Re: file to uri to file madness

Subject: Re: file to uri to file madness

From: Francois-Paul Servant <email@hidden>

Date: Thu, 24 Aug 2006 03:41:49 +0200

Delivered-to: email@hidden

Hi,

thank you for the answer.

Le 23 août 06 à 03:50, Greg Guerin a écrit :

You specifically say to create two files ending in ".htm", but you then say
the URLs you use end in ".html". I'm not sure if this difference is a typo
or not.

a typo, sorry.

Based on the above, I suspect a problem with the Apache web server, but I
don't know exactly what the problem might be. Since it appears on 10.4.7,
it may be fairly obscure.

The URL-escaped form of the decomposed-accents worked (e%cc%81), so you
might use that as a work-around, but please file a bug against Apple's Apache
because it looks like a repeatable fail-case to me.

Hum, in fact, thinking more about that, I arrived to the conclusion that the problem is in the way the OS converts from "composed accents" to "uncomposed".

As you say:

it makes no sense why 2 will work but 3 or
4 will not.

Too bad that it' took so much time to me to realize that we have the same situation when we call

(new File(dir, "éé.html")).exists();

and

(new File(dir, "ééé.html")).exists();

Second line returns false, but the first one returns true (and - apologies - I said the contrary in my previous message)

What do you think? Do you agree that both calls should return true? (in an user point of view, no doubt ;-))

Best Regards,

fps

In case someone want to test it, here is a class that do the job:

package net.semanlink.test;

import java.io.*;

import java.net.*;

/** Shows a problem when trying to access files containing certain sequences of non ASCII characters.

* You can access "éé.html" using a filename using "composed accents" (such as you get with

* a string constant), but you can't access "ééé.html".

* To run the test, pass to main a directory containing one file called "éé.html" and another called "ééé.html".

public class DiacriticsInFilenamesTest {

/** Pass as sole arg a dir containing one file called "éé.html" and another called "ééé.html". */

public static void main(String[] args) {

new DiacriticsInFilenamesTest(args[0]);

}

/** Pass a dir containing one file called "éé.html" and another called "ééé.html". */

public DiacriticsInFilenamesTest(String dirPath) {

try {

test(dirPath);

} catch (Exception e) {

e.printStackTrace();

}

void test(String dirPath) throws UnsupportedEncodingException {

File dir = new File(dirPath);

String[] names = dir.list();

for (int i = 0; i < names.length; i++) {

testOne("Short name as returned by dir.list() (non composed accents)", dir, names[i], false);

}

testOne("Short name as a java constant String (composed accents)", dir, "éé.html", true);

testOne("Short name as a java constant String (composed accents)", dir, "ééé.html", true);

}

void testOne(String mess, File dir, String shortName, boolean accentsAreComposed) throws UnsupportedEncodingException {

File file = new File(dir, shortName);

if (file.isHidden()) return;

System.out.println("\n****** TESTING " + mess +": " + shortName);

System.out.println("short name: " + shortName);

System.out.println("short name as nums: " + toNums(shortName));

String s = URLEncoder.encode(shortName,"UTF-8");

System.out.println("URLEncoder.encode(shortName,UTF-8): "+s);

System.out.println("filename: " + file.getAbsolutePath());

boolean b = file.exists();

if (b) {

System.out.println("File.exists: " + b);

} else {

System.out.println("File.exists: *** " + b + "***");

}

private String toNums(String s) {

StringBuffer x = new StringBuffer();

for (int j = 0; j < s.length(); j++) {

x.append( (int) s.charAt(j) + " " );

}

return x.toString();

}

} //

_______________________________________________ Do not post admin requests to the list. They will be ignored. Java-dev mailing list (email@hidden) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/java-dev/email@hidden This email sent to email@hidden

References:

>Re: file to uri to file madness (From: Greg Guerin <email@hidden>)