zig-skills/references/std-uri.md

std.Uri Reference (Zig 0.16.0)

Primary Zig 0.16 release-note source: https://ziglang.org/download/0.16.0/release-notes.html

URI parsing remains mostly independent of the I/O migration. For network or file operations derived from URIs, pass/use std.Io.

URI parsing and formatting that roughly adhere to RFC 3986, with percent-encoding/decoding and resolution support. The parser intentionally does not perform complete grammar or character-class validation.

Table of Contents

• Parsing URIs
• URI Components
• Formatting URIs
• Percent Encoding/Decoding
• URI Resolution
• Common Patterns

Parsing URIs

Basic Parsing

const std = @import("std");

pub fn main() !void {
    const uri = try std.Uri.parse("https://user:pass@example.com:8080/path?query=1#fragment");

    std.debug.print("Scheme: {s}\n", .{uri.scheme});                    // "https"
    std.debug.print("User: {s}\n", .{uri.user.?.percent_encoded});      // "user"
    std.debug.print("Password: {s}\n", .{uri.password.?.percent_encoded}); // "pass"
    std.debug.print("Host: {s}\n", .{uri.host.?.percent_encoded});      // "example.com"
    std.debug.print("Port: {d}\n", .{uri.port.?});                      // 8080
    std.debug.print("Path: {s}\n", .{uri.path.percent_encoded});        // "/path"
    std.debug.print("Query: {s}\n", .{uri.query.?.percent_encoded});    // "query=1"
    std.debug.print("Fragment: {s}\n", .{uri.fragment.?.percent_encoded}); // "fragment"
}

Parse After Scheme

For URIs where scheme is already known (e.g., HTTP redirects):

// Parse "//example.com/path" as an HTTP URI
const uri = try std.Uri.parseAfterScheme("http", "//example.com/path");
std.debug.print("Scheme: {s}, Host: {s}\n", .{
    uri.scheme,
    uri.host.?.percent_encoded,
});

Error Handling

const uri = std.Uri.parse(input) catch |err| switch (err) {
    error.UnexpectedCharacter => {
        std.debug.print("Invalid character in URI\n", .{});
        return err;
    },
    error.InvalidFormat => {
        std.debug.print("Malformed URI\n", .{});
        return err;
    },
    error.InvalidPort => {
        std.debug.print("Port not a valid u16\n", .{});
        return err;
    },
    error.InvalidHostName => {
        std.debug.print("Invalid host name\n", .{});
        return err;
    },
};

URI Components

Uri Struct

const Uri = struct {
    scheme: []const u8,
    user: ?Component = null,
    password: ?Component = null,
    host: ?Component = null,
    port: ?u16 = null,
    path: Component = Component.empty,
    query: ?Component = null,
    fragment: ?Component = null,
};

Component Union

Components can be raw (needs encoding) or already percent-encoded:

const Component = union(enum) {
    /// Needs percent encoding before use in URI
    raw: []const u8,
    /// Already percent-encoded, can be used directly
    percent_encoded: []const u8,

    pub const empty: Component = .{ .percent_encoded = "" };

    pub fn isEmpty(component: Component) bool;
};

Getting Host

var buffer: [std.Io.net.HostName.max_len]u8 = undefined;
const host = uri.getHost(&buffer) catch |err| switch (err) {
    error.UriMissingHost => return error.NoHost,
};

The result is a std.Io.net.HostName. A URI host is validated during parsing, so the fixed buffer is sized to the hostname limit and getHost exposes only error.UriMissingHost.

With allocation:

var arena = std.heap.ArenaAllocator.init(allocator);
defer arena.deinit();

const host = try uri.getHostAlloc(arena.allocator());

Creating URIs Programmatically

const uri: std.Uri = .{
    .scheme = "https",
    .host = .{ .raw = "example.com" },
    .port = 8080,
    .path = .{ .raw = "/api/users" },
    .query = .{ .raw = "page=1&limit=10" },
};

Component Methods

const component: std.Uri.Component = .{ .percent_encoded = "hello%20world" };

// Check if empty
if (component.isEmpty()) {
    // ...
}

// Get raw (decoded) value with buffer
var buf: [256]u8 = undefined;
const raw = try component.toRaw(&buf);  // "hello world"

// Get raw (decoded) value with allocation (only allocates if needed)
const raw_alloc = try component.toRawMaybeAlloc(allocator);  // "hello world"

toRawMaybeAlloc may return the component's original borrowed slice or arena-allocated decoded bytes. Treat the result as tied to both lifetimes and do not free it individually.

