From: <kabe@>
Subject: dblatex: do not error on gtk-doc tests/gobject tests

For unknow reason, python3 bytes (b'...') comes in where not should,
and str ("...") where bytes are expected.
One or other bombs out on gtk-doc package %check .

cd tests/gobject/docs
/bin/dblatex -o tester.pdf -I ./images -x '--path /home/kabe/c8builds/gtk-doc-c8/BUILD/gtk-doc-1.28/tests/gobject/docs:/home/kabe/c8builds/gtk-doc-c8/BUILD/gtk-doc-1.28/tests/gobject/examples' tester-docs.xml
...
Missing character &#x2570;
Missing character &#x2500;
Missing character &#x2500;
Unexpected error occured
Error: sequence item 0: expected str instance, bytes found

diff -up /usr/lib/python3.6/site-packages/dbtexmf/dblatex/rawtex.py.dist /usr/lib/python3.6/site-packages/dbtexmf/dblatex/rawtex.py
--- /usr/lib/python3.6/site-packages/dbtexmf/dblatex/rawtex.py.dist	2019-05-14 08:22:19.000000000 +0900
+++ /usr/lib/python3.6/site-packages/dbtexmf/dblatex/rawtex.py	2021-08-20 00:33:27.770242877 +0900
@@ -94,6 +94,9 @@ class RawLatex:
             fig = m[1]
             newfig = self.image.convert(fig)
 
+            # convert to bytes if str was returned
+            if isinstance(fig, str): fig = fig.encode("latin-1", "ignore")
+            if isinstance(newfig, str): newfig = newfig.encode("latin-1", "ignore")
             # If something done, replace the figure in the tex file
             if newfig != fig:
                 line = re.sub(br"{%s}" % fig, br"{%s}" % newfig, line)
diff -up /usr/lib/python3.6/site-packages/dbtexmf/dblatex/texcodec.py.dist /usr/lib/python3.6/site-packages/dbtexmf/dblatex/texcodec.py
--- /usr/lib/python3.6/site-packages/dbtexmf/dblatex/texcodec.py.dist	2019-05-14 08:22:19.000000000 +0900
+++ /usr/lib/python3.6/site-packages/dbtexmf/dblatex/texcodec.py	2021-08-20 00:13:59.219730001 +0900
@@ -13,6 +13,10 @@ tex_handler_installed = {}
 tex_handler_counter = {}
 
 def latex_char_replace(exc, pre, post, name):
+    # for unknown reason, sometimes bytes are passed as pre, post
+    # otherwise join(l) below bombs out
+    if isinstance(pre, bytes):   pre = pre.decode("latin-1")
+    if isinstance(post, bytes): post = post.decode("latin-1")
     global tex_handler_counter
     if not isinstance(exc, UnicodeEncodeError):
         raise TypeError("don't know how to handle %r" % exc)