Changeset 35878

Timestamp:

2022-01-04T17:45:39+13:00 (2 years ago)

Author:

davidb

Message:

Further development of script. Now at the point of printing out what the desired copy commands are

File:

• main/trunk/model-sites-dev/eurovision-lod/collect/eurovision/prepare/08b-COPY-AUDIO-FEATURES-TO-IMPORT.py (modified) (7 diffs)

main/trunk/model-sites-dev/eurovision-lod/collect/eurovision/prepare/08b-COPY-AUDIO-FEATURES-TO-IMPORT.py

@@ -3 +3 @@
 import os
 import re
+import shutil
 
 prep_dir = "essentia-audio-features"
@@ -29 +30 @@
      audio_year_dir = os.path.join(audio_dir,year)
 
-     json_files = [os.path.join(year,f) for f in os.listdir(audio_year_dir) if re.search(r"\.json$",f)]
-
-     all_af_json_files += json_files
-
+     for f in os.listdir(audio_year_dir):
+          if re.search(r"\.json$",f):
+               year_entrant = os.path.join(year,f)
+               all_af_json_files.append(year_entrant)
+          
 
 #print("json files = " + repr(all_af_json_files))
@@ -38 +40 @@
 
 entrant_recs = []
+
 
 for year_and_entrant in all_af_json_files:
@@ -61 +64 @@
 
 
+#
 # af = audio-features
+#
 all_af_nul_files = []
+src_af_lookup    = {}
 
-print()
-print("Nul files derived from year/country_title_artist audio feature JSON files:")
+#print()
+#print("Nul files derived from year/country_title_artist audio feature JSON files:")
 
 for entrant_rec in entrant_recs:
-     year =    entrant_rec["year"]
-     country = entrant_rec["country"].replace("&","and").replace(" ","")
+     year    = entrant_rec["year"]
+     country = entrant_rec["country"]
+     title   = entrant_rec["title"]
+     artist  = entrant_rec["artist"]
+
+     full_year_entrant = os.path.join(audio_dir,year,country+"_" + title + "_" + artist)
+     
+     # Tweak the values so they align with those used in SPARQL resultsset     
+     country = country.replace("&","and").replace(" ","")
 
      if year == "1956":
@@ -79 +92 @@
 
      all_af_nul_files.append(nul_file)
+     src_af_lookup[nul_file] = full_year_entrant
+     # print("**## storing nul_file = " + nul_file + " => " + full_year_entrant);
      
-print(", ".join(all_af_nul_files))
+# print(", ".join(all_af_nul_files))
 
 
@@ -146 +161 @@
           if af_nul_file_errata is not None:
                print("  Errata: Fixing " + af_nul_file + " => " + af_nul_file_errata)
+
+               src_af_lookup[af_nul_file_errata] = src_af_lookup[af_nul_file]
+               del src_af_lookup[af_nul_file]
+
                all_af_nul_files[i] = af_nul_file_errata
+               
                
 print("Matching Audio-Feature files to nul files in 'import':")
@@ -154 +174 @@
 print("The following files have Audio Features, but did not match in to a .nul file in the 'import' diretory")
 
+matched_af_nul_files = []
+
 for af_nul_file in all_af_nul_files:
-     if not af_nul_file in entrant_nul_files_lookup:
+     if af_nul_file in entrant_nul_files_lookup:
+          matched_af_nul_files.append(af_nul_file)
+     else:
           print("  " + af_nul_file)
                
 
+print()
+print("Copying the following Essentia audio feature JSON files into .nul matched positions in 'import'")
 
-#for af_nul_file in all_af_nul_files:
-#     if not af_nul_file in entrant_nul_files_lookup:
-#          if re.search(r"^NorthMacedonia",af_nul_file):
-#               af_nul_file = af_nul_file.replace("North","")
-#               if not af_nul_file in entrant_nul_files_lookup:
-#                    print("**** Failed to find lookup for: " + af_nul_file)
-#          else:
-#               print("**** Failed to find lookup for: " + af_nul_file)
-               
-               
+for af_nul_file in matched_af_nul_files:
+     import_files = entrant_nul_files_lookup[af_nul_file]
+     for import_file in import_files:
+          # print("** looking up " + af_nul_file)
+          src_json_file = src_af_lookup[af_nul_file]
+          dst_json_file = import_file.replace(".nul","-Essentia.json")
+
+          print("  Copying: " + src_json_file + " => " + dst_json_file)
+          # shutil.copy(src_json_file,dst_json_file)
+          
+