Allow 6 digit dossier number in url parsing and add test

src/iiif/parsing.py

@@ -89,10 +89,10 @@ def get_info_from_iiif_url(iiif_url, source_file):
         }
         if source == "edepot":  # aka pre-wabo
             # ST-00015-ST00000126_00001.jpg=relevant_url_part  ST=stadsdeel  00015=dossier  ST00000126=document_barcode  00001=file/bestand
-            # SQ1452-SQ-01452%20(2)-SQ10079651_00001.jpg=relevant_url_part  SQ=stadsdeel  01425=dossier  SQ100796511=document_barcode  00001=file/bestand
+            # SQ1452-SQ-01452%20(2)-SQ10079651_00001.jpg=relevant_url_part  SQ=stadsdeel  01452=dossier  SQ10079651=document_barcode  00001=file/bestand
             try:
                 stadsdeel, dossier, document_barcode, file = re.match(
-                    r"^.*?([A-Z]{2})-(\d{5}).*?-(.+)_(.*?)\.\w+$", relevant_url_part
+                    r"^.*?([A-Z]{2})-(\d*).*?-(.+)_(.*?)\.\w+$", relevant_url_part
                 ).groups()
             except Exception as e:
                 raise InvalidIIIFUrlError(

tests/test_settings.py

@@ -1,7 +1,9 @@
 from django.conf import settings
 
-PRE_WABO_IMG_URL_BASE = "2/edepot:ST-00015-ST00000126_00001.jpg/"
+EDEPOT_PREFIX = "2/edepot:"
+WABO_PREFIX = "2/wabo:"
 
+PRE_WABO_IMG_URL_BASE = EDEPOT_PREFIX + "ST-00015-ST00000126_00001.jpg/"
 PRE_WABO_INFO_JSON_URL = PRE_WABO_IMG_URL_BASE + "info.json"
 
 PRE_WABO_IMG_URL_WITH_SCALING = PRE_WABO_IMG_URL_BASE + "full/50,50/0/default.jpg"
@@ -12,19 +14,21 @@
     PRE_WABO_IMG_URL_BASE + "10000,10000,48,48/full/0/default.jpg"
 )
 
-
 PRE_WABO_IMG_URL_SOURCE_FILE = (
-    "2/edepot:ST-00015-ST00000126_00001.jpg/?source_file=true&"
+    EDEPOT_PREFIX + "ST-00015-ST00000126_00001.jpg/?source_file=true&"
 )
 PRE_WABO_IMG_URL_NO_SCALING = (
-    "2/edepot:ST-00015-ST00000126_00001.jpg/full/full/0/default.jpg"
+    EDEPOT_PREFIX + "ST-00015-ST00000126_00001.jpg/full/full/0/default.jpg"
 )
 
+PRE_WABO_IMG_URL_WITH_EXTRA_DOSSIER_DIGIT = (
+    EDEPOT_PREFIX + "SA-100732-SA00509506_00003.jpg/info.json"
+)
 PRE_WABO_IMG_URL_WITH_EXTRA_REFERENCE = (
-    "2/edepot:SQ1452-SQ-01452%20(2)-SQ10079651_00001.jpg/full/full/0/default.jpg"
+    EDEPOT_PREFIX + "SQ1452-SQ-01452%20(2)-SQ10079651_00001.jpg/full/full/0/default.jpg"
 )
 
-WABO_IMG_URL = "2/wabo:SDZ-38657-4900487_628547/full/1000,900/0/default.jpg"
+WABO_IMG_URL = WABO_PREFIX + "SDZ-38657-4900487_628547/full/1000,900/0/default.jpg"
 
 with open("test-images/test-image-96x85.jpg", "rb") as file:
     IMAGE_BINARY_DATA = file.read()

tests/test_utils.py

@@ -17,6 +17,7 @@
 from main.utils import ImmediateHttpResponse
 from tests.test_settings import (
     PRE_WABO_IMG_URL_NO_SCALING,
+    PRE_WABO_IMG_URL_WITH_EXTRA_DOSSIER_DIGIT,
     PRE_WABO_IMG_URL_WITH_EXTRA_REFERENCE,
     PRE_WABO_IMG_URL_WITH_REGION,
     PRE_WABO_IMG_URL_WITH_SCALING,
@@ -49,6 +50,40 @@ def test_get_info_json_from_pre_wabo_url(self):
         assert url_info["formatting"] is None
         assert url_info["info_json"] is True
 
+    def test_get_info_json_from_pre_wabo_url_with_extra_digit(self):
+        """2/edepot:SA-100732-SA00509506_00003.jpg/"""
+        url_info = get_info_from_iiif_url(
+            PRE_WABO_IMG_URL_WITH_EXTRA_DOSSIER_DIGIT, False
+        )
+        assert url_info["source"] == "edepot"
+        assert url_info["stadsdeel"] == "SA"
+        assert url_info["dossier"] == "100732"
+        assert url_info["document_barcode"] == "SA00509506"
+        assert url_info["file"] == "00003"
+        assert url_info["region"] is None
+        assert url_info["scaling"] is None
+        assert url_info["source_filename"] == "SA/100732/SA00509506_00003.jpg"
+        assert url_info["filename"] == "SA-100732-SA00509506_00003.jpg"
+        assert url_info["formatting"] is None
+        assert url_info["info_json"] is True
+
+    def test_get_info_json_from_pre_wabo_url_with_extra_reference(self):
+        """2/edepot:SA-100732-SA00509506_00003.jpg/"""
+        url_info = get_info_from_iiif_url(PRE_WABO_IMG_URL_WITH_EXTRA_REFERENCE, False)
+        assert url_info["source"] == "edepot"
+        assert url_info["stadsdeel"] == "SQ"
+        assert url_info["dossier"] == "01452"
+        assert url_info["document_barcode"] == "SQ10079651"
+        assert url_info["file"] == "00001"
+        assert url_info["region"] == "full"
+        assert url_info["scaling"] == "full"
+        assert (
+            url_info["source_filename"] == "SQ1452/SQ/01452%20(2)-SQ10079651_00001.jpg"
+        )
+        assert url_info["filename"] == "SQ1452-SQ-01452%20(2)-SQ10079651_00001.jpg"
+        assert url_info["formatting"] == "full/full/0/default.jpg"
+        assert url_info["info_json"] is False
+
     def test_get_info_from_pre_wabo_url_vanilla(self):
         url_info = get_info_from_iiif_url(PRE_WABO_IMG_URL_WITH_SCALING, False)
         assert url_info["source"] == "edepot"