Merge pull request #14732 from NixOS/optimize-nar-cat

nix nar {ls,cat}: Optimize, make nix nar cat work on pipes too

src/libutil/file-system.cc

@@ -260,8 +260,7 @@ std::string readFile(const Path & path)
     AutoCloseFD fd = toDescriptor(open(
         path.c_str(),
         O_RDONLY
-// TODO
-#ifndef _WIN32
+#ifdef O_CLOEXEC
             | O_CLOEXEC
 #endif
         ));
@@ -294,8 +293,7 @@ void readFile(const Path & path, Sink & sink, bool memory_map)
     AutoCloseFD fd = toDescriptor(open(
         path.c_str(),
         O_RDONLY
-// TODO
-#ifndef _WIN32
+#ifdef O_CLOEXEC
             | O_CLOEXEC
 #endif
         ));
@@ -309,8 +307,7 @@ void writeFile(const Path & path, std::string_view s, mode_t mode, FsSync sync)
     AutoCloseFD fd = toDescriptor(open(
         path.c_str(),
         O_WRONLY | O_TRUNC | O_CREAT
-// TODO
-#ifndef _WIN32
+#ifdef O_CLOEXEC
             | O_CLOEXEC
 #endif
         ,
@@ -344,8 +341,7 @@ void writeFile(const Path & path, Source & source, mode_t mode, FsSync sync)
     AutoCloseFD fd = toDescriptor(open(
         path.c_str(),
         O_WRONLY | O_TRUNC | O_CREAT
-// TODO
-#ifndef _WIN32
+#ifdef O_CLOEXEC
             | O_CLOEXEC
 #endif
         ,

src/libutil/include/nix/util/fs-sink.hh

@@ -12,7 +12,7 @@ namespace nix {
  *
  * See `FileSystemObjectSink::createRegularFile`.
  */
-struct CreateRegularFileSink : Sink
+struct CreateRegularFileSink : virtual Sink
 {
     /**
      * If set to true, the sink will not be called with the contents

src/libutil/include/nix/util/nar-accessor.hh

@@ -32,8 +32,16 @@ using GetNarBytes = std::function<std::string(uint64_t, uint64_t)>;
  */
 GetNarBytes seekableGetNarBytes(const Path & path);
 
+GetNarBytes seekableGetNarBytes(Descriptor fd);
+
 ref<SourceAccessor> makeLazyNarAccessor(const nlohmann::json & listing, GetNarBytes getNarBytes);
 
+/**
+ * Creates a NAR accessor from a given stream and a GetNarBytes getter.
+ * @param source Consumed eagerly. References to it are not persisted in the resulting SourceAccessor.
+ */
+ref<SourceAccessor> makeLazyNarAccessor(Source & source, GetNarBytes getNarBytes);
+
 struct NarListingRegularFile
 {
     /**

src/libutil/nar-accessor.cc

@@ -111,6 +111,7 @@ struct NarAccessor : public SourceAccessor
                 path,
                 NarMember{.stat = {.type = Type::tRegular, .fileSize = 0, .isExecutable = false, .narOffset = 0}});
             NarMemberConstructor nmc{nm, pos};
+            nmc.skipContents = true; /* Don't care about contents. */
             func(nmc);
         }
 
@@ -141,6 +142,13 @@ struct NarAccessor : public SourceAccessor
         parseDump(indexer, indexer);
     }
 
+    NarAccessor(Source & source, GetNarBytes getNarBytes)
+        : getNarBytes(std::move(getNarBytes))
+    {
+        NarIndexer indexer(*this, source);
+        parseDump(indexer, indexer);
+    }
+
     NarAccessor(const nlohmann::json & listing, GetNarBytes getNarBytes)
         : getNarBytes(getNarBytes)
     {
@@ -249,24 +257,35 @@ ref<SourceAccessor> makeLazyNarAccessor(const nlohmann::json & listing, GetNarBy
     return make_ref<NarAccessor>(listing, getNarBytes);
 }
 
+ref<SourceAccessor> makeLazyNarAccessor(Source & source, GetNarBytes getNarBytes)
+{
+    return make_ref<NarAccessor>(source, getNarBytes);
+}
+
 GetNarBytes seekableGetNarBytes(const Path & path)
 {
-    return [path](uint64_t offset, uint64_t length) {
-        AutoCloseFD fd = toDescriptor(open(
-            path.c_str(),
-            O_RDONLY
-#ifndef _WIN32
-                | O_CLOEXEC
+    AutoCloseFD fd = toDescriptor(open(
+        path.c_str(),
+        O_RDONLY
+#ifdef O_CLOEXEC
+            | O_CLOEXEC
 #endif
-            ));
-        if (!fd)
-            throw SysError("opening NAR cache file '%s'", path);
+        ));
+    if (!fd)
+        throw SysError("opening NAR cache file '%s'", path);
 
