((unsigned char *) data.ptr, data.size);
-        return n;
-    } catch (...) {
-        abort();
-    }
-}
-
 }
diff --git a/src/libutil/rust-ffi.hh b/src/libutil/rust-ffi.hh
index 3b51661c2..469a5fba3 100644
--- a/src/libutil/rust-ffi.hh
+++ b/src/libutil/rust-ffi.hh
@@ -131,18 +131,6 @@ struct String : Vec<char, ffi_String_drop>
 
 std::ostream & operator << (std::ostream & str, const String & s);
 
-struct Source
-{
-    size_t (*fun)(void * source_this, rust::Slice<uint8_t> data);
-    nix::Source * _this;
-
-    Source(nix::Source & _this)
-        : fun(sourceWrapper), _this(&_this)
-    {}
-
-    static size_t sourceWrapper(void * _this, rust::Slice<uint8_t> data);
-};
-
 /* C++ representation of Rust's Result<T, CppException>. */
 template<typename T>
 struct Result
diff --git a/src/libutil/tarfile.cc b/src/libutil/tarfile.cc
index 1be0ba24c..68e918891 100644
--- a/src/libutil/tarfile.cc
+++ b/src/libutil/tarfile.cc
@@ -1,38 +1,132 @@
-#include "rust-ffi.hh"
-#include "compression.hh"
+#include <archive.h>
+#include <archive_entry.h>
 
-extern "C" {
-    rust::Result<std::tuple<>> *
-    unpack_tarfile(rust::Source source, rust::StringSlice dest_dir, rust::Result<std::tuple<>> & out);
-}
+#include "serialise.hh"
 
 namespace nix {
 
+struct TarArchive {
+    struct archive *archive;
+    Source *source;
+    std::vector<unsigned char> buffer;
+
+    void check(int err, const char *reason = "Failed to extract archive (%s)") {
+        if (err == ARCHIVE_EOF)
+            throw EndOfFile("reached end of archive");
+        else if (err != ARCHIVE_OK)
+            throw Error(reason, archive_error_string(this->archive));
+    }
+
+    TarArchive(Source& source) : buffer(4096) {
+        this->archive = archive_read_new();
+        this->source = &source;
+
+        archive_read_support_filter_all(archive);
+        archive_read_support_format_all(archive);
+        check(archive_read_open(archive, (void *)this, TarArchive::callback_open, TarArchive::callback_read, TarArchive::callback_close), "Failed to open archive (%s)");
+    }
+
+    TarArchive(const Path &path) {
+        this->archive = archive_read_new();
+
+        archive_read_support_filter_all(archive);
+        archive_read_support_format_all(archive);
+        check(archive_read_open_filename(archive, path.c_str(), 16384), "Failed to open archive (%s)");
+    }
+
+    // disable copy constructor
+    TarArchive(const TarArchive&) = delete;
+
+    void close() {
+        check(archive_read_close(archive), "Failed to close archive (%s)");
+    }
+
+    ~TarArchive() {
+        if (this->archive) archive_read_free(this->archive);
+    }
+
+private:
+    static int callback_open(struct archive *, void *self) {
+        return ARCHIVE_OK;
+    }
+
+    static ssize_t callback_read(struct archive *archive, void *_self, const void **buffer) {
+        TarArchive *self = (TarArchive *)_self;
+        *buffer = self->buffer.data();
+
+        try {
+            return self->source->read(self->buffer.data(), 4096);
+        } catch (EndOfFile &) {
+            return 0;
+        } catch (std::exception &err) {
+            archive_set_error(archive, EIO, "Source threw exception: %s", err.what());
+
+            return -1;
+        }
+    }
+
+    static int callback_close(struct archive *, void *self) {
+        return ARCHIVE_OK;
+    }
+};
+
+struct PushD {
+    char * oldDir;
+
+    PushD(const std::string &newDir)  {
+        oldDir = getcwd(0, 0);
+        if (!oldDir) throw SysError("getcwd");
+        int r = chdir(newDir.c_str());
+        if (r != 0) throw SysError("changing directory to tar output path");
+    }
+
+    ~PushD() {
+        int r = chdir(oldDir);
+        free(oldDir);
+        if (r != 0)
+            std::cerr << "warning: failed to change directory back after tar extraction";
+        /* can't throw out of a destructor */
+    }
+};
+
+static void extract_archive(TarArchive &archive, const Path & destDir) {
+    // need to chdir back *after* archive closing
+    PushD newDir(destDir);
+    struct archive_entry *entry;
+    int flags = ARCHIVE_EXTRACT_FFLAGS
+        | ARCHIVE_EXTRACT_PERM
+        | ARCHIVE_EXTRACT_SECURE_SYMLINKS
+        | ARCHIVE_EXTRACT_SECURE_NODOTDOT
+        | ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS;
+
+    for(;;) {
+        int r = archive_read_next_header(archive.archive, &entry);
+        if (r == ARCHIVE_EOF) break;
+        else if (r == ARCHIVE_WARN)
+            std::cerr << "warning: " << archive_error_string(archive.archive) << std::endl;
+        else
+            archive.check(r);
+
+        archive.check(archive_read_extract(archive.archive, entry, flags));
+    }
+
+    archive.close();
+}
+
 void unpackTarfile(Source & source, const Path & destDir)
 {
-    rust::Source source2(source);
-    rust::Result<std::tuple<>> res;
-    unpack_tarfile(source2, destDir, res);
-    res.unwrap();
+    auto archive = TarArchive(source);
+
+    createDirs(destDir);
+    extract_archive(archive, destDir);
 }
 
-void unpackTarfile(const Path & tarFile, const Path & destDir,
-    std::optional<std::string> baseName)
+void unpackTarfile(const Path & tarFile, const Path & destDir)
 {
-    if (!baseName) baseName = std::string(baseNameOf(tarFile));
-
-    auto source = sinkToSource([&](Sink & sink) {
-        // FIXME: look at first few bytes to determine compression type.
-        auto decompressor =
-            hasSuffix(*baseName, ".bz2") ? makeDecompressionSink("bzip2", sink) :
-            hasSuffix(*baseName, ".gz") ? makeDecompressionSink("gzip", sink) :
-            hasSuffix(*baseName, ".xz") ? makeDecompressionSink("xz", sink) :
-            makeDecompressionSink("none", sink);
-        readFile(tarFile, *decompressor);
-        decompressor->finish();
-    });
-
-    unpackTarfile(*source, destDir);
+    auto archive = TarArchive(tarFile);
+
+    createDirs(destDir);
+    extract_archive(archive, destDir);
 }
 
 }
diff --git a/src/libutil/tarfile.hh b/src/libutil/tarfile.hh
index ce0911e2a..89a024f1d 100644
--- a/src/libutil/tarfile.hh
+++ b/src/libutil/tarfile.hh
@@ -4,7 +4,6 @@ namespace nix {
 
 void unpackTarfile(Source & source, const Path & destDir);
 
-void unpackTarfile(const Path & tarFile, const Path & destDir,
-    std::optional<std::string> baseName = {});
+void unpackTarfile(const Path & tarFile, const Path & destDir);
 
 }