Need absolute path to cache language files

[tnbnicer]

I believe this to be an issue, or at least something that should be looked into.

phpSyllable-master/src/Cache/File.php on line 53:

file_put_contents($file, $this->encode(self::$data));

file_put_contents requires an absolute path. Instead, what it gets is a relative path.

Warning: file_put_contents(/home/customer/www/mydomain/apps/phpSyllable-master/src/cache/syllable.en-us.json): failed to open stream: No such file or directory in /home/customer/www/mydomain/apps/phpSyllable-master/src/Cache/File.php on line 53.

I was able to workaround it by defining an absolute path. File.php.

private function filename()
{
    $dir = '/home/customer/www/mydomain/apps/phpSyllable-master/src/Cache';
    return $dir.'/'.$this->getFilename(self::$language);
}

Note: I'm not sure why 'syllable.en-us.json' must be written to. For developing purposes? As far as I know, the file never changes, except the first time when it is generated.

Also "file_put_contents" works fine with a relative path on Windows WAMPServer. Only in a shared server virtual hosting environment I get an error/warning. If you cannot reproduce the issue, feel free to query.

[alexander-nitsche]

Would you paste your full code snippet here? And where should the cache directory be?

[tnbnicer]

I don't know which code snippet you mean. It's more than a snippet. I'm testing it on a live server. The only fix applied is the one referred to above. Actually the whole function, as modified, looks like this, because I began by running it on WAMP. From line 30 in File.php:

private function filename()
{
    $remote_ip = "35.214.1.93";
    $server = (strpos(getenv( "SERVER_ADDR" ), $remote_ip) !== false) ? "remote" : "local";
    $remote = ($server == "remote") ? true : false;
    if ($remote) {
        $dir = '/home/customer/www/mydomain/apps/phpSyllable-masterTest/src/Cache';
    } else {
        $dir = self::$path;
    }
    return $dir.'/'.$this->getFilename(self::$language);
    //return self::$path.'/'.$this->getFilename(self::$language); # this is the original code that returns a relative path.
}

That seems to be the only way to get a valid path, for me anyway.

Apparently all of these files do not receive new data. I uploaded them directly from my WAMP installation on Windows yesterday.

mydomain/apps/phpSyllable-master/src/Cache/syllable.it.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.de-2017.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.en-us-old.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.de-1901.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.de.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.en-gb-old.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.cs.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.pl.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.de-1996.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.grc.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.tr.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.en-gb.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.en-us-berliner.json
mydomain/apps/phpSyllable-master/src/Cache/syllable.en-us.json

Why then is it necessary for them to cache new data?

That is what this function does*:

public function close()
{
    $file = $this->filename();
    file_put_contents($file, $this->encode(self::$data));
    @chmod($file, 0777);
}

*I haven't changed it. "file_put_contents" requires a full path.

[tnbnicer]

If you'd like me to, provided Github permits it -- probably not -- I could post a link to the test page on my server. With the error.

[tnbnicer]

Your question about where the cache directory should be, is kind of throwing me off a little. Correct me, if I'm wrong, please. The cache folder is initially defined in the main page. The demo.php calls it like this:

$syllable->getCache()->setPath(dirname(__FILE__).'/cache');

For my purposes, it's defined:

$cache = $syllable->getCache();
$cache->setPath($dir.'/apps/phpSyllable-master/src/cache');

The only other caching is when the user uploads a file, as opposed to pasting text or html in the form. Here I define a different cache that optionally saves the user's file to the server:

if($uploadOk == 1) {
	$server_path = $dir.'/apps/phpSyllable-master/tmp';
	if(is_dir(($server_path))) :
		$ext = explode('.',$_FILES['userfile']['name']);
		$extension = $ext[1];
		$newname = $ext[0].'_'.time();
		$full_server_path = $server_path.'/'.$newname.'.'.$extension;
		move_uploaded_file($_FILES['userfile']['tmp_name'], $full_server_path);
		$text = htmlspecialchars(file_get_contents($full_server_path));
	else:
		$text = htmlspecialchars(file_get_contents($_FILES["userfile"]["tmp_name"]));
	endif;
}
$source = $text;

If $server_path exists, the uploaded file is saved (cached) to that location. I'm not certain if move_uploaded_file also requires a full path. I think it does, similar to file_put_contents.