Formatting URIs

Full URI

const uri = try std.Uri.parse("https://example.com:8080/path?query#frag");

var buf: [1024]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);
try uri.format(&writer);
const formatted = writer.buffered();  // "https://example.com:8080/path?query#frag"

Selective Formatting

var buf: [1024]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);

// Format only specific parts
try std.fmt.format(&writer, "{f}", .{uri.fmt(.{
    .scheme = true,
    .authority = true,
    .path = true,
    .query = true,
    .fragment = false,  // omit fragment
})});

Format Flags

const Flags = struct {
    scheme: bool = false,         // Include scheme (e.g., "https:")
    authentication: bool = false, // Include user:password@ (requires authority)
    authority: bool = false,      // Include host and port
    path: bool = false,           // Include path
    query: bool = false,          // Include ?query (requires path)
    fragment: bool = false,       // Include #fragment (requires path)
    port: bool = true,            // Include :port (requires authority)

    pub const all: Flags = .{
        .scheme = true,
        .authentication = true,
        .authority = true,
        .path = true,
        .query = true,
        .fragment = true,
        .port = true,
    };
};

Component Formatting Methods

var buf: [256]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);

const component: std.Uri.Component = .{ .raw = "hello world" };

// Different encoding rules for different URI parts
try component.formatEscaped(&writer);  // General: unreserved chars only
try component.formatUser(&writer);     // User: unreserved + sub-delims
try component.formatPassword(&writer); // Password: user chars + ':'
try component.formatHost(&writer);     // Host: password chars + '[' + ']'
try component.formatPath(&writer);     // Path: user chars + '/' + ':' + '@'
try component.formatQuery(&writer);    // Query: path chars + '?'
try component.formatFragment(&writer); // Fragment: same as query

// Get raw (decoded) output
try component.formatRaw(&writer);      // Decodes percent-encoded chars

Percent Encoding/Decoding

Decode In Place

var buffer = "hello%20world%21".*;
const decoded = std.Uri.percentDecodeInPlace(&buffer);
// decoded == "hello world!"

Decode Backwards (Conditionally Safe for Aliasing)

const input = "%48%65%6C%6C%6F";
var output: [5]u8 = undefined;
const decoded = std.Uri.percentDecodeBackwards(&output, input);
// decoded == "Hello"

The output must be large enough. Aliasing is supported only when output.ptr <= input.ptr; there is no recoverable undersized-buffer error.

Encode with Component

var buf: [256]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);

// Raw component will be percent-encoded when formatted
const component: std.Uri.Component = .{ .raw = "hello world!" };
try component.formatPath(&writer);
const encoded = writer.buffered();  // "hello%20world%21"

Custom Percent Encoding

var buf: [256]u8 = undefined;
var writer: std.Io.Writer = .fixed(&buf);

// Encode with custom character validation
try std.Uri.Component.percentEncode(&writer, "custom data", struct {
    fn isValid(c: u8) bool {
        return std.ascii.isAlphanumeric(c);
    }
}.isValid);

URI Resolution

Resolve Relative URI

Resolves a relative URI against a base URI per RFC 3986 Section 5:

const base = try std.Uri.parse("http://a/b/c/d;p?q");

var aux_buf: [1024]u8 = undefined;
var aux_slice: []u8 = &aux_buf;

// Copy relative URI to start of aux_buf
const relative = "../g";
@memcpy(aux_buf[0..relative.len], relative);

const resolved = try std.Uri.resolveInPlace(base, relative.len, &aux_slice);
// resolved.path.percent_encoded == "/a/g"

Resolution Examples (RFC 3986)

Base: http://a/b/c/d;p?q	Reference	Result
	g	http://a/b/c/g
	./g	http://a/b/c/g
	g/	http://a/b/c/g/
	/g	http://a/g
	//g	http://g
	?y	http://a/b/c/d;p?y
	g?y	http://a/b/c/g?y
	#s	http://a/b/c/d;p?q#s
	g#s	http://a/b/c/g#s
	../	http://a/b/
	../g	http://a/b/g
	../../g	http://a/g

Common Patterns

Extract Query Parameters

fn getQueryParam(uri: std.Uri, key: []const u8) ?[]const u8 {
    const query = uri.query orelse return null;
    const query_str = query.percent_encoded;

    var iter = std.mem.splitScalar(u8, query_str, '&');
    while (iter.next()) |pair| {
        if (std.mem.indexOfScalar(u8, pair, '=')) |eq_pos| {
            if (std.mem.eql(u8, pair[0..eq_pos], key)) {
                return pair[eq_pos + 1 ..];
            }
        } else if (std.mem.eql(u8, pair, key)) {
            return "";  // Key exists with no value
        }
    }
    return null;
}

