| 1 | /* -*- C++ -*- |
|---|---|
| 2 | SPDX-FileCopyrightText: 1998 Netscape Communications Corporation <developer@mozilla.org> |
| 3 | |
| 4 | SPDX-License-Identifier: MIT |
| 5 | */ |
| 6 | |
| 7 | // for S-JIS encoding, observe characteristic: |
| 8 | // 1, kana character (or hankaku?) often have high frequency of appearance |
| 9 | // 2, kana character often exist in group |
| 10 | // 3, certain combination of kana is never used in japanese language |
| 11 | |
| 12 | #include "nsGB2312Prober.h" |
| 13 | |
| 14 | namespace kencodingprober |
| 15 | { |
| 16 | void nsGB18030Prober::Reset(void) |
| 17 | { |
| 18 | mCodingSM->Reset(); |
| 19 | mState = eDetecting; |
| 20 | mDistributionAnalyser.Reset(); |
| 21 | // mContextAnalyser.Reset(); |
| 22 | } |
| 23 | |
| 24 | nsProbingState nsGB18030Prober::HandleData(const char *aBuf, unsigned int aLen) |
| 25 | { |
| 26 | if (aLen == 0) { |
| 27 | return mState; |
| 28 | } |
| 29 | |
| 30 | for (unsigned int i = 0; i < aLen; i++) { |
| 31 | const nsSMState codingState = mCodingSM->NextState(c: aBuf[i]); |
| 32 | if (codingState == eError) { |
| 33 | mState = eNotMe; |
| 34 | break; |
| 35 | } |
| 36 | if (codingState == eItsMe) { |
| 37 | mState = eFoundIt; |
| 38 | break; |
| 39 | } |
| 40 | if (codingState == eStart) { |
| 41 | unsigned int charLen = mCodingSM->GetCurrentCharLen(); |
| 42 | |
| 43 | if (i == 0) { |
| 44 | mLastChar[1] = aBuf[0]; |
| 45 | mDistributionAnalyser.HandleOneChar(aStr: mLastChar, aCharLen: charLen); |
| 46 | } else { |
| 47 | mDistributionAnalyser.HandleOneChar(aStr: aBuf + i - 1, aCharLen: charLen); |
| 48 | } |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | mLastChar[0] = aBuf[aLen - 1]; |
| 53 | |
| 54 | if (mState == eDetecting) { |
| 55 | if (mDistributionAnalyser.GotEnoughData() && GetConfidence() > SHORTCUT_THRESHOLD) { |
| 56 | mState = eFoundIt; |
| 57 | } |
| 58 | } |
| 59 | // else |
| 60 | // mDistributionAnalyser.HandleData(aBuf, aLen); |
| 61 | |
| 62 | return mState; |
| 63 | } |
| 64 | |
| 65 | float nsGB18030Prober::GetConfidence(void) |
| 66 | { |
| 67 | float distribCf = mDistributionAnalyser.GetConfidence(); |
| 68 | |
| 69 | return (float)distribCf; |
| 70 | } |
| 71 | } |
| 72 |