synthetic: make bytes generation more flexible (#9204)

src/synthetic/Bytes.zig

@@ -1,4 +1,4 @@
-/// Generates bytes.
+//! Generates bytes.
 const Bytes = @This();
 
 const std = @import("std");
@@ -7,9 +7,7 @@ const Generator = @import("Generator.zig");
 /// Random number generator.
 rand: std.Random,
 
-/// The minimum and maximum length of the generated bytes. The maximum
-/// length will be capped to the length of the buffer passed in if the
-/// buffer length is smaller.
+/// The minimum and maximum length of the generated bytes.
 min_len: usize = 1,
 max_len: usize = std.math.maxInt(usize),
 
@@ -18,23 +16,79 @@ max_len: usize = std.math.maxInt(usize),
 /// side effect of the generator, not an intended use case.
 alphabet: ?[]const u8 = null,
 
-/// Predefined alphabets.
-pub const Alphabet = struct {
-    pub const ascii = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!@#$%^&*()_+-=[]{}|;':\\\",./<>?`~";
-};
+/// Generate an alphabet given a function that returns true/false for a
+/// given byte.
+pub fn generateAlphabet(comptime func: fn (u8) bool) []const u8 {
+    @setEvalBranchQuota(3000);
+    var count = 0;
+    for (0..256) |c| {
+        if (func(c)) count += 1;
+    }
+    var alphabet: [count]u8 = undefined;
+    var i = 0;
+    for (0..256) |c| {
+        if (func(c)) {
+            alphabet[i] = c;
+            i += 1;
+        }
+    }
+    const result = alphabet;
+    return &result;
+}
 
 pub fn generator(self: *Bytes) Generator {
     return .init(self, next);
 }
 
-pub fn next(self: *Bytes, writer: *std.Io.Writer, max_len: usize) Generator.Error!void {
-    std.debug.assert(max_len >= 1);
-    const len = @min(
-        self.rand.intRangeAtMostBiased(usize, self.min_len, self.max_len),
-        max_len,
-    );
+/// Return a copy of the Bytes, but with a new alphabet.
+pub fn newAlphabet(self: *const Bytes, new_alphabet: ?[]const u8) Bytes {
+    return .{
+        .rand = self.rand,
+        .alphabet = new_alphabet,
+        .min_len = self.min_len,
+        .max_len = self.max_len,
+    };
+}
+
+/// Return a copy of the Bytes, but with a new min_len. The new min
+/// len cannot be more than the previous max_len.
+pub fn atLeast(self: *const Bytes, new_min_len: usize) Bytes {
+    return .{
+        .rand = self.rand,
+        .alphabet = self.alphabet,
+        .min_len = @min(self.max_len, new_min_len),
+        .max_len = self.max_len,
+    };
+}
+
+/// Return a copy of the Bytes, but with a new max_len. The new max_len cannot
+/// be more the previous max_len.
+pub fn atMost(self: *const Bytes, new_max_len: usize) Bytes {
+    return .{
+        .rand = self.rand,
+        .alphabet = self.alphabet,
+        .min_len = @min(self.min_len, @min(self.max_len, new_max_len)),
+        .max_len = @min(self.max_len, new_max_len),
+    };
+}
+
+pub fn next(self: *const Bytes, writer: *std.Io.Writer, max_len: usize) std.Io.Writer.Error!void {
+    _ = try self.atMost(max_len).write(writer);
+}
+
+pub fn format(self: *const Bytes, writer: *std.Io.Writer) std.Io.Writer.Error!void {
+    _ = try self.write(writer);
+}
+
+/// Write some random data and return the number of bytes written.
+pub fn write(self: *const Bytes, writer: *std.Io.Writer) std.Io.Writer.Error!usize {
+    std.debug.assert(self.min_len >= 1);
+    std.debug.assert(self.max_len >= self.min_len);
+
+    const len = self.rand.intRangeAtMostBiased(usize, self.min_len, self.max_len);
 
     var buf: [8]u8 = undefined;
+
     var remaining = len;
     while (remaining > 0) {
         const data = buf[0..@min(remaining, buf.len)];
@@ -45,6 +99,8 @@ pub fn next(self: *Bytes, writer: *std.Io.Writer, max_len: usize) Generator.Erro
         try writer.writeAll(data);
         remaining -= data.len;
     }
+
+    return len;
 }
 
 test "bytes" {
@@ -52,9 +108,11 @@ test "bytes" {
     var prng = std.Random.DefaultPrng.init(0);
     var buf: [256]u8 = undefined;
     var writer: std.Io.Writer = .fixed(&buf);
-    var v: Bytes = .{ .rand = prng.random() };
-    v.min_len = buf.len;
-    v.max_len = buf.len;
+    var v: Bytes = .{
+        .rand = prng.random(),
+        .min_len = buf.len,
+        .max_len = buf.len,
+    };
     const gen = v.generator();
     try gen.next(&writer, buf.len);
     try testing.expectEqual(buf.len, writer.buffered().len);

src/synthetic/Osc.zig

@@ -35,19 +35,26 @@ p_valid: f64 = 1.0,
 p_valid_kind: std.enums.EnumArray(ValidKind, f64) = .initFill(1.0),
 p_invalid_kind: std.enums.EnumArray(InvalidKind, f64) = .initFill(1.0),
 
