Pattern: illegal escape character

Hi everyone,

I have been trying to create a regular expression that effectively encompasses the following rules:

starts with \( or \[
ends with \) or \]

i came up with the following

Code:

---

Pattern latex = Pattern.compile(".\\[\[(](.+\\[)\[]");

---

However this produces and illegal escape character exception, prefixing it with two backslashes or none has the same effect and treats the "[" as an opening of a unity statement (probably not the official name) and a "]" as a closing.

Help would be appreciated.

Thanks.

Quote:

---

Originally Posted by coltx

Hi everyone,

I have been trying to create a regular expression that effectively encompasses the following rules:

starts with \( or \[
ends with \) or \]

i came up with the following

Code:

---

Pattern latex = Pattern.compile(".\\[\[(](.+\\[)\[]");

---

However this produces and illegal escape character exception, prefixing it with two backslashes or none has the same effect and treats the "[" as an opening of a unity statement (probably not the official name) and a "]" as a closing.

Help would be appreciated.

Thanks.

---

Note that you have to get that String literal passed through two compilers: first javac.exe parses the String (and changes it) and second the regular expression compiles the (changed) String. If you write:

Code:

---

".\\[\[(](.+\\[)\[]"

---

javac.exe parses and changes the String to:

Code:

---

".\[\[(](.+\[)\[]"

---

... and complains about it: the second and last \[ are illegal escape sequences. As a rule of thumb: you have to type two backslashes to 'send' a single backslash to the regular expression compiler (because javac.exe changes \\ to a single \)

The regular expression compiler treats \[ as an escaped square left bracket that has 'lost' its special meaning.

kind regards,

Jos

Thanks for the reply

I am still unsuccessful though, does the \\[ then got to \[ and escape the first bracket?

because if i double backslash the backslash (".\\\[\[(](.+\\[)\[]") it produces the same exception. Also if i prepend the [ which i want taken literally as a double \\ i get a runtime exception saying group not closed at index 16 (the end of the regex) ".\\[\\[(](.+\\[)\\[]"

".\\[[(](.+\\[)[]" this yields the same results as the last.

and ".\\\[[(](.+\\\[)[]" wont compile with an illegal escape character error

Quote:

---

Originally Posted by coltx

Thanks for the reply

I am still unsuccessful though, does the \\[ then got to \[ and escape the first bracket?

because if i double backslash the backslash (".\\\[\[(](.+\\[)\[]") it produces the same exception. Also if i prepend the [ which i want taken literally as a double \\ i get a runtime exception saying group not closed at index 16 (the end of the regex) ".\\[\\[(](.+\\[)\\[]"

".\\[[(](.+\\[)[]" this yields the same results as the last.

and ".\\\[[(](.+\\\[)[]" wont compile with an illegal escape character error

---

I always work backwards with the awful regular expression Strings. e.g. I want to escape a left square bracket, so I want to send \[ to the regular expression compiler. Javac.exe treats the backslash as a special character so I have to make it keep its mouth shut so I send \\[ to the java compiler. The result is the literal String "\\[".

kind regards,

Jos

Quote:

---

Originally Posted by coltx

starts with \( or \[

---

here's an example for your first rule. You don't have to use a class range. here's how

Code:

---

Pattern.compile("^(\\[|\\()")

---

Use a capturing group, then use the alternation "|" (logical OR). The caret ^ you should already know is match at the beginning. Similarly, the check for ending character, use $

On second thought: if you want to check whether or not a String starts with ( or [ and ends with ) or ] there is no need to fire up the entire regular expression moloch: check if the length of the String is at least 2 and check if the first character equals a ( or a [ and do the same with the last character for ) or ].

It'll be a bit more code on your side but under the hood the regular expression classes won't be loaded.

kind regards,

Jos

I just wanted to mention that I have found it useful to read regex patterns from a file or from a text box. Hard coding them requires you to double escape as other users have mentioned, which is VERY confusing to look at. If you write your regex pattern into a text file and just read the text file, no double escaping is needed, its done behind the scenes! Good luck!