From 5f9465614bd1263a3a979e068cd640bf47a85347 Mon Sep 17 00:00:00 2001
From: ruslandoga <doga.ruslan@gmail.com>
Date: Thu, 21 Mar 2024 18:35:42 +0800
Subject: [PATCH] Include domain and dates in zip archive filename (#3921)

* include domain and dates in zip archive filename

* adapt to comments
---
 lib/plausible/exports.ex        | 16 ++++++++++++++++
 lib/plausible/s3.ex             | 16 +++++++++++++---
 lib/plausible/sites.ex          |  8 ++++++++
 lib/workers/export_csv.ex       | 11 ++++++++++-
 test/plausible/exports_test.exs |  2 ++
 5 files changed, 49 insertions(+), 4 deletions(-)

diff --git a/lib/plausible/exports.ex b/lib/plausible/exports.ex
index 4f2094de3..dbabcd7ff 100644
--- a/lib/plausible/exports.ex
+++ b/lib/plausible/exports.ex
@@ -6,6 +6,22 @@ defmodule Plausible.Exports do
   require Plausible
   import Ecto.Query
 
+  @doc """
+  Renders filename for the Zip archive containing the exported CSV files.
+
+  Examples:
+
+      iex> archive_filename("plausible.io", ~D[2021-01-01], ~D[2024-12-31])
+      "plausible.io_20210101_20241231.zip"
+
+      iex> archive_filename("Bücher.example", ~D[2021-01-01], ~D[2024-12-31])
+      "Bücher.example_20210101_20241231.zip"
+
+  """
+  def archive_filename(domain, min_date, max_date) do
+    "#{domain}_#{Calendar.strftime(min_date, "%Y%m%d")}_#{Calendar.strftime(max_date, "%Y%m%d")}.zip"
+  end
+
   @doc """
   Builds Ecto queries to export data from `events_v2` and `sessions_v2`
   tables  into the format of `imported_*` tables for a website.
diff --git a/lib/plausible/s3.ex b/lib/plausible/s3.ex
index 2768ce9ef..12be32fc2 100644
--- a/lib/plausible/s3.ex
+++ b/lib/plausible/s3.ex
@@ -84,15 +84,25 @@ defmodule Plausible.S3 do
 
   In the current implementation the bucket always goes into the path component.
   """
-  @spec export_upload_multipart(Enumerable.t(), String.t(), Path.t(), keyword) ::
+  @spec export_upload_multipart(Enumerable.t(), String.t(), Path.t(), String.t(), keyword) ::
           :uri_string.uri_string()
-  def export_upload_multipart(stream, s3_bucket, s3_path, config_overrides \\ []) do
+  def export_upload_multipart(stream, s3_bucket, s3_path, filename, config_overrides \\ []) do
     config = ExAws.Config.new(:s3)
 
+    encoded_filename = URI.encode(filename)
+    disposition = ~s[attachment; filename="#{encoded_filename}"]
+
+    disposition =
+      if encoded_filename != filename do
+        disposition <> "; filename*=utf-8''#{encoded_filename}"
+      else
+        disposition
+      end
+
     # 5 MiB is the smallest chunk size AWS S3 supports
     chunk_into_parts(stream, 5 * 1024 * 1024)
     |> ExAws.S3.upload(s3_bucket, s3_path,
-      content_disposition: ~s|attachment; filename="Plausible.zip"|,
+      content_disposition: disposition,
       content_type: "application/zip"
     )
     |> ExAws.request!(config_overrides)
diff --git a/lib/plausible/sites.ex b/lib/plausible/sites.ex
index 12d4235ad..b41591e3f 100644
--- a/lib/plausible/sites.ex
+++ b/lib/plausible/sites.ex
@@ -23,6 +23,14 @@ defmodule Plausible.Sites do
     Repo.get_by!(Site, domain: domain)
   end
 
+  def get_domain!(site_id) do
+    Plausible.Repo.one!(
+      from s in Plausible.Site,
+        where: [id: ^site_id],
+        select: s.domain
+    )
+  end
+
   @spec toggle_pin(Auth.User.t(), Site.t()) ::
           {:ok, Site.UserPreference.t()} | {:error, :too_many_pins}
   def toggle_pin(user, site) do
diff --git a/lib/workers/export_csv.ex b/lib/workers/export_csv.ex
index 5b46d4fd5..be7d6d395 100644
--- a/lib/workers/export_csv.ex
+++ b/lib/workers/export_csv.ex
@@ -43,6 +43,10 @@ defmodule Plausible.Workers.ExportCSV do
         )
       )
     else
+      domain = Plausible.Sites.get_domain!(site_id)
+      export_archive_filename = Plausible.Exports.archive_filename(domain, min_date, max_date)
+      s3_config_overrides = s3_config_overrides(args)
+
       download_url =
         DBConnection.run(
           ch,
@@ -55,7 +59,12 @@ defmodule Plausible.Workers.ExportCSV do
               ),
               format: "CSVWithNames"
             )
-            |> Plausible.S3.export_upload_multipart(s3_bucket, s3_path, s3_config_overrides(args))
+            |> Plausible.S3.export_upload_multipart(
+              s3_bucket,
+              s3_path,
+              export_archive_filename,
+              s3_config_overrides
+            )
           end,
           timeout: :infinity
         )
diff --git a/test/plausible/exports_test.exs b/test/plausible/exports_test.exs
index 7f0f1e01f..24ac8fd1d 100644
--- a/test/plausible/exports_test.exs
+++ b/test/plausible/exports_test.exs
@@ -1,6 +1,8 @@
 defmodule Plausible.ExportsTest do
   use Plausible.DataCase, async: true
 
+  doctest Plausible.Exports, import: true
+
   # for e2e export->import tests please see Plausible.Imported.CSVImporterTest
 
   describe "export_queries/2" do