@@ -57,8 +57,8 @@ module CharDet if order >= 0 @_mTotalChars += 1 # order is valid - if order < @_mTableSize: - if 512 > @_mCharToFreqOrder[order]: + if order < @_mTableSize + if 512 > @_mCharToFreqOrder[order] @_mFreqChars += 1 end end @@ -72,7 +72,7 @@ module CharDet return SURE_NO end - if @_mTotalChars != @_mFreqChars: + if @_mTotalChars != @_mFreqChars r = @_mFreqChars / ((@_mTotalChars - @_mFreqChars) * @_mTypicalDistributionRatio) if r < SURE_YES return r @@ -227,10 +227,10 @@ module CharDet # first byte range: 0xa0 -- 0xfe # second byte range: 0xa1 -- 0xfe # no validation needed here. State machine has done that - if aStr[0..0] >= "\xA0": - return 94 * (aStr[0] - 0xA1) + aStr[1] - 0xa1 + if aStr[0..0] >= "\xA0" + return 94 * (aStr[0] - 0xA1) + aStr[1] - 0xa1 else - return -1 + return -1 end end end lib/rchardet/chardistribution.rb @@ -40,7 +40,7 @@ module CharDet super @_mActiveNum = 0 - for prober in @_mProbers: + for prober in @_mProbers if prober prober.reset() prober.active = true lib/rchardet/charsetgroupprober.rb @@ -41,7 +41,7 @@ module CharDet def reset super() - for codingSM in @_mCodingSM: + for codingSM in @_mCodingSM next if not codingSM codingSM.active = true codingSM.reset() lib/rchardet/escprober.rb @@ -56,7 +56,7 @@ module CharDet elsif codingState == EItsMe @_mState = EFoundIt break - elsif codingState == EStart: + elsif codingState == EStart charLen = @_mCodingSM.get_current_charlen() if i == 0 @_mLastChar[1] = aBuf[0..0] lib/rchardet/eucjpprober.rb @@ -150,9 +150,9 @@ module CharDet @_mNeedToSkipCharNum = i - aLen @_mLastCharOrder = -1 else - if (order != -1) and (@_mLastCharOrder != -1): + if (order != -1) and (@_mLastCharOrder != -1) @_mTotalRel += 1 - if @_mTotalRel > MAX_REL_THRESHOLD: + if @_mTotalRel > MAX_REL_THRESHOLD @_mDone = true break end @@ -169,7 +169,7 @@ module CharDet def get_confidence # This is just one way to calculate confidence. It works well for me. - if @_mTotalRel > MINIMUM_DATA_THRESHOLD: + if @_mTotalRel > MINIMUM_DATA_THRESHOLD return (@_mTotalRel - @_mRelSample[0]) / @_mTotalRel else return DONT_KNOW @@ -208,7 +208,7 @@ module CharDet return -1, 1 unless aStr # find out current char's byte length aStr = aStr[0..1].join if aStr.class == Array - if (aStr[0..0] == "\x8E") or ((aStr[0..0] >= "\xA1") and (aStr[0..0] <= "\xFE")): + if (aStr[0..0] == "\x8E") or ((aStr[0..0] >= "\xA1") and (aStr[0..0] <= "\xFE")) charLen = 2 elsif aStr[0..0] == "\x8F" charLen = 3 lib/rchardet/jpcntx.rb @@ -71,7 +71,7 @@ module CharDet @_mLastChar[0] = aBuf[aLen - 1.. aLen-1] - if get_state() == EDetecting: + if get_state() == EDetecting if @_mContextAnalyzer.got_enough_data() and (get_confidence() > SHORTCUT_THRESHOLD) @_mState = EFoundIt end lib/rchardet/sjisprober.rb @@ -91,11 +91,11 @@ module CharDet end @_mGotData = true - if @result['encoding'] and (@result['confidence'] > 0.0): + if @result['encoding'] and (@result['confidence'] > 0.0) @done = true return end - if @_mInputState == EPureAscii: + if @_mInputState == EPureAscii if @_highBitDetector =~ (aBuf) @_mInputState = EHighbyte elsif (@_mInputState == EPureAscii) and @_escDetector =~ (@_mLastChar + aBuf) @@ -138,12 +138,12 @@ module CharDet end @done = true - if @_mInputState == EPureAscii: + if @_mInputState == EPureAscii @result = {'encoding' => 'ascii', 'confidence' => 1.0} return @result end - if @_mInputState == EHighbyte: + if @_mInputState == EHighbyte confidences = {} @_mCharSetProbers.each{ |prober| confidences[prober] = prober.get_confidence } maxProber = @_mCharSetProbers.max{ |a,b| confidences[a] <=> confidences[b] } lib/rchardet/universaldetector.rb @@ -63,7 +63,7 @@ module CharDet end end - if get_state() == EDetecting: + if get_state() == EDetecting if get_confidence() > SHORTCUT_THRESHOLD @_mState = EFoundIt end lib/rchardet/utf8prober.rb 945438bc28ed6c20fc8f5175b98b2b87e13afdeb Michael Fellinger m.fellinger@gmail.com http://github.com/jmhodges/rchardet/commit/f54c49e6067a3403cadb2329947d3627490b7b39 f54c49e6067a3403cadb2329947d3627490b7b39 2009-09-09T19:55:10-07:00 2009-08-22T09:37:23-07:00 Remove syntax errors Signed-off-by: Jeff Hodges <jeff@somethingsimilar.com> 8b47a43e8071fa8f649124721b238c9d56a69ffc Jeff Hodges jeff@somethingsimilar.com