feat: zig rewrite

9 files changed, 568 insertions, 0 deletions

diff --git a/src/cgi.zig b/src/cgi.zig
new file mode 100755
index 0000000..849ef8c
--- /dev/null
+++ b/src/cgi.zig
@@ -0,0 +1,79 @@
+const std = @import("std");
+const Io = std.Io;
+const File = Io.File;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const ArrayList = std.ArrayList;
+const StringArrayHashMap = std.StringArrayHashMapUnmanaged;
+
+const HTMLua = @import("htmlua");
+
+const Config = struct {
+    component_path: []const u8,
+    template_path: []const u8,
+};
+
+var component_dir: Io.Dir = undefined;
+
+pub fn main(init: std.process.Init) !void {
+    const io = init.io;
+
+    var arena = std.heap.ArenaAllocator.init(init.gpa);
+    defer arena.deinit();
+    const alloc = arena.allocator();
+
+    var args = init.minimal.args.iterate();
+
+    const cwd = std.Io.Dir.cwd();
+
+    if (!args.skip()) unreachable;
+    const config_file_path = args.next().?;
+    const config_buf = try cwd.readFileAllocOptions(io, config_file_path, alloc, .unlimited, std.mem.Alignment.@"1", '\x00');
+    const config = try std.zon.parse.fromSliceAlloc(
+        Config,
+        alloc,
+        config_buf,
+        null,
+        .{},
+    );
+
+    component_dir = try cwd.openDir(io, config.component_path, .{});
+    const template_dir = try cwd.openDir(io, config.template_path, .{});
+
+    var state: HTMLua.Parser = .{
+        .tags = &[_]HTMLua.ParserTag{
+            HTMLua.tags.Include,
+            HTMLua.tags.Export,
+            HTMLua.tags.Use,
+            HTMLua.tags.Syntax,
+        },
+        .tag_stack = try ArrayList([]const u8).initCapacity(alloc, 256),
+        .exported_parts = try StringArrayHashMap([]const u8).init(alloc, &[0][]const u8{}, &[0][]const u8{}),
+        .component_resolver = struct {
+            pub fn resolve(arena_inner: *ArenaAllocator, io_inner: Io, path: []const u8) anyerror!*Io.Reader {
+                const read_buff = try arena_inner.allocator().alloc(u8, 16384);
+                const clean_path = try HTMLua.utils.sanitizePath(arena_inner.allocator(), path);
+                var file = try component_dir.openFile(io_inner, clean_path, .{ .mode = .read_only });
+                const reader = try arena_inner.allocator().create(File.Reader);
+                reader.* = file.reader(io_inner, read_buff);
+                return &reader.interface;
+            }
+        }.resolve,
+    };
+
+    const request_path = init.environ_map.get("PATH_INFO").?;
+    const clean_request_path = try HTMLua.utils.sanitizePath(alloc, request_path);
+    const file = try template_dir.openFile(io, clean_request_path, .{ .mode = .read_only });
+    defer file.close(io);
+
+    const in_buf = try alloc.alloc(u8, 16384);
+    var in = file.reader(io, in_buf);
+
+    const out_buf = try alloc.alloc(u8, 16384);
+    var out = File.stdout().writer(io, out_buf);
+
+    try out.interface.writeAll("Content-type: text/html\n");
+    try HTMLua.parseTemplate(&arena, io, &state, &in.interface, &out.interface);
+
+    try out.flush();
+}
diff --git a/src/main.zig b/src/main.zig
new file mode 100755
index 0000000..18b8872
--- /dev/null
+++ b/src/main.zig
@@ -0,0 +1,59 @@
+const std = @import("std");
+const Io = std.Io;
+const File = Io.File;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const ArrayList = std.ArrayList;
+const StringArrayHashMap = std.StringArrayHashMapUnmanaged;
+
+const HTMLua = @import("root.zig");
+
+pub fn main(init: std.process.Init) !void {
+    const io = init.io;
+
+    var arena = std.heap.ArenaAllocator.init(init.gpa);
+    defer arena.deinit();
+    const alloc = arena.allocator();
+
+    var state: HTMLua.Parser = .{
+        .component_resolver = &resolver,
+        .tags = &[_]HTMLua.ParserTag{
+            HTMLua.tags.Include,
+            HTMLua.tags.Export,
+            HTMLua.tags.Use,
+        },
+        .tag_stack = try ArrayList([]const u8).initCapacity(alloc, 256),
+        .exported_parts = try StringArrayHashMap([]const u8).init(alloc, &[0][]const u8{}, &[0][]const u8{}),
