Start process_scrobbles job

This commit is contained in:
mitteneer 2024-11-30 12:44:01 -05:00
parent 32c89bc12c
commit e32d863354
9 changed files with 172 additions and 22 deletions

View file

@ -20,7 +20,7 @@ pub const Album = jetquery.Model(
id: i32,
title: []const u8,
song_num: i32,
length: f64,
length: f32,
play_count: i32,
holiday: bool,
compilation: bool,
@ -95,6 +95,9 @@ pub const Scrobble = jetquery.Model(
struct {
id: i32,
date: jetquery.DateTime,
song_id: i32,
album_id: ?i32,
artist_id: i32,
created_at: jetquery.DateTime,
updated_at: jetquery.DateTime,
},
@ -113,7 +116,7 @@ pub const Song = jetquery.Model(
struct {
id: i32,
title: []const u8,
length: f64,
length: f32,
hidden: bool,
holiday: bool,
play_count: i32,
@ -163,7 +166,7 @@ pub const Rating = jetquery.Model(
struct {
id: i32,
reference_id: i32,
score: f64,
score: f32,
date: jetquery.DateTime,
created_at: jetquery.DateTime,
updated_at: jetquery.DateTime,
@ -171,12 +174,17 @@ pub const Rating = jetquery.Model(
.{},
);
pub const RawScrobble = jetquery.Model(@This(), "raw_scrobbles", struct {
id: i32,
track: []const u8,
artist: []const u8,
album: []const u8,
date: i32,
created_at: jetquery.DateTime,
updated_at: jetquery.DateTime,
}, .{});
pub const RawScrobble = jetquery.Model(
@This(),
"raw_scrobbles",
struct {
id: i32,
track: []const u8,
artist: []const u8,
album: []const u8,
date: i32,
created_at: jetquery.DateTime,
updated_at: jetquery.DateTime,
},
.{},
);

View file

@ -8,6 +8,9 @@ pub fn up(repo: anytype) !void {
&.{
t.primaryKey("id", .{}),
t.column("date", .datetime, .{}),
t.column("song_id", .integer, .{}),
t.column("album_id", .integer, .{ .optional = true }),
t.column("artist_id", .integer, .{}),
t.timestamps(.{}),
},
.{},

