From 3737dd29e690caec67622fcefb3bc434381aff8b Mon Sep 17 00:00:00 2001
From: Quentin Guilloteau <Quentin.Guilloteau@unibas.ch>
Date: Thu, 11 Jul 2024 13:37:07 +0200
Subject: [PATCH] add start of workflow to extract non blacklisted artifacts

---
 blacklist.csv                     |  1 +
 workflow/1_artifacts_to_build.smk | 28 ++++++++++++++++++++++++++++
 2 files changed, 29 insertions(+)
 create mode 100644 blacklist.csv
 create mode 100644 workflow/1_artifacts_to_build.smk

diff --git a/blacklist.csv b/blacklist.csv
new file mode 100644
index 0000000..442168d
--- /dev/null
+++ b/blacklist.csv
@@ -0,0 +1 @@
+artifacts/sc24_test.yaml, IMAGE_NOT_FOUND, 0
diff --git a/workflow/1_artifacts_to_build.smk b/workflow/1_artifacts_to_build.smk
new file mode 100644
index 0000000..d62505d
--- /dev/null
+++ b/workflow/1_artifacts_to_build.smk
@@ -0,0 +1,28 @@
+import csv
+import os
+
+ARTIFACTS_FOLDER = "artifacts"
+BLACKLIST = "blacklist.csv"
+
+def get_blacklisted_paths(blacklist_csv_path):
+  blacklisted = set()
+  with open(blacklist_csv_path, "r") as csv_file:
+    spamreader = csv.reader(csv_file, delimiter=",")
+    for row in spamreader:
+      blacklisted.add(row[0])
+  return blacklisted
+
+def get_artifacts_to_build(artifacts_folder, blacklist_csv_path):
+  blacklisted = get_blacklisted_paths(blacklist_csv_path)
+  all_artifacts = set([os.path.join(artifacts_folder, a) for a in os.listdir(artifacts_folder) if not os.path.isdir(os.path.join(artifacts_folder, a))])
+  return list(all_artifacts.difference(blacklisted))
+
+rule all:
+  input:
+    BLACKLIST,
+  output:
+    "all.csv",
+  params:
+    artifacts = get_artifacts_to_build(ARTIFACTS_FOLDER, BLACKLIST),
+  shell:
+    "echo {params.artifacts} > {output}"