Re^2: Looking for a cleaner regex ( trie since 5.10 ! )

[LanX]

> See Regexp::Optimizer

The author should add a prominent remark that Perl is supporting trie optimization for over 10 years already. ..

perl5100delta#Trie-optimisation-of-literal-string-alternations

So what's the benefit of this module with current Perl, unless you generate a regex for another language?

update: see also Re^3: Looking for a cleaner regex (trie) ff

[Anonymous Monk]

making the regex shorter and easier to read?

[choroba]

Do you really find /foo(?:[bx]ar|zap)/ easier to read than /foobar|fooxar|foozap/? (Example copied from the documentation).

Or,

/(?^:(?^x:(?:s(?:e(?:t(?:(?:(?:(?:hos|ne)t|gr)en|s(?:erven|ockop))t|p(
+?:r(?:iority|otoent)|went|grp))|m(?:ctl|get|op)|ek(?:dir)?|lect|nd)|y
+(?:s(?:write|call|open|read|seek|tem)|mlink)|h(?:m(?:write|read|ctl|g
+et)|utdown)|ocket(?:pair)?|(?:qr|ta)t|printf|ubstr|leep|rand|in)|g(?:
+e(?:t(?:p(?:r(?:oto(?:byn(?:umber|ame)|ent)|iority)|w(?:ent|nam|uid)|
+eername|grp|pid)|s(?:erv(?:by(?:name|port)|ent)|ock(?:name|opt))|host
+(?:by(?:addr|name)|ent)|net(?:by(?:addr|name)|ent)|gr(?:ent|gid|nam)|
+login|c))?|mtime|lob|t)|r(?:e(?:ad(?:lin[ek]|pipe|dir)?|(?:quir|vers|
+nam)e|winddir|set|cv|f)|index|mdir|and)|c(?:h(?:r(?:oot)?|dir|mod|own
+)|o(?:n(?:tinue|nect)|s)|lose(?:dir)?|aller|rypt|mp)|l(?:o(?:c(?:alti
+me|k)|g)|i(?:sten|nk)|c(?:first)?|e(?:ngth)?|(?:sta)?t)|e(?:nd(?:(?:h
+os|ne)t|p(?:roto|w)|serv|gr)ent|x(?:ec|it|p)|of|q)|u(?:(?:n(?:lin|pac
+)|mas)k|c(?:first)?|time)|f(?:or(?:mline|k)|c(?:ntl)?|ileno|lock)|w(?
+:a(?:it(?:pid)?|ntarray|rn)|rite)|m(?:sg(?:ctl|get|rcv|snd)|kdir)|t(?
+:ell(?:dir)?|runcate|imes?)|d(?:bm(?:close|open)|ump|ie)|a(?:ccept|la
+rm|tan2|bs|nd)|__(?:(?:FIL|LIN)E|SUB)__|o(?:pen(?:dir)?|rd?|ct)|b(?:i
+n(?:mode|d)|less)|i(?:n(?:dex|t)|octl)|v(?:alues|ec)|p(?:ack|ipe)|n(?
+:ot|e)|quotemeta|x(?:or)?|CORE|join|kill|hex)))/

versus

/CORE|__FILE__|__LINE__|__SUB__|abs|accept|alarm|and|atan2|bind|binmod
+e|
bless|caller|chdir|chmod|chown|chr|chroot|close|closedir|cmp|connect|
continue|cos|crypt|dbmclose|dbmopen|die|dump|endgrent|endhostent|endne
+tent|
endprotoent|endpwent|endservent|eof|eq|exec|exit|exp|fc|fcntl|fileno|f
+lock|
fork|formline|ge|getc|getgrent|getgrgid|getgrnam|gethostbyaddr|
gethostbyname|gethostent|getlogin|getnetbyaddr|getnetbyname|getnetent|
getpeername|getpgrp|getppid|getpriority|getprotobyname|getprotobynumbe
+r|
getprotoent|getpwent|getpwnam|getpwuid|getservbyname|getservbyport|
getservent|getsockname|getsockopt|glob|gmtime|gt|hex|index|int|ioctl|j
+oin|
kill|lc|lcfirst|le|length|link|listen|localtime|lock|log|lstat|lt|mkdi
+r|
msgctl|msgget|msgrcv|msgsnd|ne|not|oct|open|opendir|or|ord|pack|pipe|
quotemeta|rand|read|readdir|readline|readlink|readpipe|recv|ref|rename
+|
require|reset|reverse|rewinddir|rindex|rmdir|seek|seekdir|select|semct
+l|
semget|semop|send|setgrent|sethostent|setnetent|setpgrp|setpriority|
setprotoent|setpwent|setservent|setsockopt|shmctl|shmget|shmread|shmwr
+ite|
shutdown|sin|sleep|socket|socketpair|sprintf|sqrt|srand|stat|substr|
symlink|syscall|sysopen|sysread|sysseek|system|syswrite|tell|telldir|t
+ime|
times|truncate|uc|ucfirst|umask|unlink|unpack|utime|values|vec|wait|
waitpid|wantarray|warn|write|x|xor/x

[LanX]

Especially since you can keep the alternatives inside an qw/ / or an @array and simply join "|", for the regex string.

Maintainability at its best.

See also Re: How to tokenize string by custom dictionary? (+code) for an example with a %hash .

[Anonymous Monk]

It isnt limited by trie max limit maybe?