7/20/2007

中國的食品污染

0 comments

今天讀到大紀元的一篇報導:毒牙膏再演鬧劇 質檢總局大變臉
http://epochtimes.com/b5/7/7/20/n1778129.htm

該文提到因中國外銷的含毒牙膏所引發的一連串新聞。文中提到的「二甘醇」是一種抗凝劑,若水中含有二甘醇,則在低溫時會防止水分子結成冰。二甘醇的含量越高,越不易結冰,可能在零下二、三十度時都還保持液體狀態不會變成冰塊。中國製的牙膏,就是利用二甘醇的這種特性,使其即使保存很久也不會硬化。這是對人體有害的有毒物質,因此像牙膏這種不小心會吃進人體的東西不應該含有二甘醇。

其他重點:

  1. 中國生產的幾種牙膏含有0.21%至7.5%的二甘醇。
  2. 中國衛生部不久公佈了該部「專家」的評估結論:「長期使用二甘醇含量低於15.6%的牙膏不會對人體健康產生不良影響」,但報告未顯示專家姓名,也沒說明該試驗的具體方法。
  3. 7月16日,中國質檢總局局長公開批評境外媒體對中國出口產品質量問題的報導是"顛倒黑白",他表示,中國食品合格率要高於美國。
  4. 中國工廠即使自己想要合法生產也很困難,因為他們購買的材料本身就是非法的有毒物(可能買不到無毒的)。譬如說,生產用的水,前幾年由於污染的加劇,中國政府不是採取措施防治並減少污染,反而降低水質標準,把以前的污染水當成合格水賣給企業
  5. 今年七月,據中國質檢總局對食品、化肥和農用機械等眾多消費品質量抽查,結果發現19.1%的產品達不到標準。其中最嚴重的是食品問題,特別是罐裝食品、乾果和果汁飲料。
  6. 中國常用油鹽醬醋米麵的合格率平均低於40%,特別是低價食品,其衛生質量根本無法保障。
  7. 中國進口到美國的食品只有 0.9% 有被抽查。即使在如此低的抽樣率下,中國裝船的食品每個月有 200 次在港口被拒收,排名世界第一。
  8. 中國食品藥品質量監測管理局2005年批准了1113個新藥,這數目等於美國近十年來新藥的「受理」總量(看清楚,不是批准量)。

About Subclassing in Javascript

0 comments

Let the superclass be:

function Person(sex){alert('I am Person')}

and subclasses:

function Man(){this.sex='M'}
function Woman(){this.sex='F'}

The following two subclassing approaches don't seem to make any difference:

Man.prototype = new Person
Woman.prototype = new Person()
m = new Man()
w = new Woman()

In both cases:
  • The function Person will be executed
  • The constructor of both m and w are Person


4/04/2007

IE's TextRange.compareEndPoints()

0 comments

TextRange object

The other day I was coding a web page that needs to make use of some javascript functions of textarea element. I did a little dig-out and ran into this compareEndPoints function. It is a function attached to a TextRange object. The IE-specific TextRange object serves as a mediator for the text input element (<input> with type="text", <textarea>...) to manupilate text-related operations (search, replace, copy, paste, insert ...).

Now assuming we have a <textarea> element object called oTextArea, a TextRange object can be created in one of the following ways:
  1. var oTR1 = oTextArea.createTextRange()
  2. var oTR2 = document.selection.createRange()
These code should appear inside an eventhandler, for example:

  oTextArea.onselect = function(){
...
var oTR1= this.createTextRange()
var oTR2= document.selection.createRange()
...
}

Both oTR1 and oTR2 are a TextRange object that has the following properties:

PropertiesoTR1oTR2
boundingHeight1616
boundingLeft15141
boundingTop124124
boundingWidth25256
offsetLeft15141
offsetTop124124
textThis is the test
<textarea> element
textarea
htmlTextThis is the test
&lt;textarea&gt; element.
textarea

Note that:
  1. What createTextRange returns (=oTR1) covers the entire value range of <textarea> while createRange covers only the range of text that is selected;
  2. All the numerical property values in TextRange object are in pixel. No information of index (the start and end point for the selected text range) is available;

compareEndPoints (type, textRange)

I won't go through all the methods of the TextRange object but this one compareEndPoints(type, textRange). This function checks the relationship between two TextRange objects by comparing the starting and ending indices of textRange.text :

  textRange1.compareEndPoints("StartToStart", textRange2)

The argument type can be one of the fillowing 4 : "StartToStart", "StartToEnd", "EndToStart" and "EndToEnd". When it is "StartToStart", it compares the starting index of textRange1.text with the that of textRange2.text. The return is either -1, 0 or 1.