+    };
+
+    // I dont like this but this is temporary so its ok
+    const args = try init.minimal.args.toSlice(alloc);
+    const file_path = if (args.len != 1) args[1] else null;
+    const file = if (file_path) |fp|
+        try Io.Dir.cwd().openFile(io, fp, .{ .mode = .read_only })
+    else
+        null;
+    defer if (file) |f| f.close(io);
+
+    const in_buf = try alloc.alloc(u8, 1024);
+    var in =
+        if (file) |f| f.reader(io, in_buf) else File.stdin().reader(io, in_buf);
+
+    const out_buf = try alloc.alloc(u8, 1024);
+    var out = File.stdout().writer(io, out_buf);
+
+    try HTMLua.parseTemplate(&arena, io, &state, &in.interface, &out.interface);
+
+    try out.flush();
+}
+
+// https://ziggit.dev/t/zig-0-15-1-reader-writer-dont-make-copies-of-fieldparentptr-based-interfaces/11719
+fn resolver(arena: *ArenaAllocator, io: Io, path: []const u8) anyerror!*Io.Reader {
+    const read_buff = try arena.allocator().alloc(u8, 1024);
+    const cwd = std.Io.Dir.cwd();
+
+    var file = try cwd.openFile(io, path, .{ .mode = .read_only });
+    const reader = try arena.allocator().create(Io.File.Reader);
+    reader.* = file.reader(io, read_buff);
+    return &reader.interface;
+}
diff --git a/src/root.zig b/src/root.zig
new file mode 100755
index 0000000..3ddfe78
--- /dev/null
+++ b/src/root.zig
@@ -0,0 +1,76 @@
+const std = @import("std");
+const Io = std.Io;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const SplitIterator = std.mem.SplitIterator;
+const DelimiterType = std.mem.DelimiterType;
+const ArrayList = std.ArrayList;
+const StringArrayHashMap = std.StringArrayHashMapUnmanaged;
+
+pub const tags = @import("tags.zig");
+pub const utils = @import("utils.zig");
+
+pub const ParserFn =
+    fn (
+        arena: *ArenaAllocator,
+        io: Io,
+        state: *Parser,
+        reader: *Io.Reader,
+        writer: *Io.Writer,
+        attrs: *SplitIterator(u8, DelimiterType.scalar),
+    ) anyerror!void;
+
+pub const ParserTag = struct {
+    name: []const u8,
+    parse_fn: *const ParserFn,
+};
+
+pub const Parser = struct {
+    component_resolver: *const fn (alloc: *ArenaAllocator, io: Io, path: []const u8) anyerror!*Io.Reader,
+    exported_parts: StringArrayHashMap([]const u8),
+    tags: []const ParserTag,
+    tag_stack: ArrayList([]const u8),
+};
+
+pub fn parseTemplate(arena: *ArenaAllocator, io: Io, state: *Parser, reader: *Io.Reader, writer: *Io.Writer) !void {
+    outer: while (true) {
+        _ = reader.streamDelimiter(writer, '<') catch |err| {
+            if (err == Io.Reader.Error.EndOfStream) break else return err;
+        };
+        // consume '<'
+        _ = try reader.takeByte();
+        if (try reader.peekByte() == '/') {
+            // consume '/'
+            _ = try reader.takeByte();
+            const last_tag = state.tag_stack.getLastOrNull() orelse "";
+            // if reached closing tag, send execution back up the stack
+            if (std.mem.eql(u8, last_tag, try reader.peekDelimiterExclusive('>'))) {
+                // consume closing tag
+                _ = try reader.takeDelimiterInclusive('>');
+                return;
+            } else {
+                try writer.writeAll("</");
+                continue :outer;
+            }
+        }
+        const inside = reader.takeDelimiterExclusive('>') catch |err| {
+            if (err == error.StreamTooLong) {
+                try writer.writeByte('<');
+                break :outer;
+            } else return err;
+        };
+        var it = std.mem.splitScalar(u8, inside, ' ');
+        const tag_name = it.first();
+        for (state.tags) |tag| {
+            if (std.mem.eql(u8, tag.name, tag_name) == true) {
+                // consume '>'
+                _ = try reader.takeByte();
+                try tag.parse_fn(arena, io, state, reader, writer, &it);
+                continue :outer;
+            }
+        }
+
+        try writer.writeByte('<');
+        try writer.writeAll(inside);
+    }
+}
