#
# This file is free software, which comes along with NIT. This software is
# distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY;
-# without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
# PARTICULAR PURPOSE. You can modify it is you want, provided this header
# is kept unaltered, and a notification of the changes is added.
# You are allowed to redistribute it and sell it, alone or is a part of
# Search `self` into `s` from a certain position.
# Return the position of the first character of the matching section.
# Return -1 if not found.
- fun search_index_in(s: String, from: Int): Int is abstract
+ #
+ # assert 'l'.search_index_in("hello world", 0) == 2
+ # assert 'l'.search_index_in("hello world", 3) == 3
+ # assert 'z'.search_index_in("hello world", 0) == -1
+ #
+ # This method is usually faster than `search_in` if what is
+ # required is only the index.
+ # Note: in most implementation, `search_in` is implemented with this method.
+ protected fun search_index_in(s: Text, from: Int): Int is abstract
# Search `self` into `s` from a certain position.
# Return null if not found.
- fun search_in(s: String, from: Int): nullable Match is abstract
+ #
+ # assert 'l'.search_in("hello world", 0).from == 2
+ # assert 'l'.search_in("hello world", 3).from == 3
+ # assert 'z'.search_in("hello world", 0) == null
+ #
+ # If only the index of the first character if required, see `search_index_in`.
+ #
+ # Note: Is used by `String::search`, `String::search_from`, and others.
+ protected fun search_in(s: Text, from: Int): nullable Match is abstract
# Search all `self` occurrences into `s`.
- fun search_all_in(s: String): Array[Match]
+ #
+ # assert 'l'.search_all_in("hello world").length == 3
+ # assert 'z'.search_all_in("hello world").length == 0
+ #
+ # Note: Is used by `String::search_all`.
+ protected fun search_all_in(s: Text): Array[Match]
do
var res = new Array[Match] # Result
var match = search_in(s, 0)
end
# Split `s` using `self` is separator.
- fun split_in(s: String): Array[Match]
+ #
+ # Returns an array of matches that are between each occurence of `self`.
+ # If self is not present, an array with a single match on `s` is retunred.
+ #
+ # assert 'l'.split_in("hello world").join("|") == "he||o wor|d"
+ # assert 'z'.split_in("hello world").join("|") == "hello world"
+ #
+ # Note: is used by `String::split`
+ protected fun split_in(s: Text): Array[Match]
do
var res = new Array[Match] # Result
var i = 0 # Cursor
while match != null do
# Compute the splited part length
var len = match.from - i
- res.add(new Match(s, i, len))
+ res.add(new Match(s.to_s, i, len))
i = match.after
match = search_in(s, i)
end
# Add the last part
- res.add(new Match(s, i, s.length - i))
+ res.add(new Match(s.to_s, i, s.length - i))
return res
end
+
+ # Is `self` in `s`?
+ protected fun is_in(s: Text): Bool do return search_index_in(s, 0) != -1
end
# BM_Pattern are pre-compiled string motif for the Boyer-Moore algorithm.
# (cf. A Fast String Searching Algorithm, with R.S. Boyer. Communications
# of the Association for Computing Machinery, 20(10), 1977, pp. 762-772.)
# http://www.cs.utexas.edu/users/moore/best-ideas/string-searching/index.html
+#
+# var pat = new BM_Pattern("hello")
+# assert "I say hello to the world!".search(pat).from == 6
+# assert "I say goodbye to the world!".search(pat) == null
class BM_Pattern
super Pattern
if to < 0 then
return null
else
- return new Match(s, to, _length)
+ return new Match(s.to_s, to, _length)
end
end
- # Compile a new motif
- init(motif: String)
+ init
do
- _motif = motif
- _length = motif.length
+ _length = _motif.length
_gs = new Array[Int].with_capacity(_length)
- _bc_table = new ArrayMap[Char, Int]
compute_gs
compute_bc
end
# searched motif
- var _motif: String
+ private var motif: String
# length of the motif
- var _length: Int
+ private var length: Int is noinit
private fun bc(e: Char): Int
- do
+ do
if _bc_table.has_key(e) then
return _bc_table[e]
else
end
# good shifts
- var _gs: Array[Int]
+ private var gs: Array[Int] is noinit
# bad characters
- var _bc_table: Map[Char, Int]
+ private var bc_table = new ArrayMap[Char, Int]
private fun compute_bc
do
private fun compute_gs
do
- var x = _motif
var m = _length
var suff = suffixes
var i = 0
redef fun ==(o) do return o isa BM_Pattern and o._motif == _motif
end
-# Matches are a part of a string.
