zig-search-engine

Search-engine integrations for Zig services — a Zig take on @stacksjs/search-engine.

Collections are declared as model-style Schemas (searchable / filterable / sortable attributes), and a driver turns that declaration into the engine-native data structure. Typesense is the first driver.

Usage

const se = @import("search-engine");

// The model: which attributes are searchable, filterable (facets), sortable.
const messages_schema = se.Schema{
    .name = "messages",
    .searchable = &.{ .{ .name = "subject" }, .{ .name = "body" } },
    .filterable = &.{ .{ .name = "username" }, .{ .name = "mailbox" } },
    .sortable = &.{ .{ .name = "date", .type = .int64 } },
};

const engine = se.Typesense.init(.{ .api_key = key }); // host/port default to 127.0.0.1:8108
try engine.ensureCollection(allocator, messages_schema);

// Index a document
var doc = se.DocBuilder.init(allocator);
defer doc.deinit();
try doc.putString("id", "msg-1");
try doc.putString("subject", "hello world");
const json = try doc.toOwned();
defer allocator.free(json);
try engine.upsertDocument(allocator, "messages", json);

// Search
var result = try engine.search(allocator, "messages", .{
    .query = "hello",
    .query_by = "subject,body",
    .filter_by = "username:=`chris`",
});
defer result.deinit();

var it = result.hits();
while (it.next()) |hit| {
    std.debug.print("{s}\n", .{se.SearchResponse.docString(hit, "subject")});
}

Also provided: importDocuments (JSONL bulk upsert), deleteDocument, deleteCollection, healthy, and JSON/URL encoding helpers under se.http.

Requirements

• Zig 0.16.0-dev+ (developed against 0.17.0-dev)
• libc (the HTTP client uses libc sockets)

Install

zig fetch --save https://github.com/zig-utils/zig-search-engine/archive/<commit>.tar.gz

Then in build.zig:

const search_engine = b.dependency("search_engine", .{ .target = target, .optimize = optimize });
exe.root_module.addImport("search-engine", search_engine.module("search-engine"));

License

MIT