how to replace parts of string using regular expressions - java

I am not a beginner to regular expressions, but their use in perl seems a bit different than in Java.
Anyways, I basically have a dictionary of shorthand words and their definitions. I want to iterate over words in the dictionary and replace them with their meanings. what is the best way to do this in JAVA?
I have seen String.replaceAll(), String.replace(), as well as the Pattern/Matcher classes. I wish to do a case insensitive replacement along the lines of:
word =~ s/\s?\Q$short_word\E\s?/ \Q$short_def\E /sig
While I am at it, do you think that it is best to extract all the words from the string and then apply my dictionary or just apply the dictionary to the string? I know that I need to be careful, because the shorthand words could match parts of other shorthand meanings.
Hopefully this all makes sense.
Thanks.
Clarification:
Dictionary is something like:
lol:laugh out loud, rofl:rolling on the floor laughing, ll:like lemons
string is:
lol, i am rofl
replaced text:
laugh out loud, i am rolling on the floor laughing
notice how the ll wasnt added anywhere

The danger is false positives inside of normal words. "fell" != "felikes lemons"
One way is to split the words on whitespace (do multiple spaces need to be conserved?) then loop over the List performing the 'if contains() { replace } else { output original } idea above.
My output class would be a StringBuffer
StringBuffer outputBuffer = new StringBuffer();
for(String s: split(inputText)) {
outputBuffer.append( dictionary.contains(s) ? dictionary.get(s) : s);
}
Make your split method smart enough to return word delimiters also:
split("now is the time") -> now,<space>,is,<space>,the,<space><space>,time
Then you don't have to worry about conserving white space - the loop above will just append anything that isn't a dictionary word to the StringBuffer.
Here's a recent SO thread on retaining delimiters when regexing.

If you insist on using regex, this would work (taking Zoltan Balazs' dictionary map approach):
Map<String, String> substitutions = loadDictionaryFromSomewhere();
int lengthOfShortestKeyInMap = 3; //Calculate
int lengthOfLongestKeyInMap = 3; //Calculate
StringBuffer output = new StringBuffer(input.length());
Pattern pattern = Pattern.compile("\\b(\\w{" + lengthOfShortestKeyInMap + "," + lengthOfLongestKeyInMap + "})\\b");
Matcher matcher = pattern.matcher(input);
while (matcher.find()) {
String candidate = matcher.group(1);
String substitute = substitutions.get(candidate);
if (substitute == null)
substitute = candidate; // no match, use original
matcher.appendReplacement(output, Matcher.quoteReplacement(substitute));
}
matcher.appendTail(output);
// output now contains the text with substituted words
If you plan to process many inputs, pre-compiling the pattern is more efficient than using String.split(), which compiles a new Pattern each call.
(edit) Compiling all of the keys into a single pattern yields a more efficient approach, like so:
Pattern pattern = Pattern.compile("\\b(lol|rtfm|rofl|wtf)\\b");
// rest of the method unchanged, don't need the shortest/longest key stuff
This allows the regex engine to skip over any words that happen to be short enough but aren't in the list, saving you a lot of map accesses.

The first thing, that comes into my mind is this:
...
// eg: lol -> laugh out loud
Map<String, String> dictionatry;
ArrayList<String> originalText;
ArrayList<String> replacedText;
for(String string : originalText) {
if(dictionary.contains(string)) {
replacedText.add(dictionary.get(string));
} else {
replacedText.add(string);
}
...
Or you could use a StringBuffer instead of the replacedText.

