Changeset 38072 for gs3-installations/thewillow

Timestamp:

2023-09-10T17:13:54+12:00 (9 months ago)

Author:

davidb

Message:

Changes after the latest version of the Google spreadsheet was passed on

Location:

gs3-installations/thewillow/trunk/sites/thewillow/collect/community-contributions/prepare

Files:

• 02-XLSX-TO-GSFRIENDLY-CSV.sh (modified) (2 diffs)
• xlsx-to-csv--thewillow-directorysheet.py (modified) (3 diffs)

gs3-installations/thewillow/trunk/sites/thewillow/collect/community-contributions/prepare/02-XLSX-TO-GSFRIENDLY-CSV.sh

@@ -1 +1 @@
 #!/bin/bash
 
-echo ""
-echo "Away to delete: downloads/*"
-echo ""
+echo "----"
+echo "- Away to delete: downloads/*"
+echo "----"
 
 echo "[Continuing in 5 seconds ... press enter to continue immediately]"
@@ -14 +14 @@
 echo "[... finished sleep]"
 
-echo "Removing: downloads/*"
+echo ""
+echo "**** Removing: downloads/*"
 /bin/rm -rf downloads
 mkdir downloads
 
+echo ""
+echo "Running xlsx-to-csv--thewillow-directorysheet.py"
+#echo "With a transcript of stdout and stderrr saved to: 02-output.log"
 
 python3 xlsx-to-csv--thewillow-directorysheet.py
+
+# ??? >02-output.log 2>&1 &
+
+# ????? 2>&1 | tee 02-output.log

gs3-installations/thewillow/trunk/sites/thewillow/collect/community-contributions/prepare/xlsx-to-csv--thewillow-directorysheet.py

@@ -30 +30 @@
     "Ref. No."          : { "gsheading": "RefNum"         },
     "Title"             : { "gsheading": "Title"          },
-    "Creator"           : { "gsheading": "Creator"        },
+    "Document ID"       : { "gsheading": "WCADocumentID"  },
+    "Contributor"       : { "gsheading": "Creator"        },
     "Description"       : { "gsheading": "Description"    },
     "Creation Date"     : { "gsheading": "CreationDate"   },
@@ -212 +213 @@
     download_filename = gdown.download(url, output=downloads_outputdir, fuzzy=True, quiet=True)
 
-    print("Downloaded filename:", download_filename)
+    if download_filename != None:
+        print("Downloaded filename:", download_filename)
 
     return download_filename
@@ -319 +321 @@
             downloaded_filename = downloadGoogleDoc(doc_url)
 
-            orig_file = os.path.basename(downloaded_filename)
-
-            # Determine if this will be the primary (doc) file, or an assoicated file
-            orig_filename_colpos = HeadingsTransformation["OrigFilename"]["colpos"]
-            orig_filename_value = worksheet.cell(row=row_i, column=orig_filename_colpos).value
-
-            downloaded_filename = optConvertDocToPdf(downloaded_filename)
-            
-            if orig_filename_value is None or orig_filename_value == "":
-                # Make this the primary file/doc
-                worksheet.cell(row=row_i, column=orig_filename_colpos, value=orig_file)
-
-                renamed_filename = fileRenameToDLIdentifier(downloaded_filename,dl_identifier_str)
-
-                filename_colpos = HeadingsTransformation["Filename"]["colpos"]
-                worksheet.cell(row=row_i, column=filename_colpos, value=renamed_filename)                
+            if downloaded_filename == None:
+                print("====", file=sys.stderr)
+                print("= Warning: Failed to download URL:", file=sys.stderr)
+                print("=   " + doc_url, file=sys.stderr)
+                print("= Processing row " + str(row_i) + " as a document with only metadata", file=sys.stderr)
+                print("====", file=sys.stderr)
+
             else:
-                orig_drivename_colpos = HeadingsTransformation["OrigDrivename"]["colpos"]
+                orig_file = os.path.basename(downloaded_filename)
+
+                # Determine if this will be the primary (doc) file, or an assoicated file
+                orig_filename_colpos = HeadingsTransformation["OrigFilename"]["colpos"]
+                orig_filename_value = worksheet.cell(row=row_i, column=orig_filename_colpos).value
+
+                downloaded_filename = optConvertDocToPdf(downloaded_filename)
+            
+                if orig_filename_value is None or orig_filename_value == "":
+                    # Make this the primary file/doc
+                    worksheet.cell(row=row_i, column=orig_filename_colpos, value=orig_file)
+                    
+                    renamed_filename = fileRenameToDLIdentifier(downloaded_filename,dl_identifier_str)
+                    
+                    filename_colpos = HeadingsTransformation["Filename"]["colpos"]
+                    worksheet.cell(row=row_i, column=filename_colpos, value=renamed_filename)                
+                else:
+                    orig_drivename_colpos = HeadingsTransformation["OrigDrivename"]["colpos"]
                 
-                worksheet.cell(row=row_i, column=orig_drivename_colpos, value=orig_file)
-
-                dl_identifier_assocwith_str = "tw-contrib-%04d-assocwith" % dl_identifier
+                    worksheet.cell(row=row_i, column=orig_drivename_colpos, value=orig_file)
+                    
+                    dl_identifier_assocwith_str = "tw-contrib-%04d-assocwith" % dl_identifier
                 
-                renamed_filename = fileRenameToDLIdentifier(downloaded_filename,dl_identifier_assocwith_str)
-
-                # We store this in the spreasheet mostly for auditing purposes
-                # The actual 'assocwith' DL behaviour is triggered by the plugin option,
-                # driven by a regular expression matching on the two filenames
-                filename_assocwith_colpos = HeadingsTransformation["FilenameAssocwith"]["colpos"]
-                worksheet.cell(row=row_i, column=filename_assocwith_colpos, value=renamed_filename)                
+                    renamed_filename = fileRenameToDLIdentifier(downloaded_filename,dl_identifier_assocwith_str)
+
+                    # We store this in the spreasheet mostly for auditing purposes
+                    # The actual 'assocwith' DL behaviour is triggered by the plugin option,
+                    # driven by a regular expression matching on the two filenames
+                    filename_assocwith_colpos = HeadingsTransformation["FilenameAssocwith"]["colpos"]
+                    worksheet.cell(row=row_i, column=filename_assocwith_colpos, value=renamed_filename)                
 
         dl_identifier += 1