Set a minimum threshold even for the detector's preferred language. Bug 547487, r=emk

This commit is contained in:
Simon Montagu 2010-02-23 01:46:15 -08:00
parent e38251afc0
commit c38fa14fa2
14 changed files with 88 additions and 30 deletions

View file

@ -46,16 +46,13 @@
#define SURE_YES 0.99f
#define SURE_NO 0.01f
#define MINIMUM_DATA_THRESHOLD 4
//return confidence base on received data
float CharDistributionAnalysis::GetConfidence(PRBool aIsPreferredLanguage)
float CharDistributionAnalysis::GetConfidence(void)
{
//if we didn't receive any character in our consideration range, or the
// number of frequent characters is below the minimum threshold, return
// negative answer
if (mTotalChars <= 0 ||
!aIsPreferredLanguage && mFreqChars <= MINIMUM_DATA_THRESHOLD)
if (mTotalChars <= 0 || mFreqChars <= mDataThreshold)
return SURE_NO;
if (mTotalChars != mFreqChars) {

View file

@ -42,10 +42,12 @@
#define ENOUGH_DATA_THRESHOLD 1024
#define MINIMUM_DATA_THRESHOLD 4
class CharDistributionAnalysis
{
public:
CharDistributionAnalysis() {Reset();}
CharDistributionAnalysis() {Reset(PR_FALSE);}
//feed a block of data and do distribution analysis
void HandleData(const char* aBuf, PRUint32 aLen) {}
@ -71,14 +73,15 @@ public:
}
//return confidence base on existing data
float GetConfidence(PRBool aIsPreferredLanguage);
float GetConfidence(void);
//Reset analyser, clear any state
void Reset(void)
void Reset(PRBool aIsPreferredLanguage)
{
mDone = PR_FALSE;
mTotalChars = 0;
mFreqChars = 0;
mDataThreshold = aIsPreferredLanguage ? 0 : MINIMUM_DATA_THRESHOLD;
}
//This function is for future extension. Caller can use this function to control
@ -104,6 +107,9 @@ protected:
//Total character encounted.
PRUint32 mTotalChars;
//Number of hi-byte characters needed to trigger detection
PRUint32 mDataThreshold;
//Mapping table to get frequency order from char order (get from GetOrder())
const PRInt16 *mCharToFreqOrder;

View file

@ -170,7 +170,7 @@ void JapaneseContextAnalysis::HandleData(const char* aBuf, PRUint32 aLen)
return;
}
void JapaneseContextAnalysis::Reset(void)
void JapaneseContextAnalysis::Reset(PRBool aIsPreferredLanguage)
{
mTotalRel = 0;
for (PRUint32 i = 0; i < NUM_OF_CATEGORY; i++)
@ -178,13 +178,14 @@ void JapaneseContextAnalysis::Reset(void)
mNeedToSkipCharNum = 0;
mLastCharOrder = -1;
mDone = PR_FALSE;
mDataThreshold = aIsPreferredLanguage ? 0 : MINIMUM_DATA_THRESHOLD;
}
#define DONT_KNOW (float)-1
float JapaneseContextAnalysis::GetConfidence(PRBool aIsPreferredLanguage)
float JapaneseContextAnalysis::GetConfidence(void)
{
//This is just one way to calculate confidence. It works well for me.
if (aIsPreferredLanguage || mTotalRel > MINIMUM_DATA_THRESHOLD)
if (mTotalRel > mDataThreshold)
return ((float)(mTotalRel - mRelSample[0]))/mTotalRel;
else
return (float)DONT_KNOW;

View file

@ -51,7 +51,7 @@ extern const PRUint8 jp2CharContext[83][83];
class JapaneseContextAnalysis
{
public:
JapaneseContextAnalysis() {Reset();}
JapaneseContextAnalysis() {Reset(PR_FALSE);}
void HandleData(const char* aBuf, PRUint32 aLen);
@ -74,8 +74,8 @@ public:
mLastCharOrder = order;
}
float GetConfidence(PRBool aIsPreferredLanguage);
void Reset(void);
float GetConfidence(void);
void Reset(PRBool aIsPreferredLanguage);
void SetOpion(){}
PRBool GotEnoughData() {return mTotalRel > ENOUGH_REL_THRESHOLD;}
@ -89,6 +89,9 @@ protected:
//total sequence received
PRUint32 mTotalRel;
//Number of sequences needed to trigger detection
PRUint32 mDataThreshold;
//The order of previous char
PRInt32 mLastCharOrder;

View file

@ -41,7 +41,7 @@ void nsBig5Prober::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mDistributionAnalyser.Reset();
mDistributionAnalyser.Reset(mIsPreferredLanguage);
}
nsProbingState nsBig5Prober::HandleData(const char* aBuf, PRUint32 aLen)
@ -81,7 +81,7 @@ nsProbingState nsBig5Prober::HandleData(const char* aBuf, PRUint32 aLen)
float nsBig5Prober::GetConfidence(void)
{
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence();
return (float)distribCf;
}

