Move charset to the beginning of the list.

[bookmarks_db.git] / Robots / parse_html.py

diff --git a/Robots/parse_html.py b/Robots/parse_html.py
index 8ffca5c81bde3a095d57d697a0b8b234cd6f7fb7..347ed225ecee372695abdea8db686ad96a77971f 100755 (executable)
--- a/Robots/parse_html.py
+++ b/Robots/parse_html.py
@@ -11,6 +11,14 @@ universal_charset = "utf-8"
 DEFAULT_CHARSET = "cp1251" # Stupid default for Russian Cyrillic
 
 parsers = []
+
+try:
+   import parse_html_html5
+except ImportError:
+   pass
+else:
+   parsers.append(parse_html_html5.parse_html)
+
 try:
    import parse_html_beautifulsoup
    parse_html_beautifulsoup.DEFAULT_CHARSET = DEFAULT_CHARSET
@@ -62,8 +70,9 @@ def parse_html(filename, charset=None, log=None):
    charsets = [universal_charset, DEFAULT_CHARSET]
    if charset:
       charset = charset.lower().replace("windows-", "cp")
-      if charset not in charsets:
-         charsets.insert(0, charset)
+      if charset in charsets:
+         charsets.remove(charset)
+      charsets.insert(0, charset)
 
    for p in parsers:
       parser = None