Epub 3.2 Collection Tag support (#308)

* Hooked up logic for collections based on EPUB3.2 Spec and Fixed improper tags in EPUBs since it is XML and we are using HTML to parse it.

* Fixed a bug with src:url url replacing so that it's much cleaner regex
This commit is contained in:
Joseph Milazzo 2021-06-15 09:51:37 -05:00 committed by GitHub
parent 584348c6ad
commit d02d2d3cb5
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 44 additions and 6 deletions

View file

@ -15,7 +15,7 @@ namespace API.Parser
public const string ArchiveFileExtensions = @"\.cbz|\.zip|\.rar|\.cbr|\.tar.gz|\.7zip|\.7z|.cb7";
public const string BookFileExtensions = @"\.epub";
public const string ImageFileExtensions = @"^(\.png|\.jpeg|\.jpg)";
public static readonly Regex FontSrcUrlRegex = new Regex("(src:url\\(\"?'?)([a-z0-9/\\._]+)(\"?'?\\))", RegexOptions.IgnoreCase | RegexOptions.Compiled);
public static readonly Regex FontSrcUrlRegex = new Regex(@"(src:url\(.{1})" + "([^\"']*)" + @"(.{1}\))", RegexOptions.IgnoreCase | RegexOptions.Compiled);
public static readonly Regex CssImportUrlRegex = new Regex("(@import\\s[\"|'])(?<Filename>[\\w\\d/\\._-]+)([\"|'];?)", RegexOptions.IgnoreCase | RegexOptions.Compiled);
private static readonly string XmlRegexExtensions = @"\.xml";