diff --git a/src/tags.zig b/src/tags.zig
new file mode 100755
index 0000000..3f4169e
--- /dev/null
+++ b/src/tags.zig
@@ -0,0 +1,4 @@
+pub const Include = @import("tags/Include.zig").Include;
+pub const Export = @import("tags/Export.zig").Export;
+pub const Use = @import("tags/Use.zig").Use;
+pub const Syntax = @import("tags/Syntax.zig").Syntax;
diff --git a/src/tags/Export.zig b/src/tags/Export.zig
new file mode 100755
index 0000000..4b64048
--- /dev/null
+++ b/src/tags/Export.zig
@@ -0,0 +1,52 @@
+const std = @import("std");
+const Io = std.Io;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const SplitIterator = std.mem.SplitIterator;
+const DelimiterType = std.mem.DelimiterType;
+
+const parser = @import("../root.zig");
+const Parser = parser.Parser;
+
+pub const ExportTagError = error{
+    NoPartAttribute,
+};
+
+pub const Export: parser.ParserTag = .{
+    .name = "export",
+    .parse_fn = &parse,
+};
+
+fn parse(
+    arena: *ArenaAllocator,
+    io: Io,
+    state: *Parser,
+    reader: *Io.Reader,
+    writer: *Io.Writer,
+    attrs: *SplitIterator(u8, DelimiterType.scalar),
+) anyerror!void {
+    _ = writer;
+
+    const raw_part_attr: []const u8 = while (attrs.next()) |attr| {
+        var it = std.mem.splitScalar(u8, attr, '=');
+        if (std.mem.eql(u8, it.first(), "part")) {
+            break it.next() orelse return ExportTagError.NoPartAttribute;
+        }
+    } else return ExportTagError.NoPartAttribute;
+
+    const part = try arena.allocator().dupe(u8, raw_part_attr);
+
+    var part_buffer = try Io.Writer.Allocating.initCapacity(arena.allocator(), 1024);
+
+    try state.tag_stack.append(arena.allocator(), "export");
+    try parser.parseTemplate(
+        arena,
+        io,
+        state,
+        reader,
+        &part_buffer.writer,
+    );
+    _ = state.tag_stack.pop();
+
+    try state.exported_parts.put(arena.allocator(), part, try part_buffer.toOwnedSlice());
+}
diff --git a/src/tags/Include.zig b/src/tags/Include.zig
new file mode 100755
index 0000000..87f0b3f
--- /dev/null
+++ b/src/tags/Include.zig
@@ -0,0 +1,58 @@
+const std = @import("std");
+const Io = std.Io;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const SplitIterator = std.mem.SplitIterator;
+const DelimiterType = std.mem.DelimiterType;
+
+const parser = @import("../root.zig");
+const Parser = parser.Parser;
+
+pub const IncludeTagError = error{
+    NoPathAttribute,
+    PathOpenFailure,
+};
+
+pub const Include: parser.ParserTag = .{
+    .name = "include",
+    .parse_fn = &parse,
+};
+
+fn parse(
+    arena: *ArenaAllocator,
+    io: Io,
+    state: *Parser,
+    reader: *Io.Reader,
+    writer: *Io.Writer,
+    attrs: *SplitIterator(u8, DelimiterType.scalar),
+) anyerror!void {
+    const raw_path_attr: []const u8 = while (attrs.next()) |attr| {
+        var it = std.mem.splitScalar(u8, attr, '=');
+        if (std.mem.eql(u8, it.first(), "path")) {
+            const raw_path = it.next() orelse return IncludeTagError.NoPathAttribute;
+            if (raw_path.len == 0) return IncludeTagError.NoPathAttribute;
+            if ((raw_path[0] == '"' and raw_path[raw_path.len - 1] == '"') or
+                (raw_path[0] == '\'' and raw_path[raw_path.len - 1] == '\''))
+            {
+                if (raw_path.len <= 2) return IncludeTagError.NoPathAttribute;
+                break raw_path[1 .. raw_path.len - 1];
+            } else {
+                break raw_path;
+            }
+        }
+    } else return IncludeTagError.NoPathAttribute;
+
+    const path = try arena.allocator().dupe(u8, raw_path_attr);
+
+    // parse through to gather <export> tags
+    try state.tag_stack.append(arena.allocator(), "include");
+    try parser.parseTemplate(arena, io, state, reader, writer);
+    _ = state.tag_stack.pop();
+
+    const included_reader = state.component_resolver(arena, io, path) catch {
+        std.log.err("Include Tag: failed to open file {s}", .{path});
+        return IncludeTagError.PathOpenFailure;
+    };
+
+    try parser.parseTemplate(arena, io, state, included_reader, writer);
