Unicode in Regex

[S-gw]

This is mostly a Ruby thing, and partly a Rails thing.

I’m expecting a validate_format_of with a regex like this

/^[a-zA-Z\xC0-\xD6\xD9-\xF6\xF9-\xFF.’-\ ]*?$/

to allow many of the normal characters like ö é å to be submitted via
web form.

However, the extended characters are being rejected.

This works just fine though (which is just a-zA-Z)

/^[\x41-\x5A\x61-\x7A.’-\ ]*?$/

It also seems to fail with full \x0000 numbers, is there limit at \xFF?

Some plain Ruby tests seem to suggest unicode characters don’t work at
all??

p ‘abvHgtwHFuG’.scan(/[a-z]/)
p ‘abvHgtwHFuG’.scan(/[A-Z]/)
p ‘abvHgtwHFuG’.scan(/[\x41-\x5A]/)
p ‘abvHgtwHFuG’.scan(/[\x61-\x7A]/)
p ‘aébvHögtÃ¥wHÃ…FuG’.scan(/[\xC0-\xD6\xD9-\xF6\xF9-\xFF]/)

[“a”, “b”, “v”, “g”, “t”, “w”, “u”]
[“H”, “H”, “F”, “G”]
[“H”, “H”, “F”, “G”]
[“a”, “b”, “v”, “g”, “t”, “w”, “u”]
["\303", “\303”, “\303”, “\303”]

So, what’s the secret to using unicode character ranges in Ruby regex
(or Rails validations)?

–
def gw
acts_as_n00b
writes_at(www.railsdev.ws)
end

[S-gw]

On Nov 30, 2:18 pm, Greg W. [email protected] wrote:

So, what’s the secret to using unicode character ranges in Ruby regex
(or Rails validations)?

Tim B. gave a great talk about I18N, M17N and Unicode at the 2006
Ruby Conference. His presentation can be found at:

rubyconf2006.pdf

He described how many member functions have trouble dealing with these
character sets. He made special reference to regular expressions.

–Dale

[S-gw]

Dale M. wrote:

On Nov 30, 2:18 pm, Greg W. [email protected] wrote:

So, what’s the secret to using unicode character ranges in Ruby regex
(or Rails validations)?

Tim B. gave a great talk about I18N, M17N and Unicode at the 2006
Ruby Conference. His presentation can be found at:

http://www.tbray.org/talks/rubyconf2006.pdf

He described how many member functions have trouble dealing with these
character sets. He made special reference to regular expressions.

That’s just beyond sad.

I’ve been using Lasso for several years now, and 2003 it provided
complete support for Unicode. I know there’s some esoterics it may not
deal with, but for all practical purposes we can round-trip data in
western and eastern languages with Lasso quite easily.

How can all these other languages be so far behind?

Pretty bad if I can’t even allow Mr. Muños or Göran to enter their names
in a web form with proper server side validations. Aargh.

– gw

[S-gw]

On Nov 30, 4:00 pm, Greg W. [email protected] wrote:

How can all these other languages be so far behind?

Pretty bad if I can’t even allow Mr. Muños or Göran to enter their names
in a web form with proper server side validations. Aargh.

– gw

Posted viahttp://www.ruby-forum.com/.

Ruby 1.8 doesn’t have unicode support (1.9 is starting to get it).
Everything in ruby is a bytestring.

irb(main):001:0> ‘aébvHögtåwHÅFuG’.scan(/./)
=> [“a”, “\303”, “\251”, “b”, “v”, “H”, “\303”, “\266”, “g”, “t”,
“\303”, “\245”, “w”, “H”, “\303”, “\205”, “F”, “u”, “G”]

So your character class is matching the first byte of the composite
characters (which is \303 in octal), and skipping the next (since it’s
below the range). You probably want something like…

reg = /[\xc0-\xd6\xd9-\xf6\xf9-\xff][\x80-\xbc]/
‘aébvHögtåwHÅFuG’.scan(reg)

irb(main):006:0* reg = /[\xc0-\xd6\xd9-\xf6\xf9-\xff][\x80-\xbc]/
=> /[\xc0-\xd6\xd9-\xf6\xf9-\xff][\x80-\xbc]/
irb(main):007:0> ‘aébvHögtåwHÅFuG’.scan(reg)
=> [“\303\251”, “\303\266”, “\303\245”, “\303\205”]
irb(main):008:0> “å” == “\303\245”
=> true

Ps. I’m not entirely sure the value of the second character class is
right.

