Real files, annotated.

347      try {
348        $mysqli = new mysqli(DB_SERVER, DB_USER, DB_PASSWORD, DB_NAME, DB_PORT, DB_SOCKET);
349      } catch (mysqli_sql_exception $e) {
350        error_log('[AMAPI Map] database connection failed: ' . $e->getMessage());
351        return [];
352      }
353      $mysqli->set_charset('utf8mb4');
354
355      // Only the columns actually used below, instead of SELECT *.
356      $columns = 'article, title, artist, lat, lon, nature, wikidata';
357
358      try {
359        if (is_array($nature) && sizeof($nature) > 0) {
360          $placeholders = implode(', ', array_fill(0, sizeof($nature), '?'));
361          $stmt = $mysqli->prepare(
362            'SELECT ' . $columns . ' FROM ' . DATABASE_MAP . ' WHERE nature IN (' . $placeholders . ')'
363          );
364          // All bound values are strings; str_repeat builds the type string.
365          $stmt->bind_param(str_repeat('s', sizeof($nature)), ...array_values($nature));
366          $stmt->execute();
367          $result = $stmt->get_result();
368        } else {
369          $result = $mysqli->query('SELECT ' . $columns . ' FROM ' . DATABASE_MAP);
370        }
371      } catch (mysqli_sql_exception $e) {

143// 3) Rewrite am_map inside a transaction (atomic; the map is never empty
144//    mid-rebuild). DELETE + INSERT rather than TRUNCATE so it can roll back.
145$mysqli = @new mysqli( DB_SERVER, DB_USER, DB_PASSWORD, DB_NAME, DB_PORT, DB_SOCKET );
146if ( $mysqli->connect_errno ) {
147    rm_log( 'Database connection failed: ' . $mysqli->connect_error );
148    flock( $lock, LOCK_UN );
149    exit( 1 );
150}
151$mysqli->set_charset( 'utf8' );
152
153$mysqli->begin_transaction();
154try {
155    $mysqli->query( 'DELETE FROM ' . DATABASE_MAP . ';' );
156
157    $batchSize = 100;
158    for ( $j = 0; $j < $total; $j += $batchSize ) {
159        $values = [];
160        for ( $i = $j; $i < $j + $batchSize && $i < $total; $i++ ) {
161            $row = $dataAM[ $i ];
162
163            // Keep each value within its column width so a stray oversized value
164            // can never abort the whole rebuild. Wikidata must be a Q-id.
165            $wd = (string)( $row['wikidata'] ?? '' );
166            if ( !preg_match( '/^Q[0-9]{1,14}$/', $wd ) ) {
167                $wd = '';
168            }
169            $article = mb_substr( (string)( $row['article'] ?? '' ), 0, 255 );
170            $title   = mb_substr( (string)( $row['title'] ?? '' ), 0, 255 );
171            $artist  = mb_substr( (string)( $row['artist'] ?? '' ), 0, 255 );
172            $nature  = mb_substr( (string)( $row['nature'] ?? '' ), 0, 20 );
173            $image   = mb_substr( (string)( $row['image'] ?? '' ), 0, 255 );
174
175            $values[] = sprintf(
176                '(%d,"%s","%s","%s",%f,%f,"%s","%s","%s")',
177                $i + 1,
178                $mysqli->real_escape_string( $article ),
179                $mysqli->real_escape_string( $title ),
180                $mysqli->real_escape_string( $artist ),
181                isset( $row['lat'] ) ? (float)$row['lat'] : 0,
182                isset( $row['lon'] ) ? (float)$row['lon'] : 0,
183                $mysqli->real_escape_string( $nature ),
184                $mysqli->real_escape_string( $wd ),
185                $mysqli->real_escape_string( $image )
186            );
187        }
188        if ( $values ) {
189            $sql = 'INSERT INTO ' . DATABASE_MAP
190                . ' (id,article,title,artist,lat,lon,nature,wikidata,image) VALUES '
191                . implode( ',', $values );
192            if ( !$mysqli->query( $sql ) ) {
193                throw new RuntimeException( $mysqli->error );
194            }
195        }
196    }
197
198    $mysqli->commit();
199} catch ( \Throwable $e ) {
200    $mysqli->rollback();
201    rm_log( 'Rebuild failed, rolled back: ' . $e->getMessage() );
202    flock( $lock, LOCK_UN );
203    exit( 1 );
204}
205
206$ms = round( ( microtime( true ) - $t0 ) * 1000 );
207rm_log( "Map rebuilt: $total points written to " . DATABASE_MAP . " in {$ms} ms." );

