Function main

packages/plugin/scripts/backfill-embeddings.ts:29–134

()

}

async function main() {
    const directory = getArg("--directory") ?? process.cwd();
    const projectFilter = getArg("--project");
    const forceProjectPath = process.argv.includes("--force-project-path");
    const projectIdentity = resolveProjectIdentity(directory);

    if (projectFilter && projectFilter !== projectIdentity && !forceProjectPath) {
        console.error(
            `--project ${projectFilter} does not match identity for --directory ${directory}: ${projectIdentity}. ` +
                "Pass --force-project-path to override.",
        );
        process.exit(1);
    }

    const db = new Database(DB_PATH);
    db.exec("PRAGMA journal_mode=WAL");
    const config = loadPluginConfig(directory);
    registerProjectEmbedding(
        db,
        projectIdentity,
        config.embedding,
        {
            memoryEnabled: config.memory.enabled,
            gitCommitEnabled: config.memory.git_commit_indexing.enabled,
        },
        directory,
    );

    const snapshot = getProjectEmbeddingSnapshot(projectIdentity);
    if (!snapshot?.enabled) {
        console.error("Embedding is disabled for this project.");
        db.close();
        process.exit(1);
    }

    // Find memories without embeddings for the current model (optionally filtered to one project).
    const effectiveProject = projectFilter ?? projectIdentity;
    const query = effectiveProject
        ? `SELECT m.id, m.content, m.category, m.project_path
           FROM memories m
           LEFT JOIN memory_embeddings me ON me.memory_id = m.id AND me.model_id = ?
           WHERE m.status != 'deleted' AND me.memory_id IS NULL AND m.project_path = ?`
        : `SELECT m.id, m.content, m.category, m.project_path
           FROM memories m
           LEFT JOIN memory_embeddings me ON me.memory_id = m.id AND me.model_id = ?
           WHERE m.status != 'deleted' AND me.memory_id IS NULL`;
    const stmt = db.prepare(query);
    const allMemories = (
        effectiveProject ? stmt.all(snapshot.modelId, effectiveProject) : stmt.all(snapshot.modelId)
    ) as Array<{ id: number; content: string; category: string; project_path: string }>;

    console.log(
        `Found ${allMemories.length} memories without embeddings${effectiveProject ? ` in project ${effectiveProject}` : ""}`,
    );

    if (allMemories.length === 0) {
        console.log("Nothing to do.");
        db.close();