Related

How to find a String of last 2 items in colon separated string

I have a string = ab:cd:ef:gh. On this input, I want to return the string ef:gh (third colon intact).
The string apple:orange:cat:dog should return cat:dog (there's always 4 items and 3 colons).
I could have a loop that counts colons and makes a string of characters after the second colon, but I was wondering if there exists some easier way to solve it.
You can use the split() method for your string.
String example = "ab:cd:ef:gh";
String[] parts = example.split(":");
System.out.println(parts[parts.length-2] + ":" + parts[parts.length-1]);
String example = "ab:cd:ef:gh";
String[] parts = example.split(":",3); // create at most 3 Array entries
System.out.println(parts[2]);
The split function might be what you're looking for here. Use the colon, like in the documentation as your delimiter. You can then obtain the last two indexes, like in an array.
Yes, there is easier way.
First, is by using method split from String class:
String txt= "ab:cd:ef:gh";
String[] arr = example.split(":");
System.out.println(arr[arr.length-2] + " " + arr[arr.length-1]);
and the second, is to use Matcher class.
Use overloaded version of lastIndexOf(), which takes the starting index as 2nd parameter:
str.substring(a.lastIndexOf(":", a.lastIndexOf(":") - 1) + 1)
Another solution would be using a Pattern to match your input, something like [^:]+:[^:]+$. Using a pattern would probably be easier to maintain as you can easily change it to handle for example other separators, without changing the rest of the method.
Using a pattern is also likely be more efficient than String.split() as the latter is also converting its parameter to a Pattern internally, but it does more than what you actually need.
This would give something like this:
String example = "ab:cd:ef:gh";
Pattern regex = Pattern.compile("[^:]+:[^:]+$");
final Matcher matcher = regex.matcher(example);
if (matcher.find()) {
// extract the matching group, which is what we are looking for
System.out.println(matcher.group()); // prints ef:gh
} else {
// handle invalid input
System.out.println("no match");
}
Note that you would typically extract regex as a reusable constant to avoid compiling the pattern every time. Using a constant would also make the pattern easier to change without looking at the actual code.

How to find a whole word in a String in Java?