-        if (lseek(fromDescriptorReadOnly(fd.get()), offset, SEEK_SET) != (off_t) offset)
-            throw SysError("seeking in '%s'", path);
+    return [inner = seekableGetNarBytes(fd.get()), fd = make_ref<AutoCloseFD>(std::move(fd))](
+               uint64_t offset, uint64_t length) { return inner(offset, length); };
+}
+
+GetNarBytes seekableGetNarBytes(Descriptor fd)
+{
+    return [fd](uint64_t offset, uint64_t length) {
+        if (::lseek(fromDescriptorReadOnly(fd), offset, SEEK_SET) == -1)
+            throw SysError("seeking in file");
 
         std::string buf(length, 0);
-        readFull(fd.get(), buf.data(), length);
+        readFull(fd, buf.data(), length);
 
         return buf;
     };

src/nix/cat.cc

@@ -1,5 +1,6 @@
 #include "nix/cmd/command.hh"
 #include "nix/store/store-api.hh"
+#include "nix/util/archive.hh"
 #include "nix/util/nar-accessor.hh"
 #include "nix/util/serialise.hh"
 #include "nix/util/source-accessor.hh"
@@ -79,9 +80,40 @@ struct CmdCatNar : StoreCommand, MixCat
         if (!fd)
             throw SysError("opening NAR file '%s'", narPath);
         auto source = FdSource{fd.get()};
-        auto narAccessor = makeNarAccessor(source);
-        nlohmann::json listing = listNarDeep(*narAccessor, CanonPath::root);
-        cat(makeLazyNarAccessor(listing, seekableGetNarBytes(narPath)), CanonPath{path});
+
+        struct CatRegularFileSink : NullFileSystemObjectSink
+        {
+            CanonPath neededPath = CanonPath::root;
+            bool found = false;
+
+            void createRegularFile(const CanonPath & path, std::function<void(CreateRegularFileSink &)> crf) override
+            {
+                struct : CreateRegularFileSink, FdSink
+                {
+                    void isExecutable() override {}
+                } crfSink;
+
+                crfSink.fd = INVALID_DESCRIPTOR;
+
+                if (path == neededPath) {
+                    logger->stop();
+                    crfSink.skipContents = false;
+                    crfSink.fd = STDOUT_FILENO;
+                    found = true;
+                } else {
+                    crfSink.skipContents = true;
+                }
+
+                crf(crfSink);
+            }
+        } sink;
+
+        sink.neededPath = CanonPath(path);
+        /* NOTE: We still parse the whole file to validate that it's a correct NAR. */
+        parseDump(sink, source);
+
+        if (!sink.found)
+            throw Error("NAR does not contain regular file '%1%'", path);
     }
 };
 

src/nix/ls.cc

@@ -154,9 +154,7 @@ struct CmdLsNar : Command, MixLs
         if (!fd)
             throw SysError("opening NAR file '%s'", narPath);
         auto source = FdSource{fd.get()};
-        auto narAccessor = makeNarAccessor(source);
-        nlohmann::json listing = listNarDeep(*narAccessor, CanonPath::root);
-        list(makeLazyNarAccessor(listing, seekableGetNarBytes(narPath)), CanonPath{path});
+        list(makeLazyNarAccessor(source, seekableGetNarBytes(fd.get())), CanonPath{path});
     }
 };
 

tests/functional/nar-access.sh

@@ -23,6 +23,8 @@ diff -u data.cat-nar "$storePath/foo/data"
 # Check that file contents of baz match.
 nix nar cat "$narFile" /foo/baz > baz.cat-nar
 diff -u baz.cat-nar "$storePath/foo/baz"
+nix nar cat /dev/stdin /foo/baz < "$narFile" > baz.cat-nar-pipe
+expect 1 nix nar cat "$narFile" /foo/baz/doesntexist 2>&1 | grep "NAR does not contain regular file '/foo/baz/doesntexist'"
 
 nix store cat "$storePath/foo/baz" > baz.cat-nar
 diff -u baz.cat-nar "$storePath/foo/baz"