import re

regtitlepreprocess = re.compile(r'<article>([^<]*)</article>',re.MULTILINE | re.DOTALL)
regdeflist = re.compile(r'(\n;[^:\n]*:)')
regsentinitialbracket = re.compile(r'([^M][^rs]\.)\s+(\[\[([^|\]]+\|)?([A-Z])([^|\]]+)\]\])',re.MULTILINE | re.DOTALL)
regneosfirst = re.compile(r'(<NEOS />)+\s*(<EOS\s?/>)+\s*(<NEOS />)*\s?', re.MULTILINE | re.DOTALL)
regneosfinal = re.compile(r'(<NEOS />)*\s*(<EOS\s?/>)+\s*(<NEOS />)+\s?', re.MULTILINE | re.DOTALL)
regneosremove = re.compile(r'<NEOS />', re.MULTILINE | re.DOTALL)
regeos = re.compile(r'<EOS\s?/>\s?', re.MULTILINE | re.DOTALL)
regeosTwoOrMore = re.compile(r'((\s*<EOS\s?/>)|(\s?\n)){3-10}', re.MULTILINE | re.DOTALL)
regredirect = re.compile(r'#[rR][eE][dD][iI][rR][eE][cC][tT]\s?\[\[([^\]]+)\]\]', re.MULTILINE | re.DOTALL )
regcurly1 = re.compile(r'(\({{[^}^{]*?}}\)|\'*{{[^}^{]*?}}\'*)', re.MULTILINE | re.DOTALL)

regwikitable1 = re.compile(r'{[^}]*?class="?wikitable"?[^}]*?}', re.MULTILINE | re.DOTALL)
regwikitable2 = re.compile(r'[^{]{\|[^\\][^}^{]*?}', re.MULTILINE | re.DOTALL)

regtablehardcode = re.compile('{\|\sclass=\"wikitable\"\s!English\s\|\|\sFrench\s\|\|\sIPA pronunciation\s\(Canadian accent\).*?\|}', re.MULTILINE | re.DOTALL)
regeos = re.compile(r'<EOS\s?/>\s?', re.MULTILINE | re.DOTALL)
regbr = re.compile(r'<br\s?/?>', re.MULTILINE | re.DOTALL)
regref = re.compile(r'<[Rr]ef.*?/[Rr]ef>', re.MULTILINE | re.DOTALL)
regref2 = re.compile(r'<[Rr]ef[^>]*?/>', re.MULTILINE | re.DOTALL)
regcomment = re.compile(r'<!--.*?-->', re.MULTILINE | re.DOTALL)
regblockquote = re.compile(r'<blockquote\s?>(.*?)</blockquote\s?>',re.MULTILINE | re.DOTALL)
regsource = re.compile(r'<source\s?.*?</source\s?>',re.MULTILINE | re.DOTALL)
regdiv2 = re.compile(r'</?div[^>]*?>',re.MULTILINE | re.DOTALL)
reggallery = re.compile(r'<gallery.*?</gallery\s?>',re.MULTILINE | re.DOTALL)
regsingleast = re.compile(r'^\*$')
reghyphen = re.compile(r'----?',re.MULTILINE | re.DOTALL)
regcenter = re.compile(r'<center.*?</center\s?>',re.MULTILINE | re.DOTALL)
regsealso = re.compile(r'==+\s?See also\s?==.*', re.MULTILINE | re.DOTALL)
regnotes = re.compile(r'==+\s?Notes\s?.*', re.MULTILINE | re.DOTALL)
regreferences = re.compile(r'==+\s?References.*', re.MULTILINE | re.DOTALL)
regsources = re.compile(r'==+\s?Sources.*', re.MULTILINE | re.DOTALL)

regsourcelookahead = re.compile(r'==+\s?(?=Sources.*?==+\s?Sources)', re.MULTILINE | re.DOTALL)
regseealsolookahead = re.compile(r'==+\s?(?=See also.*?==+\s?See also)', re.MULTILINE | re.DOTALL)
regnoteslookahead = re.compile(r'==+\s?(?=Notes.*?==+\s?Notes)', re.MULTILINE | re.DOTALL)
regreflookahead = re.compile(r'==+\s?(?=References.*?==+\s?References)', re.MULTILINE | re.DOTALL)

regsourcelookaheadrestore = re.compile(r'___(Sources)(.*?==+)',re.MULTILINE | re.DOTALL)
regnoteslookaheadrestore = re.compile(r'___(Notes)(.*?==+)',re.MULTILINE | re.DOTALL)
regreflookaheadrestore = re.compile(r'___(References)(.*?==+)',re.MULTILINE | re.DOTALL)
regseealsolookaheadrestore = re.compile(r'___(See also)(.*?==+)',re.MULTILINE | re.DOTALL)