The point however is that $cache->setPath($dir.'/apps/phpSyllable-masterTest/src/cache') doesn't do the trick for PHP's file_put_contents on line 53 of File.php. That's my interpretation.

If I can assist, in any way, I'm happy to help. I'd tried a Perl hyphenator first. phpSyllable is orders of magnitude better. Brilliant achievement by Mr. Van Der Lee.

[alexander-nitsche]

You need to make sure the phpSyllable cache directory exists. The error of file_put_contents() suggests it does not. Run something like

$cacheDir = $dir.'/apps/phpSyllable-master/src/cache';
if (!is_dir($cacheDir)) {
  mkdir($cacheDir)
}

in the main script somewhere before initialising and using the Syllable object.

Note: You should place the Syllable cache in a folder outside of the phpSyllable app folder and not touch the phpSyllable app folder at all, as this makes updates of the package more difficult.

[tnbnicer]

You're right. Before the initialization is where I should make sure the folder and files exist. Technically, the folder does exist, but on the server for some reason it isn't found, even causing an error.

On the other hand, there are threads on "Stack Overflow" among other online communities, that point out that you must pass the exact absolute folder path with filename to file_put_contents().

The stumbling block for me is that no method at all writes to the folder, error or no error.

I take on board your suggestion to move the "cache" folder. I've been in touch with my web-hosting service. They inform me that because the cache folder is not in the publicly accessible area, the files are automatically cached. That, of course, will not do, if the aim is to "update" the *.json files stored in cache. So I am very tempted to move the "cache" folder to the publicly accessible site area. Maybe that will fix something. It would be nice to close this issue. Issues are bad publicity and it might be entirely my fault.

Thanks for the reply.

[alexander-nitsche]

Use some project file structure like

<project>/
 ├── apps/phpSyllable-master/
 ├── cache/phpSyllable-master/
 └── main.php

and have some main.php like

<?php
use Vanderlee\Syllable\Syllable;

function hyphenateText($text, $language) {
  $cacheDir = __DIR__.'/cache/phpSyllable-master';
  $languageDir = __DIR__.'/apps/phpSyllable-master/languages';

  if (!is_dir($cacheDir)) {
    mkdir($cacheDir, 0777, true);
  }
  
  $syllable = new Syllable($language);
  $syllable->getCache()->setPath($cacheDir);
  $syllable->getSource()->setPath($languageDir);
  
  return $syllable->hyphenateText($text);
}

echo hyphenateText(
    'WampServer ist eine komplette Webserver-Software inklusive Datenbank. ' .
    'Die Software ist ein Alternative zu XAMPP und enthält den Apache-Webserver, ' .
    'MySQL-Datenbank-Server, PHP sowie PHPmyAdmin.', 
    'de'
);

and you will be fine.

Note: The path in $cacheDir is an absolute path, as required by file_put_contents().

[vanderlee]

@alexander-nitsche do you think it would be possible for the language update script to also run a version of the Caching to generate files? That would eliminate the need for write permissions for most use cases unless somebody want to provide their own .tex files.

[tnbnicer]

Hi. To get involved, if I may. There's money in it. If users could upload their own .tex files, or vocabulary lists to splice onto personalized .tex files, it would allow a hyphenation service, maybe for budding e-book authors. Amazon's hyphenation is a disgrace. Authors can do a lot better, if they try to do it manually themselves.

At the moment my hosting provider is being stubborn. I haven't even gotten as far as writing/caching *.json files. Moving the "Cache" to the public_html area has helped, but there's still some kind of issue with caching and write permissions.

A useful thing I found along the way is that the LibreOffice English hyphenation .tex dictionaries are somewhat better than the existing ones. Should anyone be interested, you can download them from https://github.com/LibreOffice/dictionaries.

As soon as (if) I get the *.json caching fixed on my server, I'll post the result. Note the fix, if there is one, wouldn't be applicable to this issue, which is partially only to do with shared hosting. WAMP works great.

[alexander-nitsche]

Hi @vanderlee , what do you want to achieve? I am not sure if i could follow here:)

Hi @tnbnicer , what do you want to achieve? Who would want to load his/her own tex files? And you can do so now already. You could simply create a specific language files directory, e.g. __DIR__.'/user/phpSyllable-master/languages', that contains the custom tex files, and point your Syllable object to it via

<?php
..
function hyphenateText($text, $language) {
  ..
  $languageDir = __DIR__.'/user/phpSyllable-master/languages';
  ..
}
..

