X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=parse_html%2Fbkmk_parse_html.py;h=07fe32e3e7e6a658ec82f0123acff746a7ac0b01;hb=9faa13f6f8199790cf01533e857c593520559649;hp=7bdc5664470b2c2ae4499be6f7519689daae4ae6;hpb=4255ead7de9ed5069f94b90cb134a077387d43c1;p=bookmarks_db.git diff --git a/parse_html/bkmk_parse_html.py b/parse_html/bkmk_parse_html.py index 7bdc566..07fe32e 100644 --- a/parse_html/bkmk_parse_html.py +++ b/parse_html/bkmk_parse_html.py @@ -69,6 +69,7 @@ from htmlentitydefs import name2codepoint entity_re = re.compile("(&\w+;)") num_entity_re = re.compile("(&#[0-9]+;)") + def recode_entities(title, charset): output = [] for part in entity_re.split(title): @@ -95,6 +96,7 @@ def recode_entities(title, charset): import os BKMK_DEBUG_HTML_PARSERS = os.environ.get("BKMK_DEBUG_HTML_PARSERS") + def parse_html(html_text, charset=None, log=None): if not parsers: return None @@ -211,6 +213,7 @@ def parse_html(html_text, charset=None, log=None): parser.icon = icon.encode(parser.charset) return parser + def parse_filename(filename, charset=None, log=None): fp = open(filename, 'r') try: