Changeset 34952

Timestamp:

2021-03-01T23:47:30+13:00 (3 years ago)

Author:

davidb

Message:

Country-label to URI lookup introduced

File:

• main/trunk/model-sites-dev/eurovision-lod/collect/eurovision/prepare/voting-excel/xlsx-tocountry-jsonmetadata.py (modified) (7 diffs)

main/trunk/model-sites-dev/eurovision-lod/collect/eurovision/prepare/voting-excel/xlsx-tocountry-jsonmetadata.py

@@ -12 +12 @@
 
 import util
+import csvutil
 import xlsxutil
 
@@ -67 +68 @@
 
 
-def fileset_voting_for_esc_country_in_year(to_country_year_votes):
+def fileset_voting_for_esc_country_in_year(to_country_year_votes,countrylabel_to_uri_mapping):
 
     # Looking to build data-structure (for output as JSON) in the form
@@ -104 +105 @@
     tl_to_country_id = re.sub(r'\s+', '', tl_to_country) + str(tl_year)
 
-    metadata_array.append({ "name": "Final","content": tl_final })
-        
+    metadata_array.append({ "name": "Country","content": tl_to_country })
+    metadata_array.append({ "name": "Year",   "content": tl_year })
+    metadata_array.append({ "name": "Final",  "content": tl_final })
+
+    # tl_to_country_lc = tl_to_country.lower()
+    tl_to_country_uri = countrylabel_to_uri_mapping.get(tl_to_country)
+    if (tl_to_country_uri != None):
+        metadata_array.append({ "name": "CountryDBURI","content": tl_to_country_uri })
+    else:
+        util.eprint("Warning: Failed to find DBpedia Country URI match for: " + tl_to_country)
+                
     for to_country_year_vote in to_country_year_votes:
         to_country   = to_country_year_vote.get('To country')
@@ -150 +160 @@
 
 
-def gs_directory_metadata(to_country_year_voting_groups):
+def gs_directory_metadata(to_country_year_voting_groups,countrylabel_to_uri_mapping):
     # Next step is to express the grouped to-country voting data
     # in the Greenstone JSON metadata format:
@@ -178 +188 @@
     for to_country_year_votes in to_country_year_voting_groups:
         
-        fileset = fileset_voting_for_esc_country_in_year(to_country_year_votes)
+        fileset = fileset_voting_for_esc_country_in_year(to_country_year_votes,countrylabel_to_uri_mapping)
         directory_metadata.append(fileset)
 
@@ -229 +239 @@
     if (json_output_filename == None):
         json_output_filename = os.path.splitext(excel_input_filename)[0]+'.json'
-        
+
+    countrylabel_to_uri_mapping = csvutil.load_and_tidy_csv("dbpedia--european-countries.csv")
+    
     worksheet = xlsxutil.load_xslx_sheet(excel_input_filename,sheetname)
 
@@ -261 +273 @@
     print()
     print("Generating Greenstone JSON to-country voting metadata received by:")
-    greenstone_metadata_json = gs_directory_metadata(to_country_year_voting_groups)
+    greenstone_metadata_json = gs_directory_metadata(to_country_year_voting_groups,countrylabel_to_uri_mapping)
     
     print("Saving output as: " + json_output_filename)