From eadd8a01ea94e629cdb5a9fbf21e04464f258075 Mon Sep 17 00:00:00 2001
From: Dendy Faist <dendy@fai.st>
Date: Sun, 9 Feb 2025 23:15:57 +0900
Subject: [PATCH] feat: Separate SentenceNote into its own extension of Note

---
 src/Controller/AnkiController.php |  32 +++++-
 src/Entity/Note.php               | 166 ++++------------------------
 src/Entity/SentenceNote.php       | 177 ++++++++++++++++++++++++++++++
 src/Service/AnkiService.php       |  24 +++-
 4 files changed, 249 insertions(+), 150 deletions(-)
 create mode 100644 src/Entity/SentenceNote.php

diff --git a/src/Controller/AnkiController.php b/src/Controller/AnkiController.php
index 951e3f7..6b2f521 100644
--- a/src/Controller/AnkiController.php
+++ b/src/Controller/AnkiController.php
@@ -3,6 +3,8 @@
 namespace App\Controller;
 
 use App\Entity\Note;
+use App\Entity\SentenceNote;
+use App\Entity\Term;
 use App\Service\AnkiService;
 use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;
 use Symfony\Component\HttpFoundation\Response;
@@ -23,11 +25,35 @@ class AnkiController extends AbstractController
     #[Route('/', name: 'main')]
     public function index(): Response
     {
-        $allIds = $this->ankiService->getAllNoteIds();
-
+        $allIds = $this->ankiService->getAllSentenceNoteIds();
         $allNotes = $this->ankiService->getNotes($allIds);
 
-        dd($allNotes);
+        $kanjiNotes = [];
+        foreach ($allNotes as $note) {
+            if (!$note instanceof SentenceNote) throw new \Exception(sprintf(
+                'Expected SentenceNote, got %s',
+                $note::class,
+            ));
+
+            foreach ($note->getTerms() as $term) {
+                assert($term instanceof Term);
+
+                if (key_exists($term->getKanji(), $kanjiNotes)) continue;
+
+                $newNote = new Note();
+
+                echo $note->getFields()['SentKanji'];
+                echo '<br><small>';
+                echo var_dump($note->getHighlightedKanji());
+                echo '</small><br><br>';
+                //echo $term->getKanji();
+                //echo ' | ';
+
+                $kanjiNotes[$term->getKanji()] = $newNote;
+            }
+        }
+
+        die();
     }
 
     #[Route('/nonconforming', name: 'nonconforming')]
diff --git a/src/Entity/Note.php b/src/Entity/Note.php
index d16c7d4..4e2df4b 100644
--- a/src/Entity/Note.php
+++ b/src/Entity/Note.php
@@ -2,58 +2,47 @@
 
 namespace App\Entity;
 
-//use App\Repository\NoteRepository;
-//use Doctrine\ORM\Mapping as ORM;
-
-//#[ORM\Entity(repositoryClass: NoteRepository::class)]
 class Note
 {
-    //#[ORM\Id]
-    //#[ORM\GeneratedValue]
-    //#[ORM\Column]
-
-    private int    $id;
-    private int    $mod;
-    private array  $terms = [];
-    private string $profile;
-    private array  $tags = [];
-    private string $model;
-    // Maybe these doesn't make sense to keep but leaving it here just in
-    // case for handiness' sake 
-    private array  $fields = [];
-    private ?array $mediaInfo = null;
-    private array  $cardIds;
-
     const HIGHLIGHT_PATTERN = '/<span\s+([^>]*)>(.*?)<\/span>/i';
     const HIGHLIGHT_ATTR_KANJI = 'style="color: rgb(255, 78, 8);"';
 
+    private   readonly int    $id;
+    private   readonly int    $mod;
+    private   readonly string $model;
+    private            string $profile;
+    private            array  $cardIds   = [];
+    protected          array  $fields    = [];
+    private            array  $tags      = [];
+
+
+    // -------------------------------------------------- Getters & setters ---
+
     public function getId(): int
     {
         return $this->id;
     }
-    public function getTerms(): array
+    public function getModel(): string
     {
-        return $this->terms;
+        return $this->model;
     }
+
     public function getFields(): array
     {
         return $this->fields;
     }
-
-    public function hasTerm(string $kanji): bool
+    public function setFields(array $fields): static
     {
-        foreach ($this->terms as $term) {
-            assert($term instanceof Term);
-
-            if ($term->kanji == $kanji) return true;
-        }
-
-        return false;
+        $this->fields = $fields;
+        return $this;
     }
 
