Text.Regex.XMLSchema.Generic

grep like filter for lists of strings

The regular expression may be prefixed with the usual context spec "^" for start of string, and "\<" for start of word. and suffixed with "$" for end of text and "\>" end of word. Word chars are defined by the multi char escape sequence "\w"

Examples

grep "a"    ["_a_", "_a", "a_", "a", "_"]      => ["_a_", "_a", "a_", "a"]
grep "^a"   ["_a_", "_a", "a_", "a", "_"]      => ["a_", "a"]
grep "a$"   ["_a_", "_a", "a_", "a", "_"]      => ["_a", "a"]
grep "^a$"  ["_a_", "_a", "a_", "a", "_"]      => ["a"]
grep "\\<a" ["x a b", " ax ", " xa ", "xab"]   => ["x a b", " ax "]
grep "a\\>" ["x a b", " ax ", " xa ", "xab"]   => ["x a b", " xa "]

grep with extended regular expressions

grep with already prepared Regex (ususally with parseContextRegex)

grep with Regex and line numbers

convenient function for matchRE

Examples:

match "x*" "xxx" = True
match "x" "xxx"  = False
match "[" "xxx"  = False

match with extended regular expressions

convenient function for matchRE

Examples:

matchSubex "({1}x*)"                 "xxx"      = [("1","xxx")]
matchSubex "({1}x*)"                 "y"        = []
matchSubex "({w}[0-9]+)x({h}[0-9]+)" "800x600"  = [("w","800"),("h","600")]
matchSubex "[" "xxx"                            = []

convenient function for sedRE

examples:

sed (const "b") "a" "xaxax"       = "xbxbx"
sed (\ x -> x ++ x) "a" "xax"     = "xaax"
sed undefined       "[" "xxx"     = "xxx"

convenient function for splitRE

examples:

split "a*b" "abc" = ("ab","c")
split "a*"  "bc"  = ("", "bc")    -- "a*" matches ""
split "a+"  "bc"  = ("", "bc")    -- "a+" does not match, no split
split "["   "abc" = ("", "abc")   -- "["  syntax error, no split

split with extended syntax

convenient function for splitSubex, uses extended syntax

examples:

splitSubex "({1}a*)b"  "abc" = ([("1","a")],"c")
splitSubex "({2}a*)"   "bc"  = ([("2","")], "bc")
splitSubex "({1}a|b)+" "abc" = ([("1","a"),("1","b")],"c")        -- subex 1 matches 2 times

splitSubex ".*({x}a*)" "aa"  = ([("x",""),("x","a"),("x","aa")],"")
                                                                  -- nondeterminism: 3 matches for a*

splitSubex "({1}do)|({2}[a-z]+)" "do you know"
                               = ([("1","do"),("2","do")]," you know")
                                                                  -- nondeterminism: 2 matches for do

splitSubex "({1}do){|}({2}[a-z]+)" "do you know"
                               = ([("1","do")]," you know")
                                                                  -- no nondeterminism with {|}: 1. match for do

splitSubex "({1}a+)"   "bcd" = ([], "bcd")                        -- no match
splitSubex "["         "abc" = ([], "abc")                        -- syntax error

split a string into tokens (words) by giving a regular expression which all tokens must match.

Convenient function for tokenizeRE

This can be used for simple tokenizers. It is recommended to use regular expressions where the empty word does not match. Else there will appear a lot of probably useless empty tokens in the output. All none matching chars are discarded. If the given regex contains syntax errors, Nothing is returned

examples:

tokenize "a" "aabba"      = ["a","a","a"]
tokenize "a*" "aaaba"     = ["aaa","a"]
tokenize "a*" "bbb"       = ["","",""]
tokenize "a+" "bbb"       = []

tokenize "a*b" ""         = []
tokenize "a*b" "abc"      = ["ab"]
tokenize "a*b" "abaab ab" = ["ab","aab","ab"]

tokenize "[a-z]{2,}|[0-9]{2,}|[0-9]+[.][0-9]+" "ab123 456.7abc"
                          = ["ab","123","456.7","abc"]

tokenize "[a-z]*|[0-9]{2,}|[0-9]+[.][0-9]+" "cab123 456.7abc"
                          = ["cab","123","456.7","abc"]

tokenize "[^ \t\n\r]*" "abc def\t\n\rxyz"
                          = ["abc","def","xyz"]

tokenize ".*"   "\nabc\n123\n\nxyz\n"
                          = ["","abc","123","","xyz"]

tokenize ".*"             = lines

tokenize "[^ \t\n\r]*"    = words

tokenize with extended syntax

convenient function for tokenizeRE'