Regards,
Jordan

[S-gw]

Unicode in Regex
Posted by Greg W. (-gw-) on 30.11.2007 21:18
This is mostly a Ruby thing, and partly a Rails thing.

I’m expecting a validate_format_of with a regex like this

/^[a-zA-Z\xC0-\xD6\xD9-\xF6\xF9-\xFF.'-\ ]*?$/

to allow many of the normal characters like ö é å to be submitted via
web form.

How about the utf8 validation regex here:
http://snippets.dzone.com/posts/show/4527 ?

[S-gw]

MonkeeSage wrote:

Ruby 1.8 doesn’t have unicode support (1.9 is starting to get it).

I enrages me to see this kind of FUD. Through regular expressions, ruby
1.8 has 80-90% complete utf8 support. And oniguruma makes utf8 support
well-near 100% complete.

‘aébvHögtÃ¥wHÃ…FuG’.scan(/./u)
=> [“a”, “é”, “b”, “v”, “H”, “ö”, “g”, “t”, “Ã¥”, “w”, “H”, “Ã…”, “F”,
“u”, “G”]

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[éöåÅ]/u)
=> [“é”, “ö”, “Ã¥”, “Ã…”]

Ok, sometimes you have to take a weird approach because of the missing
10-20%, but it’s still workable

‘aébvHögtÃ¥wHÃ…FuG’.scan(/(?:\303\251|\303\266|\303\245|\303\205)/u)
=> [“é”, “ö”, “Ã¥”, “Ã…”]

Everything in ruby is a bytestring.

YES! And that’s exactyly how it should be. Who is it that spread the
flawed idea that strings are fundamentally made of characters? I’d like
to slap him around a little. Fundamentally, ever since the word “string”
was applied to computing, strings were made of 8-BIT CHARS, not n-bit
characters. If only the creators of C has called that datatype “byte”
instead of “char” it would have saved us so many misunderstandings.

Usually the complaint about the support lack of unicode support is that
something like “日本語”.length returns 9 instead of 3, or that "日本語
".index(“語”) returns 6 instead of 2. It’s nice that people want to
completely redefine the API to return character positions and all that,
but please don’t complain that it’s broken just because you happen to be
using it incorrectly. Use the right tool for the job. SQL for database
queries, non-home-brewed crypto libraries for security, regular
expressions for string manipulation.

I’m terribly sorry for the rant but I had to get it off my chest.

Dan

[S-gw]

Greg W. wrote:

characters. I have discovered that I can get a little further by doing

But… this fails /^[a-zA-Z\xE4-\xE6]*?&/u

But… this works /^[a-zA-Z\xE4\xE5\xE6]*?&/u

I’ve boiled the experiments down to realizing I can’t define a range
with \x

Is this just one of those things that just doesn’t work yet WRT Ruby/
Rails/UTF8, or is there another syntax? I’ve scoured all the regex
docs I can find, and they seem to indicate a range should work.

Let me try to explain that in order to redeem myself from my previous
angry post.

Basically, \xE4 is counted as the byte value 0xE4, not the unicode
character U+00E4. And in a range expression, each escaped value is taken
as one character within the range. Which results in not-immediately
obvious situations:

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[\303\251]/u)
=> []

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[#{"\303\251"}]/u)
=> [“é”]

What is happening in the first case is that the string does not contain
characters \303 or \251 because those are invalid utf8 sequences. But
when the value “\303\251” is inlined into the regex, that is
recognized as the utf8 character “é” and a match is found.

