src/protocols.xfst

define URLChar [Char|[Sym - ["<"|">"|%"]]];

define Dot "."|[["["|"("] "d" "o" "t" [")"|"]"]] .o. Caseinsensitive;
define At "@"|[["["|"("] "a" "t" [")"|"]"]] .o. Caseinsensitive;

define TldEnd [{org}|{de}|{com}] .o. Caseinsensitive;

! Very relaxed URL scheme, not based on the strict Lucene implementation
define URL [ [ [{http} (s) | {ftp} | {file}] ":" "/" "/"] | [{www} Dot] ]
URLChar [URLChar|SP]* URLChar
.o. Caseinsensitive;

define Domain Char+ [Dash Char+]* Dot TldEnd;

! Email addresses
define Email URLChar+ At URLChar+ [Dot URLChar+]+;

! Twitter user, hashtag, Google+
define SNS ["@"|"#"|"+"] Char+;

define FileEnd [
                [{htm} ("l")]|
                [{doc} ("x")]|
                {pdf}|
                ["j" "p" ("e") "g"]|
                ["m" "p" ["3"|"4"]]|
                {ogg}|
                {png}|
                [{ppt} ("x")]|
                {avi}|
                {txt}|
                {xls}|
                {xml}|
                {aac}|
                {gif}|
                {exe}
                ] .o. Caseinsensitive;

define File (( AsciiLetter ":" %\ | "/" ) [ Char | "_" | "-" | Char [ %\ | "/" ] ]*) [Char | "-" | "_" ]+ "." FileEnd;