java - Regular Expression negate whole regex -

i want parse string regular expression groups:

{4: :35b:isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word :16s:confdet :16r:setdet :22f::setr//trad :11a::fxib//eur :16r:amt :19a::deal//eur222, :16s:amt :16r:amt :19a::loco//eur555 :16s:amt :16r:amt :19a::othr//eur444 :16s:amt :16r:amt :19a::sett//eur333,33 :16s:amt :16s:setdet -} 

i created regex (:\d\d[a-za-z]:*(\w*\/\/)?|:\d\d:)([^:]+) matches in of cases, not in one. want extract groups this:

:35b: => isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (xxxx t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word  :16s: => confdet  :16r: => setdet 

...

i expected there no ':' in second group. maybe can me. need extact whole sting until next :\d\d\w: block.

edit: input string has key-value structure. example :35b: key , behind until next key value (in case of example value 'isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (xxxx t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word'). want extract key-value pairs of input string. here small code example of want:

charsequence swiftmessage = "{4: :35b:isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word :16s:confdet :16r:setdet :22f::setr//trad :11a::fxib//eur :16r:amt :19a::deal//eur222, :16s:amt :16r:amt :19a::loco//eur555 :16s:amt :16r:amt :19a::othr//eur444 :16s:amt :16r:amt :19a::sett//eur333,33 :16s:amt :16s:setdet -}";  pattern pattern = pattern.compile("(:\\d\\d([a-za-z]):*(\\w*//)?|:\\d\\d:)([^:]+)"); matcher matcher = pattern.matcher(swiftmessage);  while( matcher.find() ) {     string key = matcher.group(1);     string value = matcher.group(4);      system.out.println(key + "=>" + value);  } 

expected output (the stucture key=>value):

:35b:=>isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word :16s:=>confdet  :16r:=>setdet  :22f::setr//=>trad  :11a::fxib//=>eur  :16r:=>amt  :19a::deal//=>eur222,  :16s:=>amt  :16r:=>amt  :19a::loco//=>eur555  :16s:=>amt  :16r:=>amt  :19a::othr//=>eur444  :16s:=>amt  :16r:=>amt  :19a::sett//=>eur333,33  :16s:=>amt  :16s:=>setdet -} 

in regex value of key :35b: 'isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f' because regex looks next colon. expexted value should 'isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word'

hopefully it's better understand.

it looks want find tokens separated (space): , treat part before first : in each token key , rest value.

in case can try

(?<key>(?<=\\s):\\d\\d[a-za-z]):(?<value>.*?)(?=\\s:|$) 

which try

• find :\\d\\d[a-za-z] part preceded space (?<=\\s) , put group named key
• find minimal (since *? quantifier reluctant) set of characters until next \\s: or end of string found, , place part in group named value.

so code can

pattern pattern = pattern.compile("(?<key>(?<=\\s):\\d\\d[a-za-z]):(?<value>.*?)(?=\\s:|$)"); matcher matcher = pattern.matcher(swiftmessage); while( matcher.find() ) {     string key = matcher.group("key");     string value = matcher.group("value");      system.out.println(key + "=>" + value);  } 

---

other approach splitting on \\s: data in parts like

{4: 35b:isin de000xxxxxxx disc.z 11.11.11 xxxx90 1234 (hsbc t r.? b.) /f:12345/r:n/w:n/c:n/s:n/g:n/a:n/f:n /xx/any word 16s:confdet ... 16s:setdet -} 

and split again each part on : limited number of splits 2 (so "foo:bar:baz:".split(":",2) becomes ["foo", "bar:baz"]).

with approach code can

for (string token : swiftmessage.tostring().split("\\s:")){     //system.out.println(token);     //lets ignore first `{4:` part     //maybe     if (token.length()<=3) continue;      string[] key_value = token.split(":",2);     system.out.println(":"+key_value[0]+"=>"+key_value[1]); }