So ranges do work in utf8 but you have to be careful:

“à âäçèéêîïôü”.scan(/[ä-î]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]

“à âäçèéêîïôü”.scan(/[\303\244-\303\256]/u)
=> ["\303", “\303”, “\303”, “\244”, “\303”, “\247”, “\303”, “\250”,
“\303”, “\251”, “\303”, “\252”, “\303”, “\256”, “\303”, “\257”, “\303”,
“\264”, “\303”, “\274”]

“à âäçèéêîïôü”.scan(/[#{"\303\244-\303\256"}]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]

Hope this helps.

Dan

[S-gw]

Greg W. wrote:

I’m expecting a validate_format_of with a regex like this
/^[a-zA-Z\xC0-\xD6\xD9-\xF6\xF9-\xFF.’-\ ]*?$/
to allow many of the normal characters like ö é å to be submitted via
web form. However, the extended characters are being rejected.

So, I’ve been pounding the web for info on UTF8 in Ruby and Rails the
past couple days to concoct some validations that allow UTF8
characters. I have discovered that I can get a little further by doing
the
following:

• declaring $KCODE = ‘UTF8’
• adding /u to regex expressions.

The only thing not working now is the ability to define a range of \x
characters in a regex.

So, this /^[a-zA-Z\xE4]*?&/u will validate that a string is allowed
to have an ä in it. Perfect.

But… this fails /^[a-zA-Z\xE4-\xE6]*?&/u

But… this works /^[a-zA-Z\xE4\xE5\xE6]*?&/u

I’ve boiled the experiments down to realizing I can’t define a range
with \x

Is this just one of those things that just doesn’t work yet WRT Ruby/
Rails/UTF8, or is there another syntax? I’ve scoured all the regex
docs I can find, and they seem to indicate a range should work.

For now, I just have all the characters I want included < \xFF listed
individually.

utf_accents = ‘\xC0\xC1\xC2…’

Is_person_name = /^[a-zA-Z#{utf_accents}.’-\ ]*?$/u

But I’d like to solve the range notation if I can.

–
def gw
acts_as_n00b
writes_at(www.railsdev.ws)
end

[S-gw]

Greg W. wrote:

So, this /^[a-zA-Z\xE4]*?&/u will validate that a string is allowed
to have an ä in it. Perfect.

If that actually works, it means you are really using ISO-8859-1
strings, not UTF-8.

utf_accents = ‘\xC0\xC1\xC2…’

Nope, that’s not UTF-8. UTF-8 characters ÀÁÂ would look like
utf_accents = “\xC3\x80\xC3\x81\xC3\x82…”

Dan

[S-gw]

On Dec 2, 2:35 pm, Greg W. [email protected] wrote:

following:

individually.
writes_at(www.railsdev.ws)
end

Posted viahttp://www.ruby-forum.com/.

This seems to work…

$KCODE = “UTF8”
p /^[a-zA-Z\xC0-\xD6\xD9-\xF6\xF9-\xFF.'-\ ]*?/u =~ “Jäsp…it
works”

=> 0

However, it looks to me like it would be more robust to use a slightly
modified version of UTF8REGEX (found in the link Jimmy posted
above)…

UTF8REGEX = /\A(?:
[a-zA-Z.-'\ ]
| [\xC2-\xDF][\x80-\xBF]
| \xE0[\xA0-\xBF][\x80-\xBF]
| [\xE1-\xEC\xEE\xEF][\x80-\xBF]{2}
| \xED[\x80-\x9F][\x80-\xBF]
| \xF0[\x90-\xBF][\x80-\xBF]{2}
| [\xF1-\xF3][\x80-\xBF]{3}
| \xF4[\x80-\x8F][\x80-\xBF]{2}
)*\z/mnx

p UTF8REGEX =~ “Jäsp…it works here too”

=> 0

Look at the link to see the explanation of the alternations.