View file

@ -1,18 +1,77 @@
const std = @import("std");
const jetzig = @import("jetzig");
const jetquery = @import("jetzig").jetquery;
// The `run` function for a job is invoked every time the job is processed by a queue worker
// (or by the Jetzig server if the job is processed in-line).
//
// Arguments:
// * allocator: Arena allocator for use during the job execution process.
// * params: Params assigned to a job (from a request, any values added to `data`).
// * params: Params assigned to a job (from a request, values added to response data).
// * env: Provides the following fields:
// - logger: Logger attached to the same stream as the Jetzig server.
// - environment: Enum of `{ production, development }`.
pub fn run(allocator: std.mem.Allocator, params: *jetzig.data.Value, env: jetzig.jobs.JobEnv) !void {
_ = allocator;
_ = params;
// Job execution code goes here. Add any code that you would like to run in the background.
try env.logger.INFO("Running a job.", .{});
// Get all scrobbles from the RawScrobbles table
const query = jetzig.database.Query(.RawScrobble).select(.{});
const scrobbles = try env.repo.all(query);
defer env.repo.free(scrobbles);
for (scrobbles) |scrobble| {
// Make hashes
const album_hash = std.hash.Fnv1a_64.hash(scrobble.album);
const artist_hash = std.hash.Fnv1a_64.hash(scrobble.artist);
const song_hash = std.hash.Fnv1a_64.hash(scrobble.track);
var album_id: u64 = 0;
const song_id = (song_hash ^ artist_hash ^ album_hash) % 99999989;
if (artist_hash == album_hash) {
album_id = album_hash % 99999989;
} else {
album_id = (artist_hash ^ album_hash) % 99999989;
}
const artist_id = artist_hash % 99999989;
// ID start - I think we can use SERIAL for this
// We don't compare intermediate IDs to anything,
// so keeping it a SERIAL is probably fine
const artistalbum_offset = try jetzig.database.Query(.ArtistAlbum).select(.{}).count().execute(env.repo) orelse unreachable;
const albumsong_offset = try jetzig.database.Query(.AlbumSong).select(.{}).count().execute(env.repo) orelse unreachable;
const artistsong_offset = try jetzig.database.Query(.ArtistSong).select(.{}).count().execute(env.repo) orelse unreachable;
// Inserts
const artistalbum_insert = jetzig.database.Query(.ArtistAlbum).insert(.{ .id = 1 + artistalbum_offset, .artist_id = artist_id, .album_id = album_id });
const albumsong_insert = jetzig.database.Query(.AlbumSong).insert(.{ .id = 1 + albumsong_offset, .song_id = song_id, .album_id = album_id });
const artistsong_insert = jetzig.database.Query(.ArtistSong).insert(.{ .id = 1 + artistsong_offset, .artist_id = artist_id, .song_id = song_id });
const album_insert = jetzig.database.Query(.Album).insert(.{ .id = album_id, .title = scrobble.album, .song_num = 0, .length = 0.0, .play_count = 0, .holiday = false, .compilation = false, .deluxe = false, .live = false });
const artist_insert = jetzig.database.Query(.Artist).insert(.{ .id = artist_id, .name = scrobble.artist, .album_num = 0, .song_num = 0, .play_count = 0 });
const song_insert = jetzig.database.Query(.Song).insert(.{ .id = song_id, .title = scrobble.track, .length = 0.0, .hidden = false, .holiday = false, .play_count = 0 });
// Checks
const artistalbum_check = try jetzig.database.Query(.ArtistAlbum).where(.{ .{ .artist_id = artist_id }, .AND, .{ .album_id = album_id } }).count().execute(env.repo);
const albumsong_check = try jetzig.database.Query(.AlbumSong).where(.{ .{ .album_id = album_id }, .AND, .{ .song_id = song_id } }).count().execute(env.repo);
const artistsong_check = try jetzig.database.Query(.ArtistSong).where(.{ .{ .artist_id = artist_id }, .AND, .{ .song_id = song_id } }).count().execute(env.repo);
const album_check = try jetzig.database.Query(.Album).where(.{.{ .id = album_id }}).count().execute(env.repo);
const artist_check = try jetzig.database.Query(.Artist).where(.{.{ .id = artist_id }}).count().execute(env.repo);
const song_check = try jetzig.database.Query(.Song).where(.{.{ .id = song_id }}).count().execute(env.repo);
// Insert into Intermediate Tables
if (artistalbum_check == 0) try env.repo.execute(artistalbum_insert);
if (albumsong_check == 0) try env.repo.execute(albumsong_insert);
if (artistsong_check == 0) try env.repo.execute(artistsong_insert);
if (album_check == 0) try env.repo.execute(album_insert);
if (artist_check == 0) try env.repo.execute(artist_insert);
if (song_check == 0) try env.repo.execute(song_insert);
const scrobble_offset = try jetzig.database.Query(.Scrobble).select(.{}).count().execute(env.repo) orelse unreachable;
try jetzig.database.Query(.Scrobble).insert(.{ .id = scrobble_offset + 1, .song_id = song_id, .album_id = album_id, .artist_id = artist_id }).execute(env.repo);
}
// Clear RawScrobbles when done processing
try jetzig.database.Query(.RawScrobble).deleteAll().execute(env.repo);
}

