From bb1bd62250822cacc694236a686e66b228c438d2 Mon Sep 17 00:00:00 2001
From: crunk <dvanderkleij@gmail.com>
Date: Mon, 13 May 2024 21:02:23 +0200
Subject: [PATCH] file crawler for annotations,descriptions,tags

---
 verse/file_crawler.py | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/verse/file_crawler.py b/verse/file_crawler.py
index 8ba8024..9e5456b 100644
--- a/verse/file_crawler.py
+++ b/verse/file_crawler.py
@@ -1 +1,22 @@
+import os
+
+import magic
+from distribusi.mappings import CODE_TYPES, FILE_TYPES, SUB_TYPES
+
 from models.distribusi_file_model import DistribusiFiles
+
+MIME_TYPE = magic.Magic(mime=True)
+
+
+def distribusi_file_with_type(full_path):
+    mime = MIME_TYPE.from_file(full_path)
+    type_, subtype = mime.split("/")
+    if type_ in FILE_TYPES:
+        print(f"distribusi file:{full_path}  type:{type_}")
+
+
+for root, dirs, files in os.walk("stash", topdown=True):
+    files = list(filter(lambda f: not f.startswith("."), files))
+    for file in files:
+        full_path = os.path.join(root, file)
+        distribusi_file_with_type(full_path)