PDF metadata for multiple authors

Author: mccallc

dvcurator/gui.py

@@ -165,8 +165,7 @@ def make_github(self):
 
 	def rename(self):
 		self.disable_buttons()
-		prefix = dvcurator.rename.last_name_prefix(self.citation)
-		t = threading.Thread(target=dvcurator.rename.basic_rename, args=(self.subfolder_path, prefix))
+		t = threading.Thread(target=dvcurator.rename.basic_rename, args=(self.subfolder_path, self.citation))
 		t.start()
 		self.schedule_check(t)
 
@@ -178,7 +177,7 @@ def convert(self):
 
 	def set_metadata(self):
 		self.disable_buttons()
-		t = threading.Thread(target=dvcurator.pdf_metadata.standard_metadata, args=(self.subfolder_path, self.citation['author'][0]['authorName']['value']))
+		t = threading.Thread(target=dvcurator.pdf_metadata.standard_metadata, args=(self.subfolder_path, self.citation))
 		t.start()
 		self.schedule_check(t)
 
@@ -201,21 +200,22 @@ def __init__(self, parent, *args, **kwargs):
 		self.filemenu = tk.Menu(self.menubar, tearoff=False)
 		self.filemenu.add_command(label="Open config", command=self.load_config)
 		self.filemenu.add_command(label="Save current config", command=self.save_config)
-		self.filemenu.add_command(label="Select project subfolder manually", command=self.set_subfolder)
 		self.filemenu.add_command(label="Exit dvcurator", command=parent.destroy)
 		self.menubar.add_cascade(label="File", menu=self.filemenu)
 
 		self.editmenu = tk.Menu(self.menubar, tearoff=False)
-		self.editmenu.add_command(label="Open Dropbox subfolder", command=self.open_explorer)
 		self.editmenu.add_command(label="Basic file rename", command=self.rename)
 		self.editmenu.add_command(label="Convert docx to pdf", command=self.convert)
 		self.editmenu.add_command(label="Set PDF metadata", command=self.set_metadata)
+		self.editmenu.add_separator()
+		self.editmenu.add_command(label="Open Dropbox subfolder", command=self.open_explorer)
+		self.editmenu.add_command(label="Select project subfolder manually", command=self.set_subfolder)
 		self.menubar.add_cascade(label="Edit", menu=self.editmenu)
 		self.menubar.entryconfig("Edit", state=tk.DISABLED)
 		parent.config(menu=self.menubar)
 
 		# Checklist of tickets included in the .md files
-		checklist = tk.Frame(self)
+		checklist = tk.LabelFrame(self, text="Project issues:")
 		if getattr(sys, 'frozen', False) and hasattr(sys, '_MEIPASS'):
 			self.issues = os.listdir(os.path.join(sys._MEIPASS, "issues"))
 		else:
@@ -230,17 +230,7 @@ def __init__(self, parent, *args, **kwargs):
 
 		# Settings
 		settings = tk.Frame(self)
-		#config_file=tk.StringVar()
-		#config_label = tk.Label(settings, text="Config file: ")
-		#config_options = tk.Frame(settings)
-		#config_entry = tk.Button(config_options, text="Open", command=self.load_config)
-		#config_save = tk.Button(config_options, text="Save", command=self.save_config)
 
-		#config_label.grid(column=1, row=1)
-		#config_entry.grid(column=1, row=1)
-		#config_save.grid(column=2, row=1)
-		#config_options.grid(column=2, row=1)
-		
 		self.doi=tk.StringVar()
 		doi_label = tk.Label(settings, text="Persistent ID (DOI): ")
 		self.doi_entry = tk.Entry(settings, textvariable=self.doi)
@@ -289,7 +279,7 @@ def __init__(self, parent, *args, **kwargs):
 		self.reset_button.pack()
 
 		settings.grid(column=1, row=1)
-		checklist.grid(column=2, row=2)
+		checklist.grid(column=2, row=2, padx=10)
 		process.grid(column=2, row=1)
 
 		from tkinter import scrolledtext
@@ -299,15 +289,10 @@ def __init__(self, parent, *args, **kwargs):
 		sys.stderr = redir
 		self.out.grid(column=1, row=2)
 
-
-
-
 def main():
 	root=tk.Tk()
 	root.title("dvcurator " + dvcurator.version.version)
 	MainApp(root).pack(side="top", fill="both", expand=True)
-
-	
 	root.mainloop()
 
 if __name__ == "__main__":

dvcurator/pdf_metadata.py

@@ -11,8 +11,18 @@ def find_pdfs(path):
 				pdfs += [os.path.join(root, name)]
 
 	return pdfs
-	
-def standard_metadata(folder, author):
+
+# Generate the string for the author metadata field
+# separate with semicolons between full names for > 1 author
+def combine_author_names(citation):
+	author_string = citation['author'][0]['authorName']['value']
+	if (len(citation['author']) > 1):
+		for author in citation['author'][1:]:
+			author_string += "; " + author['authorName']['value']
+	return author_string
+
+# This is the function run from the GUI
+def standard_metadata(folder, citation):
 	import pikepdf, os, shutil
 	import dvcurator.fs
 
