From e0322e10016725c59d90fe1e2ce0f51a605f983f Mon Sep 17 00:00:00 2001
From: Anders Lorentsen <anders.lorentsen@gmail.com>
Date: Mon, 27 Jan 2025 09:04:01 +0100
Subject: [PATCH] import registyr repath-script to this repo, with added
 comment

---
 repath.py | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)
 create mode 100644 repath.py

diff --git a/repath.py b/repath.py
new file mode 100644
index 0000000..caec700
--- /dev/null
+++ b/repath.py
@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+
+"""
+Small script to change the pointed-to-data paths inside the regsitry files
+to match the new setup on gtweb-02. Inside the container image, the CWB files
+are mounted as /corpora/registry/ and /corpora/data/
+
+The script may need adjusting for the various places it is run. The paths are
+different on local machines, as opposed to servers.
+"""
+
+# the lines we want to change are like this:
+# HOME /corpora/gt_cwb/data/xxx_category_yyyymmdd
+
+import re
+from pathlib import Path
+
+cwd = Path(".").resolve()
+
+for path in Path("registry").iterdir():
+    contents = path.read_text()
+    name = path.name
+    contents = re.sub("HOME .*", f"HOME /corpora/data/{name}", contents)
+    contents = re.sub("INFO .*", f"INFO /corpora/data/{name}/.info", contents)
+    path.write_text(contents)
+
+print("done")