feat(lexer): supports link and image

3 files changed, 25 insertions, 5 deletions

diff --git a/grammar.ebnf b/grammar.ebnf
index d0259a1..664f68b 100644
--- a/grammar.ebnf
+++ b/grammar.ebnf
@@ -6,7 +6,7 @@ strong-delimiter = "\n\n", { "\n" };
 
 (* blocks *)
 block = title | paragraph | quote | callout | code-block | math-block | image | footnotes;
-title = ? #{1,6} ?, content;
+title = ? #{1,6} ?, " ", content;
 paragraph = content, { weak-delimiter, content };
 quote = ">", content, { weak-delimiter, ">", content }, [ paragraph ];
 callout = ":::", [ ? [a-z]+ ? ], { delimiter, content }, weak-delimiter, ":::";
diff --git a/src/lexer/lexed.zig b/src/lexer/lexed.zig
index a572c2a..59b6e31 100644
--- a/src/lexer/lexed.zig
+++ b/src/lexer/lexed.zig
@@ -13,6 +13,8 @@ pub const Kind = enum {
     bold,
     italic,
     ref,
+    list_ordored,
+    list_unordored,
 };
 
 pub const Lexed = struct {
diff --git a/src/lexer/lexer.zig b/src/lexer/lexer.zig
index a29384e..b76cd50 100644
--- a/src/lexer/lexer.zig
+++ b/src/lexer/lexer.zig
@@ -30,14 +30,14 @@ pub const Lexer = struct {
                 self.force_lit = true;
                 current_kind = .literal;
             } else {
-                current_kind = self.getCurrentKind(rune, acc.items);
+                current_kind = self.getCurrentKind(current_kind, rune, acc.items);
                 self.force_lit = false;
                 try acc.appendSlice(alloc, rune);
             }
             // conds here to avoid creating complex condition in while
             const next_rune = self.iter.peek(1);
             if (next_rune.len > 0) {
-                if (self.getCurrentKind(next_rune, acc.items) != current_kind.?) {
+                if (self.getCurrentKind(current_kind, next_rune, acc.items) != current_kind.?) {
                     if (!requiresSpace(current_kind.?)) break;
                     if (std.mem.eql(u8, next_rune, " ")) {
                         // consume next space
@@ -55,13 +55,15 @@ pub const Lexer = struct {
         return lexed.Lexed.init(alloc, kind, acc);
     }
 
-    fn getCurrentKind(self: *Self, rune: []const u8, acc: []const u8) lexed.Kind {
+    fn getCurrentKind(self: *Self, before: ?lexed.Kind, rune: []const u8, acc: []const u8) lexed.Kind {
         if (self.force_lit) return .literal;
         if (std.mem.eql(u8, rune, ">")) return .quote;
         if (std.mem.eql(u8, rune, "\n")) return .delimiter;
+        if (std.mem.eql(u8, rune, "!")) return .image;
         if (is('#', 6, rune, acc)) return .title;
         if (is('`', 3, rune, acc)) return .code;
         if (is('$', 3, rune, acc)) return .math;
+        if (isIn(links, before, .link, rune, acc)) return .link;
         return .literal;
     }
 };
@@ -72,6 +74,21 @@ fn is(v: u8, maxLen: usize, rune: []const u8, acc: []const u8) bool {
     return std.mem.eql(u8, rune, &[_]u8{v});
 }
 
+const links = &[_][]const u8{ "[", "](", ")" };
+
+fn isIn(ops: []const []const u8, before: ?lexed.Kind, now: lexed.Kind, rune: []const u8, p: []const u8) bool {
+    var acc = p;
+    if (before) |b| {
+        if (now != b) acc = &[_]u8{};
+    }
+    for (ops) |op| {
+        const ln = acc.len + rune.len;
+        if (op.len >= ln and std.mem.eql(u8, acc, op[0..acc.len]) and std.mem.eql(u8, rune, op[acc.len..ln]))
+            return true;
+    }
+    return false;
+}
+
 fn requiresSpace(k: lexed.Kind) bool {
     return switch (k) {
         .title => true,
@@ -98,7 +115,8 @@ test "lexer common" {
     var l = try Lexer.init("# hello world :)");
 
     try doTest(alloc, &l, .title, "#");
-    try doTest(alloc, &l, .literal, "hello world :)");
+    try doTest(alloc, &l, .literal, "hello world :");
+    try doTest(alloc, &l, .link, ")");
 
     try expect(try l.next(alloc) == null);
 }