On Oct 10, 5:02 pm, Chouser <[EMAIL PROTECTED]> wrote:
> On Fri, Oct 10, 2008 at 7:48 AM, Chouser <[EMAIL PROTECTED]> wrote:
>
> > Of course that means I need to do this for you when printing...
>
> I've attached an updated patch with a new print method, against the
> latest SVN 1058.
>
> If you really want to dig into the ugliness, here's a test I've been
> using. Apply the attached patch, rebuild clojure, and then load up
> this code:
>
> (import '(java.util.regex Pattern)
> '(java.io PushbackReader StringReader))
> (defn test-re [s & [ts]]
> (let [p1 (Pattern/compile s)
> p2 (read (PushbackReader. (StringReader. (prn-str p1))))]
> (println (str "raw str: " s))
> (print (str "prn1: " (prn-str p1)))
> (print (str "prn2: " (prn-str p2)))
> (println (if (= (prn-str p1) (prn-str p2)) "PASS" "FAIL"))
> (when ts
> (println (str "match1: " (re-find p1 ts)))
> (println (str "match2: " (re-find p2 ts)))
> (println (if (= (re-find p1 ts) (re-find p2 ts)) "PASS" "FAIL")))
> (println)))
>
> You run it by passing in a string to be compiled by Pattern. This is
> essentially the "old format" without the leading # char:
>
> user=> (test-re "foo")
> raw str: foo
> prn1: #"foo"
> prn2: #"foo"
> PASS
>
> If my print method is correct, prn1 and prn2 should always be the
> same, and thus PASS. Also what you're seeing there is the "new
> format". In this case, both the old and new formats are the same.
>
> If you pass in an optional second string, test-re will try to match it
> with both patterns:
>
> user=> (test-re "a \\\"\\w*\\\" please" "a \"word\" please")
> raw str: a \"\w*\" please
> prn1: #"a \"\w*\" please"
> prn2: #"a \"\w*\" please"
> PASS
> match1: a "word" please
> match2: a "word" please
> PASS
>
> Again, if my patch if my patch is correct, match1 and match2 should be
> the same, and thus another PASS. Here you can see how the new format
> (prn1 and prn2) are simpler than the old format, as well as being
> identical to what Pattern actually operates on (raw str).
>
> Here's a truly nasty example prompted by Christophe Grand's comment:
>
> user=> (test-re "a\"\\Qb\"c\\d\\Ee\"f" "a\"b\"c\\de\"f")
> raw str: a"\Qb"c\d\Ee"f
> prn1: #"a\"\Qb\E\"\Qc\d\Ee\"f"
> prn2: #"a\"\Qb\E\"\Qc\d\Ee\"f"
> PASS
> match1: a"b"c\de"f
> match2: a"b"c\de"f
> PASS
>
> If you can find *any* input that produces FAIL, please let me know.
>
Patch applied - SVN rev 1070 - thanks!
If you use regex literals, this is a breaking change - you must fix
them as described above (basically, \ doesn't need to be escaped
anymore).
Rich
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To post to this group, send email to clojure@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
-~----------~----~----~----~------~----~------~--~---
