search

valu-digital / wp-graphql-polylang

WPGraphQL Polylang Extension for WordPress
https://packagist.org/packages/valu/wp-graphql-polylang
GNU General Public License v2.0
129 stars 22 forks source link

Language prefixes do not work in `idType: URI` queries #35

Open adriandmitroca opened 4 years ago

adriandmitroca commented 4 years ago

Hey,

I think at the current time it is literally impossible to fetch home page of non default language. Everything works fine for pages. Take look at full report listed below.

Fetching default home page

query Page {
  pageBy(uri: "/") {
    title
  }
  page(id: "/", idType: URI) {
    title
  }
}

Fetching home page for selected language

Disclaimer: Expected page exists, uses directory based permalinks and is available through WordPress through URL like: wordpress-instance.com/pl

query Page {
  pageBy(uri: "/pl") {
    title
  }
  page(id: "/pl", idType: URI) {
    title
  }
}

Fetching normal page for selected language

query Page {
  pageBy(uri: "/pl/kontakt") {
    title
  }
  page(id: "/pl/kontakt", idType: URI) {
    title
  }
}

Fetching normal page for selected language without language prefix

query Page {
  pageBy(uri: "/kontakt") {
    title
  }
  page(id: "/kontakt", idType: URI) {
    title
  }
}

Unless I am doing something wrong, I think there could be a great benefit by exposing additional language parameter under pageBy entity, alongside already existing id, pageId, and uri so that individual page could be safely retrieved in expected language. This should also fix fetching home page for different language issue at the same time.

What do you think?

PS. Excellent package by the way!

gu-stav commented 3 years ago

@esamattis Could you provide a pointer here, where to start/ fix that? 🙏🏽

adriandmitroca commented 3 years ago

@gustavpursche

Hey, I ultimately use REST API altogether with GraphQL (yes, that means multiple network requests) to fetch page IDs.

esamattis commented 3 years ago

Thanks for the report!

The id: "/kontakt", idType: URI args are fairly new in wp-graphql and this plugin has been written before those, so there's no build-in support yet.

Quick look at the wp-graphql code points the issue here:

https://github.com/wp-graphql/wp-graphql/blob/f15bbb5044066c3d09dc96cd5c23acbade1a5727/src/Data/NodeResolver.php#L303

The frontpage logic is hard coded. We need to inject our own custom logic to detect the translated frontpages. Unfortunately the resolve_uri() does not seem have any filters which we could use. There might be a way to process the id field earlier and transform it to a format supported by the resolve_uri().

But @jasonbahl of wp-graphql has been very open to adding filters to wp-graphql to support things like this.

For example a short circuiting filter like graphql_pre_resolver_uri' at the very begining of resolve_uri would allow us to solve this like this:

        $front_page_post = get_option('page_on_front');
        foreach (pll_languages_list('slug') as $lang_slug) {
            if ($parsed_url['path'] === "/$lang_slug") {
                $post = pll_get_post($front_page_post, $lang_slug);
                if ( $post ) {
                    return $post;
                }
            }
        }
esamattis commented 3 years ago

We might be able to wrap the existing page resolver function with the graphql_Page_fields filter and first check if the the request is for a translated front page and then call the original resolver if not 🤔

Here's an example how to override it https://github.com/valu-digital/wp-graphql-polylang/issues/38

simplenotezy commented 3 years ago

Ah, this would be nice. For now I will have to hard-code the frontpage slug in the code, but full support for the URI would be great.

esamattis commented 2 years ago

There's an user land based workaround by @gorkalaucirica here: https://github.com/valu-digital/wp-graphql-polylang/issues/46#issuecomment-852849101

Maybe we can build something like that in to this plugin.

lewebsimple commented 2 years ago

Any update on this ?

simplenotezy commented 2 years ago

A temporary workaround. This fixed it for my use-case:

/**
 * Bugfix: prevent fetching pages from other languages. Without below filter, /de/about-us would resolve
 * to the english post called /en/about-us, which will give SEO issues. It also fixes an issue if two
 * posts in two different languages had the same post_name, the wrong one would/could be resolved
 * depending on what language you was requesting
 *
 * https://github.com/valu-digital/wp-graphql-polylang/issues/35
 */

add_filter( 'request', function ( $query_args) {
    if(isset($query_args['uri']) && $query_args['uri'] && isset($query_args['lang'])) {
        $explodedUri = explode('/', $query_args['uri']);
        if($explodedUri[0] !== $query_args['lang']) {
            // query was made without language slug in URI
            return $query_args;
        }

        unset($explodedUri[0]); // remove language slug from URL
        $uriWithoutLang = implode('/', $explodedUri); // rebuild the URI without the slug in front

        if(function_exists('pll_get_post_language')) {
            $post = get_page_by_path($uriWithoutLang);

            if(!$post || $post && pll_get_post_language($post->ID) !== $query_args['lang']) {
                $query_args = [];
            }
        }
    }
    return $query_args;
},  10, 2);

It's not the prettiest, but it does it's job. We had an issue if two pages was called the same, e.g. /de/about us and /en/about-us, when requesting the /de/about-us, it would resolve the english version.

Also we had an issue where if you requested an english slug, on the german prefix, if a german page did not exist with that name, it would return the english page - on the german site. Above filter also fixes that.

MKlblangenois commented 2 years ago

Hey !