View file

@ -0,0 +1,71 @@
/// Demo middleware. Assign middleware by declaring `pub const middleware` in the
/// `jetzig_options` defined in your application's `src/main.zig`.
///
/// Middleware is called before and after the request, providing full access to the active
/// request, allowing you to execute any custom code for logging, tracking, inserting response
/// headers, etc.
///
/// This middleware is configured in the demo app's `src/main.zig`:
///
/// ```
/// pub const jetzig_options = struct {
/// pub const middleware: []const type = &.{@import("app/middleware/DemoMiddleware.zig")};
/// };
/// ```
const std = @import("std");
const jetzig = @import("jetzig");
const jetquery = @import("jetzig").jetquery;
/// Define any custom data fields you want to store here. Assigning to these fields in the `init`
/// function allows you to access them in various middleware callbacks defined below, where they
/// can also be modified.
my_custom_value: []const u8,
const Self = @This();
/// Initialize middleware.
pub fn init(request: *jetzig.http.Request) !*Self {
var middleware = try request.allocator.create(Self);
middleware.my_custom_value = "initial value";
return middleware;
}
/// Invoked immediately after the request is received but before it has started processing.
/// Any calls to `request.render` or `request.redirect` will prevent further processing of the
/// request, including any other middleware in the chain.
//pub fn afterRequest(self: *Self, request: *jetzig.http.Request) !void {
// const Scrobble = struct {
// track: []u8,
// artist: []u8,
// album: []u8,
// date: i64
// };
//
// const lastfm = struct {
// username: u8,
// scrobbles: []Scrobble,
// };
//
// var root = try request.data(.object);
//
// if (try request.file("upload")) |file| {
// const parsed = try std.json.parseFromSlice(lastfm, request.allocator, file.content, .{});
// const history = parsed.value;
// var scrobbles = try root.put("scrobbles", .array);
//
// for (history.scrobbles) |scrobble| {
// try scrobbles.append(scrobble);
// }
//
// //var scrobble = try root.put("scrobbles", .array);
// }
//}
/// Invoked immediately before the response renders to the client.
/// The response can be modified here if needed.
pub fn beforeResponse(self: *Self, request: *jetzig.http.Request, response: *jetzig.http.Response) !void {
try request.server.logger.DEBUG(
"[DemoMiddleware:beforeResponse] my_custom_value: {s}, response status: {s}",
.{ self.my_custom_value, @tagName(response.status_code) },
);
}

View file

@ -27,25 +27,31 @@ pub fn post(request: *jetzig.Request) !jetzig.View {
};
var root = try request.data(.object);
var job = try request.job("process_scrobbles");
var counter: u16 = 0;
if (try request.file("upload")) |file| {
const parsed = try std.json.parseFromSlice(lastfm, request.allocator, file.content, .{});
const history = parsed.value;
//std.debug.print("{s}", .{history.scrobbles[19].artist});
var scrobbles = try root.put("scrobbles", .array);
for (history.scrobbles) |scrobble| {
try scrobbles.append(scrobble);
//const song_hash: u64 = std.hash.Fnv1a_64.hash(scrobble.track) % 99999989;
//job.params.put(scrobble.song, song_hash);
//std.debug.print("{d}\n", .{song_hash});
const database_update = jetzig.database.Query(.RawScrobble)
.insert(.{ .track = scrobble.track, .album = scrobble.album, .artist = scrobble.artist, .date = @divFloor(scrobble.date, 1000) });
.insert(.{ .id = counter, .track = scrobble.track, .album = scrobble.album, .artist = scrobble.artist, .date = @divFloor(scrobble.date, 1000) });
try request.repo.execute(database_update);
counter += 1;
}
}
try job.schedule();
var upload_table = try root.put("upload_table", .array);
try upload_table.append("Track");
try upload_table.append("Artist");

View file

@ -16,11 +16,10 @@ pub const jetzig_options = struct {
jetzig.middleware.HtmxMiddleware,
// Demo middleware included with new projects. Remove once you are familiar with Jetzig's
// middleware system.
@import("app/middleware/DemoMiddleware.zig"),
};
// Maximum bytes to allow in request body.
// pub const max_bytes_request_body: usize = std.math.pow(usize, 2, 16);
pub const max_bytes_request_body: usize = std.math.pow(usize, 2, 32);
// Maximum filesize for `public/` content.
// pub const max_bytes_public_content: usize = std.math.pow(usize, 2, 20);