@@ -25,12 +35,14 @@ def standard_metadata(folder, author):
 		print("Error: no PDFs detected in: " + edit_path)
 		return None
 
+	author_string = combine_author_names(citation)
+
 	# Ideally, we would just edit the files in place
 	# Some versions of pikepdf can't do this though
 	# so we copy them to a separate folder, then save back to the orginal place
 	old_path = os.path.join(edit_path, "originals")
 	os.mkdir(old_path)
-		
+
 	for path in pdfs:
 		original = os.path.join(old_path, os.path.basename(path))
 		os.rename(path, original)
@@ -48,7 +60,7 @@ def standard_metadata(folder, author):
 				print("!! Warning !! PDF is PDF/A")
 			meta['dc:title'] = os.path.basename(path)
 			#meta['dc:creator'] = author
-			meta['pdf:Author'] = author
+			meta['pdf:Author'] = author_string
 			meta['dc:description'] = "QDR Data Project"
 			meta['pdf:Subject'] = "QDR Data Project"
 			meta['pdf:Keywords'] = "-"
@@ -57,8 +69,7 @@ def standard_metadata(folder, author):
 		pdf.close()
 		print("Metadata written to: %s" %os.path.basename(path))
 
-	#os.rmdir(old_path)
-	#shutil.rmtree(old_path)
+	shutil.rmtree(old_path)
 	print("PDF metadata process complete!")
 
 	return edit_path

dvcurator/rename.py

@@ -53,9 +53,10 @@ def remove_all_accents(folder):
             unicodedata.normalize('NFKD', f).encode('ascii', 'ignore').decode('ascii')))
 
 # This is the function we call from the GUI, which calls all the above
-def basic_rename(folder, prefix):
+def basic_rename(folder, citation):
     import dvcurator.fs
     print("Renaming files", end="... ")
+    prefix = last_name_prefix(citation)
     new_path = dvcurator.fs.copy_new_step(folder, "rename")
     if (not new_path):
         return None

test.py

@@ -34,6 +34,23 @@ def test_check(self):
 	def test_search(self):
 		self.assertTrue(github.search_existing("Karcher - Anonymous Peer Review", "QualitativeDataRepository/testing-demos"))
 
+class TestRename(unittest.TestCase):
+
+	def test_rename(self):
+		f = tempfile.TemporaryDirectory()
+		first_folder = os.path.join(f.name, "QDR Prepared", "1_extract")
+		os.makedirs(first_folder) 
+
+		fake_file = "foobar.txt"
+		with open(os.path.join(first_folder, fake_file), 'w') as fp:
+			pass
+		
+		citation = dataverse.get_citation(host, doi)
+		new_path = rename.basic_rename(f.name, citation)
+		new_file = os.listdir(new_path)[0]
+		self.assertEqual(rename.last_name_prefix(citation) + "_" + fake_file,
+			new_file)
+
 class TestPDFMetadata(unittest.TestCase):
 
 	def test_makedir(self):
@@ -44,10 +61,14 @@ def test_makedir(self):
 		f.cleanup()
 
 	def test_pdfmetadata(self):
-		# This test is to make sure test_string gets written
+		# This test is to make sure author string gets written
 		# We read it back out from one of the files
 		import pikepdf
-		test_string = "Unit Test"
+
+		# Get author string from online citation
+		citation = dataverse.get_citation(host, doi)
+		author_string = pdf_metadata.combine_author_names(citation)
+
 		d = tempfile.TemporaryDirectory()
 		temp_structure = os.path.normpath(os.path.join(d.name, "QDR Prepared/5_rename"))
 		os.makedirs(temp_structure) 
@@ -57,11 +78,11 @@ def test_pdfmetadata(self):
 		for i in range(1, 11):
 			empty_pdf.save(os.path.join(temp_structure, f'test{i}.pdf'))
 
-		edit_path = pdf_metadata.standard_metadata(d.name, test_string)
+		edit_path = pdf_metadata.standard_metadata(d.name, citation)
 		one_file = os.path.join(edit_path, os.listdir(edit_path)[4])
 		example = pikepdf.open(one_file)
 		meta = example.open_metadata()
-		self.assertEqual(meta['pdf:Author'], test_string)
+		self.assertEqual(meta['pdf:Author'], author_string)
 		example.close()
 
 		d.cleanup()