-    public static function fromAnki(array $noteInfo): self
+
+    // ------------------------------------------------------- Anki-related ---
+
+    public static function fromAnki(array $noteInfo): static
     {
-        $note = new self();
+        $note = new static();
 
         [
             'noteId'    => $note->id,
@@ -72,127 +61,18 @@ class Note
         //         the order would be advisable.
         $note->fields = array_map(fn($x) => $x['value'], $noteInfo['fields']);
 
-        $note->mediaInfo = $note->parseMediaInfo($note->fields['Notes']);
-
-        // Set VocabKanji field
-        $note->terms = Term::fromNoteFields($note->fields);
-
-        // If unable to, create them from the highlighted parts in the sentence
-        if (empty($note->terms)) {
-            foreach ($note->getHighlightedKanji() as $highlighedKanji) {
-                $term = new Term();
-                $term->kanji = $highlighedKanji;
-                $term->definitionEn = null;
-                $term->definitionJp = null;
-                $note->terms[] = $term;
-            }
-        }
-
-        // Set to null whatever is null
-        $readings = array_map(
-            fn($x) => in_array($x, ['_', '＿', '']) ? null : $x,
-            explode('｜', $note->fields['VocabFurigana']),
-        );
-
-        // Set readings from furigana field
-        foreach ($note->terms as $key => &$term) {
-            if (null === $term->getReading()) {
-                if (null !== ($readings[$key] ?? null)) {
-                    $term->kanji .= '[' . $readings[$key] . ']';
-                }
-            }
-        }
-
         return $note;
     }
 
-    /** Return an array of strings with the highlighted kanji in the SentKanji */
-    public function getHighlightedKanji(): array
-    {
-        $ret = [];
-        $matches = [];
-
-        // 1. Get all spans in the text
-        preg_match_all(
-            self::HIGHLIGHT_PATTERN,
-            $this->fields['SentKanji'],
-            $matches,
-            PREG_SET_ORDER,
-        );
-
-        // 2. Check the ones that match with the kanji color
-        foreach ($matches as $match) {
-            if ($match[1] === self::HIGHLIGHT_ATTR_KANJI) {
-                $ret[] = mb_trim($match[2]);
-            }
-        }
-
-        return $ret;
-    }
-
     public function toAnki(): array
     {
         return [
             'id' => $this->id,
-            'fields' => [
-                'VocabKanji' => join('｜', array_map(
-                    fn(Term $x) => $x->getKanji(),
-                    $this->terms,
-                )),
-                'VocabFurigana' => join('｜', array_map(
-                    fn(Term $x) => $x->getReading() ?? '＿',
-                    $this->terms,
-                )),
-                'VocabDef' => join("<br>\n", array_map(
-                    fn(Term $x) => $x->toAnkiVocabDef(),
-                    $this->terms,
-                )),
-            ],
         ];
     }
 
-    public function parseMediaInfo(string $notes): ?array
-    {
-        $matches = null;
 
-        // Parse the notes fields. It can be in the form of
-        //   series-name_S01 EP07 (11h22m33s44ms)
-        // or
-        //   movie-name EP (11h22m33s44ms)
-        if (1 !== preg_match(
-            '/(?<name>[a-z\-_]+)(_S)?(?<season>\d*) EP(?<episode>\d*) \((?<time>.*)\)/n',
-            $notes,
-            $matches,
-        )) {
-            return null;
-        }
-
-        // Remove number-indexed matches, cast numbers to integers
-        $matches = [
-            'name' => $matches['name'],
-            'time' => $matches['time'],
-            // NOTE: intval returns 0 if not a number, which is false-like
-            'season'  => intval($matches['season'])  ?: null,
-            'episode' => intval($matches['episode']) ?: null,
-        ];
-
-        // Parse time into a DateInterval and replace it in the matches array
-        $time = new \DateInterval('PT0S');
-
-        preg_match('/(\d+)ms/', $matches['time'], $milliseconds);
-        preg_match('/(\d+)s/',  $matches['time'], $seconds);
-        preg_match('/(\d+)m/',  $matches['time'], $minutes);
-        preg_match('/(\d+)h/',  $matches['time'], $hours);
-
-        if ($milliseconds[1] ?? false) $time->f = $milliseconds[1] * 1000;
-        if ($seconds[1]      ?? false) $time->s = $seconds[1];
-        if ($minutes[1]      ?? false) $time->i = $minutes[1];
-        if ($hours[1]        ?? false) $time->h = $hours[1];
-
-        $matches['time'] = $time;
-
-        return $matches;
-    }
+    // ---------------------------------------------------- Derived methods ---
 