Regards,
Jordan

[S-gw]

Daniel DeLorme wrote:

Greg W. wrote:

But… this fails /^[a-zA-Z\xE4-\xE6]?&/u
But… this works /^[a-zA-Z\xE4\xE5\xE6]?&/u

I’ve boiled the experiments down to realizing I can’t define a range
with \x

Let me try to explain that in order to redeem myself from my previous
angry post.

Basically, \xE4 is counted as the byte value 0xE4, not the unicode
character U+00E4. And in a range expression, each escaped value is taken
as one character within the range. Which results in not-immediately
obvious situations:

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[\303\251]/u)
=> []

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[#{"\303\251"}]/u)
=> [“é”]

OK, I see oniguruma docs refer to \x as encoded byte value and \x{} as a
character code point – which with your explanation I can finally tie
together what that means.

Took me a second to recognize the #{} as Ruby and not some new regex I’d
never seen

And I realize now too I wasn’t picking up on the use of octal vs
decimal.

Seems like Ruby doesn’t like to use the hex \x{7HHHHHHH} variant?

What is happening in the first case is that the string does not contain
characters \303 or \251 because those are invalid utf8 sequences. But
when the value “\303\251” is inlined into the regex, that is
recognized as the utf8 character “é” and a match is found.

So ranges do work in utf8 but you have to be careful:

“à âäçèéêîïôü”.scan(/[ä-î]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]

“à âäçèéêîïôü”.scan(/[\303\244-\303\256]/u)
=> ["\303", “\303”, “\303”, “\244”, “\303”, “\247”, “\303”, “\250”,
“\303”, “\251”, “\303”, “\252”, “\303”, “\256”, “\303”, “\257”, “\303”,
“\264”, “\303”, “\274”]

“à âäçèéêîïôü”.scan(/[#{"\303\244-\303\256"}]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]

Hope this helps.

Yes!

– gw

[S-gw]

Basically, \xE4 is counted as the byte value 0xE4, not the unicode
character U+00E4. And in a range expression, each escaped value is taken
as one character within the range. Which results in not-immediately
obvious situations:

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[\303\251]/u)
=> []

‘aébvHögtÃ¥wHÃ…FuG’.scan(/[#{"\303\251"}]/u)
=> [“é”]

OK, one thing I’m still confused about – when I look up é in any table,
it’s DEC is 233 which converted to OCT is 351, yet you’re using 251 (and
indeed it seems like reducing the OCTs I come up with by 100 is what
actually works).

Where is this 100 difference coming from?

– gw

[S-gw]

On Dec 3, 2:07 pm, Greg W. [email protected] wrote:

Where is this 100 difference coming from?
Unicode Character 'LATIN SMALL LETTER E WITH ACUTE' (U+00E9)

The UTF-16 value is 233 (decimal), but the UTF-8 value is 0xC3 0xA9,
which is 195 169 in decimal, or 0303 0251 in octal.

[S-gw]

Daniel DeLorme wrote:

Usually the complaint about the support lack of unicode support is that
something like “日本語”.length returns 9 instead of 3, or that "日本語
".index(“語”) returns 6 instead of 2. It’s nice that people want to
completely redefine the API to return character positions and all that,
but please don’t complain that it’s broken just because you happen to be
using it incorrectly. Use the right tool for the job. SQL for database
queries, non-home-brewed crypto libraries for security, regular
expressions for string manipulation.

I’m terribly sorry for the rant but I had to get it off my chest.

Regular expressions for all character work would be a terribly slow
way to get things done. If you want to get the nth character, should you
do a match for n-1 characters and a group to grab the nth? Or would it
be better if you could just index into the string and have it do the
right thing? How about if you want to iterate over all characters in a
string? Should the iterating code have to know about the encoding?
Should you use a regex to peel off one character at a time? Absurd.

Regex for string access goes a long way, but’s just about the heaviest
way to do it. Strings should be aware of their encoding and should be
able to provide you access to characters as easily as bytes. That’s what
1.9 (and upcoming changes in JRuby) fixes.

