How to remove duplicate words from text

[Voetleuce en fênsievry]

Hello everyone.

I'm not a JavaScript author myself, but I'm looking for a method to
remove duplicate words from a piece of text. This text would
presumably be pasted into a text box.

I have, for example, a list of town names, but there are hundreds of
duplicates, like:

"Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness etc."

Can you direct me to a script (or write one, please) that will remove
the excess names and leave just one instance of each name, like:

"Aberdeen Edinburg Inverness etc."

Thanks in advance!
[E-mail answers appreciated, but I'll also check the newsgroup.]

 

[Richard Cornford]

I'm not a JavaScript author myself, but I'm looking for a method to
remove duplicate words from a piece of text. This text would
presumably be pasted into a text box.

I have, for example, a list of town names, but there are hundreds of
duplicates, like:

"Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness etc."

Can you direct me to a script (or write one, please) that will remove
the excess names and leave just one instance of each name, like:

"Aberdeen Edinburg Inverness etc."

You need to be clearer about the nature of your list. How does a
space-separated list of city/town names cope with 'Chipping Sudbury' or
'Milton Kenes'. Having decided how the list is going to distinguish one
distinct place name from the next (comma separated?) you could use the
String.split method to produce an array of strings - var placeNameArray
= listText.split( [ delimiter string or regular expression] ); - and
then create a new Object and assign a new (maybe boolean true) property
to the Object using each place name from the array as the property name.
Duplicates would only server to re-set an existing property. Then a -
for(var propName in Obj) loop would enumerate the created properties of
the object and could be used to create a duplicate-free list.

Richard.

 

[Lasse Reichstein Nielsen]

I'm not a JavaScript author myself, but I'm looking for a method to
remove duplicate words from a piece of text. This text would
presumably be pasted into a text box.

I have, for example, a list of town names, but there are hundreds of
duplicates, like:

"Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness etc."

So your list is a string, and not, e.g., an array.
We need to be able to delimiter the names, so for now I'll assume that
spaces delimiter town names.

Can you direct me to a script (or write one, please) that will remove
the excess names and leave just one instance of each name, like:

"Aberdeen Edinburg Inverness etc."

---
function uniqueTowns(towns){
var arrTowns = towns.split(" ");
var arrNewTowns = [];
var seenTowns = {};
for(var i=0;i<arrTowns.length;i++) {
if (!seenTowns[arrTowns]) {
seenTowns[arrTowns]=true;
arrNewTowns.push(arrTowns);
}
}
return arrNewTowns.join(" ");
}
---
This function splits the argument string "towns" into an array
of strings, then runs through them and adds the string to a new
array the first time it is encountered. This new array is then
joined to a string again and returned.

[E-mail answers appreciated, but I'll also check the newsgroup.]

Good choice
/L

 

[Dr John Stockton]

JRS: In article <[email protected]>, seen in

function uniqueTowns(towns){
var arrTowns = towns.split(" ");
var arrNewTowns = [];
var seenTowns = {};
for(var i=0;i<arrTowns.length;i++) {
if (!seenTowns[arrTowns]) {
seenTowns[arrTowns]=true;
arrNewTowns.push(arrTowns);
}
}
return arrNewTowns.join(" ");
}

That failed in MSIE 4 - no .push().

The following is simpler, and assumes a sorted list as per example :

T = "Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness etc."

function uniqueTowns(towns){
var arrTowns = towns.split(/\s+/), ans = "", Had = ""
for (var i=0 ; i < arrTowns.length ; i++)
if ( Had != (Had=arrTowns) ) ans += Had + " "
return ans }

uniqueTowns(T)

However : it's Edinburgh - and what about Store Heddinge, St. Ives, Cape
Town, & New York?

 

[Lasse Reichstein Nielsen]

That failed in MSIE 4 - no .push().

And in IE/Mac IIRC. So, add it:
---
if (!Array.prototype.push) {
Array.prototype.push = function(elem) {
this[this.length]=elem;
}
}
---

The following is simpler, and assumes a sorted list as per example :

T = "Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness etc."

function uniqueTowns(towns){
var arrTowns = towns.split(/\s+/), ans = "", Had = ""
for (var i=0 ; i < arrTowns.length ; i++)

if ( Had != (Had=arrTowns) )

You depend on the evaluation of the arguments to "!=" to be
left-to-right. I know the specification requires it, but it would be
an easy place for an implementation to break unnoticed, so I would be
weary using it. It's hard to read as well.

ans += Had + " "
return ans }

uniqueTowns(T)

However : it's Edinburgh - and what about Store Heddinge, St. Ives, Cape
Town, & New York?

My version would work for "New York", as long as "York" didn't come
first
It is unsolvable if there is no clear delimiter between towns. I.e.,
if both "York" and "New York" are legal names, and the names are space
separated as well, any solution is bound to break.

/L

 

[Voetleuce en fênsievry]

However : it's Edinburgh - and what about Store Heddinge,
St. Ives, Cape Town, & New York?

Thanks for everyone's answers -- I'll definitely give it a try.