Now, the reason that I talk about this compareEndPoints specifically. Since it is used upon two TextRange objects, and as we saw earlier that both createRange() and createTextRange() return a TextRange object, instinctly it would work like this:

   var oTR1= oTextArea.createTextRange()
var oTR2= document.selection.createRange()
var x = oTR1.compareEndPoints("StartToStart", oTR2)

However, it won't work. IE rejects this usage by showing "illegal argument" error. It took me a while of searching and code digging to figure out a way to do this:

   var oTR1 = document.body.createTextRange(); // Create TextRange on entire body 
oTR1.moveToElementText(oTextArea) // Use moreToElementText to move
// the TextRange to the textarea

var oTR2= document.selection.createRange();
var x = oTR1.compareEndPoints("StartToStart", oTR2)

Only when the createTextRange( ) is used in this way can the resulting TextRange object be used together in compareEndPoints( ) with that generated by createRange( ).

3/31/2007

My javascript addEvent

1 comments

In the end of 2005 several web programming gurus, Peter-Paul Koch (ppk), Dean Edward and Scott Andrew LePera host a "javascript recoding contest" in which they asked participants to submit their code for adding and removing events in javascript. A template page was designed to test the code. The winner was John Resig, whose original code is here. The final code was slightly modified version:

function addEvent( obj, type, fn )
{
if (obj.addEventListener)
obj.addEventListener( type, fn, false );
else if (obj.attachEvent)
{
obj["e"+type+fn] = fn;
obj[type+fn] = function() { obj["e"+type+fn]( window.event ); }
obj.attachEvent( "on"+type, obj[type+fn] );
}
}
function removeEvent( obj, type, fn )
{
if (obj.removeEventListener)
obj.removeEventListener( type, fn, false );
else if (obj.detachEvent)
{
obj.detachEvent( "on"+type, obj[type+fn] );
obj[type+fn] = null;
obj["e"+type+fn] = null;
}
}

I personally don't feel comfortable with these code. First of all, as one of the comments mentioned (#2 Posted by Tino Zijdel on 18 October 2005) that in cases of multiple-eventhandlers in IE, the eventhandler added last will be executed first:

... since this is for IE a wrapper around attachEvent, the execution order of the events is different ('random' according to Microsofts documentation, but LIFO in practice against FIFO using the W3C event model)

A simple test reveals this problem:

function init(){
var btn=document.createElement('input')
btn.value = 'Click me'
btn.type = 'button'
document.body.appendChild( btn )
addEvent( btn, "click", function(){alert("#1")} )
addEvent( btn, "click", function(){alert("#2")} )
}

Run the init() function as the onload eventhandler (<body onload="init()">) and load with browser. Firefox alerts "#1" then "#2". But IE alerts "#2" first then "#1", which is not what it should be.

Another part making me uncomfortable is that it uses the entire function content as the key name of a hash:

   obj["e"+type+fn] = fn;
obj[type+fn] = function() { obj["e"+type+fn]( window.event ); }

Although none obvioiusly problem is mentioned or discussed, it could result in extremely long and complex key name, which, IMHO, is not what a hash is supposed to be.

I therefore came up with my own version:

function addEvent( obj, type, fn ) 
{
obj.eventHooks = obj.eventHooks || {} // set .eventHooks for obj if not exists
var evs = obj.eventHooks[type] || [] // set .eventHooks[type] if not exists
obj.eventHooks[type] = evs.concat( fn.concat? fn:[fn]) // this allows for multiple fns added
obj['on'+type] = function (e) // like addEvent(obj,"click",[f1,f2])
{
for (var i=0;i<this.eventHooks[type].length;i++)
{
this.tmp = this.eventHooks[type][i] // attach to the obj such that the this
this.tmp(e) // inside the function points to the obj correctly
this.tmp = undefined
}
}
obj = null
}
function removeEvent( obj, type, fn )
{
if (obj.eventHooks && obj.eventHooks[type])
{
var evs = obj.eventHooks[type]
for (var i=0; i<evs.length;i++)
{
if (evs[i]==fn)
{
obj.eventHooks[type] = evs.slice(0,i).concat(evs.slice(i+1))
break
}
}
}
obj = null
}

