Update (keywords): new logic

This commit is contained in:
2023-11-23 02:15:41 +08:00
parent d2c2059531
commit b2762ecef2
15 changed files with 211 additions and 198 deletions

View File

@@ -25,7 +25,7 @@ public function home(Request $request)
// $query->whereNotIn('id', $featured_posts->pluck('id')->toArray());
// })->where('status', 'publish')->where('published_at', '<=', now())->orderBy('published_at', 'desc')->limit(10)->get();
$rss_posts = RssPost::where('status', 'published')->orderBy('published_at', 'desc')->paginate(15);
$rss_posts = RssPost::with('entities_keywords')->where('status', 'published')->orderBy('published_at', 'desc')->paginate(15);
return response(view('front.welcome', compact('rss_posts')), 200);
}

View File

@@ -40,7 +40,7 @@ public function searchResults(Request $request, $query)
SEOTools::jsonLd();
SEOTools::setTitle($title, false);
$rss_posts = RssPost::with('category')
$rss_posts = RssPost::with('category', 'entities_keywords')
->where('status', 'published')
->whereRaw("to_tsvector('english', title || ' ' || bites || ' ' || keyword_list) @@ plainto_tsquery('english', ?)", [trim(preg_replace('/\s+/', ' ', $query))])
->where('published_at', '<=', now())
@@ -81,7 +81,7 @@ public function index(Request $request)
SEOTools::jsonLd();
SEOTools::setTitle($title, false);
$rss_posts = RssPost::with('category')->where('status', 'published')
$rss_posts = RssPost::with('category', 'entities_keywords')->where('status', 'published')
->where('published_at', '<=', now())
->orderBy('published_at', 'desc')
->cursorPaginate(60) ?? collect();
@@ -128,7 +128,7 @@ public function category(Request $request, $category_slug)
SEOTools::jsonLd();
SEOTools::setTitle($title, false);
$rss_posts = RssPost::with('category')->where('status', 'published')
$rss_posts = RssPost::with('category', 'entities_keywords')->where('status', 'published')
->where('category_id', $category->id)
->where('published_at', '<=', now())
->orderBy('published_at', 'desc')

View File

@@ -25,13 +25,11 @@ public function prm(Request $request)
}
ParseRssPostMetadataTask::handle($id);
return 'ok';
}
public function crawlTask(Request $request)
{
$id = $request->input('id');
@@ -41,12 +39,11 @@ public function crawlTask(Request $request)
}
CrawlRssPostTask::handle($id);
return 'ok';
}
public function opml(Request $request)
{
$raw_posts = BrowseRSSLatestNewsTask::handleSingle('https://hnrss.org/newest?q=ai', 240);