To John Stockton and Richard Cornford, sorry, I should have given more
details (I didn't want to clutter the post with irrelevant details).
I have a postal code list in XLS format in four colums, the fourth of
which contains town names. I extracted the town names to plaintext,
replaced all the spaces with underscores (eg New_York, Cape_Town),
removed all the fullstops and other non-alphabetic characters (eg
St_Yves, St_George_s_Mall) and sorted it alphabetically using a word
processor. Now I'm stuck with a list of names with hundreds of
duplicates (many of these towns have several postal codes, you see).

I'm a translator (EN-AF-EN) and I'd like to use the list as a spelling
reference (chances are that the Post Office will know the correct
spelling of a place name).

I see many applications which are useful to translators which can
easily be solved using JavaScript (and JavaScript is handy too because
it works offline and on most browsers, and even the ones that are
browser specific are specific to commonly available browsers). I have
seen a handy word-count facility, for example. I have also seen a
spell checker (which doesn't work in Opera and which chokes the
computer when trying to use more than 100 k words, heh-heh).
Something to sort words in alphabetical order would be equally useful.

Some might say "but don't MS Word or Trados or Wordfast have all these
features?" -- yes but not all translators have access to a computer
with MS Word installed on it al the time, and most of these JavaScript
scripts would fit on a stiffy for handy use. Plus, I'm sure there are
other functions not currently supported by MS Word which can easily be
accomplished using JavaScript (jargon extraction, for example).

Thanks again for your replies. I don't know JavaScript myself and I
often regret it (they say it's easy, but I find it hard).

 

[Mark Szlazak]

Now for a change in pace, see if this or a similar regex solution will
fit your needs.

inText = 'Aberdeen Aberdeen Aberdeen Edinburg Edinburg Inverness
etc.';
rxDoubled = /\b(\w+)(?\s+)\1\b)+/g
outText = inText.replace(rxDoubled,'$1$2');

If the non-capture group (?: is not supported, change the regex and
ordinals to:

rxDoubled = /\b(\w+)((\s+)\1\b)+/g
outText = inText.replace(rxDoubled,'$1$3');

Also, if you need other non-word-character stuff in between like a tag
then replace (\s+) with ((?:\s|<[^>]+>)+)

 

[Richard Cornford]

Thanks for everyone's answers -- I'll definitely give it a try.

<snip>

This is an example using some of the methods suggested. I did make one
or two changes; lasse's split call used a space character, I changed
that to a regular expression, Dr John Stockton's function requires
sorted input so I added a call to the array sort method (if the input is
pre-sorted then this is not needed and can be removed). On the subject
of sorting, Dr John Stockton's function always produces sorted output,
Lasse's will produce sorted output if the input is sorted but a call to
the array sort method could be added (either just before the final join
or following the split). My - for(var prop in seenTowns) - method cannot
guarantee the order of the output.

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<title></title>
<script type="text/javascript">

function uniqueTownsLRN(frm){
var towns = frm.elements['inputText'].value;
var arrTowns = towns.split(/\s+/);
var arrNewTowns = [];
var seenTowns = {};
for(var i=0;i<arrTowns.length;i++) {
if (!seenTowns[arrTowns]) {
seenTowns[arrTowns]=true;
arrNewTowns.push(arrTowns);
}
}
frm.elements['outputText'].value = arrNewTowns.join(" ");
}

function uniqueTownsDJS(frm){
var towns = frm.elements['inputText'].value;
var arrTowns = towns.split(/\s+/).sort(), ans = "", Had = "";
for(var i=0;i < arrTowns.length;i++)
if( Had != (Had=arrTowns) )ans += Had + " ";
frm.elements['outputText'].value = ans;
}

function uniqueTownsRC(frm){
var towns = frm.elements['inputText'].value;
var st = '', seenTowns = {}, arrTowns = towns.split(/\s+/);
for(var c = arrTowns.length;c--{
seenTowns[arrTowns[c]] = true;
}
for(var prop in seenTowns){
st += prop+' ';
}
frm.elements['outputText'].value = st;
}

</script>

</head>
<body>
<form name="testForm" action="" method="get"
onsubmit="return subTests()">
<textarea name="inputText" cols="70" rows="8">St_Yves St_George_s_Mall
Aberdeen Aberdeen New_York Cape_Town Aberdeen Edinburgh
Edinburgh Inverness St_Yves St_George_s_Mall New_York
Aberdeen Aberdeen New_York Cape_Town Aberdeen Edinburgh
Edinburgh Inverness St_Yves St_George_s_Mall New_York
Cape_Town New_York Cape_Town</textarea><br>
<input type="button" value="Lasse Reichstein Nielsen Version"
onclick="uniqueTownsLRN(this.form);"><br>
<input type="button" value="Dr John Stockton Version"
onclick="uniqueTownsDJS(this.form);"><br>
<input type="button" value="Richard Cornford Version"
onclick="uniqueTownsRC(this.form);"><br>
<textarea name="outputText" cols="70" rows="8"></textarea>
</form>
</body>
</html>

Richard.