Unlike most addEvent inventions, in which either addEventListener or attachEvent is used, the above code simply stores eventhandlers into a buffer that has the following structure:

  obj.eventHooks = { click    : [f1, f2]
, keyup : []
, mouseover: [f3]
....
}

For each event, say, click, a function is assigned to the .onclick. This function carries out the following steps:
  1. Check if obj.eventHooks exists. If not, create one
  2. Check if obj.eventHooks.click exists. If not, create one
  3. Loop through each functions in obj.eventHooks.click
  4. For each function, attaches it to obj as a temporary obj.tmp, then executes obj.tmp(e). This makes sure that any this keyword inside that function point to the obj correctly.
A test page, using the contest template, is here. I am sure that there will be some downside along with this code, but at least at this moment it seems to solve the problems came with the winning code.

3/20/2007

Javascript string replacing function

0 comments

In python there's a very convinient string replacing function that comes with every string. The following example show how it works:

To construct a string:   "firstname:Runsun, lastname:Pan, project:wysiwyc"

firstname='Runsun'
lastname ='Pan'
project ='wysiwyc'
s= 'firstname:<b>%s</b>, lastname:<b>%s</b>, project:<b>%s</b>'%(firstname, lastname, project)

The symbol % attaching to the end of a string is a function to insert a series of strings (firstname, lastname, project) into the original string. This not only makes it easier to read but also easier to type, as compared to:

'firstname:<b>'+ firstname +'</b>, lastname:<b>'+ lastname+ '</b>, project:<b>'+ project + '</b>'

In javascript you don't have that luxury. I came up with a tiny function that does just that:

function ___(s)
{
var ss = arguments[0].split('___')
var out= [ss[0]]
for (var i=1;i<ss.length;i++){
out.push(arguments[i])
out.push(ss[i])
}
return out.join('')
}

So now you can do this:

s= ___('firstname:<b>___</b>, lastname:<b>___</b>, project:<b>___</b>',firstname, lastname, project)

I found that this syntax is actually better than its python cousin, because the symbol "___" implies that there's something to be filled, just like what we used to do when filling up a form with a pen.

It would look even more clear if we attach this "___()" function to the string prototype:

String.prototype._= function()
{

var ss = this.valueOf().split('___')
var out= [ss.shift()]
for (var i=0;i<ss.length;i++){
out.push(arguments[i])
out.push(ss[i])
}
return out.join('')
}

which allows this:

s= 'firstname:<b>___</b>, lastname:<b>___</b>, project:<b>___</b>'._(firstname, lastname, project)

Some other uses of this function:

Example-1:

alert('myArray=[___]'._(myArray))

Example-2:

tag = '<___ style="___">___</___>'
DIV = tag._('div','___','___','div')
SPAN = tag._('span','___','___','span')
adiv = DIV._("border:1px dashed blue; color:blue", "div example")
aspan= SPAN._("color:red;background-color:yellow", "span example")

The 2nd example gives:

div example
span example

Replace with a data in a hash


The above ___() function is already very helpful. But the original python % has one other feature:

data= {'fname':'Runsun'
,'lname':'Pan'
,'proj' :'wysiwyc'}
s= 'firstname:<b>%(fname)s</b>, lastname:<b>%(lname)s</b>, project:<b>%(proj)s</b>'%(data)

By inserting a (fname) into %s, you allow the data be in hash form. Now lets put this feature into the ___() function.

Stand-alone ___() function:

function ___(s)
{
var isObj=false
try {
isObj = arguments[1].constructor.toString().split('(')[0].split(' ')[1]=='Object'
}catch(e) {}

if (isObj)
{
var args = arguments[1]
for (var k in args) { s=s.split( '_'+k+'_').join( args[k] ) }
return s

}else{

var ss = arguments[0].split('___')
var out= [ss[0]]
for (var i=1;i<ss.length;i++){
out.push(arguments[i])
out.push(ss[i])
}
return out.join("")
}
}

Attached to string prototype:

String.prototype._= function()
{
var ss= this.valueOf()
var isObj=false
try {
isObj = arguments[0].constructor.toString().split('(')[0].split(' ')[1]=='Object'
}catch(e) {}

if (isObj)
{
var args = arguments[0]
for (var k in args) { ss=ss.split( '_'+k+'_').join( args[k] ) }
return ss

}else{

var ss = ss.split('___')
var out= [ss.shift()]
for (var i=0;i<ss.length;i++){
out.push(arguments[i])
out.push(ss[i])
}
return out.join("")
}
}

Now we can do this in javascript:

data= {fname:'Runsun'
,lname:'Pan'
,proj :'wysiwyc'}
s= 'firstname:<b>_fname_</b>, lastname:<b>_lname_</b>, project:<b>_proj_</b>'._(data)

Again, IMO, this looks better than its python cousin does.