142	public static function resolveLabels( array $qids ): array {
143		$out = [];
144		$fetch = [];
145		$dir = sys_get_temp_dir();
146		foreach ( $qids as $q ) {
147			$cf = $dir . '/amrepair-' . $q . '.txt';
148			if ( is_file( $cf ) && ( time() - filemtime( $cf ) < 2592000 ) ) {
149				$v = @file_get_contents( $cf );
150				if ( $v !== false && $v !== '' ) { $out[ $q ] = $v; continue; }
151			}
152			$fetch[] = $q;
153		}
154		// Wikidata allows up to 50 ids per call.
155		foreach ( array_chunk( $fetch, 50 ) as $chunk ) {
156			$url = 'https://www.wikidata.org/w/api.php?action=wbgetentities&props=labels'
157				. '&languages=fr%7Cen&format=json&ids=' . implode( '%7C', $chunk );
158			$raw = self::httpGet( $url );
159			$json = $raw ? json_decode( $raw ) : null;
160			if ( !$json || !isset( $json->entities ) ) { continue; }
161			foreach ( $chunk as $q ) {
162				$label = '';
163				if ( isset( $json->entities->{$q}->labels->fr->value ) ) {
164					$label = $json->entities->{$q}->labels->fr->value;
165				} elseif ( isset( $json->entities->{$q}->labels->en->value ) ) {
166					$label = $json->entities->{$q}->labels->en->value;
167				}
168				if ( $label !== '' ) {
169					$out[ $q ] = $label;
170					@file_put_contents( $dir . '/amrepair-' . $q . '.txt', $label );
171				}
172			}
173		}
174		return $out;
175	}
176
177	/** Simple GET with a descriptive User-Agent (Wikidata requires one). */

86	public static function rebuild( ?callable $progress = null ): array {
87		$config   = MediaWikiServices::getInstance()->getMainConfig();
88		$category = $config->get( 'AtlasQualiteCategory' );
89		$checks   = self::checks();
90
91		$rows        = self::querySmw( $category, $checks );
92		$records     = [];
93		$issueCounts = [];
94		foreach ( $checks as $c ) {
95			$issueCounts[ $c['key'] ] = 0;
96		}
97		$complete = 0;
98		$done     = 0;
99
100		foreach ( $rows as $page => $values ) {
101			$missing    = [];
102			$maxWeight  = 0;
103			$gotWeight  = 0;
104
105			foreach ( $checks as $c ) {
106				$maxWeight += $c['weight'];
107				$val = $values[ $c['key'] ] ?? '';
108				$ok  = self::passes( $c, $val );
109				if ( $ok ) {
110					$gotWeight += $c['weight'];
111				} else {
112					$missing[] = $c['key'];
113					$issueCounts[ $c['key'] ]++;
114				}
115			}
116
117			$score = $maxWeight > 0 ? (int) round( 100 * $gotWeight / $maxWeight ) : 100;
118			if ( $score === 100 ) {
119				$complete++;
120			}
121
122			$records[] = [
123				'page'    => $page,
124				'score'   => $score,
125				'missing' => $missing,
126			];
127
128			$done++;
129			if ( $progress && ( $done % 200 === 0 ) ) {
130				$progress( $done );
131			}
132		}
133