View File

@@ -2,7 +2,6 @@
namespace App\Jobs;
use App\Jobs\Tasks\CrawlRssPostTask;
use App\Models\RssPost;
use App\Models\RssPostKeyword;
use Illuminate\Bus\Queueable;
@@ -32,114 +31,105 @@ public function __construct($rss_post_id)
*/
public function handle(): void
{
$rss_post = RssPost::find($this->rss_post_id);
$rss_post = RssPost::find($this->rss_post_id);
if (is_null($rss_post))
{
return ;
}
if (is_null($rss_post)) {
return;
}
if ($rss_post->keyword_saved == true)
{
return ;
}
if ($rss_post->keyword_saved == true) {
return;
}
$words_to_add_in_keyword_list = [];
$words_to_save = [];
$words_to_add_in_keyword_list = [];
$words_to_save = [];
$first_keyword_found = false;
$first_keyword_found = false;
// Entities
if (isset($rss_post->entities)) {
if (count($rss_post->entities) > 0) {
// Entities
if (isset($rss_post->entities)) {
if (count($rss_post->entities) > 0) {
foreach ($rss_post->entities as $key => $word) {
foreach ($rss_post->entities as $key => $word) {
$word = trim($word);
$word = trim($word);
$words_to_save[] = (object) [
'is_main' => ($key == 0) ? true : false,
'type' => 'entity',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_save[] = (object) [
'is_main' => ($key == 0) ? true : false,
'type' => 'entity',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_add_in_keyword_list[] = $word;
}
}
}
// Keywords
if (isset($rss_post->keywords)) {
if (count($rss_post->keywords) > 0) {
foreach ($rss_post->keywords as $word) {
$word = trim($word);
foreach($words_to_save as $saved_word)
{
if (strtolower($word) == $saved_word->value_lowercased)
{
continue 2;
}
}
$words_to_save[] = (object) [
'is_main' => !$first_keyword_found,
'type' => 'keyword',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_add_in_keyword_list[] = $word;
if ($first_keyword_found == false) {
$first_keyword_found = true;
}
}
}
}
$rss_post->keyword_list = implode(',', $words_to_add_in_keyword_list);
$rss_post->status = 'published';
if($rss_post->save())
{
$has_saved_keyword = false;
$deleted_rpk = RssPostKeyword::where('rss_post_id', $rss_post->id)->delete();
foreach ($words_to_save as $word_to_save)
{
$new_rpk = new RssPostKeyword;
$new_rpk->rss_post_id = $rss_post->id;
$new_rpk->type = $word_to_save->type;
$new_rpk->is_main = $word_to_save->is_main;
$new_rpk->value = $word_to_save->value;
$new_rpk->value_lowercased = $word_to_save->value_lowercased;
$new_rpk->created_at = $rss_post->published_at;
$new_rpk->updated_at = $rss_post->published_at;
if($new_rpk->save())
{
if (!$has_saved_keyword)
{
$has_saved_keyword = true;
$words_to_add_in_keyword_list[] = $word;
}
}
}
}
if ($has_saved_keyword)
{
$rss_post->keyword_saved = true;
$rss_post->save();
// Keywords
if (isset($rss_post->keywords)) {
if (count($rss_post->keywords) > 0) {
foreach ($rss_post->keywords as $word) {
$word = trim($word);
foreach ($words_to_save as $saved_word) {
if (strtolower($word) == $saved_word->value_lowercased) {
continue 2;
}
}
$words_to_save[] = (object) [
'is_main' => ! $first_keyword_found,
'type' => 'keyword',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_add_in_keyword_list[] = $word;
if ($first_keyword_found == false) {
$first_keyword_found = true;
}
}
}
}
$rss_post->keyword_list = implode(',', $words_to_add_in_keyword_list);
$rss_post->status = 'published';
if ($rss_post->save()) {
$has_saved_keyword = false;
$deleted_rpk = RssPostKeyword::where('rss_post_id', $rss_post->id)->delete();
foreach ($words_to_save as $word_to_save) {
$new_rpk = new RssPostKeyword;
$new_rpk->rss_post_id = $rss_post->id;
$new_rpk->type = $word_to_save->type;
$new_rpk->is_main = $word_to_save->is_main;
$new_rpk->value = $word_to_save->value;
$new_rpk->value_lowercased = $word_to_save->value_lowercased;
$new_rpk->created_at = $rss_post->published_at;
$new_rpk->updated_at = $rss_post->published_at;
if ($new_rpk->save()) {
if (! $has_saved_keyword) {
$has_saved_keyword = true;
}
}
}
if ($has_saved_keyword) {
$rss_post->keyword_saved = true;
$rss_post->save();
}
}
}
}
}