Any update about this issue ? I'm facing the same issue, with postQuery and contentNodeQuery

With "Remove ID from slug home" settings checked in Polylang : Capture d’écran 2022-07-08 à 14 06 58 Capture d’écran 2022-07-08 à 14 07 11

Without "Remove ID from slug home" settings checked in Polylang : Capture d’écran 2022-07-08 à 14 05 45 Capture d’écran 2022-07-08 à 14 06 00

maximilliangeorge commented 1 year ago

I am also experiencing this issue. It is not possible to request the translated homepage via a /<lang> uri, however /<lang>/some/path works just fine.

maximilliangeorge commented 1 year ago

Needs some polish but I think this will work for querying nodeByUri.

<?php

add_filter('graphql_resolve_field', function ($result, $source, $args, $context, $info, $type_name, $field_key, $field, $field_resolver) {

  if ($type_name === 'RootQuery' && $field_key === 'nodeByUri') {

    // explode uri and get only the middle part

    $exploded = explode('/', $args['uri']);
    $lang = $exploded[1];

    // if $exploded length is more than 3 it means that the uri is not a front page

    if (count($exploded) > 3) {
      return $result;
    }

    // if $exploded length is less than 2 it means that the uri is not a front page 

    if (count($exploded) < 2) {
      return $result;
    }

    // check if language exists in pll

    $valid_languages = pll_languages_list();

    if (in_array($lang, $valid_languages)) {

      // get translated post object

      $homepage_id = get_option('page_on_front');
      $frontpage = pll_get_post($homepage_id, $lang);
      $page = get_post($frontpage);

      // 💖

      $result = new \WPGraphQL\Model\Post($page);

    }

  }

  return $result;

}, 10, 9);
assef commented 11 months ago

I've improved @maximilliangeorge code, this way I was able to solve all my problems. To make this work you have to query using NodeByUri and pass the URL always with a leading slash ( / ).

My use case was:

Also, after getting the basic infos I needed from the page, I've used it's ID to query all the other infos (I'm querying each components props with its own query), which allowed me to use the Page query type.

add_filter('graphql_resolve_field', function ($result, $source, $args, $context, $info, $type_name, $field_key, $field, $field_resolver) {
  if ($type_name === 'RootQuery' && $field_key === 'nodeByUri') {
        // if is defaultLanguage home returns
        if($args["uri"] === "/") {
            return $result;
        }

        // explode uri and get only the first string
    $exploded = explode('/', substr($args['uri'], 1));
    $lang = $exploded[0];

    // if $exploded length is different than 1 (["slug"]) there's no chance to be a different language home
    if (count($exploded) != 1) {
      return $result;
    }

    // check if $lang is really a polylng aviable language (if not, it can be a normal page)
    $valid_languages = pll_languages_list();

    if (in_array($lang, $valid_languages)) {
      // get homepage ID in default language
      $homepage_id = get_option('page_on_front');
            // get all translations for homepage
            $translations = pll_get_post_translations($homepage_id);

            // get requested lang page ID or null if not available
            $post_id = $translations[$lang] ?? null;

            // checks if requested lang homePage exists and is published
            if (!$post_id || get_post_status($post_id) !== "publish") {
                    return null;
            }

            // returns pageObject
            return new \WPGraphQL\Model\Post(
                    \WP_Post::get_instance($post_id)
            );
    }
  }

  return $result;
}, 10, 9);

Edit: Improved the code to check if page is published because I had problems with draft pages.

maximilliangeorge commented 8 months ago

@assef Thanks, I was having trouble with my aforementioned solution after upgrading WPGraphQL. However, your altered solution above works! Thanks for saving me a bunch of time.

LarsEjaas commented 5 months ago

I've improved @maximilliangeorge code, this way I was able to solve all my problems. To make this work you have to query using NodeByUri and pass the URL always with a leading slash ( / ).

My use case was:

  • Home Page:

    • /
    • /en

  • Other Pages (E.g.: About Us)

    • /slug
    • /en/slug

Also, after getting the basic infos I needed from the page, I've used it's ID to query all the other infos (I'm querying each components props with its own query), which allowed me to use the Page query type.

add_filter('graphql_resolve_field', function ($result, $source, $args, $context, $info, $type_name, $field_key, $field, $field_resolver) {
  if ($type_name === 'RootQuery' && $field_key === 'nodeByUri') {
      // if is defaultLanguage home returns
      if($args["uri"] === "/") {
          return $result;
      }

      // explode uri and get only the first string
    $exploded = explode('/', substr($args['uri'], 1));
    $lang = $exploded[0];

    // if $exploded length is different than 1 (["slug"]) there's no chance to be a different language home
    if (count($exploded) != 1) {
      return $result;
    }

    // check if $lang is really a polylng aviable language (if not, it can be a normal page)
    $valid_languages = pll_languages_list();

    if (in_array($lang, $valid_languages)) {
      // get homepage ID in default language
      $homepage_id = get_option('page_on_front');
          // get all translations for homepage
          $translations = pll_get_post_translations($homepage_id);

          // get requested lang page ID or null if not available
          $post_id = $translations[$lang] ?? null;

          // checks if requested lang homePage exists and is published
          if (!$post_id || get_post_status($post_id) !== "publish") {
                  return null;
          }

          // returns pageObject
          return new \WPGraphQL\Model\Post(
                  \WP_Post::get_instance($post_id)
          );
    }
  }

  return $result;
}, 10, 9);

Edit: Improved the code to check if page is published because I had problems with draft pages.

I really love your solution: Thanks so much for sharing! 🙏

I sometimes get empty strings in the $exploded variable, so I rewrote the code like this:

// explode uri and get only the first string
$exploded = explode('/', substr($args['uri'], 1));
// Filter out empty strings
$exploded = array_filter($exploded, function ($value) {
return strlen($value) > 0;
});