module re
import text
-intrude import text::flat
+import text::flat
import gc
import error
# Cache of a single `regmatch_t` to prevent many calls to `malloc`
private var native_match: NativeMatchArray is lazy do
native_match_is_init = true
- return new NativeMatchArray.malloc(native.re_nsub+1)
+ return new NativeMatchArray.malloc(native.as(not null).re_nsub+1)
end
private var native_match_is_init = false
private fun get_error(errcode: Int): String
do
+ var native = native
+ assert native != null
+
# Error, should be out of memory but we cover any possible error anyway
var error_cstr = native.regerror(errcode)
var comp_res = compile
assert comp_res == null else "Regex compilation failed with: {comp_res.message}\n".output
+ var native = native
+ assert native != null
+
# Actually execute
var eflags = gather_eflags
var res = native.regexec_match_only(text.to_cstring, eflags)
var comp_res = compile
assert comp_res == null else "Regex compilation failed with: {comp_res.message}\n".output
+ var native = native
+ assert native != null
+
# Actually execute
text = text.to_s
var cstr = text.substring_from(from).to_cstring
# assert "el+o".to_re.search_in("hello world", 0).from == 1
# assert "l+".to_re.search_in("hello world", 3).from == 3
# assert "z".to_re.search_in("hello world", 0) == null
- redef fun search_in(text, from)
+ # assert "cd(e)".to_re.search_in("abcdef", 2)[1].to_s == "e"
+ redef fun search_in(text, charfrom)
do
assert not optimize_has
var comp_res = compile
assert comp_res == null else "Regex compilation failed with: {comp_res.message}\n".output
+ var native = native
+ assert native != null
+
# Actually execute
- text = text.to_s
- var sub = text.substring_from(from)
- var cstr = sub.to_cstring
- var bstr = new FlatString.full(cstr, sub.bytelen, 0, sub.bytelen - 1, text.length - from)
+ var cstr = text.to_cstring
+ var rets = cstr.to_s_with_length(text.bytelen)
+ var bytefrom = cstr.char_to_byte_index_cached(charfrom, 0, 0)
+ var subcstr = cstr.fast_cstring(bytefrom)
var eflags = gather_eflags
var native_match = self.native_match
var nsub = native.re_nsub
- var res = native.regexec(cstr, nsub+1, native_match, eflags)
+ var res = native.regexec(subcstr, nsub + 1, native_match, eflags)
# Found one?
if res == 0 then
- var bso = bstr.byte_to_char_index(native_match.rm_so)
- var ln = bstr.byte_to_char_index(native_match.rm_eo - native_match.rm_so - 1)
- var match = new Match(text,
- from + bso,
- ln + 1)
+ var bfrom = native_match.rm_so + bytefrom
+ var bto = native_match.rm_eo - 1 + bytefrom
+ var cpos = cstr.byte_to_char_index_cached(bfrom, charfrom, bytefrom)
+ var len = cstr.utf8_length(bfrom, bto - bfrom + 1)
+ var match = new Match(rets, cpos, len)
+ var subs = match.subs
# Add sub expressions
for i in [1 .. nsub] do
- bso = bstr.byte_to_char_index(native_match[i].rm_so)
- ln = bstr.byte_to_char_index(native_match[i].rm_eo - native_match[i].rm_so - 1)
- match.subs.add new Match( text,
- bso ,
- ln + 1)
+ if native_match[i].rm_so < 0 then
+ subs.add null
+ continue
+ end
+ var sub_bfrom = native_match[i].rm_so + bytefrom
+ var sub_bto = native_match[i].rm_eo - 1 + bytefrom
+ var sub_cpos = cstr.byte_to_char_index_cached(sub_bfrom, cpos, bfrom)
+ var sub_len = cstr.utf8_length(sub_bfrom, sub_bto - sub_bfrom + 1)
+ subs.add(new Match(rets, sub_cpos, sub_len))
end
return match
var comp_res = compile
assert comp_res == null else "Regex compilation failed with: {comp_res.message}\n".output
+ var native = native
+ assert native != null
+
# Actually execute
- text = text.to_s
var cstr = text.to_cstring
+ var subcstr = cstr
+ var rets = cstr.to_s_with_length(text.bytelen)
var eflags = gather_eflags
var eflags_or_notbol = eflags | flag_notbol
var native_match = self.native_match
var matches = new Array[Match]
var nsub = native.re_nsub
- var res = native.regexec(cstr, nsub+1, native_match, eflags)
- var d = 0
+ var res = native.regexec(subcstr, nsub + 1, native_match, eflags)
+ var bytesub = 0
+ var charsub = 0
while res == 0 do
- var match = new Match(text,
- d + native_match.rm_so,
- native_match.rm_eo - native_match.rm_so)
+ var bfrom = native_match.rm_so + bytesub
+ var bto = native_match.rm_eo - 1 + bytesub
+ var cstart = cstr.byte_to_char_index_cached(bfrom, charsub, bytesub)
+ var len = cstr.utf8_length(bfrom, bto - bfrom + 1)
+ var match = new Match(rets, cstart, len)
matches.add match
+ var subs = match.subs
# Add sub expressions
- for i in [1..nsub] do
- match.subs.add new Match( text,
- d + native_match[i].rm_so,
- native_match[i].rm_eo - native_match[i].rm_so)
+ for i in [1 .. nsub] do
+ if native_match[i].rm_so < 0 then
+ subs.add null
+ continue
+ end
+ var sub_bfrom = native_match[i].rm_so + bytesub
+ var sub_bto = native_match[i].rm_eo - 1 + bytesub
+ var sub_cstart = cstr.byte_to_char_index_cached(sub_bfrom, cstart, bfrom)
+ var sub_len = cstr.utf8_length(sub_bfrom, sub_bto - sub_bfrom + 1)
+ subs.add(new Match(rets, sub_cstart, sub_len))
end
- if d == native_match.rm_eo then
- d += 1
- else d = d + native_match.rm_eo
- cstr = cstr.substring_from(native_match.rm_eo)
- res = native.regexec(cstr, nsub+1, native_match, eflags_or_notbol)
+ bytesub = bto + 1
+ charsub = cstart + len
+ subcstr = cstr.fast_cstring(bytesub)
+ res = native.regexec(subcstr, nsub + 1, native_match, eflags_or_notbol)
end
# No more match?
# assert match.subs.length == 1
# assert match.subs.first.to_s == "d eee"
# ~~~
- var subs = new Array[Match] is lazy
+ var subs = new Array[nullable Match] is lazy
# Get the `n`th expression in this match
#
# assert match[0].to_s == "c d eee f"
# assert match[1].to_s == "d eee"
# ~~~
- fun [](n: Int): Match do
+ fun [](n: Int): nullable Match do
if n == 0 then return self
assert n > 0 and n <= subs.length
return subs[n-1]
nitlanguage / nit.git / blobdiff