I have a String that I have to parse for different keywords.
For example, I have the String:
"I will come and meet you at the 123woods"
And my keywords are
'123woods'
'woods'
I should report whenever I have a match and where. Multiple occurrences should also be accounted for.
However, for this one, I should get a match only on '123woods', not on 'woods'. This eliminates using String.contains() method. Also, I should be able to have a list/set of keywords and check at the same time for their occurrence. In this example, if I have '123woods' and 'come', I should get two occurrences. Method execution should be somewhat fast on large texts.
My idea is to use StringTokenizer but I am unsure if it will perform well. Any suggestions?
The example below is based on your comments. It uses a List of keywords, which will be searched in a given String using word boundaries. It uses StringUtils from Apache Commons Lang to build the regular expression and print the matched groups.
String text = "I will come and meet you at the woods 123woods and all the woods";
List<String> tokens = new ArrayList<String>();
tokens.add("123woods");
tokens.add("woods");
String patternString = "\\b(" + StringUtils.join(tokens, "|") + ")\\b";
Pattern pattern = Pattern.compile(patternString);
Matcher matcher = pattern.matcher(text);
while (matcher.find()) {
System.out.println(matcher.group(1));
}
If you are looking for more performance, you could have a look at StringSearch: high-performance pattern matching algorithms in Java.
Use regex + word boundaries as others answered.
"I will come and meet you at the 123woods".matches(".*\\b123woods\\b.*");
will be true.
"I will come and meet you at the 123woods".matches(".*\\bwoods\\b.*");
will be false.
Hope this works for you:
String string = "I will come and meet you at the 123woods";
String keyword = "123woods";
Boolean found = Arrays.asList(string.split(" ")).contains(keyword);
if(found){
System.out.println("Keyword matched the string");
}
http://codigounico.blogspot.com/
How about something like Arrays.asList(String.split(" ")).contains("xx")?
See String.split() and How can I test if an array contains a certain value.
Got a way to match Exact word from String in Android:
String full = "Hello World. How are you ?";
String one = "Hell";
String two = "Hello";
String three = "are";
String four = "ar";
boolean is1 = isContainExactWord(full, one);
boolean is2 = isContainExactWord(full, two);
boolean is3 = isContainExactWord(full, three);
boolean is4 = isContainExactWord(full, four);
Log.i("Contains Result", is1+"-"+is2+"-"+is3+"-"+is4);
Result: false-true-true-false
Function for match word:
private boolean isContainExactWord(String fullString, String partWord){
String pattern = "\\b"+partWord+"\\b";
Pattern p=Pattern.compile(pattern);
Matcher m=p.matcher(fullString);
return m.find();
}
Done
public class FindTextInLine {
String match = "123woods";
String text = "I will come and meet you at the 123woods";
public void findText () {
if (text.contains(match)) {
System.out.println("Keyword matched the string" );
}
}
}
Try to match using regular expressions. Match for "\b123wood\b", \b is a word break.
The solution seems to be long accepted, but the solution could be improved, so if someone has a similar problem:
This is a classical application for multi-pattern-search-algorithms.
Java Pattern Search (with Matcher.find) is not qualified for doing that. Searching for exactly one keyword is optimized in java, searching for an or-expression uses the regex non deterministic automaton which is backtracking on mismatches. In worse case each character of the text will be processed l times (where l is the sum of the pattern lengths).
Single pattern search is better, but not qualified, too. One will have to start the whole search for every keyword pattern. In worse case each character of the text will be processed p times where p is the number of patterns.
Multi pattern search will process each character of the text exactly once. Algorithms suitable for such a search would be Aho-Corasick, Wu-Manber, or Set Backwards Oracle Matching. These could be found in libraries like Stringsearchalgorithms or byteseek.
// example with StringSearchAlgorithms
AhoCorasick stringSearch = new AhoCorasick(asList("123woods", "woods"));
CharProvider text = new StringCharProvider("I will come and meet you at the woods 123woods and all the woods", 0);
StringFinder finder = stringSearch.createFinder(text);
List<StringMatch> all = finder.findAll();
A much simpler way to do this is to use split():
String match = "123woods";
String text = "I will come and meet you at the 123woods";
String[] sentence = text.split();
for(String word: sentence)
{
if(word.equals(match))
return true;
}
return false;
This is a simpler, less elegant way to do the same thing without using tokens, etc.
You can use regular expressions.
Use Matcher and Pattern methods to get the desired output
You can also use regex matching with the \b flag (whole word boundary).
To Match "123woods" instead of "woods" , use atomic grouping in regular expresssion.
One thing to be noted is that, in a string to match "123woods" alone , it will match the first "123woods" and exits instead of searching the same string further.
\b(?>123woods|woods)\b
it searches 123woods as primary search, once it got matched it exits the search.
Looking back at the original question, we need to find some given keywords in a given sentence, count the number of occurrences and know something about where. I don't quite understand what "where" means (is it an index in the sentence?), so I'll pass that one... I'm still learning java, one step at a time, so I'll see to that one in due time :-)
It must be noticed that common sentences (as the one in the original question) can have repeated keywords, therefore the search cannot just ask if a given keyword "exists or not" and count it as 1 if it does exist. There can be more then one of the same. For example:
// Base sentence (added punctuation, to make it more interesting):
String sentence = "Say that 123 of us will come by and meet you, "
+ "say, at the woods of 123woods.";
// Split it (punctuation taken in consideration, as well):
java.util.List<String> strings =
java.util.Arrays.asList(sentence.split(" |,|\\."));
// My keywords:
java.util.ArrayList<String> keywords = new java.util.ArrayList<>();
keywords.add("123woods");
keywords.add("come");
keywords.add("you");
keywords.add("say");
By looking at it, the expected result would be 5 for "Say" + "come" + "you" + "say" + "123woods", counting "say" twice if we go lowercase. If we don't, then the count should be 4, "Say" being excluded and "say" included. Fine. My suggestion is:
// Set... ready...?
int counter = 0;
// Go!
for(String s : strings)
{
// Asking if the sentence exists in the keywords, not the other
// around, to find repeated keywords in the sentence.
Boolean found = keywords.contains(s.toLowerCase());
if(found)
{
counter ++;
System.out.println("Found: " + s);
}
}
// Statistics:
if (counter > 0)
{
System.out.println("In sentence: " + sentence + "\n"
+ "Count: " + counter);
}
And the results are:
Found: Say
Found: come
Found: you
Found: say
Found: 123woods
In sentence: Say that 123 of us will come by and meet you, say, at the woods of 123woods.
Count: 5
If you want to identify a whole word in a string and change the content of that word you can do this way. Your final string stays equals, except the word you treated. In this case "not" stays "'not'" in final string.
StringBuilder sb = new StringBuilder();
String[] splited = value.split("\\s+");
if(ArrayUtils.isNotEmpty(splited)) {
for(String valor : splited) {
sb.append(" ");
if("not".equals(valor.toLowerCase())) {
sb.append("'").append(valor).append("'");
} else {
sb.append(valor);
}
}
}
return sb.toString();