// Usage
const uri = try std.Uri.parse("https://example.com?name=alice&age=30");
const name = getQueryParam(uri, "name");  // "alice"

Build URL with Query Parameters

fn buildUrl(allocator: std.mem.Allocator, base: []const u8, params: []const [2][]const u8) ![]u8 {
    var result: std.Io.Writer.Allocating = .init(allocator);
    errdefer result.deinit();

    try result.writer.writeAll(base);

    for (params, 0..) |param, i| {
        try result.writer.writeByte(if (i == 0) '?' else '&');
        try (std.Uri.Component{ .raw = param[0] }).formatEscaped(&result.writer);
        try result.writer.writeByte('=');
        try (std.Uri.Component{ .raw = param[1] }).formatEscaped(&result.writer);
    }

    return result.toOwnedSlice();
}

Serialize a Parsed URI

fn serializeUri(allocator: std.mem.Allocator, uri_str: []const u8) ![]u8 {
    const uri = try std.Uri.parse(uri_str);

    var buf: [4096]u8 = undefined;
    var writer: std.Io.Writer = .fixed(&buf);

    // Parsed percent-encoded components are preserved when formatted.
    try uri.format(&writer);

    return try allocator.dupe(u8, writer.buffered());
}

This is serialization, not general normalization: existing escape spelling is preserved. Formatting with all components also emits / for an empty included path.

Validate URI

fn isAcceptedByUriParser(str: []const u8) bool {
    _ = std.Uri.parse(str) catch return false;
    return true;
}

fn hasHttpSchemeAndHost(str: []const u8) bool {
    const uri = std.Uri.parse(str) catch return false;
    return uri.host != null and
        (std.mem.eql(u8, uri.scheme, "http") or std.mem.eql(u8, uri.scheme, "https"));
}

These remain lightweight parser/scheme checks, not complete URI or HTTP-URL validation.

Join Path Segments

const OwnedUri = struct {
    value: std.Uri,
    path_storage: []u8,

    fn deinit(self: *OwnedUri, allocator: std.mem.Allocator) void {
        allocator.free(self.path_storage);
        self.* = undefined;
    }
};

fn joinPath(allocator: std.mem.Allocator, base_uri: std.Uri, segments: []const []const u8) !OwnedUri {
    var path: std.Io.Writer.Allocating = .init(allocator);
    errdefer path.deinit();

    // Preserve/encode the existing component according to path rules.
    try base_uri.path.formatPath(&path.writer);

    // Each segment is escaped as a segment, so '/' inside a segment becomes %2F.
    for (segments) |seg| {
        const current = path.written();
        if (current.len == 0 or current[current.len - 1] != '/')
            try path.writer.writeByte('/');
        try (std.Uri.Component{ .raw = seg }).formatEscaped(&path.writer);
    }

    const owned_path = try path.toOwnedSlice();
    var result = base_uri;
    result.path = .{ .percent_encoded = owned_path };
    result.query = null;
    result.fragment = null;
    return .{ .value = result, .path_storage = owned_path };
}

The returned URI owns its assembled path; call OwnedUri.deinit after use.

Extract Base URL

fn getBaseUrl(allocator: std.mem.Allocator, uri: std.Uri) ![]u8 {
    var buf: [1024]u8 = undefined;
    var writer: std.Io.Writer = .fixed(&buf);

    try std.fmt.format(&writer, "{f}", .{uri.fmt(.{
        .scheme = true,
        .authority = true,
        .port = true,
    })});

    return try allocator.dupe(u8, writer.buffered());
}

// Usage
const uri = try std.Uri.parse("https://example.com:8080/path?query#frag");
const base = try getBaseUrl(allocator, uri);  // "https://example.com:8080"

Error Types

ParseError

pub const ParseError = error{
    UnexpectedCharacter,  // Invalid character in URI component
    InvalidFormat,        // Malformed URI structure
    InvalidPort,          // Port not a valid u16
    InvalidHostName,      // Host did not pass std.Io.net.HostName validation
};

ResolveInPlaceError

pub const ResolveInPlaceError = ParseError || error{
    NoSpaceLeft,  // Auxiliary buffer too small
};

Component Errors

// getHost errors
error.UriMissingHost   // URI has no host component

// toRaw errors
error.NoSpaceLeft      // Buffer too small for decoded string