When the regular expression parses as Zero, [Left input] is returned, that means no tokens are found

convenient function for tokenizeSubexRE a string

examples:

tokenizeSubex "({name}[a-z]+)|({num}[0-9]{2,})|({real}[0-9]+[.][0-9]+)"
                "cab123 456.7abc"
                                 = [("name","cab")
                                   ,("num","123")
                                   ,("real","456.7")
                                   ,("name","abc")]

tokenizeSubex "({real}({n}[0-9]+)([.]({f}[0-9]+))?)"
                "12.34"          = [("real","12.34")
                                   ,("n","12")
                                   ,("f","34")]

tokenizeSubex "({real}({n}[0-9]+)([.]({f}[0-9]+))?)"
                 "12 34"         = [("real","12"),("n","12")
                                   ,("real","34"),("n","34")]

tokenizeSubex "({real}({n}[0-9]+)(([.]({f}[0-9]+))|({f})))"
                 "12 34.56"      = [("real","12"),("n","12"),("f","")
                                   ,("real","34.56"),("n","34"),("f","56")]

match a string with a regular expression

match a string with a regular expression and extract subexpression matches

sed like editing function

All matching tokens are edited by the 1. argument, the editing function, all other chars remain as they are

split a string by taking the longest prefix matching a regular expression

Nothing is returned in case there is no matching prefix, else the pair of prefix and rest is returned

split a string by removing the longest prefix matching a regular expression and then return the list of subexpressions found in the matching part

Nothing is returned in case of no matching prefix, else the list of pairs of labels and submatches and the rest is returned

The function, that does the real work for tokenize

split a string into tokens and delimierter by giving a regular expression which all tokens must match

This is a generalisation of the above tokenizeRE functions. The none matching char sequences are marked with Left, the matching ones are marked with Right

If the regular expression contains syntax errors Nothing is returned

The following Law holds:

concat . map (either id id) . tokenizeRE' re == id

split a string into tokens (pair of labels and words) by giving a regular expression containing labeled subexpressions.

This function should not be called with regular expressions without any labeled subexpressions. This does not make sense, because the result list will always be empty.

Result is the list of matching subexpressions This can be used for simple tokenizers. At least one char is consumed by parsing a token. The pairs in the result list contain the matching substrings. All none matching chars are discarded. If the given regex contains syntax errors, Nothing is returned

construct the r.e. for the empty set. An (error-) message may be attached

construct the r.e. for the set containing the empty word

construct an r.e. for a single char set

construct an r.e. for an intervall of chars

mkSym generaized for strings

construct an r.e. for the set of all Unicode chars

construct r.e. for r*

construct an r.e. for the set of all Unicode words

construct the r.e for r1|r2

construct the r.e. for r1{|}r2 (r1 orElse r2).

This represents the same r.e. as r1|r2, but when collecting the results of subexpressions in (...) and r1 succeeds, the subexpressions of r2 are discarded, so r1 matches are prioritized

example

splitSubex "({1}x)|({2}.)"   "x" = ([("1","x"),("2","x")], "")

splitSubex "({1}x){|}({2}.)" "x" = ([("1","x")], "")

Construct the sequence r.e. r1.r2

mkSeq extened to lists

Construct repetition r{i,}

Construct range r{i,j}

Construct option r?

Construct difference r.e.: r1 {\} r2

example

match "[a-z]+{\\}bush" "obama"     = True
match "[a-z]+{\\}bush" "clinton"   = True
match "[a-z]+{\\}bush" "bush"      = False     -- not important any more

Construct r.e. for intersection: r1 {&} r2

example

match ".*a.*{&}.*b.*" "-a-b-"  = True
match ".*a.*{&}.*b.*" "-b-a-"  = True
match ".*a.*{&}.*b.*" "-a-a-"  = False
match ".*a.*{&}.*b.*" "---b-"  = False

Construct r.e. for exclusive or: r1 {^} r2

example

match "[a-c]+{^}[c-d]+" "abc"  = True
match "[a-c]+{^}[c-d]+" "acdc" = False
match "[a-c]+{^}[c-d]+" "ccc"  = False
match "[a-c]+{^}[c-d]+" "cdc"  = True

Construct the Complement of an r.e.: whole set of words - r

Construct a labeled subexpression: ({label}r)

parse a standard W3C XML Schema regular expression

parse an extended syntax W3C XML Schema regular expression

The Syntax of the W3C XML Schema spec is extended by further useful set operations, like intersection, difference, exor. Subexpression match becomes possible with "named" pairs of parentheses. The multi char escape sequence \a represents any Unicode char, The multi char escape sequence \A represents any Unicode word, (\A = \a*). All syntactically wrong inputs are mapped to the Zero expression representing the empty set of words. Zero contains as data field a string for an error message. So error checking after parsing becomes possible by checking against Zero (isZero predicate)

parse a regular expression surrounded by contenxt spec

a leading ^ denotes start of text, a trailing $ denotes end of text, a leading \< denotes word start, a trailing \> denotes word end.

The 1. param ist the regex parser (parseRegex or parseRegexExt)