1
|
dic = [[u'(?u)(?<![-\\w\u2013.,\xad])and and(?![-\\w\u2013\xad])', u'and', u'Did you mean:', False], [u'(?u)(?<![-\\w\u2013.,\xad])or or(?![-\\w\u2013\xad])', u'or', u'Did you mean:', False], [u'(?u)(?<![-\\w\u2013.,\xad])for for(?![-\\w\u2013\xad])', u'for', u'Did you mean:', False], [u'(?u)(?<![-\\w\u2013.,\xad])the the(?![-\\w\u2013\xad])', u'the', u'Did you mean:', False], [u'(?iu)(?<![-\\w\u2013.,\xad])[Yy][Ii][Nn][Gg] [Aa][Nn][Dd] [Yy][Aa][Nn][Gg](?![-\\w\u2013\xad])', u'yin and yang', u'Did you mean:', False], [u'(?iu)(?<![-\\w\u2013.,\xad])[Ss][Cc][Oo][Tt] [Ff][Rr][Ee][Ee](?![-\\w\u2013\xad])', u'scot-free\\nscotfree', u'Did you mean:', False], [u"(?iu)(?<![-\\w\u2013.,\xad])([Yy][Oo][Uu][Rr]|[Hh][Ee][Rr]|[Oo][Uu][Rr]|[Tt][Hh][Ee][Ii][Rr])['\u2019][Ss](?![-\\w\u2013\xad])", u'\\1s', u'Possessive pronoun:', False], [u'(?u)(?<![-\\w\u2013.,\xad])(?P<a_1>[Aa])n(?P<_>[ ][\'\u2018"\u201c]?)(?P<vow_1>[aeiouAEIOU]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'\\g<a_1>\\g<_>\\g<vow_1>\\g<etc_1>', u'Did you mean:', u'm.group("vow_1") in aA or m.group("vow_1").lower() in aA'], [u'(?u)(?<![-\\w\u2013.,\xad])a(?P<_>[ ][\'\u2018"\u201c]?)(?P<vow_1>[aeiouAEIOU]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'an\\g<_>\\g<vow_1>\\g<etc_1>', u'Bad article?', u'(m.group("vow_1") <> m.group("vow_1").upper()) and not (m.group("vow_1") in aA or m.group("vow_1").lower() in aA) and spell(LOCALE,m.group("vow_1"))'], [u'(?u)(?<![-\\w\u2013.,\xad])a(?P<_>[ ][\'\u2018"\u201c]?)(?P<con_1>[bcdfghj-np-tv-zBCDFGHJ-NP-TV-Z]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'an\\g<_>\\g<con_1>\\g<etc_1>', u'Did you mean:', u'm.group("con_1") in aAN or m.group("con_1").lower() in aAN'], [u'(?u)(?<![-\\w\u2013.,\xad])(?P<a_1>[Aa])n(?P<_>[ ][\'\u2018"\u201c]?)(?P<con_1>[bcdfghj-np-tv-zBCDFGHJ-NP-TV-Z]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'\\g<a_1>\\g<_>\\g<con_1>\\g<etc_1>', u'Bad article?', u'(m.group("con_1") <> m.group("con_1").upper()) and not (m.group("con_1") in aA or m.group("con_1").lower() in aAN) and not m.group("con_1") in aB and spell(LOCALE,m.group("con_1"))'], [u'(?u)((?<=[!?.] )|^)A(?P<_>[ ][\'\u2018"\u201c]?)(?P<vow_1>[aeiouAEIOU]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'An\\g<_>\\g<vow_1>\\g<etc_1>', u'Bad article?', u'(m.group("vow_1") <> m.group("vow_1").upper()) and not (m.group("vow_1") in aA or m.group("vow_1").lower() in aA) and spell(LOCALE,m.group("vow_1"))'], [u'(?u)((?<=[!?.] )|^)A(?P<_>[ ][\'\u2018"\u201c]?)(?P<con_1>[bcdfghj-np-tv-zBCDFGHJ-NP-TV-Z]\\w*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'An\\g<_>\\g<con_1>\\g<etc_1>', u'Did you mean:', u'm.group("con_1") in aAN or m.group("con_1").lower() in aAN'], [u'(?u)(?<![-\\w\u2013.,\xad])a(?P<_>[ ][\'\u2018"\u201c]?)(?P<nvow_1>(8[0-9]*|1[18](000)*)(th)?)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'an\\g<_>\\g<nvow_1>\\g<etc_1>', u'Did you mean:', False], [u'(?u)((?<=[!?.] )|^)A(?P<_>[ ][\'\u2018"\u201c]?)(?P<nvow_1>(8[0-9]*|1[18](000)*)(th)?)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'An\\g<_>\\g<nvow_1>\\g<etc_1>', u'Did you mean:', False], [u'(?u)(?