use Illuminate\Support\Facades\DB;

// Insert <br> after each sentence-ending period in question_text + solutions,
// for the ingested past-paper set (qa_status ai_generated / unreviewed).
// Safe: never touches math ($..$, $$..$$, \(..\), \[..\]) or html tags; keeps
// decimals/money (no space after the dot) and common abbreviations intact.
// Idempotent: ".<br>" has no trailing space so it never re-matches.
//
// Set $DRY = true to preview only (no writes). false = backup + update.

$DRY = false;
$LIMIT_SAMPLES = 10;

$ABBR = ['mr','mrs','ms','mdm','dr','prof','sr','jr','st','vs','etc','no','fig','approx','e.g','i.e'];

$sb = function($s) use ($ABBR) {
    if ($s === null || $s === '') return $s;
    // single pass: group 1 = math/tag (kept verbatim); else groups 2/3 = a
    // token + sentence-ending period + following spaces.
    $rx = '/(\$\$.*?\$\$|\$[^\$\n]+\$|\\\\\(.*?\\\\\)|\\\\\[.*?\\\\\]|<[^>]+>)|([A-Za-z0-9.]*)\.([ \t]+)/su';
    $out = preg_replace_callback($rx, function($m) use ($ABBR) {
        if (isset($m[1]) && $m[1] !== '') return $m[1];       // math / tag: leave alone
        $w = strtolower($m[2] ?? '');
        if (in_array($w, $ABBR, true)) return $m[0];          // abbreviation: no break
        return ($m[2] ?? '') . '.<br>';
    }, $s);
    if ($out === null) return $s;                             // regex failure: no-op
    return preg_replace('/(?:<br>\s*){2,}/', '<br>', $out);   // collapse doubled breaks
};

$ids = DB::table('questions')->whereIn('qa_status',['ai_generated','unreviewed'])->pluck('id')->toArray();
echo "scope questions: ".count($ids)."\n";

if ($DRY) {
    $shown = 0;
    foreach (DB::table('questions')->whereIn('id',$ids)
                ->where('question','like','% %. %')->inRandomOrder()->limit(200)->get(['id','question']) as $q) {
        $nq = $sb($q->question);
        if ($nq !== $q->question) {
            echo "--- Q#{$q->id} ---\nBEFORE: ".mb_substr($q->question,0,200)."\nAFTER : ".mb_substr($nq,0,220)."\n";
            if (++$shown >= $LIMIT_SAMPLES) break;
        }
    }
    // count how many WOULD change
    $qc=0; foreach (array_chunk($ids,1000) as $c){
        foreach (DB::table('questions')->whereIn('id',$c)->get(['question','explanation']) as $q){
            if ($sb($q->question)!==$q->question || $sb($q->explanation)!==$q->explanation) $qc++;
        }
    }
    echo "\nWOULD update questions: $qc (dry-run, no writes)\n";
    return;
}

// real run
$ts = date('Ymd_His');
$bk = fopen(storage_path("app/_sbreak_backup_$ts.jsonl"), 'w');
$qn=0; $sn=0;
foreach (array_chunk($ids,500) as $chunk) {
    foreach (DB::table('questions')->whereIn('id',$chunk)->get(['id','question','explanation']) as $q) {
        $nq=$sb($q->question); $ne=$sb($q->explanation);
        if ($nq!==$q->question || $ne!==$q->explanation) {
            fwrite($bk, json_encode(['t'=>'q','id'=>$q->id,'q'=>$q->question,'e'=>$q->explanation])."\n");
            DB::table('questions')->where('id',$q->id)->update(['question'=>$nq,'explanation'=>$ne]);
            $qn++;
        }
    }
    foreach (DB::table('solutions')->whereIn('question_id',$chunk)->get(['id','solution']) as $s) {
        $ns=$sb($s->solution);
        if ($ns!==$s->solution) {
            fwrite($bk, json_encode(['t'=>'s','id'=>$s->id,'s'=>$s->solution])."\n");
            DB::table('solutions')->where('id',$s->id)->update(['solution'=>$ns]);
            $sn++;
        }
    }
}
fclose($bk);
echo "DONE questions_updated=$qn solutions_updated=$sn  backup=storage/app/_sbreak_backup_$ts.jsonl\n";