srctree

Gregory Mullen parent 3abb4a2b b63a27bc
add some basic bot detection code

inline split

build.zig added: 84, removed: 10, total 74

@@ -4,11 +4,20 @@ pub fn build(b: *std.Build) !void {

const target = b.standardTargetOptions(.{});

const optimize = b.standardOptimizeOption(.{});

//if (b.args) |args| for (args) |arg| std.debug.print("arg {s}\n", .{arg});

//std.debug.print("default: {s}\n", .{b.default_step.name});

// root build options

const template_path = b.option(std.Build.LazyPath, "template-path", "path for the templates generated at comptime") orelse b.path("examples/templates/");

const bot_detection = b.option(bool, "bot-detection", "path for the templates generated at comptime") orelse

false;

const ver = version(b);

const options = b.addOptions();

options.addOption([]const u8, "version", ver);

options.addOption(bool, "botdetection", true);

options.addOption(bool, "botdetection", bot_detection);

const verse_lib = b.addModule("verse", .{

.root_source_file = b.path("src/verse.zig"),

@@ -25,8 +34,6 @@ pub fn build(b: *std.Build) !void {

});

lib_unit_tests.root_module.addOptions("verse_buildopts", options);

const template_path = b.option(std.Build.LazyPath, "template-path", "path for the templates generated at comptime") orelse b.path("examples/templates/");

var compiler = Compiler.init(b);

compiler.addDir(template_path);

src/bot-detection.zig added: 84, removed: 10, total 74

@@ -3,10 +3,51 @@

//! It does something, what that something is? who know, but it's big!

bot: bool,

malicious: bool,

const BotDetection = @This();

pub fn init() BotDetection {}

pub fn init(r: *const Request) BotDetection {

if (r.user_agent == null) return .{ .bot = true, .malicious = true };

const ua = r.user_agent.?;

switch (ua.resolved) {

.bot => {

return .{

.bot = true,

.malicious = false,

};

.browser => |browser| {

switch (browser.name) {

.chrome => {

return .{

.bot = true,

.malicious = false,

};

else => {

return .{

.bot = true,

.malicious = false,

};

}

.script => {

return .{

.bot = true,

.malicious = false,

};

.unknown => {

return .{

.bot = true,

.malicious = std.mem.startsWith(u8, ua.string, "Mozilla/"),

};

}

comptime unreachable;

}

pub const Browsers = struct {

const Date = i64;

@@ -86,3 +127,4 @@ test Browsers {

const std = @import("std");

const UA = @import("user-agent.zig");

const Request = @import("request.zig");

src/request.zig added: 84, removed: 10, total 74

@@ -13,6 +13,7 @@ referer: ?Referer,

accept: ?Accept,

accept_encoding: Encoding = .default,

authorization: ?Authorization,

protocol: Protocol,

headers: Headers,

/// Default API, still unstable, but unlike to drastically change

@@ -91,6 +92,11 @@ pub const Methods = enum(u9) {

}

};

pub const Protocol = struct {

// TODO split name and version

str: []const u8,

};

fn initCommon(

a: Allocator,

remote_addr: RemoteAddr,

@@ -104,6 +110,7 @@ fn initCommon(

authorization: ?Authorization,

headers: Headers,

cookies: ?[]const u8,

proto: []const u8,

data: Data,

raw: RawReq,

) !Request {

@@ -127,6 +134,7 @@ fn initCommon(

.remote_addr = remote_addr,

.uri = uri,

.user_agent = if (ua) |u| .init(u) else null,

.protocol = .{ .str = proto },

};

}

@@ -142,6 +150,7 @@ pub fn initZWSGI(a: Allocator, zwsgi: *zWSGIRequest, data: Data) !Request {

var encoding: Encoding = Encoding.default;

var authorization: ?Authorization = null;

var cookie_header: ?[]const u8 = null;

var proto: []const u8 = "ERROR";

for (zwsgi.vars) |v| {

try headers.addCustom(v.key, v.val);

@@ -168,6 +177,8 @@ pub fn initZWSGI(a: Allocator, zwsgi: *zWSGIRequest, data: Data) !Request {

authorization = v.val;

} else if (eqlIgnoreCase("HTTP_COOKIE", v.key)) {

cookie_header = v.val;

} else if (eqlIgnoreCase("SERVER_PROTOCOL", v.key)) {

proto = v.val;

}

@@ -184,6 +195,7 @@ pub fn initZWSGI(a: Allocator, zwsgi: *zWSGIRequest, data: Data) !Request {

authorization,

headers,

cookie_header,

proto,

data,

.{ .zwsgi = zwsgi },

);

@@ -200,6 +212,7 @@ pub fn initHttp(a: Allocator, http: *std.http.Server.Request, data: Data) !Reque

var encoding: Encoding = Encoding.default;

var authorization: ?Authorization = null;

var cookie_header: ?[]const u8 = null;

const proto: []const u8 = @tagName(http.head.version);

while (itr.next()) |head| {

try headers.addCustom(head.name, head.value);

@@ -241,6 +254,7 @@ pub fn initHttp(a: Allocator, http: *std.http.Server.Request, data: Data) !Reque

authorization,

headers,

cookie_header,

proto,

data,

.{ .http = http },

);

src/user-agent.zig added: 84, removed: 10, total 74

@@ -8,6 +8,14 @@ resolved: Resolved,

const UserAgent = @This();

pub fn botDetectionDump(ua: UserAgent, r: *const Request) void {

if (comptime !BOTDETC_ENABLED) @compileError("Bot Detection is currently disabled");

const bd: BotDetection = .init(r);

std.debug.print("ua detection: {} \n", .{ua});

std.debug.print("bot detection: {} \n", .{bd});

}

pub const Resolved = union(enum) {

bot: Bot,

browser: Browser,

@@ -141,14 +149,14 @@ pub const Browser = struct {

version_string: []const u8 = "",

pub fn age(b: Browser) !i64 {

if (comptime !verse_buildopts.botdetection) @compileError("Bot Detection is currently disabled");

if (comptime !BOTDETC_ENABLED) @compileError("Bot Detection is currently disabled");

const versions = BotDetection.Browsers.Versions[@intFromEnum(b.name)];

if (b.version >= versions.len) return error.UnknownVersion;

return std.time.timestamp() - versions[b.version];

}

test age {

if (!verse_buildopts.botdetection) return error.SkipZigTest;

if (!BOTDETC_ENABLED) return error.SkipZigTest;

const browser = Browser{ .name = .chrome, .version = 134 };

try std.testing.expect(try browser.age() < 86400 * 3650); // breaks in 10 years, good luck future me!

try std.testing.expect(try browser.age() > 3148551);

@@ -184,12 +192,15 @@ pub fn init(ua_str: []const u8) UserAgent {

const Request = @import("request.zig");

const BotDetection = @import("bot-detection.zig");

const BOTDETC_ENABLED: bool = verse_buildopts.botdetection or builtin.is_test;

test UserAgent {

std.testing.refAllDecls(@This());

_ = &BotDetection;

}

const std = @import("std");

const builtin = @import("builtin");

const verse_buildopts = @import("verse_buildopts");

const startsWith = std.mem.startsWith;

const endsWith = std.mem.endsWith;