     public function getCreatedAt(): \DateTimeImmutable
     {
diff --git a/src/Entity/SentenceNote.php b/src/Entity/SentenceNote.php
new file mode 100644
index 0000000..0c04d4b
--- /dev/null
+++ b/src/Entity/SentenceNote.php
@@ -0,0 +1,177 @@
+<?php
+
+namespace App\Entity;
+
+class SentenceNote extends Note
+{
+    const MODEL_NAME = 'Japanese sentences';
+
+    private ?array $mediaInfo = [];
+    private array  $terms = [];
+
+    // -------------------------------------------------- Getters & setters ---
+
+    public function getTerms(): array
+    {
+        return $this->terms;
+    }
+    public function setTerms(array $terms): static
+    {
+        $this->terms = $terms;
+        return $this;
+    }
+
+
+    // ------------------------------------------------------- Anki-related ---
+
+    public static function fromAnki(array $noteInfo): static
+    {
+        $note = parent::fromAnki($noteInfo);
+
+        if ($note->getModel() !== self::MODEL_NAME) {
+            throw new \Exception('Trying to parse wrong model');
+        }
+
+        $note->mediaInfo = self::parseMediaInfo($note->fields['Notes']);
+
+        // Set VocabKanji field
+        $note->terms = Term::fromNoteFields($note->fields);
+
+        // If unable to, create them from the highlighted parts in the sentence
+        if (empty($note->terms)) {
+            foreach ($note->getHighlightedKanji() as $highlighedKanji) {
+                $term = new Term();
+                $term->kanji = $highlighedKanji;
+                $term->definitionEn = null;
+                $term->definitionJp = null;
+                $note->terms[] = $term;
+            }
+        }
+
+        // Set to null whatever is null
+        $readings = array_map(
+            fn($x) => in_array($x, ['_', '＿', '']) ? null : $x,
+            explode('｜', $note->fields['VocabFurigana']),
+        );
+
+        // Set readings from furigana field
+        foreach ($note->terms as $key => &$term) {
+            if (null === $term->getReading()) {
+                if (null !== ($readings[$key] ?? null)) {
+                    $term->kanji .= '[' . $readings[$key] . ']';
+                }
+            }
+        }
+
+        return $note;
+    }
+
+    public function toAnki(): array
+    {
+        return array_merge(parent::toAnki(), [
+            'fields' => [
+                'VocabKanji' => join('｜', array_map(
+                    fn(Term $x) => $x->getKanji(),
+                    $this->terms,
+                )),
+                'VocabFurigana' => join('｜', array_map(
+                    fn(Term $x) => $x->getReading() ?? '＿',
+                    $this->terms,
+                )),
+                'VocabDef' => join("<br>\n", array_map(
+                    fn(Term $x) => $x->toAnkiVocabDef(),
+                    $this->terms,
+                )),
+            ],
+        ]);
+    }
+
+
+    // ---------------------------------------------------- Derived methods ---
+
+    public function hasTerm(string $kanji): bool
+    {
+        foreach ($this->terms as $term) {
+            assert($term instanceof Term);
+
+            if ($term->kanji == $kanji) return true;
+        }
+
+        return false;
+    }
+
+    public function isSentKanjiHighlighted(): bool
+    {
+        return str_contains(
+            $this->fields['SentKanji'],
+            self::HIGHLIGHT_ATTR_KANJI,
+        );
+    }
+
+    /** Return an array of strings with the highlighted kanji in the SentKanji */
+    public function getHighlightedKanji(): array
+    {
+        $ret = [];
+        $matches = [];
+
+        // 1. Get all spans in the text
+        preg_match_all(
+            self::HIGHLIGHT_PATTERN,
+            $this->fields['SentKanji'],
+            $matches,
+            PREG_SET_ORDER,
+        );
+
+        // 2. Check the ones that match with the kanji color
+        foreach ($matches as $match) {
+            if ($match[1] === self::HIGHLIGHT_ATTR_KANJI) {
+                $ret[] = mb_trim($match[2]);
+            }
+        }
+
+        return $ret;
+    }
+
+    private static function parseMediaInfo(string $notes): ?array
+    {
+        $matches = null;
+
+        // Parse the notes fields. It can be in the form of
+        //   series-name_S01 EP07 (11h22m33s44ms)
+        // or
+        //   movie-name EP (11h22m33s44ms)
+        if (1 !== preg_match(
+            '/(?<name>[0-9A-Za-z\-_]+)(_S)?(?<season>\d*) EP(?<episode>\d*) \((?<time>.*)\)/n',
+            $notes,
+            $matches,
+        )) {
+            return null;
+        }
+
+        // Remove number-indexed matches, cast numbers to integers
+        $matches = [
+            'name' => $matches['name'],
+            'time' => $matches['time'],
+            // NOTE: intval returns 0 if not a number, which is false-like
+            'season'  => intval($matches['season'])  ?: null,
+            'episode' => intval($matches['episode']) ?: null,
+        ];
+
+        // Parse time into a DateInterval and replace it in the matches array
+        $time = new \DateInterval('PT0S');
+
+        preg_match('/(\d+)ms/', $matches['time'], $milliseconds);
+        preg_match('/(\d+)s/',  $matches['time'], $seconds);
+        preg_match('/(\d+)m/',  $matches['time'], $minutes);
+        preg_match('/(\d+)h/',  $matches['time'], $hours);
+
+        if ($milliseconds[1] ?? false) $time->f = $milliseconds[1] * 1000;
+        if ($seconds[1]      ?? false) $time->s = $seconds[1];
+        if ($minutes[1]      ?? false) $time->i = $minutes[1];
+        if ($hours[1]        ?? false) $time->h = $hours[1];
+
+        $matches['time'] = $time;
+
+        return $matches;
+    }
+}
diff --git a/src/Service/AnkiService.php b/src/Service/AnkiService.php
index 10faa09..370dbc0 100644
--- a/src/Service/AnkiService.php
+++ b/src/Service/AnkiService.php
@@ -3,6 +3,7 @@
 namespace App\Service;
 