• Charlie

[S-gw]

On Dec 2, 7:40 pm, Daniel DeLorme [email protected] wrote:

characters. I have discovered that I can get a little further by doing

‘aébvHögtåwHÅFuG’.scan(/[#{“\303\251”}]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]
“àâäçèéêîïôü”.scan(/[\303\244-\303\256]/u)
=> [“\303”, “\303”, “\303”, “\244”, “\303”, “\247”, “\303”, “\250”,
“\303”, “\251”, “\303”, “\252”, “\303”, “\256”, “\303”, “\257”, “\303”,
“\264”, “\303”, “\274”]
“àâäçèéêîïôü”.scan(/[#{“\303\244-\303\256”}]/u)
=> [“ä”, “ç”, “è”, “é”, “ê”, “î”]

Hope this helps.

Dan

I missed your ranting.

Firstly, ruby doesn’t have unicode support in 1.8, since unicode IS
a standard mapping of bytes to characters. That’s what unicode is.
I’m sorry you don’t like that, but don’t lie and say ruby 1.8 supports
unicode when it knows nothing about that standard mapping and treats
everything as individual bytes (and any byte with a value greater than
126 just prints an octal escape); and please don’t accuse others of
spreading FUD when they state the facts.

Secondly, as I said in my first post to this thread, the characters
trying to be matched are composite characters, which requires you to
match both bytes. You can try to using a unicode regexp, but then you
run into the problem you mention–the regexp engine expects the pre-
composed, one-byte form…

“ò”.scan(/[\303\262]/u) # => []
“ò”.scan(/[\xf2]/u) # => [“\303\262”]

…which is why I said it’s more robust to use something like the the
regexp that Jimmy linked to and I reposted, instead of a unicode
regexp.

Regards,
Jordan

[S-gw]

On Dec 3, 12:47 pm, Greg W. [email protected] wrote:

[S-gw]

On Dec 3, 1:47 pm, Greg W. [email protected] wrote:

Seems like Ruby doesn’t like to use the hex \x{7HHHHHHH} variant?
Oniguruma is not in ruby 1.8 (though you can install it as a gem). It
is in 1.9.

[S-gw]

Charles Oliver N. wrote:

Regular expressions for all character work would be a terribly slow
way to get things done. If you want to get the nth character, should you
do a match for n-1 characters and a group to grab the nth? Or would it
be better if you could just index into the string and have it do the

Ok, I’m not very familiar with the internal working of strings in 1.9,
but it seems to me that for character sets with variable byte size, it
is logically impossible to directly index into the string. Unless
there’s some trick I’m unaware of, you have to count from the
beginning of the string for utf8 strings.

right thing? How about if you want to iterate over all characters in a
string? Should the iterating code have to know about the encoding?
Should you use a regex to peel off one character at a time?

That is certainly one possible way of doing things…
string.scan(/./){ |char| do_someting_with(char) }

Regex for string access goes a long way, but’s just about the heaviest
way to do it.

Heavy compared to what? Once compiled, regex are orders of magnitude
faster than jumping in and out of ruby interpreted code.

Strings should be aware of their encoding and should be
able to provide you access to characters as easily as bytes. That’s what
1.9 (and upcoming changes in JRuby) fixes.

Overall I agree that the encoding stuff in 1.9 is very nice.
Encapsulating the encoding with the string is very OO. Very intuitive.
No need to think about encoding anymore, now it “just works” for
encoding-ignorant programmers (at least until the abstraction leaks). It
allows to shut up one frequent complaint about ruby; a clear political
victory. Overall it is more robust and less error-prone than the 1.8
way.

But my point was that there is a 1.8 way. The thing that riled me up
and that I was responding to was the claim that 1.8 did not have unicode
support AT ALL. Unequivocally, it does, and it works pretty well for me.
IMHO there is a certain minimalist elegance in considering strings as
encoding-agnostic and using regex to get encoding-specific views. I
could do str[/./n] and str[/./u]; I can’t do that anymore.

