From e18ec6c1e98273b23eed6733c9905529d0ee371d Mon Sep 17 00:00:00 2001
From: Jarno Bakker <jarno.bakker@di.huc.knaw.nl>
Date: Tue, 28 Jan 2025 13:39:10 +0100
Subject: [PATCH] feat: allow tweaking vocabularies when importing

---
 entrypoint.py       |  3 +++
 src/graphdb.py      | 10 +++++++---
 src/vocabularies.py |  4 ++--
 3 files changed, 12 insertions(+), 5 deletions(-)

diff --git a/entrypoint.py b/entrypoint.py
index d36436c..4713387 100755
--- a/entrypoint.py
+++ b/entrypoint.py
@@ -75,6 +75,9 @@ def main() -> None:
             if reload:
                 print(f"Loading vocabulary {vocab}")
                 load_vocabulary(vocab_config['source'], data, graph)
+                if "tweaks" in vocab_config:
+                    print(f"Tweaks found for {vocab}. Loading")
+                    load_vocabulary(vocab_config['tweaks'], data, graph, True)
                 if graph in loaded_vocabs:
                     update_timestamp(graph, int(time.time()))
                 else:
diff --git a/src/graphdb.py b/src/graphdb.py
index 02cb5d7..b8657e1 100644
--- a/src/graphdb.py
+++ b/src/graphdb.py
@@ -2,6 +2,7 @@
 This file contains functions for interacting with GraphDB
 """
 import os
+import urllib
 from typing import TextIO
 
 import requests
@@ -124,12 +125,13 @@ def get_type(extension: str) -> str:
     return "application/x-turtle"
 
 
-def add_vocabulary(graph: TextIO, graph_name: str, extension: str) -> None:
+def add_vocabulary(graph: TextIO, graph_name: str, extension: str, append: bool = False) -> None:
     """
     Add a vocabulary to GraphDB
     :param graph:       File
     :param graph_name:  String representing the name of the graph
-    :param extension:    String representing the extension
+    :param extension:   String representing the extension
+    :param append:      Append data instead of replacing
     :return:
     """
     print(f"Adding vocabulary {graph_name}")
@@ -142,7 +144,9 @@ def add_vocabulary(graph: TextIO, graph_name: str, extension: str) -> None:
     headers = {
         'Content-Type': get_type(extension),
     }
-    response = requests.put(
+    method = requests.post if append else requests.put
+
+    response = method(
         f"{endpoint}/statements",
         data=content,
         headers=headers,
diff --git a/src/vocabularies.py b/src/vocabularies.py
index 179fba5..e330f54 100644
--- a/src/vocabularies.py
+++ b/src/vocabularies.py
@@ -97,7 +97,7 @@ def get_file_from_config(config_data: dict, data_dir: str) -> TextIO:
     raise InvalidConfigurationException("Unknown type")
 
 
-def load_vocabulary(source_data: dict, data_dir: str, graph_name: str) -> None:
+def load_vocabulary(source_data: dict, data_dir: str, graph_name: str, append: bool = False) -> None:
     """
     Load a vocabulary using the source data from the yaml.
     :param source_data:
@@ -106,7 +106,7 @@ def load_vocabulary(source_data: dict, data_dir: str, graph_name: str) -> None:
     :return:
     """
     with get_file_from_config(source_data, data_dir) as vocab_file:
-        add_vocabulary(vocab_file, graph_name, get_vocab_format(source_data))
+        add_vocabulary(vocab_file, graph_name, get_vocab_format(source_data), append)
 
 
 def get_graph(fp: IO) -> str: