From 9023627fe50f4222b214444d36531fa3e24fb12c Mon Sep 17 00:00:00 2001
From: Lars Kruse <devel@sumpfralle.de>
Date: Tue, 8 Aug 2023 20:21:56 +0200
Subject: [PATCH] fix: handle non-ascii characters in page names and
 attachments

---
 doku.py      |  3 +--
 moin2doku.py | 17 +++++++++++------
 2 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/doku.py b/doku.py
index c750a19..0ecac1a 100755
--- a/doku.py
+++ b/doku.py
@@ -37,8 +37,7 @@ class DokuWiki:
 				uargs.append(arg.encode('utf-8'))
 		key = "%s:%s" % (method, ",".join(uargs))
 		if not self.callcache.has_key(key):
-			cmd = ['php', './doku.php', method ] + uargs
+			cmd = ['php', './doku.php', method] + [arg.encode("utf-8") for arg in uargs]
 			res = subprocess.Popen(cmd, stdin = None, stdout = subprocess.PIPE, stderr = sys.stderr, close_fds = False).communicate()
 			self.callcache[key] = unicode(res[0].decode('utf-8'))
-			print "%s->%s" % (cmd, self.callcache[key])
 		return self.callcache[key]
diff --git a/moin2doku.py b/moin2doku.py
index fd60983..8d59c0b 100755
--- a/moin2doku.py
+++ b/moin2doku.py
@@ -65,6 +65,11 @@ def writefile(filename, content, overwrite=False):
 	f.writelines([line + u'\n' for line in content])
 	f.close()
 
+
+def encode_relaxed(text):
+	return text.encode("ascii", errors="ignore")
+
+
 # page = MoinMoin Page oject
 # ns = DokuWiki namespace where attachments to copy
 def copy_attachments(page, ns,randomID):
@@ -84,7 +89,7 @@ def copy_attachments(page, ns,randomID):
 			copyfile(src, dst)
 			copystat(src, dst)
 		except UnicodeDecodeError:
-			print 'ERROR: unable to convert attachment "%s"' % attachment
+			print 'ERROR: unable to convert attachment "%s"' % encode_relaxed(attachment)
 
 def print_help():
 	program = sys.argv[0]
@@ -168,12 +173,12 @@ def convertfile(page, output = None, overwrite = False):
 	if not output:
 		output = pagename
 
-	print "Converting %s" % pagename
+	print "Converting %s" % encode_relaxed(pagename)
 
 	if page.isUnderlayPage():
 		print "underlay: %s" % page.request.cfg.data_underlay_dir
 		print "underlay: %s" % request.cfg.data_underlay_dir
-		print "SKIP UNDERLAY: %s" % pagename
+		print "SKIP UNDERLAY: %s" % encode_relaxed(pagename)
 		return False
 
 	current_exists = page.exists()
@@ -202,7 +207,7 @@ def convertfile(page, output = None, overwrite = False):
 				print "recovered %s: %s" % (rev, mtime)
 
 			if not mtime:
-				print "NO REVISION: for %s" % pagefile
+				print "NO REVISION: for %s" % encode_relaxed(pagefile)
 				continue
 
 		if rev == current_rev:
@@ -237,7 +242,7 @@ def convertfile(page, output = None, overwrite = False):
 		if old_page != ID:
 			redirect_map[old_page] = ID
 
-	print "Converted %s as %s" % (pagename, dw.wikiFN(output))
+	print "Converted %s as %s" % (encode_relaxed(pagename), dw.wikiFN(output))
 
 	return True
 
@@ -324,7 +329,7 @@ else:
 
 print "--------------------------------------------------"
 for output, pagename in pages.items():
-	print " - %s" % pagename
+	print " - %s" % encode_relaxed(pagename)
 print "--------------------------------------------------"
 
 converted = 0