Identifiant de la source (traçabilité)

etalab · Dec 20, 2024 · 87bcf13 · 87bcf13
1 parent 2d53a9a
commit 87bcf13
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 18 deletions.
diff --git a/apps/transport/lib/registry/engine.ex b/apps/transport/lib/registry/engine.ex
@@ -5,6 +5,7 @@ defmodule Transport.Registry.Engine do
 
   alias Transport.Registry.GTFS
   alias Transport.Registry.NeTEx
+  alias Transport.Registry.Model.DataSource
   alias Transport.Registry.Model.Stop
   alias Transport.Registry.Result
 
@@ -45,15 +46,17 @@ defmodule Transport.Registry.Engine do
   end
 
   def prepare_extractor(%DB.Resource{} = resource) do
+    data_source_id = "PAN:resource:#{resource.id}"
+
     case resource.format do
-      "GTFS" -> {:ok, {GTFS, resource.url}}
-      "NeTEx" -> {:ok, {NeTEx, resource.url}}
+      "GTFS" -> {:ok, {GTFS, data_source_id, resource.url}}
+      "NeTEx" -> {:ok, {NeTEx, data_source_id, resource.url}}
       _ -> {:error, "Unsupported format"}
     end
   end
 
-  def download({extractor, url}) do
-    Logger.debug("download #{extractor} #{url}")
+  def download({extractor, data_source_id, url}) do
+    Logger.debug("download #{extractor} #{data_source_id} #{url}")
     tmp_path = System.tmp_dir!() |> Path.join("#{Ecto.UUID.generate()}.dat")
 
     safe_error = fn msg ->
@@ -75,7 +78,7 @@ defmodule Transport.Registry.Engine do
       {:ok, %{status: status}} ->
         cond do
           status >= 200 && status < 300 ->
-            {:ok, {extractor, tmp_path}}
+            {:ok, {extractor, data_source_id, tmp_path}}
 
           status > 400 ->
             safe_error.("Error #{status} while downloading the resource from #{url}")
@@ -86,10 +89,10 @@ defmodule Transport.Registry.Engine do
     end
   end
 
-  @spec extract_from_archive({module(), Path.t()}) :: Result.t([Stop.t()])
-  def extract_from_archive({extractor, file}) do
-    Logger.debug("extract_from_archive #{extractor} #{file}")
-    extractor.extract_from_archive(file)
+  @spec extract_from_archive({module(), DataSource.data_source_id(), Path.t()}) :: Result.t([Stop.t()])
+  def extract_from_archive({extractor, data_source_id, file}) do
+    Logger.debug("extract_from_archive #{extractor} #{data_source_id} #{file}")
+    extractor.extract_from_archive(data_source_id, file)
   end
 
   def dump_to_csv(enumerable, output_file) do

diff --git a/apps/transport/lib/registry/extractor.ex b/apps/transport/lib/registry/extractor.ex
@@ -5,8 +5,10 @@ defmodule Transport.Registry.Extractor do
 
   require Logger
 
+  alias Transport.Registry.Model.DataSource
   alias Transport.Registry.Model.Stop
   alias Transport.Registry.Result
 
-  @callback extract_from_archive(path :: Path.t()) :: Result.t([Stop.t()])
+  @callback extract_from_archive(data_source_id :: DataSource.data_source_id(), path :: Path.t()) ::
+              Result.t([Stop.t()])
 end
diff --git a/apps/transport/lib/registry/gtfs.ex b/apps/transport/lib/registry/gtfs.ex
@@ -15,7 +15,7 @@ defmodule Transport.Registry.GTFS do
   @doc """
   Extract stops from GTFS ressource.
   """
-  def extract_from_archive(archive) do
+  def extract_from_archive(data_source_id, archive) do
     case file_stream(archive) do
       {:error, error} ->
         Logger.error(error)
@@ -26,13 +26,13 @@ defmodule Transport.Registry.GTFS do
 
         content
         |> Utils.to_stream_of_maps()
-        |> Stream.flat_map(&handle_stop/1)
+        |> Stream.flat_map(&handle_stop(data_source_id, &1))
         |> Enum.to_list()
         |> Result.ok()
     end
   end
 
-  defp handle_stop(record) do
+  defp handle_stop(data_source_id, record) do
     latitude = Utils.fetch_position(record, "stop_lat")
     longitude = Utils.fetch_position(record, "stop_lon")
 
@@ -44,7 +44,9 @@ defmodule Transport.Registry.GTFS do
           latitude: latitude,
           longitude: longitude,
           projection: :utm_wgs84,
-          stop_type: record |> Utils.csv_get_with_default("location_type", "0") |> to_stop_type()
+          stop_type: record |> Utils.csv_get_with_default("location_type", "0") |> to_stop_type(),
+          data_source_format: :gtfs,
+          data_source_id: data_source_id
         }
       ]
     else

diff --git a/apps/transport/lib/registry/netex.ex b/apps/transport/lib/registry/netex.ex
@@ -13,10 +13,7 @@ defmodule Transport.Registry.NeTEx do
   @doc """
   Extract stops from a NeTEx archive.
   """
-  def extract_from_archive(archive) do
-    # FIXME: propagate some context
-    data_source_id = nil
-
+  def extract_from_archive(data_source_id, archive) do
     archive
     |> Transport.NeTEx.read_all_stop_places()
     |> Enum.flat_map(&process_stop_places(data_source_id, &1))