From 70bc70375f8e82843830d93631ab005302057a15 Mon Sep 17 00:00:00 2001
From: owl
Date: Fri, 30 Aug 2024 12:23:33 +0200
Subject: count codepoints instead of bytes, to determine width

A complete solution would be to count grapheme clusters, but that would
require adding a dependency on something like zg.

Counting codepoints will ensure that typical non-ASCII text is
supported, but you can still throw it off with more complex Unicode
constructions, which might not be so useful in help text.

Fixes #75
---
 clap/codepoint_counting_writer.zig | 102 +++++++++++++++++++++++++++++++++++++
 1 file changed, 102 insertions(+)
 create mode 100644 clap/codepoint_counting_writer.zig

(limited to 'clap/codepoint_counting_writer.zig')

diff --git a/clap/codepoint_counting_writer.zig b/clap/codepoint_counting_writer.zig
new file mode 100644
index 0000000..e6b9d1c
--- /dev/null
+++ b/clap/codepoint_counting_writer.zig
@@ -0,0 +1,102 @@
+const std = @import("std");
+const builtin = @import("builtin");
+const native_endian = builtin.cpu.arch.endian();
+
+/// A Writer that counts how many codepoints has been written to it.
+/// Expects valid UTF-8 input, and does not validate the input.
+pub fn CodepointCountingWriter(comptime WriterType: type) type {
+    return struct {
+        codepoints_written: u64,
+        child_stream: WriterType,
+
+        pub const Error = WriterType.Error || error{Utf8InvalidStartByte};
+        pub const Writer = std.io.Writer(*Self, Error, write);
+
+        const Self = @This();
+
+        pub fn write(self: *Self, bytes: []const u8) Error!usize {
+            const bytes_and_codepoints = try utf8CountCodepointsAllowTruncate(bytes);
+            // Might not be the full input, so the leftover bytes are written on the next call.
+            const bytes_to_write = bytes[0..bytes_and_codepoints.bytes];
+            const amt = try self.child_stream.write(bytes_to_write);
+            const bytes_written = bytes_to_write[0..amt];
+            self.codepoints_written += (try utf8CountCodepointsAllowTruncate(bytes_written)).codepoints;
+            return amt;
+        }
+
+        pub fn writer(self: *Self) Writer {
+            return .{ .context = self };
+        }
+    };
+}
+
+// Like `std.unicode.utf8CountCodepoints`, but on truncated input, it returns
+// the number of codepoints up to that point.
+// Does not validate UTF-8 beyond checking the start byte.
+fn utf8CountCodepointsAllowTruncate(s: []const u8) !struct { bytes: usize, codepoints: usize } {
+    var len: usize = 0;
+
+    const N = @sizeOf(usize);
+    const MASK = 0x80 * (std.math.maxInt(usize) / 0xff);
+
+    var i: usize = 0;
+    while (i < s.len) {
+        // Fast path for ASCII sequences
+        while (i + N <= s.len) : (i += N) {
+            const v = std.mem.readInt(usize, s[i..][0..N], native_endian);
+            if (v & MASK != 0) break;
+            len += N;
+        }
+
+        if (i < s.len) {
+            const n = try std.unicode.utf8ByteSequenceLength(s[i]);
+            // Truncated input; return the current counts.
+            if (i + n > s.len) return .{ .bytes = i, .codepoints = len };
+
+            i += n;
+            len += 1;
+        }
+    }
+
+    return .{ .bytes = i, .codepoints = len };
+}
+
+pub fn codepointCountingWriter(child_stream: anytype) CodepointCountingWriter(@TypeOf(child_stream)) {
+    return .{ .codepoints_written = 0, .child_stream = child_stream };
+}
+
+const testing = std.testing;
+
+test CodepointCountingWriter {
+    var counting_stream = codepointCountingWriter(std.io.null_writer);
+    const stream = counting_stream.writer();
+
+    const utf8_text = "blåhaj" ** 100;
+    stream.writeAll(utf8_text) catch unreachable;
+    const expected_count = try std.unicode.utf8CountCodepoints(utf8_text);
+    try testing.expectEqual(expected_count, counting_stream.codepoints_written);
+}
+
+test "handles partial UTF-8 writes" {
+    var buf: [100]u8 = undefined;
+    var fbs = std.io.fixedBufferStream(&buf);
+    var counting_stream = codepointCountingWriter(fbs.writer());
+    const stream = counting_stream.writer();
+
+    const utf8_text = "ååå";
+    // `å` is represented as `\xC5\xA5`, write 1.5 `å`s.
+    var wc = try stream.write(utf8_text[0..3]);
+    // One should have been written fully.
+    try testing.expectEqual("å".len, wc);
+    try testing.expectEqual(1, counting_stream.codepoints_written);
+
+    // Write the rest, continuing from the reported number of bytes written.
+    wc = try stream.write(utf8_text[wc..]);
+    try testing.expectEqual(4, wc);
+    try testing.expectEqual(3, counting_stream.codepoints_written);
+
+    const expected_count = try std.unicode.utf8CountCodepoints(utf8_text);
+    try testing.expectEqual(expected_count, counting_stream.codepoints_written);
+
+    try testing.expectEqualSlices(u8, utf8_text, fbs.getWritten());
+}
-- 
cgit v1.2.3