UPDATE(7/23/07): See a regular expression version of the same function.

3/15/2007

HTML tablizer

0 comments

Introducing a html tablizer -- an online utilite to make simple html table.

The program has an Input Section and Output Section:


Enter your table data in the Input Section and either hit [Submit] button or simply move the cursor away from the input box. Then the resulting table will be shown immediately in the Demonstration in the Output Section. Also, the corresponding HTML code will be displayed in HTML code for copy and paste.

After the HTML code and the demo result are shown, you can modify the HTML code and see the result immediately:



If the resulting table is too wide:



You can hit the button [demo on top] to bring the table up:



Enjoy tablizer here.

3/11/2007

△ Recovering CHK files

0 comments

I have an old One-Touch external hard drive from Maxtor. It went crazy and generated 10,000 FILE????.CHK files about a year ago. My wife and I were worried about that the entire hard drive has gone unstable so we simply gave up using it. Recently I found some spare time trying to search for some tools to recover those CHK files.

RecognizeImageFiles.py


One of the most important file types to recover is our photos. The first step I tried was to find some python script that can recognize image files by file content but not by file extensions. There is indeed such a nice little script imghdr.py in python repository. I modified it a little to have it specifically useful in CHK image file recovery (RecognizeImageFiles).
Screenshot of RecognizeImageFiles.py execution

It was quite successful and about 500 image files were recovered. But then, what about other types of files? I went online to search for third-party tools.

UnCHK and FileCHK


Surprisingly there isn't too many tools online. The most popular tools seem to be UnCHK (by Eric Phelps) and FileCHK (by Martin Kratz). They seem to be neat and useful. However, UnCHK (version UnCHK3) froze at the very beginning:



and FileCHK gave an error that reads:

Runtime error "9":
Subscript out of range

It's possible that they couldn't recognize my One-Touch in drive F. Whatever, they don't work in my case.

MediaHeal


I then found a MediaHeal from officerecovery. The description reads:

Recovers files from corrupted hard disks. Supported are all common file systems, including NTFS, FAT12, VFAT12, FAT16, and FAT32.

I downloaded the MediaHeal for Hard Drives and gave it a try. Unfortunately, this program only aims at the entire hard drives but doesn't allow users to check folder(s):

MediaHeal doesn't allow you to choose a folder (image shrunk to fit screen)

Much worse, it froze at the very beginning on this screen:

The excution of MediaHeal demo version froze at the very beginning
(image shrunk to fit screen)


RecoverMyFiles


Finally, I found RecoverMyFiles from GetData. According to their website, this program allows users to recover deleted files even after the hard disk is reformatted (I didn't even know that it's possible):

Recover deleted files even if emptied from the Recycle Bin
Recover formatted hard drives, even if you have reinstalled Windows!
Recover your files after a hard disk crash
Get back files after a partitioning error
Recover documents, photos, video music and email.
Recover from hard drive, camera card, USB, Zip, floppy disk or other media

and it's able to recover 200 types of files. I downloaded the trial version and was amazed how efficient it was:

The screenshot of RecoverMyFile excution.

It ran smoothly and took less than 4 minutes to complete checking my 250GB One-Touch. You can check the content of each file to decide what file is valuable. Since this is a trial vresion so it doesn't allow real rescuing. But it already impressed me.

Other links


FILExt.com
http://filext.com/detaillist.php?extdetail=chk&Search=Search

Look for info about a file ext
http://shell.windows.com/fileassoc/0409/xml/redir.asp?Ext=jpg

A list of common file extensions
http://safari.oreilly.com/0596002491/winxpnut-APP-F

OfficeRecovery.com (data recovery software for corrupted files)
http://www.officerecovery.com/mediaheal/index.htm