+# Matches are a part of a `Text` found by a `Pattern`.
class Match
# The base string matched
- readable var _string: String
+ var string: String
# The starting position in the string
- readable var _from: Int
+ var from: Int
# The length of the matching part
- readable var _length: Int
+ var length: Int
# The position of the first character just after the matching part.
# May be out of the base string
- fun after: Int do return _from + _length
+ fun after: Int do return from + length
# The contents of the matching part
- redef fun to_s do return _string.substring(_from, _length)
+ redef fun to_s do return string.substring(from,length)
- # Matches `len` characters of `s` from `f`.
- init(s: String, f: Int, len: Int)
+ init
do
- assert positive_length: len >= 0
- assert valid_from: f >= 0
- assert valid_after: f + len <= s.length
- _string = s
- _from = f
- _length = len
+ assert positive_length: length >= 0
+ assert valid_from: from >= 0
+ assert valid_after: from + length <= string.length
end
end
if pos < 0 then
return null
else
- return new Match(s, pos, 1)
+ return new Match(s.to_s, pos, 1)
end
end
end
-redef class String
+redef class Text
super Pattern
redef fun search_index_in(s, from)
if pos < 0 then
return null
else
- return new Match(s, pos, length)
+ return new Match(s.to_s, pos, length)
end
end
- # Like `search_from` but from the first character.
+ # Search the first occurence of the pattern `p`.
+ # Return null if not found.
+ #
+ # assert "I say hello to the world!".search("hello").from == 6
+ # assert "I say goodbye to the world!".search("hello") == null
fun search(p: Pattern): nullable Match do return p.search_in(self, 0)
- # Search the given pattern into self from a.
+ # Search the first occurence of the pattern `p` after `from`.
# The search starts at `from`.
# Return null if not found.
+ #
+ # assert "I say hello to the world!".search_from("hello",4).from == 6
+ # assert "I say hello to the world!".search_from("hello",7) == null
fun search_from(p: Pattern, from: Int): nullable Match do return p.search_in(self, from)
+ # Search the last occurence of the text `t`.
+ #
+ # assert "bob".search_last("b").from == 2
+ # assert "bob".search_last("bo").from == 0
+ # assert "bob".search_last("ob").from == 1
+ # assert "bobob".search_last("ob").from == 3
+ # assert "bobbob".search_last("bb").from == 2
+ # assert "bobbob".search_last("bob").from == 3
+ # assert "bob".search_last("z") == null
+ # assert "".search_last("b") == null
+ fun search_last(t: Text): nullable Match do
+ return search_last_up_to(t, length)
+ end
+
+ # Search the last occurence of the text `t` before `up_to`.
+ #
+ # assert "bobbob".search_last_up_to("b", 3).from == 2
+ # assert "bobbob".search_last_up_to("b", 6).from == 5
+ # assert "bobbob".search_last_up_to("b", 0) == null
+ fun search_last_up_to(t: Text, up_to: Int): nullable Match do
+ var i = up_to - t.length
+
+ while i >= 0 do
+ if substring(i, t.length) == t then
+ return new Match(self.to_s, i, t.length)
+ end
+ i -= 1
+ end
+ return null
+ end
+
# Search all occurrences of p into self.
#
# var a = new Array[Int]
# @deprecated alias for `split`
fun split_with(p: Pattern): Array[String] do return self.split(p)
+ # Split `self` on the first `=`
+ #
+ # assert "hello".split_once_on('l') == ["he", "lo"]
+ # assert "a, b, c, d, e".split_once_on(", ") == ["a", "b, c, d, e"]
+ fun split_once_on(p: Pattern): Array[SELFTYPE]
+ do
+ var m = p.search_in(self, 0)
+ var res = new Array[SELFTYPE]
+ if m == null then
+ res.add self
+ else
+ res.add substring(0, m.from)
+ res.add substring_from(m.after)
+ end
+ return res
+ end
+
# Replace all occurences of a pattern with a string
#
# assert "hlelo".replace("le", "el") == "hello"
# assert "hello".replace('l', "") == "heo"
- fun replace(p: Pattern, string: String): String
+ fun replace(p: Pattern, string: SELFTYPE): String
do
return self.split_with(p).join(string)
end
- # Escape the four characters < > & and " with their html counterpart
+ # Does `self` contains at least one instance of `pattern`?
#
- # assert "a&b->\"x\"".html_escape == "a&b->"x""
- fun html_escape: String
- do
- var ret = self
- if ret.chars.has('&') then ret = ret.replace('&', "&")
- if ret.chars.has('<') then ret = ret.replace('<', "<")
- if ret.chars.has('>') then ret = ret.replace('>', ">")
- if ret.chars.has('"') then ret = ret.replace('"', """)
- return ret
- end
+ # assert "hello".has('l')
+ # assert "hello".has("ll")
+ # assert not "hello".has("lll")
+ fun has(pattern: Pattern): Bool do return pattern.is_in(self)
end
nitlanguage / nit.git / blobdiff