From 907b8955d158b57e4249d95f46fa894ab2148410 Mon Sep 17 00:00:00 2001
From: Ben Fairless <ben@oaf.org.au>
Date: Mon, 24 Aug 2026 11:30:51 +0800
Subject: [PATCH] Compress uploaded code with gzip

Wrap the tar upload in a Zlib::GzipWriter so scraper code is
gzip-compressed before being posted to the server, cutting upload time
and bandwidth. The gzip stream is finished (not closed) so the tempfile
handle stays open and rewound for the upload, matching the existing tar
trailer handling. The multipart content type becomes application/gzip.

Note: morph.io's ApiController#run_remote currently unpacks the upload
with Archive::Tar::Minitar.unpack on a plain tar, so it needs a
matching change to decompress before this can be released.

Resolves #1

Assisted-by: OpenCode:anthropic.claude-fable-5
Signed-off-by: Ben Fairless <ben@oaf.org.au>
---
 lib/morph-cli.rb       | 22 ++++++++++++----------
 spec/morph_cli_spec.rb | 24 ++++++++++++++++++------
 2 files changed, 30 insertions(+), 16 deletions(-)

diff --git a/lib/morph-cli.rb b/lib/morph-cli.rb
index ddadb4d..9fc0bd4 100644
--- a/lib/morph-cli.rb
+++ b/lib/morph-cli.rb
@@ -6,6 +6,7 @@
 require 'tempfile'
 require 'fileutils'
 require 'filesize'
+require 'zlib'
 require 'faraday'
 require 'faraday/multipart'
 require 'minitar'
@@ -42,7 +43,7 @@ def self.execute(directory, _development, env_config)
     connection.post("/run") do |req|
       req.body = {
         api_key: env_config[:api_key],
-        code: Faraday::Multipart::FilePart.new(file, "application/octet-stream")
+        code: Faraday::Multipart::FilePart.new(file, "application/gzip")
       }
       req.options.timeout = timeout
       req.options.on_data = proc do |chunk, _overall_received_bytes, env|
@@ -106,21 +107,22 @@ def self.in_directory(directory)
     FileUtils.cd(cwd)
   end
 
-  # Packs the given paths (relative to directory) into a tar file and returns
-  # an open, rewound file handle ready for upload.
+  # Packs the given paths (relative to directory) into a gzip-compressed tar
+  # file and returns an open, rewound file handle ready for upload.
   def self.create_tar(directory, paths)
-    tempfile = Tempfile.new(["morph", ".tar"])
+    tempfile = Tempfile.new(["morph", ".tar.gz"])
     tempfile.binmode
 
     in_directory(directory) do
-      output = Minitar::Output.new(tempfile)
-      paths.each do |entry|
-        Minitar.pack_file(entry, output)
-      end
+      gzip = Zlib::GzipWriter.new(tempfile)
+      output = Minitar::Output.new(gzip)
+      paths.each { |entry| Minitar.pack_file(entry, output) }
     ensure
-      # Writes the tar trailer and flushes without closing the underlying
-      # tempfile, so the returned handle stays open for reading.
+      # Closing the tar writer writes the tar trailer; finishing (not
+      # closing) the gzip stream writes the gzip trailer, leaving the
+      # underlying tempfile handle open for reading.
       output&.tar&.close
+      gzip&.finish
     end
 
     tempfile.flush
diff --git a/spec/morph_cli_spec.rb b/spec/morph_cli_spec.rb
index 66111de..79b25f5 100644
--- a/spec/morph_cli_spec.rb
+++ b/spec/morph_cli_spec.rb
@@ -30,7 +30,7 @@ def with_scraper_directory
       end
     end
 
-    it "posts the API key and the code as multipart form data" do
+    it "posts the API key and the gzipped code as multipart form data" do
       stub_request(:post, "https://morph.io/run").to_return(status: 200, body: "")
 
       with_scraper_directory do |dir|
@@ -41,7 +41,7 @@ def with_scraper_directory
       expect(WebMock).to(have_requested(:post, "https://morph.io/run").with do |req|
         req.headers["Content-Type"].start_with?("multipart/form-data") &&
           req.body.include?("secret-key") &&
-          req.body.include?("scraper.rb")
+          req.body.b.include?("\x1f\x8b".b) # gzip magic bytes
       end)
     end
 
@@ -122,7 +122,7 @@ def with_scraper_directory
   end
 
   describe ".create_tar" do
-    it "packs the given paths into a readable tar" do
+    it "packs the given paths into a readable gzip-compressed tar" do
       Dir.mktmpdir do |dir|
         File.write(File.join(dir, "scraper.rb"), "puts 'hi'\n")
         FileUtils.mkdir_p(File.join(dir, "lib"))
@@ -132,13 +132,25 @@ def with_scraper_directory
         tar = described_class.create_tar(dir, paths)
 
         names = []
-        Minitar::Input.open(tar.path) do |input|
-          input.each { |entry| names << entry.full_name }
+        Zlib::GzipReader.open(tar.path) do |gzip|
+          Minitar::Input.open(gzip) do |input|
+            input.each { |entry| names << entry.full_name }
+          end
         end
         expect(names).to contain_exactly("scraper.rb", "lib/helper.rb")
       end
     end
 
+    it "compresses the tar" do
+      Dir.mktmpdir do |dir|
+        File.write(File.join(dir, "scraper.rb"), "a" * 100_000)
+
+        tar = described_class.create_tar(dir, described_class.all_paths(dir))
+
+        expect(File.size(tar.path)).to be < 100_000
+      end
+    end
+
     it "returns an open file handle ready for reading" do
       Dir.mktmpdir do |dir|
         File.write(File.join(dir, "scraper.rb"), "puts 'hi'\n")
@@ -147,7 +159,7 @@ def with_scraper_directory
 
         expect(tar).not_to be_closed
         expect(tar.pos).to eq(0)
-        expect(tar.read).to include("scraper.rb")
+        expect(tar.read(2)).to eq("\x1f\x8b".b) # gzip magic bytes
       end
     end
   end