-/// The alphabet for random bytes (omitting 0x1B and 0x07).
-const bytes_alphabet: []const u8 = alphabet: {
-    var alphabet: [256]u8 = undefined;
-    for (0..alphabet.len) |i| {
-        if (i == 0x1B or i == 0x07) {
-            alphabet[i] = @intCast(i + 1);
-        } else {
-            alphabet[i] = @intCast(i);
-        }
-    }
-    const result = alphabet;
-    break :alphabet &result;
-};
+fn checkKvAlphabet(c: u8) bool {
+    return switch (c) {
+        std.ascii.control_code.esc, std.ascii.control_code.bel, ';', '=' => false,
+        else => std.ascii.isPrint(c),
+    };
+}
+
+/// The alphabet for random bytes in OSC key/value pairs (omitting 0x1B,
+/// 0x07, ';', '=').
+pub const kv_alphabet = Bytes.generateAlphabet(checkKvAlphabet);
+
+fn checkOscAlphabet(c: u8) bool {
+    return switch (c) {
+        std.ascii.control_code.esc, std.ascii.control_code.bel => false,
+        else => true,
+    };
+}
+
+/// The alphabet for random bytes in OSCs (omitting 0x1B and 0x07).
+pub const osc_alphabet = Bytes.generateAlphabet(checkOscAlphabet);
 
 pub fn generator(self: *Osc) Generator {
     return .init(self, next);
@@ -99,35 +106,43 @@ fn nextUnwrapped(self: *Osc, writer: *std.Io.Writer, max_len: usize) Generator.E
 
 fn nextUnwrappedValidExact(self: *const Osc, writer: *std.Io.Writer, k: ValidKind, max_len: usize) Generator.Error!void {
     switch (k) {
-        .change_window_title => {
-            try writer.writeAll("0;"); // Set window title
-            var bytes_gen = self.bytes();
-            try bytes_gen.next(writer, max_len - 2);
+        .change_window_title => change_window_title: {
+            if (max_len < 3) break :change_window_title;
+            try writer.print("0;{f}", .{self.bytes().atMost(max_len - 3)}); // Set window title
         },
 
-        .prompt_start => {
+        .prompt_start => prompt_start: {
+            if (max_len < 4) break :prompt_start;
+            var remaining = max_len;
+
             try writer.writeAll("133;A"); // Start prompt
+            remaining -= 4;
 
             // aid
-            if (self.rand.boolean()) {
-                var bytes_gen = self.bytes();
-                bytes_gen.max_len = 16;
+            if (self.rand.boolean()) aid: {
+                if (remaining < 6) break :aid;
                 try writer.writeAll(";aid=");
-                try bytes_gen.next(writer, max_len);
+                remaining -= 5;
+                remaining -= try self.bytes().newAlphabet(kv_alphabet).atMost(@min(16, remaining)).write(writer);
             }
 
             // redraw
-            if (self.rand.boolean()) {
+            if (self.rand.boolean()) redraw: {
+                if (remaining < 9) break :redraw;
                 try writer.writeAll(";redraw=");
                 if (self.rand.boolean()) {
                     try writer.writeAll("1");
                 } else {
                     try writer.writeAll("0");
                 }
+                remaining -= 9;
             }
         },
 
-        .prompt_end => try writer.writeAll("133;B"), // End prompt
+        .prompt_end => prompt_end: {
+            if (max_len < 4) break :prompt_end;
+            try writer.writeAll("133;B"); // End prompt
+        },
     }
 }
 
@@ -139,14 +154,11 @@ fn nextUnwrappedInvalidExact(
 ) Generator.Error!void {
     switch (k) {
         .random => {
-            var bytes_gen = self.bytes();
-            try bytes_gen.next(writer, max_len);
+            try self.bytes().atMost(max_len).format(writer);
         },
 
         .good_prefix => {
-            try writer.writeAll("133;");
-            var bytes_gen = self.bytes();
-            try bytes_gen.next(writer, max_len - 4);
+            try writer.print("133;{f}", .{self.bytes().atMost(max_len - 4)});
         },
     }
 }
@@ -154,7 +166,7 @@ fn nextUnwrappedInvalidExact(
 fn bytes(self: *const Osc) Bytes {
     return .{
         .rand = self.rand,
-        .alphabet = bytes_alphabet,
+        .alphabet = osc_alphabet,
     };
 }
 

src/synthetic/cli/Ascii.zig

@@ -3,12 +3,21 @@ const Ascii = @This();
 const std = @import("std");
 const assert = std.debug.assert;
 const Allocator = std.mem.Allocator;
-const synthetic = @import("../main.zig");
+const Bytes = @import("../Bytes.zig");
 
 const log = std.log.scoped(.@"terminal-stream-bench");
 
 pub const Options = struct {};
 
+fn checkAsciiAlphabet(c: u8) bool {
+    return switch (c) {
+        ' ' => false,
+        else => std.ascii.isPrint(c),
+    };
+}
+
+pub const ascii = Bytes.generateAlphabet(checkAsciiAlphabet);
+
 /// Create a new terminal stream handler for the given arguments.
 pub fn create(
     alloc: Allocator,
@@ -23,12 +32,10 @@ pub fn destroy(self: *Ascii, alloc: Allocator) void {
     alloc.destroy(self);
 }
 
-pub fn run(self: *Ascii, writer: *std.Io.Writer, rand: std.Random) !void {
-    _ = self;
-
-    var gen: synthetic.Bytes = .{
+pub fn run(_: *Ascii, writer: *std.Io.Writer, rand: std.Random) !void {
+    var gen: Bytes = .{
         .rand = rand,
-        .alphabet = synthetic.Bytes.Alphabet.ascii,
+        .alphabet = ascii,
     };
 
     while (true) {