+}
diff --git a/src/tags/Syntax.zig b/src/tags/Syntax.zig
new file mode 100755
index 0000000..03930b1
--- /dev/null
+++ b/src/tags/Syntax.zig
@@ -0,0 +1,169 @@
+const std = @import("std");
+const Io = std.Io;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const SplitIterator = std.mem.SplitIterator;
+const DelimiterType = std.mem.DelimiterType;
+
+const ezts = @import("easy-zig-tree-sitter");
+const ts = @import("tree-sitter");
+
+const parser = @import("../root.zig");
+const Parser = parser.Parser;
+
+pub const SyntaxTagError = error{InvalidLanguage};
+pub const Syntax: parser.ParserTag = .{
+    .name = "syntax",
+    .parse_fn = &parse,
+};
+
+fn parse(
+    arena: *ArenaAllocator,
+    io: Io,
+    state: *Parser,
+    reader: *Io.Reader,
+    writer: *Io.Writer,
+    attrs: *SplitIterator(u8, DelimiterType.scalar),
+) anyerror!void {
+    _ = state;
+    _ = io;
+
+    const raw_lang_attr: []const u8 = while (attrs.next()) |attr| {
+        var it = std.mem.splitScalar(u8, attr, '=');
+        if (std.mem.eql(u8, it.first(), "lang")) {
+            break it.next() orelse "";
+        }
+    } else "";
+
+    const lang_key = try arena.allocator().dupe(u8, raw_lang_attr);
+
+    _ = try reader.takeDelimiterInclusive('\n');
+    const first_line = try reader.peekDelimiterInclusive('\n');
+    const whitespace_chr = for (first_line, 0..) |chr, idx| {
+        if (!(chr == ' ' or chr == '\t')) break idx;
+    } else 0;
+
+    var code = try std.ArrayList(u8).initCapacity(arena.allocator(), 1024);
+
+    while (true) {
+        const line = try reader.takeDelimiterInclusive('\n');
+        if (std.mem.indexOf(u8, line, "</syntax>")) |idx| {
+            try code.appendSlice(arena.allocator(), line[0..idx]);
+            // TODO: tags at the end will break the parser
+            break;
+        }
+        const whitespace_lem = @min(line.len, whitespace_chr);
+
+        if (std.mem.countScalar(u8, line[0..whitespace_lem], ' ') + std.mem.countScalar(u8, line[0..whitespace_lem], '\t') < whitespace_lem) {
+            try code.appendSlice(arena.allocator(), line);
+        } else {
+            try code.appendSlice(arena.allocator(), line[whitespace_lem..line.len]);
+        }
+    }
+
+    // Alignment issues?
+    // ts.setAllocator(arena.allocator());
+    // TODO: TS Thread Mutex
+
+    // TODO: null handeling
+    var hlp = HLParser.create(lang_key, code.items) orelse {
+        std.log.warn("Failed to make Parser for {s}", .{lang_key});
+        return;
+    };
+    defer hlp.destroy();
+    var next_hl = hlp.nextHL();
+
+    var end_stack = try std.ArrayList(u32).initCapacity(arena.allocator(), 1024);
+    var code_idx: u32 = 0;
+
+    try writer.writeAll("<pre class=\"hl\"><code>");
+
+    while (code_idx != code.items.len) {
+        var next_idx: u32 = @intCast(code.items.len);
+        if (end_stack.getLastOrNull()) |last| {
+            if (last == code_idx) {
+                try writer.writeAll("</span>");
+                _ = end_stack.pop();
+                continue;
+            } else if (last < next_idx) {
+                next_idx = last;
+            }
+        }
+        if (next_hl) |hl| {
+            if (hl.start == code_idx) {
+                try writer.writeAll("<span class=\"hl");
+                // no '.' in css class names
+                var it = std.mem.splitScalar(u8, hl.name, '.');
+                while (it.next()) |n| {
+                    try writer.writeByte('-');
+                    try writer.writeAll(n);
+                }
+                try writer.writeAll("\">");
+                try end_stack.append(arena.allocator(), hl.end);
+                next_hl = hlp.nextHL();
+                continue;
+            } else if (hl.start < next_idx) {
+                next_idx = hl.start;
+            }
+        }
+        try writer.writeAll(code.items[code_idx..next_idx]);
+        code_idx = next_idx;
+    }
+
+    try writer.writeAll("</code></pre>");
+}
+
+const HLGroup = struct {
+    name: []const u8,