<![-\\w\u2013.,\xad])(?P<a_1>[Aa])n(?P<_>[ ][\'\u2018"\u201c]?)(?P<ncon_1>[0-79][0-9]*)(?P<etc_1>[-\u2013\'\u2019\\w]*)(?![-\\w\u2013\xad])', u'\\g<a_1>\\g<_>\\g<ncon_1>\\g<etc_1>', u'Did you mean:', u'not m.group("ncon_1")[:2] in ["11", "18"]'], [u'(?u)((?<=[!?.] )|^)(?P<low_1>[a-z]+)(?![-\\w\u2013\xad])', u'= m.group("low_1").capitalize()', u'Missing capitalization?', u're.match("[a-z].*[.?!] [A-Z]", TEXT) and not abbrev.search(TEXT)'], [u'(?u)((?<=[!?.] )|^)(?P<low_1>[a-z]+)(?![-\\w\u2013\xad])', u'= m.group("low_1").capitalize()', u'Missing capitalization?', u'option(LOCALE,"cap") and not abbrev.search(TEXT)'], [u'(?u) ([.?!,:;)\u201d\\[\\]])\\b', u'\\1 ', u'Reversed space and punctuation?', False], [u'(?u) +[.]', u'.', u'Extra space before the period?', u'LOCALE.Country == "US"'], [u'(?u) +[.]', u'.', u'Extra space before the full stop?', u'LOCALE.Country != "US"'], [u'(?u) +([?!,:;)\u201d\\]])', u'\\1', u'= "Extra space before the " + punct[m.group(1)] + "?"', False], [u'(?u)([([\u201c]) ', u'\\1', u'= "Extra space after the " + punct[m.group(1)] + "?"', False], [u'(?u)\\b(---?| --? )\\b', u' \u2013 \\n\u2014', u'En dash or em dash:', u'not option(LOCALE,"ndash") and not option(LOCALE,"mdash")'], [u'(?u)\\b(---?| --? |\u2014)\\b', u' \u2013 ', u'En dash:', u'option(LOCALE,"ndash") and not option(LOCALE,"mdash")'], [u'(?u)\\b(---?| --? | \u2013 )\\b', u'\u2014', u'Em dash:', u'option(LOCALE,"mdash")'], [u'(?u)(?P<number_1>\\d+([.]\\d+)?)(x| x )(?P<number_2>\\d+([.]\\d+)?)', u'\\g<number_1>\xd7\\g<number_2>', u'Multiplication sign.', u'option(LOCALE,"times")'], [u'(?u)(?P<Abc_1>[a-zA-Z]+)(?P<pun_1>[?!,:;%\u2030\u2031\u02da\u201c\u201d\u2018])(?P<Abc_2>[a-zA-Z]+)', u'\\g<Abc_1>\\g<pun_1> \\g<Abc_2>', u'Missing space?', False], [u'(?u)(?P<abc_1>[a-z]+)[.](?P<ABC_1>[A-Z]+)', u'\\g<abc_1>. \\g<ABC_1>', u'Missing space?', False], [u'(?u)[)]', u'', u'Extra closing parenthesis?', u'option(LOCALE,"pair") and not "(" in TEXT'], [u'(?u)[(]', u'', u'Extra opening parenthesis?', u'option(LOCALE,"pair") and TEXT[-1] in u"?!;:\u201d\u2019" and not ")" in TEXT'], [u'(?u)(?<![0-9])\u201d', u'', u'Extra quotation mark?', u'option(LOCALE,"pair") and not u"\u201c" in TEXT'], [u'(?u)(?<=[0-9])\u201d', u'\u2033\\n', u'Bad double prime or extra quotation mark?', u'option(LOCALE,"apostrophe") and not u"\u201c" in TEXT'], [u'(?u)\u201c', u'', u'Extra quotation mark?', u'option(LOCALE,"pair") and TEXT[-1] in u"?!;:\u201d\u2019" and not u"\u201d" in TEXT'], [u'(?u)[.]{3}', u'\u2026', u'Ellipsis.', u'option(LOCALE,"ellipsis")'], [u'(?u)\\b {2,3}(\\b|$)', u'\\1 ', u'Extra space.', u'option(LOCALE,"spaces")'], [u'(?u)(^|\\b|(?P<pun_1>[?!,:;%\u2030\u2031\u02da\u201c\u201d\u2018])|[.]) {2,3}(\\b|$)', u'\\1 ', u'Extra space.', u'option(LOCALE,"spaces2")'], [u'(?u)(^|\\b|(?P<pun_1>[?!,:;%\u2030\u2031\u02da\u201c\u201d\u2018])|[.]) {4,}(\\b|$)', u'\\1 \\n\t', u'Change multiple spaces to a single space or a tabulator:', u'option(LOCALE,"spaces3")'], [u'(?iu)[\\"\u201c\u201d\u201f\u201e]((?P<abc_1>[a-zA-Z]+)[^\\"\u201c\u201d\u201f\u201e]*)[\\"\u201c\u201f]', u'\u201c\\1\u201d', u'Quotation