Btw: there is already such a hyphenation service at https://syllable.toyls.com/ , which is unfortunately broken currently with console error

XHR
POST https://syllable.toyls.com/?path=syllable/hyphenate
[HTTP/2 500 Internal Server Error 40ms]

[tnbnicer]

@alexander-nitsche
I have yet to come across such a service. People are being duped by Amazon and other e-book publishers into thinking hyphenation is an automatic feature. It is anything but, if you do it right. Think of all the foreign words or coinages an author might have in an e-book. No way can an algorithm hyphenate them all correctly.

In the old days people who did hyphenation were called typesetters.

Another point.
We are talking about a list of perhaps 1000 or more words. Can someone who is serious about publishing a 100000 word e-book upload the list of exceptions to a service, words which he or she manually finds in the text, and upload the hyphenation exceptions list to a service that would parse his/her e-book's html files including the manually hyphenated words in a given language? Sure there are hyphenators, but adding exceptions is a manual chore first. How would like it if you were an author of a book in English with a Latin word in it hyphenated incorrectly?

I speculate of course that Amazon's hyphenation is an algorithm. Most probably it is and uses patterns just like we do.

[tnbnicer]

@vanderlee and @alexander-nitsche, hi.
How do you feel about changing the function in 'phpSyllable-master/src/Cache/File.php' on line 22 from:

    public function setPath($path)
    {
        if ($path !== self::$path) {
            self::$path = $path;
            self::$data = null;
        }
    }

to:

    public function setPath($path)
    {
        if ($path !== __DIR__) {
            self::$path = __DIR__;
            self::$data = null;
        }
    }

??
Just a suggestion, not a pull request. It avoids the error that "self::$path" is not a legitimate string.

Caching appears to have been behind my server problems. It's fixed. So long as the path is correct, and caching is gone.

[tnbnicer]

I can close the issue if you want or you can.

[alexander-nitsche]

@tnbnicer : Thanks for the details. So you would like to have two features:

1. The syllable API should be able to consider an arbitrary amount of custom tex files along the main language tex file.
2. The syllable API should be able to convert a simple user-defined array of custom hyphenations into a custom tex file. For example an array like

<?php
$myCustomLatinHyphenationsInAFrenchText = [
    'forum' => ['fo', 'rum'],
    'romanum' => ['ro', 'ma', 'num'],
];

[alexander-nitsche]

@tnbnicer : As for integrating the Syllable API into your own project: you should not modify any file of the Syllable package directly, e.g. do not add the cache paths directly into the \Vanderlee\Syllable\Cache\File::filename() method, but do it from outside the package where you instantiate the Syllable object and set the cache path via the cache object, such as.

<?php
$remote_ip = "35.214.1.93";
$server = (strpos(getenv( "SERVER_ADDR" ), $remote_ip) !== false) ? "remote" : "local";
$remote = ($server == "remote") ? true : false;

$syllable = new Syllable('nl');

if ($remote) {
  $syllable->getCache()->setPath('/home/customer/www/mydomain/apps/phpSyllable-masterTest/src/Cache');
} else {
  // do nothing and use default path
}

PHP packages are generally meant to be used by instantiating some package objects and executing their API, i.e. their public methods.

And you should not set the cache path anywhere inside the syllable package but outside, so rather choose something like

if ($remote) {
  $syllable->getCache()->setPath('/home/customer/www/mydomain/cache/phpSyllable-masterTest');
} else {
  // do nothing and use default path
}

and create that path manually or by PHP code (as shown above).

And you should not write the full path but make use of the adaptive __DIR__ or similar environment variable, that contains the path to the project root folder. For example, if your application PHP file resides in the project root /home/customer/www/mydomain/ on the server, you should use

if ($remote) {
  $syllable->getCache()->setPath(__DIR__.'/cache/phpSyllable-masterTest');
} else {
  // do nothing and use default path
}

This prevents revealing sensitive information and keeps your application flexible. You could probably use the same configuration on your local webserver too, removing the whole local/remote condition:

<?php
$syllable = new Syllable('nl');
$syllable->getCache()->setPath(__DIR__.'/cache/phpSyllable-masterTest');