Solr

zero exposes Solr (and other search engines) through a unified, type-erased ctx.Search handle.

Index, fetch, delete, and query documents with the same calls regardless of backend — add more backends (Elasticsearch, Meilisearch, …) in src/datasource/specialized/searchInterface.zig.

ctx.Search is optional: it is null until SOLR_URL (plus SOLR_DEFAULT_COLLECTION) is configured, so always guard with if (ctx.Search) |s| { ... } else { notConfigured }.

Configuration

configs/.env

# App configs
APP_ENV=dev
APP_NAME=basic
APP_VERSION=1.0.0
LOG_LEVEL=info
HTTP_PORT=8080

# Solr (search / persistence)
SOLR_URL=http://localhost:8983/solr
SOLR_DEFAULT_COLLECTION=docs
# Optional HTTP basic auth, as "user:pass"
SOLR_BASIC_AUTH=admin:secret

Env	Required	Description
SOLR_URL	yes	Base URL of the Solr node (include /solr).
SOLR_DEFAULT_COLLECTION	yes	Collection/core used when a call omits collection.
SOLR_BASIC_AUTH	no	user:pass sent as HTTP Basic auth.

API

try ctx.Search.index(ctx, "docs", "{\"id\":\"1\",\"title\":\"zero framework\"}"); // index a doc
const hits = try ctx.Search.query(ctx, "docs", "title:zero");                  // search -> JSON hits
const doc  = try ctx.Search.get(ctx, "docs", "1");                             // fetch by id (?[]const u8)
try ctx.Search.delete(ctx, "docs", "1");                                      // delete by id

Returned []const u8 slices (query/get) are allocated on the request allocator — free them with defer ctx.allocator.free(slice) when you hold a reference, or let the request-scoped allocator release them at the end of the handler.

Example handler

pub fn indexDoc(ctx: *Context) !void {
    if (ctx.Search) |s| {
        const doc = ctx.request.body() orelse "";
        try s.index(ctx, "docs", doc);
        try ctx.response.json(.{ .status = "indexed" }, .{});
    } else {
        ctx.response.setStatus(.not_implemented);
        try ctx.response.json(.{ .message = "SOLR_URL not configured" }, .{});
    }
}

pub fn search(ctx: *Context) !void {
    if (ctx.Search) |s| {
        const q = if (ctx.request.method == .POST)
            (ctx.request.body() orelse "")
        else
            (try ctx.request.query()).get("q") orelse "";
        const hits = try s.query(ctx, "docs", q);
        defer ctx.allocator.free(hits);
        ctx.response.content_type = .JSON;
        try ctx.response.writer().writeAll(hits);
    }
}

Recommendation

🚩 Use the ctx allocator wherever possible; it is tied to the request lifecycle, so its memory is released automatically and you avoid leaks.

See the runnable zero-search example and the implementation in src/datasource/specialized/solr.zig.