+    start: u32,
+    end: u32,
+};
+
+const HLParser = struct {
+    lang: *ts.Language,
+    tree: *ts.Tree,
+    query: *ts.Query,
+    cursor: *ts.QueryCursor,
+    fn create(lang_key: []const u8, code: []const u8) ?HLParser {
+        const lang = ezts.getLang(lang_key) orelse return null;
+        const query_string = ezts.getQuery(lang_key, "highlights") orelse return null;
+
+        const ps = ts.Parser.create();
+
+        ps.setLanguage(lang) catch return null;
+        const tree = ps.parseString(code, null) orelse return null;
+
+        var error_offset: u32 = 0;
+        const query = ts.Query.create(lang, query_string, &error_offset) catch return null;
+
+        const cursor = ts.QueryCursor.create();
+        cursor.exec(query, tree.rootNode());
+
+        return .{
+            .lang = lang,
+            .tree = tree,
+            .query = query,
+            .cursor = cursor,
+        };
+    }
+
+    fn destroy(self: *HLParser) void {
+        self.cursor.destroy();
+        self.query.destroy();
+        self.tree.destroy();
+        self.lang.destroy();
+    }
+
+    fn nextHL(self: *HLParser) ?HLGroup {
+        if (self.cursor.nextCapture()) |data| {
+            const idx, const match = data;
+            const c = match.captures[idx];
+
+            return .{
+                .start = c.node.startByte(),
+                .end = c.node.endByte(),
+                .name = self.query.captureNameForId(c.index) orelse return null,
+            };
+        } else return null;
+    }
+};
diff --git a/src/tags/Use.zig b/src/tags/Use.zig
new file mode 100755
index 0000000..6ba1b1b
--- /dev/null
+++ b/src/tags/Use.zig
@@ -0,0 +1,48 @@
+const std = @import("std");
+const Io = std.Io;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const SplitIterator = std.mem.SplitIterator;
+const DelimiterType = std.mem.DelimiterType;
+
+const parser = @import("../root.zig");
+const Parser = parser.Parser;
+
+pub const UseTagError = error{ NoPartAttribute, NoPartExported };
+
+pub const Use: parser.ParserTag = .{
+    .name = "use",
+    .parse_fn = &parse,
+};
+
+fn parse(
+    arena: *ArenaAllocator,
+    io: Io,
+    state: *Parser,
+    reader: *Io.Reader,
+    writer: *Io.Writer,
+    attrs: *SplitIterator(u8, DelimiterType.scalar),
+) anyerror!void {
+    _ = io;
+
+    const raw_part_attr: []const u8 = while (attrs.next()) |attr| {
+        var it = std.mem.splitScalar(u8, attr, '=');
+        if (std.mem.eql(u8, it.first(), "part")) {
+            break it.next() orelse return UseTagError.NoPartAttribute;
+        }
+    } else return UseTagError.NoPartAttribute;
+
+    const part = try arena.allocator().dupe(u8, raw_part_attr);
+
+    const part_text = state.exported_parts.get(part) orelse return UseTagError.NoPartExported;
+
+    var part_reader = Io.Reader.fixed(part_text);
+
+    _ = try part_reader.streamRemaining(writer);
+
+    while (true) {
+        _ = try reader.discardDelimiterInclusive('<');
+        const inside = try reader.takeDelimiterInclusive('>');
+        if (std.mem.eql(u8, "/use>", inside)) return;
+    }
+}
diff --git a/src/utils.zig b/src/utils.zig
new file mode 100755
index 0000000..ae03d76
--- /dev/null
+++ b/src/utils.zig
@@ -0,0 +1,23 @@
+const std = @import("std");
+const Io = std.Io;
+const File = Io.File;
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+const ArrayList = std.ArrayList;
+const StringArrayHashMap = std.StringArrayHashMapUnmanaged;
+
+pub fn sanitizePath(alloc: Allocator, path: []const u8) ![]const u8 {
+    const clean = try alloc.alloc(u8, path.len);
+    var list = ArrayList(u8).initBuffer(clean);
+    var pit = std.mem.splitScalar(u8, path, '/');
+    while (pit.next()) |part| {
+        if (part.len == 0) continue;
+        if (std.mem.eql(u8, part, "..")) continue;
+        // This should never actually allocate
+        try list.appendSlice(alloc, part);
+        try list.append(alloc, '/');
+    }
+    const new_len = list.items.len - 1;
+    _ = alloc.resize(clean, new_len);
+    return clean[0..new_len];
+}