Changeset 38704

Timestamp:

2024-02-06T14:04:44+13:00 (4 months ago)

Author:

davidb

Message:

filename to regex safe function added

File:

• gs3-installations/intermuse/trunk/sites/intermuse/collect/ephemeral-heterogeneous/prepare/pagedimagemd.py (modified) (3 diffs)

gs3-installations/intermuse/trunk/sites/intermuse/collect/ephemeral-heterogeneous/prepare/pagedimagemd.py

@@ -7 +7 @@
 MetadataDirectoryRules = {
     'RCM BMS chapters programmes and congress'  : {},
-    'YorkHMS BMS Borthwick IIIF/BMS'          : {
+    'York|HMS BMS Borthwick IIIF/BMS'          : {
         "type"       : "Programme",
         "publisher"  : "The British Music Society of York",
         "file-regex" : r'^BMS_\d+.*\.pdf$'
     },
-    'YorkHMS BMS Borthwick IIIF/HMS'          : {
+    'York|HMS BMS Borthwick IIIF/HMS'          : {
         "type": "Programme",
         "publisher" : "The Huddersfield Music Society",
@@ -34 +34 @@
     'New Zealand'                               : {},
     'York BMS 1921 single programme'            : {},
-    'BMS AnnualCat. 1920 - Bodleian copy'     : {},
+    'BMS Annual|Cat. 1920 - Bodleian copy'     : {},
     'Bradford BMS'                              : {},
     'Birmingham BMS'                            : {},
-    'BMS AnnualCat. 1922 - Lewis Foreman copy': {},
+    'BMS Annual|Cat. 1922 - Lewis Foreman copy': {},
     'Blackpool BMS'                             : {}
 }
+
+
+def make_filename_re_safe(filename):
+    
+    safe1  = re.sub(r"\.","\\.",filename) # . safe
+    safe2  = re.sub(r"\|","\\|",safe1)    # | safe
+    safe3  = re.sub(r"\(","\\(",safe2)    # (
+    safe4  = re.sub(r"\)","\\)",safe3)    # )
+
+    final_safe = "^"+safe4+"$"
+
+    return final_safe
 
 
@@ -77 +89 @@
         )
 
-        pdf_filename_re = "^"+re.sub(r"\.","\\.",pdf_filename)+"$"
+        pdf_filename_re = make_filename_re_safe(pdf_filename)
         
         metadata_fileset = { "FileSet": { "FileName": f"{pdf_filename_re}", "Descriptions": metadata_description_list } }