86export function linkHash(prevHash: Sha256, body: ChainRecordBody): Sha256 {
87  const canonical = canonicalJsonOrThrow(bodyForHashing(body));
88  return Sha256.unsafe(sha256Hex(concatBytes(utf8(prevHash), utf8(canonical))));
89}
90
91/**
92 * The hashed shape is written out field by field rather than spread from the
93 * argument. An extra property on a caller's object must never silently become
94 * part of the digest, and a renamed field must break the build rather than
95 * quietly invalidate every stored chain.
96 */
97function bodyForHashing(body: ChainRecordBody): JsonValue {
98  return {
99    chainId: body.chainId,
100    payload: body.payload,
101    recordedAt: body.recordedAt,
102    sequence: body.sequence,
103  };
104}
105

89CREATE TABLE measurements (
90    id          uuid        PRIMARY KEY,
91    run_id      uuid        NOT NULL UNIQUE REFERENCES runs (id) ON DELETE RESTRICT,
92    target_id   uuid        NOT NULL REFERENCES targets (id) ON DELETE RESTRICT,
93
94    -- The chain. `payload` is the canonical-JSON body; `hash` is
95    -- sha256(prev_hash || canonical_json(body)) as defined in packages/core.
96    chain_id    uuid        NOT NULL,
97    sequence    bigint      NOT NULL,
98    recorded_at timestamptz NOT NULL,
99    payload     jsonb       NOT NULL,
100    prev_hash   char(64)    NOT NULL,
101    hash        char(64)    NOT NULL,
102
103    created_at  timestamptz NOT NULL DEFAULT now(),
104
105    CONSTRAINT measurements_sequence_non_negative CHECK (sequence >= 0),
106    CONSTRAINT measurements_hash_hex      CHECK (hash ~ '^[0-9a-f]{64}$'),
107    CONSTRAINT measurements_prev_hash_hex CHECK (prev_hash ~ '^[0-9a-f]{64}$'),
108    -- The genesis record is the only one that may point at the zero hash.
109    CONSTRAINT measurements_genesis_is_first CHECK (
110        (sequence = 0) = (prev_hash = repeat('0', 64))
111    ),
112    -- A chain never forks: one record per position, and one successor per record.
113    CONSTRAINT measurements_one_per_position UNIQUE (chain_id, sequence),
114    CONSTRAINT measurements_one_successor    UNIQUE (chain_id, prev_hash),
115    CONSTRAINT measurements_hash_unique      UNIQUE (hash)
116);

102export async function claimJobs(
103  sql: Sql,
104  workerId: string,
105  limit = 1,
106  leaseMs: number = DEFAULT_LEASE_MS,
107): Promise<ClaimedJob[]> {
108  const rows = await sql<JobRow[]>`
109    UPDATE jobs AS j
110    SET state = 'claimed',
111        claimed_by = ${workerId},
112        claimed_at = now(),
113        lease_expires_at = now() + make_interval(secs => ${leaseMs / 1000}),
114        attempts = j.attempts + 1,
115        updated_at = now()
116    WHERE j.id IN (
117      SELECT c.id
118      FROM jobs c
119      WHERE c.state = 'pending'
120        AND c.run_at <= now()
121        -- A job with no attempts left must never be picked up: the claim
122        -- increments attempts, and jobs_attempts_bounded would reject it.
123        -- Reaching that constraint means the queue tried to run something it
124        -- had already given up on.
125        AND c.attempts < c.max_attempts
126      ORDER BY c.run_at
127      FOR UPDATE SKIP LOCKED
128      LIMIT ${limit}
129    )
130    RETURNING j.id, j.target_id, j.state, j.run_at, j.attempts, j.max_attempts,
131              j.claimed_by, j.lease_expires_at, j.last_error, j.dead_lettered_at
132  `;
133  return rows.map(toClaimed);
134}