From 81bde9197bd83ce37baac13924c782c02353a172 Mon Sep 17 00:00:00 2001 From: Marvin W Date: Sun, 29 Dec 2019 16:11:53 +0100 Subject: URL linker: count open/close parens, simplify regex --- main/src/ui/util/helper.vala | 38 +++++++++++++++++++++++++++++++++++++- 1 file changed, 37 insertions(+), 1 deletion(-) (limited to 'main/src/ui/util') diff --git a/main/src/ui/util/helper.vala b/main/src/ui/util/helper.vala index a310de6c..940d5c9b 100644 --- a/main/src/ui/util/helper.vala +++ b/main/src/ui/util/helper.vala @@ -1,3 +1,4 @@ +using Gee; using Gtk; using Dino.Entities; @@ -6,6 +7,7 @@ using Xmpp; namespace Dino.Ui.Util { private static Regex url_regex; +private static Map matching_chars; private const string[] allowed_schemes = {"http", "https", "ftp", "ftps", "irc", "ircs", "xmpp", "mailto", "sms", "smsto", "mms", "tel", "geo", "openpgp4fpr", "im", "news", "nntp", "sip", "ssh", "bitcoim", "sftp", "magnet", "vnc"}; private const string[] tango_colors_light = {"FCE94F", "FCAF3E", "E9B96E", "8AE234", "729FCF", "AD7FA8", "EF2929"}; private const string[] tango_colors_medium = {"EDD400", "F57900", "C17D11", "73D216", "3465A4", "75507B", "CC0000"}; @@ -250,11 +252,21 @@ public static bool is_24h_format() { public static Regex get_url_regex() { if (url_regex == null) { - url_regex = /\b((https?|ftps?|ircs?|xmpp|mailto|sms|smsto|mms|tel|geo|openpgp4fpr|im|news|nntp|sip|ssh|bitcoin|sftp|magnet|vnc|urn):(\/\/([^\/?#,;!)}>"'»”’\s]+)(\/([^#\s,.;!?)\]}>"'»”’]|[,.;!)\]}>"'»”’][^?#\s])*)?|([^\/#\s,.;!?)\]}>"'»”’]|[,.;!)\]}>"'»”’][^\/?#\s])*)(\?([^#\s,.;!?)\]}>"'»”’]|[,.;!?)\]}>"'»”’][^#\s])+)?(#([^\s,.;!?)\]}>"'»”’]|[,.;!?)\]}>"'»”’][^\s])+)?)/; + url_regex = /\b(((http|ftp)s?:\/\/|(ircs?|xmpp|mailto|sms|smsto|mms|tel|geo|openpgp4fpr|im|news|nntp|sip|ssh|bitcoin|sftp|magnet|vnc|urn):)([^\s,.;!?"'»”’]|[,.;!?"'»”’]\S)+)/; } return url_regex; } +public static Map get_matching_chars() { + if (matching_chars == null) { + matching_chars = new HashMap(); + matching_chars[")".get_char(0)] = "(".get_char(0); + matching_chars["]".get_char(0)] = "[".get_char(0); + matching_chars["}".get_char(0)] = "{".get_char(0); + } + return matching_chars; +} + public static string parse_add_markup(string s_, string? highlight_word, bool parse_links, bool parse_text_markup, bool already_escaped_ = false) { string s = s_; bool already_escaped = already_escaped_; @@ -266,6 +278,30 @@ public static string parse_add_markup(string s_, string? highlight_word, bool pa int start, end; match_info.fetch_pos(0, out start, out end); string link = s[start:end]; + Map matching_chars = get_matching_chars(); + unichar close_char; + int last_char_index = link.length; + while (link.get_prev_char(ref last_char_index, out close_char) && matching_chars.has_key(close_char)) { + unichar open_char = matching_chars[close_char]; + unichar char; + int index = 0; + int open = 0, close = 0; + while (link.get_next_char(ref index, out char)) { + if (char == open_char) { + open++; + } else if (char == close_char) { + close++; + } + } + if (close > open) { + // Remove last char from url + end -= close_char.to_string().length; + link = s[start:end]; + } else { + break; + } + } + if (GLib.Uri.parse_scheme(link) in allowed_schemes) { return parse_add_markup(s[0:start], highlight_word, parse_links, parse_text_markup, already_escaped) + "" + parse_add_markup(link, highlight_word, false, false, already_escaped) + "" + -- cgit v1.2.3-54-g00ecf