From cbdb9375975f7a4155c7b398bb3aa3c27a4ab5fa Mon Sep 17 00:00:00 2001
From: Magnus Hagander <magnus@hagander.net>
Date: Sat, 26 Jan 2019 17:21:37 +0100
Subject: [PATCH] Update docsload to use tidylib instead of tidy

The old library is not supported on py3. This is the same migration
we've done for other projects like pgarchives.
---
 tools/docs/docload.py | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/tools/docs/docload.py b/tools/docs/docload.py
index 7d03bffd..edf1dc5e 100755
--- a/tools/docs/docload.py
+++ b/tools/docs/docload.py
@@ -6,7 +6,7 @@ import sys
 import os
 import tarfile
 import re
-import tidy
+import tidylib
 from optparse import OptionParser
 from configparser import ConfigParser
 
@@ -34,7 +34,8 @@ def load_doc_file(filename, f):
     # Postgres 10 started using xml toolchain and now produces docmentation in utf8. So we need
     # to figure out which version it is.
     rawcontents = f.read()
-    if rawcontents.startswith('<?xml version="1.0" encoding="UTF-8"'):
+    rawfirst = rawcontents[:50].decode('utf8', errors='ignore')
+    if rawfirst.startswith('<?xml version="1.0" encoding="UTF-8"'):
         # Version 10, use utf8
         encoding = 'utf-8'
         # XML builds also don't need clean=1, and that one adds some interesting CSS properties
@@ -58,12 +59,13 @@ def load_doc_file(filename, f):
     if not quiet:
         print("--- file: %s (%s) ---" % (filename, title))
 
-    s = tidy.parseString(contents.encode('utf-8'), **tidyopts)
+    (html, errors) = tidylib.tidy_document(contents, options=tidyopts)
+
     curs.execute("INSERT INTO docs (file, version, title, content) VALUES (%(f)s, %(v)s, %(t)s, %(c)s)", {
         'f': filename,
         'v': ver,
         't': title,
-        'c': str(s),
+        'c': html,
     })
     global pagecount
     pagecount += 1
-- 
2.39.5