The best way to find out that part of the string is potencial RegEx match

how would you do this:
I have a string and some regexes. Then I iterate over the string and in every iteration I need to know if the part (string index 0 to string currently iterated index) of that string is possible full match of one or more given regexes in next iterations.
Thank you for help.
What about a code like this:
// all of *greedy* regexs into a list
List<String> regex = new ArrayList<String>();
// here is my text
String mytext = "...";
String tmp = null;
// iterate over letters of my text
for (int i = 0; i < mytext.length(); i++) {
// substring from 0. position till i. index
tmp = mytext.substring(0, i);
// append regex on sub text
for (String reg : regex ) {
Pattern p = Pattern.compile(reg);
Matcher m = p.matcher(tmp);
// if found, do smt
if (m.find() ) { bingo.. do smt! }
}
}
You could use Matcher.lookingAt() to try to match as much as possible from a given input, but not requiring the whole input to match (.matches() would require the full input to match and .find() would not require the match to start at the beginning).
I don't believe the Java regular expression API provides such "incremental" or "step-by-step" search.
What you could do however, is to formulate your expression using reluctant quantifiers.
[...] The reluctant quantifiers, however, take the opposite approach: They start at the beginning of the input string, then reluctantly eat one character at a time looking for a match. The last thing they try is the entire input string. [...]
If this isn't viable in your case, you could use the Matcher.setRegion method to incrementally increase the region used by the matcher.
So I've been searching for alternatives to Java's standart RegEx library and found one that does the job well - JRegex

String splitting

I have a string in what is the best way to put the things in between $ inside a list in java?
String temp = $abc$and$xyz$;
how can i get all the variables within $ sign as a list in java
[abc, xyz]
i can do using stringtokenizer but want to avoid using it if possible.
thx
Maybe you could think about calling String.split(String regex) ...
The pattern is simple enough that String.split should work here, but in the more general case, one alternative for StringTokenizer is the much more powerful java.util.Scanner.
String text = "$abc$and$xyz$";
Scanner sc = new Scanner(text);
while (sc.findInLine("\\$([^$]*)\\$") != null) {
System.out.println(sc.match().group(1));
} // abc, xyz
The pattern to find is:
\$([^$]*)\$
\_____/ i.e. literal $, a sequence of anything but $ (captured in group 1)
1 and another literal $
The […] is a character class. Something like [aeiou] matches one of any of the lowercase vowels. [^…] is a negated character class. [^aeiou] matches one of anything but the lowercase vowels.
(…) is used for grouping. (pattern) is a capturing group and creates a backreference.
The backslash preceding the $ (outside of character class definition) is used to escape the $, which has a special meaning as the end of line anchor. That backslash is doubled in a String literal: "\\" is a String of length one containing a backslash).
This is not a typical usage of Scanner (usually the delimiter pattern is set, and tokens are extracted using next), but it does show how'd you use findInLine to find an arbitrary pattern (ignoring delimiters), and then using match() to access the MatchResult, from which you can get individual group captures.
You can also use this Pattern in a Matcher find() loop directly.
Matcher m = Pattern.compile("\\$([^$]*)\\$").matcher(text);
while (m.find()) {
System.out.println(m.group(1));
} // abc, xyz
Related questions
Validating input using java.util.Scanner
Scanner vs. StringTokenizer vs. String.Split
Just try this one:temp.split("\\$");
I would go for a regex myself, like Riduidel said.
This special case is, however, simple enough that you can just treat the String as a character sequence, and iterate over it char by char, and detect the $ sign. And so grab the strings yourself.
On a side node, I would try to go for different demarkation characters, to make it more readable to humans. Use $ as start-of-sequence and something else as end-of-sequence for instance. Or something like I think the Bash shell uses: ${some_value}. As said, the computer doesn't care but you debugging your string just might :)
As for an appropriate regex, something like (\\$.*\\$)* or so should do. Though I'm no expert on regexes (see http://www.regular-expressions.info for nice info on regexes).
Basically I'd ditto Khotyn as the easiest solution. I see you post on his answer that you don't want zero-length tokens at beginning and end.
That brings up the question: What happens if the string does not begin and end with $'s? Is that an error, or are they optional?
If it's an error, then just start with:
if (!text.startsWith("$") || !text.endsWith("$"))
return "Missing $'s"; // or whatever you do on error
If that passes, fall into the split.
If the $'s are optional, I'd just strip them out before splitting. i.e.:
if (text.startsWith("$"))
text=text.substring(1);
if (text.endsWith("$"))
text=text.substring(0,text.length()-1);
Then do the split.
Sure, you could make more sophisticated regex's or use StringTokenizer or no doubt come up with dozens of other complicated solutions. But why bother? When there's a simple solution, use it.
PS There's also the question of what result you want to see if there are two $'s in a row, e.g. "$foo$$bar$". Should that give ["foo","bar"], or ["foo","","bar"] ? Khotyn's split will give the second result, with zero-length strings. If you want the first result, you should split("\$+").
If you want a simple split function then use Apache Commons Lang which has StringUtils.split. The java one uses a regex which can be overkill/confusing.
You can do it in simple manner writing your own code.
Just use the following code and it will do the job for you
import java.util.ArrayList;
import java.util.List;
public class MyStringTokenizer {
/**
* #param args
*/
public static void main(String[] args) {
List <String> result = getTokenizedStringsList("$abc$efg$hij$");
for(String token : result)
{
System.out.println(token);
}
}
private static List<String> getTokenizedStringsList(String string) {
List <String> tokenList = new ArrayList <String> ();
char [] in = string.toCharArray();
StringBuilder myBuilder = null;
int stringLength = in.length;
int start = -1;
int end = -1;
{
for(int i=0; i<stringLength;)
{
myBuilder = new StringBuilder();
while(i<stringLength && in[i] != '$')
i++;
i++;
while((i)<stringLength && in[i] != '$')
{
myBuilder.append(in[i]);
i++;
}
tokenList.add(myBuilder.toString());
}
}
return tokenList;
}
}
You can use
String temp = $abc$and$xyz$;
String array[]=temp.split(Pattern.quote("$"));
List<String> list=new ArrayList<String>();
for(int i=0;i<array.length;i++){
list.add(array[i]);
}
Now the list has what you want.