1.9 makes encodings easier for the english-speaking masses not used to
extended characters, but let’s remember that ruby always had support
for multibyte character sets; after all it did originate from a
country with two gazillion “characters”.

Daniel

[S-gw]

MonkeeSage wrote:

Firstly, ruby doesn’t have unicode support in 1.8, since unicode IS
a standard mapping of bytes to characters. That’s what unicode is.
I’m sorry you don’t like that, but don’t lie and say ruby 1.8 supports
unicode when it knows nothing about that standard mapping and treats
everything as individual bytes (and any byte with a value greater than
126 just prints an octal escape)

Ok, then how do you explain this:

$KCODE=‘u’
=> “u”

“abc\303\244”.scan(/./)
=> [“a”, “b”, “c”, “ä”]

This doesn’t require any libraries, and it seems to my eyes that ruby is
converting 5 bytes into 4 characters. It shows an awareness of utf8. If
that’s not some kind of unicode support then please tell me what it
is. It seem were disagreeing on some basic definition of what “unicode
support” means.

Secondly, as I said in my first post to this thread, the characters
trying to be matched are composite characters, which requires you to
match both bytes. You can try to using a unicode regexp, but then you
run into the problem you mention–the regexp engine expects the pre-
composed, one-byte form…

“ò”.scan(/[\303\262]/u) # => []
“ò”.scan(/[\xf2]/u) # => ["\303\262"]

Wow, I never knew that second one could work. Unicode support is
actually better than I thought! You learn something new every day.

…which is why I said it’s more robust to use something like the the
regexp that Jimmy linked to and I reposted, instead of a unicode
regexp.

I’m not sure what makes that huge regexp more robust than a simple
unicode regexp.

Daniel

[S-gw]

On Dec 4, 3:07 am, Daniel DeLorme [email protected] wrote:

support" means.
I guess we were talking about different things then. I never meant to
imply that the regexp engine can’t match unicode characters (it’s
“dumb” implementation though; it basically only knows that bytes above
127 can have more bytes following and should be grouped together as
candidates for a match; that’s slightly simplified, but basically
accurate).

I, like Charles (and I think most people), was referring to the
ability to index into strings by characters, find their lengths in
characters, to compose and decompose composite characters, to
normalize characters, convert them to other encodings like shift-jis,
and other such things. Ruby 1.9 has started adding such support, while
ruby 1.8 lacks it. It can be hacked together with regular expressions
(e.g., the link Jimmy posted), or even as a real, compiled extension
[1], but merely saying that you the programmer can implement it
using ruby 1.8, is not the same thing as saying ruby 1.8 supports it
(just like I could build a python VM in ruby, but that doesn’t mean
that the ruby interpreter runs python bytecode). Anyhow, I guess it’s
just a difference of opinion. I don’t mind being wrong (happens a
lot! I just don’t like being accused of spreading FUD about ruby,
which to my mind implies malice of forethought rather that simply
mistake.

[1] http://rubyforge.org/projects/char-encodings/
http://git.bitwi.se/ruby-character-encodings.git/

actually better than I thought! You learn something new every day.

…which is why I said it’s more robust to use something like the the
regexp that Jimmy linked to and I reposted, instead of a unicode
regexp.

I’m not sure what makes that huge regexp more robust than a simple
unicode regexp.

Daniel

Well, I won’t claim that you can’t get a unicode regexp to match the
same. And I only saw that large regexp when it was posted here, so
I’ve not tested it to any great length. Interestingly, 1.9 uses this
regexp (originally from jcode.rb in stdlib) to classify a string as
containing utf-8: ‘[\xc0-\xdf][\x80-\xbf]|[\xe0-\xef][\x80-\xbf][\x80-
\xbf]’. My thought was that without knowing all of the minute
intricacies of unicode and how ruby strings and regexps work with
unicode values (which I don’t, and assume the OP doesn’t), I think the
huge regexp is more likely to Just Work in more cases than a home-
brewed unicode regexp. But like I said, that’s just an initial
conclusion, I don’t claim it’s absolutely correct.

Regards,
Jordan