regbibliography = re.compile(r'==+\s?Bibliography\s?.*', re.MULTILINE | re.DOTALL)
regfootnotes = re.compile(r'==+\s?Footnotes\s?.*', re.MULTILINE | re.DOTALL)
regrelated = re.compile(r'==+\s?Related web sites\s?.*', re.MULTILINE | re.DOTALL)
regexternal = re.compile(r'==+\s?External links\s?.*', re.MULTILINE | re.DOTALL)
regmath = re.compile(r'<math>[^\>]+?</math>',re.MULTILINE | re.DOTALL)
regremovenewline = re.compile(r'(?=\n)\n',re.MULTILINE | re.DOTALL)
regparagraph = re.compile(r'(\n\n)',re.MULTILINE | re.DOTALL)
regtitle = re.compile(r'(==+?\s?[^=]+?\s?==+\s?)',re.MULTILINE | re.DOTALL)
regbullets = re.compile(r'(^[#*].*?)\n', re.MULTILINE | re.DOTALL)
regbulletscolon = re.compile(r'(?<=([\.:]\s))([*#].*?)\n') 
regbullets2 = re.compile(r'(^\*.*?)\n', re.MULTILINE | re.DOTALL)
regindentcolon = re.compile(r'(^[;:][:;]?.*?)\n', re.MULTILINE | re.DOTALL)
regcolon = re.compile(r'^:', re.MULTILINE | re.DOTALL)
regimage = re.compile(r'\[\[Image:[^\[\]]*(\[\[[^\]]*\]\][^\[\]]*)*?\]\]',re.MULTILINE | re.DOTALL)
regspan = re.compile(r'</?\s*span.*?>',re.MULTILINE | re.DOTALL)
regbracket = re.compile(r'(\[\[([^|\]]+\|)?([^|\]]+)\]\]s\.)',re.MULTILINE | re.DOTALL)
regjava = re.compile(r'{{(Javadoc:.*?)}}',re.MULTILINE | re.DOTALL)
regiast = re.compile(r'{{(IAST.*?)}}',re.MULTILINE | re.DOTALL)
regipa = re.compile(r'{{(IPA.*?)}}(?=[^}])', re.MULTILINE | re.DOTALL)
regyeareos = re.compile(r'([0-9]{2,4}s\.)', re.MULTILINE | re.DOTALL)
regorg = re.compile(r'(\.org\.\s)', re.MULTILINE | re.DOTALL)


##japanese

regtransjap = re.compile(r'{{transl\|ja\|([^}]*?)}}', re.MULTILINE | re.DOTALL)
reglangjap = re.compile(r'{{lang\|ja\|([^}]*?)}}', re.MULTILINE | re.DOTALL)
regnihongohardcode = re.compile(r'({{Nihongo\|[^|]+\|[^|]+\|)3=',re.MULTILINE | re.DOTALL)
regnihongojap5 = re.compile(r'{{[Nn]ihongo\|([^|}]+)\|([^|}]+)\|([^|}]+)\|([^|}]+)\|([^|}]+)}}', re.MULTILINE | re.DOTALL)
regnihongojap4 = re.compile(r'{{[Nn]ihongo\|([^|}]+)\|([^|}]+)\|([^|}]+)\|([^|}]+)}}', re.MULTILINE | re.DOTALL)
regnihongojap3 = re.compile(r'{{[Nn]ihongo\|([^|}]*?)\|([^|}]*?)\|([^|}]+)}}', re.MULTILINE | re.DOTALL)
regnihongojap2 = re.compile(r'{{[Nn]ihongo\|([^|}]+)\|([^|}]+)\|?}}', re.MULTILINE | re.DOTALL)





regbacktocurly1 = re.compile(r'<___', re.MULTILINE | re.DOTALL)
regbacktocurly2 = re.compile(r'___>', re.MULTILINE | re.DOTALL)



regremoveeosinmath = re.compile(r'(<math[^<]*?)<EOS />(.*<EOS />)*?(.*?</math>)', re.MULTILINE | re.DOTALL)
regremoveeosinsource = re.compile(r'(<source[^<]*?)<EOS />(.*<EOS />)*?(.*?</source>)', re.MULTILINE | re.DOTALL)
regremoveeosinsource2 = re.compile(r'(<source.*?</source>)', re.MULTILINE | re.DOTALL)
regremoveeosincode2 = re.compile(r'(<code.*?</code>)', re.MULTILINE | re.DOTALL)
regremoveeosincode = re.compile(r'(<code[^<]*?)<EOS />(.*<EOS />)*?(.*?</code>)', re.MULTILINE | re.DOTALL)
regremoveeosinnowiki = re.compile(r'(<nowiki[^<]*?)<EOS />(.*<EOS />)*?(.*?</nowiki>)', re.MULTILINE | re.DOTALL)



#regdigitandbold = re.compile(r'(^:"[0-9][0-9]\.\s+(?=\'))', re.MULTILINE | re.DOTALL)

#regcurly1 = re.compile(r'{{?[^}^{]*?}?}', re.MULTILINE | re.DOTALL)

#regsource1 = re.compile(r'<source', re.MULTILINE | re.DOTALL)

#regsource2 = re.compile(r'</source', re.MULTILINE | re.DOTALL)