Find a complete word in a string java

I am writing a piece of code in which i have to find only complete words for example if i have
String str = "today is tuesday";
and I'm searching for "t" then I should not find any word.
Can anybody tell how can I write such a program in java?
I use a regexps for such tasks. In your case it should look something like this:
String str = "today is tuesday";
return str.matches(".*?\\bt\\b.*?"); // returns "false"
String str = "today is t uesday";
return str.matches(".*?\\bt\\b.*?"); // returns "true"
A short explanation:
. matches any character, *? is for zero or more times, \b is a word boundary.
More information on regexps can be found here or specifically for java here
String sentence = "Today is Tuesday";
Set<String> words = new HashSet<String>(
Arrays.asList(sentence.split(" "))
);
System.out.println(words.contains("Tue")); // prints "false"
System.out.println(words.contains("Tuesday")); // prints "true"
Each contains(word) query is O(1), so short of implementing your own sophisticated dictionary data structure, this is the fastest most practical solution if you have many words to look for in a text.
This uses String.split to separate out the words from the sentence on the " " delimiter. Other possible variations, depending on how the problem is defined, is to use \b, the word boundary anchor. The problem is considerably more difficult if you must take every grammatical features of natural languages into consideration (e.g. "can't" is split by \b into "can" and "t").
Case insensitivity can be easily introduced by using the traditional case normalization trick: split and hash sentence.toLowerCase() instead, and see if it contains(word.toLowerCase()).
See also
regular-expressions.info -- Anchors
Wikipedia -- String searching algorithm
Wikipedia -- Patricia Trie
String[] tokens = str.split(" ");
for(String s: tokens) {
if ("t".equals(s)) {
// t exists
break;
}
}
String[] words = str.split(" ");
Arrays.sort(words);
Arrays.binarySearch(words, searchedFor);
String str = "today is tuesday";
StringTokenizer stringTokenizer = new StringTokenizer(str);
bool exists = false;
while (stringTokenizer.hasMoreTokens()) {
if (stringTokenizer.nextToken().equals("t")) {
exists = true;
break;
}
}
use a regex like "\bt\b".
you can do that by putting a regex which should end with a space.
I would recommend you use the "split" functionality for String with spaces as separators, then go through these elements one by one and make a direct comparison.
I would suggest using this regex pattern1 = ".\bt\b." instead of pattern2 = ".?\bt\b.?" . Pattern1 will help you to match the complete String if 't' occurs in that string rather than the pattern2 which just reaches the string "t" you are searching for and ignores rest of the string. There is not much difference in two approaches and for your particular use case of returning true/false will run fine both the ways. The one I suggested will help you to improvise the regex in case you make further changes in your use case

Categories

Resources