View File

@@ -63,10 +63,10 @@ public static function handle(int $rss_post_id)
$word = trim($word);
$words_to_save[] = (object) [
'is_main' => ($key == 0) ? true : false,
'type' => 'entity',
'value' => $word,
'value_lowercased' => strtolower($word),
'is_main' => ($key == 0) ? true : false,
'type' => 'entity',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_add_in_keyword_list[] = $word;
@@ -82,26 +82,23 @@ public static function handle(int $rss_post_id)
$word = trim($word);
foreach($words_to_save as $saved_word)
{
if (strtolower($word) == $saved_word->value_lowercased)
{
continue 2;
}
foreach ($words_to_save as $saved_word) {
if (strtolower($word) == $saved_word->value_lowercased) {
continue 2;
}
}
$words_to_save[] = (object) [
'is_main' => !$first_keyword_found,
'type' => 'keyword',
'value' => $word,
'value_lowercased' => strtolower($word),
'is_main' => ! $first_keyword_found,
'type' => 'keyword',
'value' => $word,
'value_lowercased' => strtolower($word),
];
$words_to_add_in_keyword_list[] = $word;
if ($first_keyword_found == false) {
$first_keyword_found = true;
$first_keyword_found = true;
}
}
@@ -145,38 +142,33 @@ public static function handle(int $rss_post_id)
$rss_post->keyword_list = implode(',', $words_to_add_in_keyword_list);
$rss_post->status = 'published';
if($rss_post->save())
{
$has_saved_keyword = false;
$deleted_rpk = RssPostKeyword::where('rss_post_id', $rss_post->id)->delete();
if ($rss_post->save()) {
$has_saved_keyword = false;
foreach ($words_to_save as $word_to_save)
{
$deleted_rpk = RssPostKeyword::where('rss_post_id', $rss_post->id)->delete();
$new_rpk = new RssPostKeyword;
$new_rpk->rss_post_id = $rss_post->id;
$new_rpk->type = $word_to_save->type;
$new_rpk->is_main = $word_to_save->is_main;
$new_rpk->value = $word_to_save->value;
$new_rpk->value_lowercased = $word_to_save->value_lowercased;
foreach ($words_to_save as $word_to_save) {
if($new_rpk->save())
{
if (!$has_saved_keyword)
{
$has_saved_keyword = true;
}
$new_rpk = new RssPostKeyword;
$new_rpk->rss_post_id = $rss_post->id;
$new_rpk->type = $word_to_save->type;
$new_rpk->is_main = $word_to_save->is_main;
$new_rpk->value = $word_to_save->value;
$new_rpk->value_lowercased = $word_to_save->value_lowercased;
if ($new_rpk->save()) {
if (! $has_saved_keyword) {
$has_saved_keyword = true;
}
}
}
}
if ($has_saved_keyword)
{
$rss_post->keyword_saved = true;
$rss_post->save();
}
if ($has_saved_keyword) {
$rss_post->keyword_saved = true;
$rss_post->save();
}
}
}

View File

@@ -69,6 +69,11 @@ class RssPost extends Model implements Feedable
'keyword_saved',
];
public function entities_keywords()
{
return $this->hasMany(RssPostKeyword::class);
}
public function category()
{
return $this->belongsTo(Category::class);

View File

@@ -11,7 +11,7 @@
/**
* Class RssPostKeyword
*
*
* @property int $id
* @property int $rss_post_id
* @property string $type
@@ -20,30 +20,27 @@
* @property string $value_lowercased
* @property Carbon|null $created_at
* @property Carbon|null $updated_at
*
* @property RssPost $rss_post
*
* @package App\Models
*/
class RssPostKeyword extends Model
{
protected $table = 'rss_post_keywords';
protected $table = 'rss_post_keywords';
protected $casts = [
'rss_post_id' => 'int',
'is_main' => 'bool'
];
protected $casts = [
'rss_post_id' => 'int',
'is_main' => 'bool',
];
protected $fillable = [
'rss_post_id',
'type',
'is_main',
'value',
'value_lowercased'
];
protected $fillable = [
'rss_post_id',
'type',
'is_main',
'value',
'value_lowercased',
];
public function rss_post()
{
return $this->belongsTo(RssPost::class);
}
public function rss_post()
{
return $this->belongsTo(RssPost::class);
}
}