build: generate reference page for config for website

src/build/webgen/main_config.zig

@@ -0,0 +1,109 @@
+const std = @import("std");
+const Config = @import("../../config/Config.zig");
+const help_strings = @import("help_strings");
+
+pub fn main() !void {
+    const output = std.io.getStdOut().writer();
+    try genConfig(output);
+}
+
+pub fn genConfig(writer: anytype) !void {
+    // Write the header
+    try writer.writeAll(
+        \\---
+        \\title: Reference
+        \\description: Reference of all Ghostty configuration options.
+        \\---
+        \\
+        \\This is a reference of all Ghostty configuration options. These
+        \\options are ordered roughly by how common they are to be used
+        \\and grouped with related options. I recommend utilizing your
+        \\browser's search functionality to find the option you're looking
+        \\for.
+        \\
+        \\In the future, we'll have a more user-friendly way to view and
+        \\organize these options.
+        \\
+        \\
+    );
+
+    @setEvalBranchQuota(3000);
+    const fields = @typeInfo(Config).Struct.fields;
+    inline for (fields, 0..) |field, i| {
+        if (field.name[0] == '_') continue;
+        if (!@hasDecl(help_strings.Config, field.name)) continue;
+
+        // Write the field name.
+        try writer.writeAll("## `");
+        try writer.writeAll(field.name);
+        try writer.writeAll("`\n");
+
+        // For all subsequent fields with no docs, they are grouped
+        // with the previous field.
+        if (i + 1 < fields.len) {
+            inline for (fields[i + 1 ..]) |next_field| {
+                if (next_field.name[0] == '_') break;
+                if (@hasDecl(help_strings.Config, next_field.name)) break;
+
+                try writer.writeAll("## `");
+                try writer.writeAll(next_field.name);
+                try writer.writeAll("`\n");
+            }
+        }
+
+        // Newline after our headers
+        try writer.writeAll("\n");
+
+        var iter = std.mem.splitScalar(
+            u8,
+            @field(help_strings.Config, field.name),
+            '\n',
+        );
+
+        // We do some really rough markdown "parsing" here so that
+        // we can fix up some styles for what our website expects.
+        var block: ?enum {
+            /// Plaintext, do nothing.
+            text,
+
+            /// Code block, wrap in triple backticks. We use indented
+            /// code blocks in our comments but the website parser only
+            /// supports triple backticks.
+            code,
+        } = null;
+
+        while (iter.next()) |s| {
+            // Empty line resets our block
+            if (std.mem.eql(u8, s, "")) {
+                if (block) |v| switch (v) {
+                    .text => {},
+                    .code => try writer.writeAll("```\n"),
+                };
+                block = null;
+
+                try writer.writeAll("\n");
+                continue;
+            }
+
+            // If we don't have a block figure out our type.
+            if (block == null) {
+                if (std.mem.startsWith(u8, s, "    ")) {
+                    block = .code;
+                    try writer.writeAll("```\n");
+                } else {
+                    block = .text;
+                }
+            }
+
+            try writer.writeAll(switch (block.?) {
+                .text => s,
+                .code => if (std.mem.startsWith(u8, s, "    "))
+                    s[4..]
+                else
+                    s,
+            });
+            try writer.writeAll("\n");
+        }
+        try writer.writeAll("\n");
+    }
+}

src/build_config.zig

@@ -172,6 +172,7 @@ pub const ExeEntrypoint = enum {
     helpgen,
     mdgen_ghostty_1,
     mdgen_ghostty_5,
+    webgen_config,
     bench_parser,
     bench_stream,
     bench_codepoint_width,

src/config/Config.zig

@@ -286,16 +286,16 @@ const c = @cImport({
 ///
 /// Valid values are:
 ///
-///   * `legacy` - Use a legacy method to determine grapheme width, such as
-///     wcswidth This maximizes compatibility with legacy programs but may result
-///     in incorrect grapheme width for certain graphemes such as skin-tone
-///     emoji, non-English characters, etc.
+/// * `legacy` - Use a legacy method to determine grapheme width, such as
+///   wcswidth This maximizes compatibility with legacy programs but may result
+///   in incorrect grapheme width for certain graphemes such as skin-tone
+///   emoji, non-English characters, etc.
 ///
-///     This is called "legacy" and not something more specific because the
-///     behavior is undefined and we want to retain the ability to modify it.
-///     For example, we may or may not use libc `wcswidth` now or in the future.
+///   This is called "legacy" and not something more specific because the
+///   behavior is undefined and we want to retain the ability to modify it.
+///   For example, we may or may not use libc `wcswidth` now or in the future.
 ///
-///   * `unicode` - Use the Unicode standard to determine grapheme width.
+/// * `unicode` - Use the Unicode standard to determine grapheme width.
 ///
 /// If a running program explicitly enables terminal mode 2027, then `unicode`
 /// width will be forced regardless of this configuration. When mode 2027 is

src/main.zig

@@ -7,6 +7,7 @@ const entrypoint = switch (build_config.exe_entrypoint) {
     .helpgen => @import("helpgen.zig"),
     .mdgen_ghostty_1 => @import("build/mdgen/main_ghostty_1.zig"),
     .mdgen_ghostty_5 => @import("build/mdgen/main_ghostty_5.zig"),
+    .webgen_config => @import("build/webgen/main_config.zig"),
     .bench_parser => @import("bench/parser.zig"),
     .bench_stream => @import("bench/stream.zig"),
     .bench_codepoint_width => @import("bench/codepoint-width.zig"),