 use App\Entity\Note;
+use App\Entity\SentenceNote;
 use Symfony\Contracts\HttpClient\HttpClientInterface;
 
 class AnkiService
@@ -30,7 +31,7 @@ class AnkiService
         return $result;
     }
 
-    public function getAllNoteIds(): array
+    public function getAllSentenceNoteIds(): array
     {
         return $this->request(
             'findNotes',
@@ -60,19 +61,34 @@ class AnkiService
 
     public function getNotes(array $nids): array
     {
-        return array_map(Note::fromAnki(...), $this->getNotesInfo($nids));
+        return array_map(self::parseNoteInfo(...), $this->getNotesInfo($nids));
     }
 
     public function getNote(int $nid): ?Note
     {
-        return Note::fromAnki($this->getNoteInfo($nid));
+        $noteInfo = $this->getNoteInfo($nid)
+            ?? throw new \Exception("Note $nid not found.");
+
+        return self::parseNoteInfo($noteInfo);
+    }
+
+    private static function parseNoteInfo(array $noteInfo): Note
+    {
+        return match ($noteInfo['modelName']) {
+            SentenceNote::MODEL_NAME => SentenceNote::fromAnki($noteInfo),
+            default => throw new \Exception(sprintf(
+                'Unrecognized Note "%s" of type "%s"',
+                $noteInfo['noteId'],
+                $noteInfo['modelName'],
+            ))
+        };
     }
 
     public function getLatestNote(): ?Note
     {
         // NoteIDs are just timestamps in milliseconds, so the latest is just
         // the biggest numerically
-        $latestId = max($this->getAllNoteIds());
+        $latestId = max($this->getAllSentenceNoteIds());
         return $this->getNote($latestId);
     }