View file

@ -46,8 +46,8 @@ void nsEUCJPProber::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mContextAnalyser.Reset();
mDistributionAnalyser.Reset();
mContextAnalyser.Reset(mIsPreferredLanguage);
mDistributionAnalyser.Reset(mIsPreferredLanguage);
}
nsProbingState nsEUCJPProber::HandleData(const char* aBuf, PRUint32 aLen)
@ -91,8 +91,8 @@ nsProbingState nsEUCJPProber::HandleData(const char* aBuf, PRUint32 aLen)
float nsEUCJPProber::GetConfidence(void)
{
float contxtCf = mContextAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float contxtCf = mContextAnalyser.GetConfidence();
float distribCf = mDistributionAnalyser.GetConfidence();
return (contxtCf > distribCf ? contxtCf : distribCf);
}

View file

@ -41,7 +41,7 @@ void nsEUCKRProber::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mDistributionAnalyser.Reset();
mDistributionAnalyser.Reset(mIsPreferredLanguage);
//mContextAnalyser.Reset();
}
@ -84,7 +84,7 @@ nsProbingState nsEUCKRProber::HandleData(const char* aBuf, PRUint32 aLen)
float nsEUCKRProber::GetConfidence(void)
{
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence();
return (float)distribCf;
}

View file

@ -41,7 +41,7 @@ void nsEUCTWProber::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mDistributionAnalyser.Reset();
mDistributionAnalyser.Reset(mIsPreferredLanguage);
//mContextAnalyser.Reset();
}
@ -84,7 +84,7 @@ nsProbingState nsEUCTWProber::HandleData(const char* aBuf, PRUint32 aLen)
float nsEUCTWProber::GetConfidence(void)
{
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence();
return (float)distribCf;
}

View file

@ -46,7 +46,7 @@ void nsGB18030Prober::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mDistributionAnalyser.Reset();
mDistributionAnalyser.Reset(mIsPreferredLanguage);
//mContextAnalyser.Reset();
}
@ -89,7 +89,7 @@ nsProbingState nsGB18030Prober::HandleData(const char* aBuf, PRUint32 aLen)
float nsGB18030Prober::GetConfidence(void)
{
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence();
return (float)distribCf;
}

View file

@ -46,8 +46,8 @@ void nsSJISProber::Reset(void)
{
mCodingSM->Reset();
mState = eDetecting;
mContextAnalyser.Reset();
mDistributionAnalyser.Reset();
mContextAnalyser.Reset(mIsPreferredLanguage);
mDistributionAnalyser.Reset(mIsPreferredLanguage);
}
nsProbingState nsSJISProber::HandleData(const char* aBuf, PRUint32 aLen)
@ -90,8 +90,8 @@ nsProbingState nsSJISProber::HandleData(const char* aBuf, PRUint32 aLen)
float nsSJISProber::GetConfidence(void)
{
float contxtCf = mContextAnalyser.GetConfidence(mIsPreferredLanguage);
float distribCf = mDistributionAnalyser.GetConfidence(mIsPreferredLanguage);
float contxtCf = mContextAnalyser.GetConfidence();
float distribCf = mDistributionAnalyser.GetConfidence();
return (contxtCf > distribCf ? contxtCf : distribCf);
}

View file

@ -32,6 +32,16 @@ function InitDetectorTests()
SetDetectorPref(gDetectorList[0]);
gTestIndex = 0;
$("testframe").onload = DoDetectionTest;
if (gExpectedCharset == "default") {
try {
gExpectedCharset = prefService
.getComplexValue("intl.charset.default",
Components.interfaces.nsIPrefLocalizedString).data;
} catch (e) {
gExpectedCharset = "ISO-8859-8";
}
}
}
function SetDetectorPref(aPrefValue)

View file

@ -66,6 +66,8 @@ _TEST_FILES = \
test_bug431054-japanese.html \
bug488426_text.html \
test_bug488426.html \
bug547487_text.html \
test_bug547487.html \
$(NULL)
libs:: $(_TEST_FILES)

View file

@ -0,0 +1 @@
The quick brown fox jumps over the lazy dog.

View file

@ -0,0 +1,38 @@
<!DOCTYPE HTML>
<html>
<!--
https://bugzilla.mozilla.org/show_bug.cgi?id=547487
-->
<head>
<title>Test for Bug 547487</title>
<script type="text/javascript"
src="chrome://mochikit/content/MochiKit/packed.js"></script>
<script type="text/javascript"
src="chrome://mochikit/content/tests/SimpleTest/SimpleTest.js">
</script>
<script type="text/javascript" src="CharsetDetectionTests.js"></script>
<link rel="stylesheet" type="text/css"
href="chrome://mochikit/content/tests/SimpleTest/test.css" />
</head>
<body>
<a target="_blank" href="https://bugzilla.mozilla.org/show_bug.cgi?id=547487">Mozilla Bug 547487</a>
<p id="display"></p>
<div id="content" style="display: none">
</div>
<iframe id="testframe"></iframe>
<pre id="test">
<script class="testbody" type="text/javascript">
/** Test for Bug 547487 **/
CharsetDetectionTests("bug547487_text.html",
"default",
new Array("zhtw_parallel_state_machine",
"zhcn_parallel_state_machine",
"ja_parallel_state_machine",
"ko_parallel_state_machine",
"zh_parallel_state_machine",
"cjk_parallel_state_machine",
"universal_charset_detector"));
</script>
</pre>
</body>
</html>