marks.', u'option(LOCALE,"quotation")'], [u'(?iu)[\\"\u201d\u201f\u201e]((?P<abc_1>[a-zA-Z]+)[^\\"\u201c\u201d\u201f\u201e]*)[\\"\u201c\u201d\u201f]', u'\u201c\\1\u201d', u'Quotation marks.', u'option(LOCALE,"quotation")'], [u"(?iu)'(?P<abc_1>[a-zA-Z]+)'", u'\u2018\\g<abc_1>\u2019', u'Quotation marks.', u'option(LOCALE,"apostrophe")'], [u'(?iu)[\\"\u201d\u201f\u201e]((?P<abc_1>[a-zA-Z]+)[^\\"\u201c\u201d\u201f\u201e]*)[\\"\u201c\u201d\u201f]', u'\u201c\\1\u201d', u'Quotation marks.', u'option(LOCALE,"apostrophe")'], [u"(?iu)(?P<Abc_1>[a-zA-ZA-Z]+)'(?P<w_1>\\w*)", u'\\g<Abc_1>\u2019\\g<w_1>', u'Replace typewriter apostrophe or quotation mark:', u'option(LOCALE,"apostrophe")'], [u"(?u)(?<= )'(?P<Abc_1>[a-zA-Z]+)", u'\u2018\\g<Abc_1>\\n\u2019\\g<Abc_1>', u'Replace typewriter quotation mark or apostrophe:', u'option(LOCALE,"apostrophe")'], [u"(?u)^'(?P<Abc_1>[a-zA-Z]+)", u'\u2018\\g<Abc_1>\\n\u2019\\g<Abc_1>', u'Replace typewriter quotation mark or apostrophe:', u'option(LOCALE,"apostrophe")'], [u'(?u)\\b(?P<d2_1>\\d\\d)(?P<d_1>\\d\\d\\d)\\b', u'\\g<d2_1>,\\g<d_1>\\n\\g<d2_1>\u202f\\g<d_1>', u'Use thousand separator (common or ISO).', u'option(LOCALE,"numsep")'], [u'(?u)\\b(?P<D_1>\\d{1,3})(?P<d_1>\\d\\d\\d)(?P<d_2>\\d\\d\\d)\\b', u'\\g<D_1>,\\g<d_1>,\\g<d_2>\\n\\g<D_1>\u202f\\g<d_1>\u202f\\g<d_2>', u'Use thousand separators (common or ISO).', u'option(LOCALE,"numsep")'], [u'(?u)\\b(?P<D_1>\\d{1,3})(?P<d_1>\\d\\d\\d)(?P<d_2>\\d\\d\\d)(?P<d_3>\\d\\d\\d)\\b', u'\\g<D_1>,\\g<d_1>,\\g<d_2>,\\g<d_3>\\n\\g<D_1>\u202f\\g<d_1>\u202f\\g<d_2>\u202f\\g<d_3>', u'Use thousand separators (common or ISO).', u'option(LOCALE,"numsep")'], [u'(?u)(?<![-\\w\u2013.,\xad])(?P<Abc_1>[a-zA-Z]+) \\1(?![-\\w\u2013\xad])', u'\\g<Abc_1>', u'Word duplication?', u'option(LOCALE,"dup")'], [u'(?u)(?<![-\\w\u2013.,\xad])([Tt])his (?P<abc_1>[a-z]+)(?![-\\w\u2013\xad])', u'\\1hese \\g<abc_1>\\n\\1his, \\g<abc_1>', u'Did you mean:', u'option(LOCALE,"grammar") and morph(LOCALE,m.group("abc_1"), "Ns")'], [u"(?u)(?<![-\\w\u2013.,\xad])with it['\u2019]s(?![-\\w\u2013\xad])", u'with its\\nwith, it\u2019s', u'Did you mean:', u'option(LOCALE,"grammar")'], [u"(?iu)(?<![-\\w\u2013.,\xad])([Ii][Tt]|[Ss]?[Hh][Ee]) [Dd][Oo][Nn]['\u2019][Tt](?![-\\w\u2013\xad])", u'\\1 doesn\u2019t', u'Did you mean:', u'option(LOCALE,"grammar")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) (\xb0F|Fahrenheit)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "F", "C", u" \xb0C", ".", ",")', u'Convert to Celsius:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) (\xb0C|Celsius)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "C", "F", u" \xb0F", ".", ",")', u'Convert to Fahrenheit:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*(?: 1/2| ?\xbd)?) (ft|foot|feet)(?! [1-9])(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "ft", "cm", " cm", ".", ",") + "\\n" + measurement(m.group(1), "ft", "m", " m", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*(?: 1/2| ?\xbd)?) ft[.]? ([0-9]+(?: 1/2| ?\xbd)?) in(?![-\\w\u2013\xad])', u'= measurement(m.group(1) + "*12+" + m.group(2), "in", "cm", " cm", ".", ",") + "\\n" + measurement(m.group(1) + "*12+" + m.group(2), "in", "m", " m", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*(?: 1/2| ?\xbd)?) in(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "in", "mm", " mm", ".", ",") + "\\n" + measurement(m.group(1), "in", "cm", " cm", ".", ",") + "\\n" + measurement(m.group(1), "in", "m", " m", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) mm(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "mm", "in", " in", ".", ",")', u'Convert from metric:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) cm(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "cm", "in", " in", ".", ",") + "\\n" + measurement(m.group(1), "cm", "ft", " ft", ".", ",")', u'Convert from metric:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) (m|meter|metre)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "m", "in", " in", ".", ",") + "\\n" + measurement(m.group(1), "m", "ft", " ft", ".", ",") + "\\n" + measurement(m.group(1), "m", "mi", " mi", ".", ",")', u'Convert from metric:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*(?: 1/2| ?\xbd)?) miles?(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "mi", "m", " m", ".", ",") + "\\n" + measurement(m.group(1), "mi", "km", " km", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) km(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "km", "mi", " mi", ".", ",")', u'Convert to miles:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:,\\d+)?) (yd|yards?)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "yd", "m", " m", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:,\\d+)?) (gal(lons?)?)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "gal", "l", " l", ".", ",") + "\\n" + measurement(m.group(1), "uk_gal", "l", " l (in UK)", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:,\\d+)?) (pint)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "pt", "dl", " dl", ".", ",") + "\\n" + measurement(m.group(1), "uk_pt", "dl", " dl (in UK)", ".", ",") + "\\n" + measurement(m.group(1), "pt", "l", " l", ".", ",") + "\\n" + measurement(m.group(1), "uk_pt", "l", " l (in UK)", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:,\\d+)?) (l|L|litres?|liters?)(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "l", "gal", " gal", ".", ",") + "\\n" + measurement(m.group(1), "l", "gal", " gal (in UK)", ".", ",")', u'Convert to gallons:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) lbs?[.]?(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "lbm", "kg", " kg", ".", ",")', u'Convert to metric:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) kg[.]?(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "kg", "lbm", " lb", ".", ",")', u'Convert to pounds:', u'option(LOCALE,"nonmetric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) mph(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "mph", "km/h", " km/h", ".", ",")', u'Convert to km/hour:', u'option(LOCALE,"metric")'], [u'(?u)(?<![-\\w\u2013.,\xad])([-\u2212]?\\d+(?:[,.]\\d+)*) km/h(?![-\\w\u2013\xad])', u'= measurement(m.group(1), "km/h", "mph", " mph", ".", ",")', u'Convert to miles/hour:', u'option(LOCALE,"nonmetric")']]
|
