feat: add parseIn and parseBetween to expr

src/expr.zig

@@ -155,3 +155,163 @@ const Parser = struct {
         const value = try rawValue.dupe(self.allocator);
         self.advance();
 
+        const cond = Condition{ .field = fieldCopy, .value = value };
+        if (std.mem.eql(u8, op, "eq")) return .{ .eq = cond };
+        if (std.mem.eql(u8, op, "ne")) return .{ .ne = cond };
+        if (std.mem.eql(u8, op, "lt")) return .{ .lt = cond };
+        if (std.mem.eql(u8, op, "lte")) return .{ .lte = cond };
+        if (std.mem.eql(u8, op, "gt")) return .{ .gt = cond };
+        if (std.mem.eql(u8, op, "gte")) return .{ .gte = cond };
+        if (std.mem.eql(u8, op, "like")) {
+            if (value != .text) return error.InvalidValue;
+            return .{ .like = cond };
+        }
+        return error.UnknownOperator;
+    }
+
+    /// `field in (v1 v2 ...)`: desugars to an `or_` of per-value `eq`
+    /// leaves against the same field, sharing the one field-name copy the
+    /// caller already made (Condition never owns `field`, so aliasing it
+    /// across every leaf is safe). A single value collapses to a bare `eq`
+    /// leaf, the same collapse `parseAnd`/`parseOr` already do for a
+    /// one-element and/or chain - lets a one-value `in` use an index like
+    /// `eq` does, since `findIndexableLeaf`/`findCompositeMatch` never look
+    /// inside `or_`. An empty list, `field in ()`, is not a syntax error:
+    /// it desugars to an `or_` over zero leaves, which `Predicate.evaluate`
+    /// already folds to `false` for an empty `preds` slice - matching SQL's
+    /// "IN () is always false" without a special case here.
+    fn parseIn(self: *Parser, field: []const u8) ParseError!Predicate {
+        if (self.cur.kind != .lparen) return if (self.cur.kind == .eof) error.UnexpectedEnd else error.UnexpectedToken;
+        self.advance();
+
+        var list = std.ArrayList(Predicate).init(self.allocator);
+        while (self.cur.kind != .rparen) {
+            if (self.cur.kind != .word) return if (self.cur.kind == .eof) error.UnexpectedEnd else error.UnexpectedToken;
+            const rawValue = rowmod.Value.parseCli(self.cur.text) catch return error.InvalidValue;
+            const value = try rawValue.dupe(self.allocator);
+            self.advance();
+            try list.append(.{ .eq = .{ .field = field, .value = value } });
+        }
+        self.advance();
+
+        if (list.items.len == 1) return list.items[0];
+        return .{ .or_ = try list.toOwnedSlice() };
+    }
+
+    /// `field between lo and hi`: desugars to `field gte lo and field lte
+    /// hi` (inclusive both ends, matching SQL BETWEEN). The `and` here is
+    /// consumed as part of this leaf's own grammar, not the outer
+    /// `parseAnd` binary connective, so it never nests a further `and_`
+    /// level - by the time control returns to the caller, `self.cur` is
+    /// whatever token follows `hi` (rparen/eof/"and"/"or"), decided purely
+    /// by having advanced past every token this leaf owns.
+    fn parseBetween(self: *Parser, field: []const u8) ParseError!Predicate {
+        if (self.cur.kind != .word) return error.UnexpectedEnd;
+        const rawLow = rowmod.Value.parseCli(self.cur.text) catch return error.InvalidValue;
+        const low = try rawLow.dupe(self.allocator);
+        self.advance();
+
+        if (!isKeyword(self.cur, "and")) return if (self.cur.kind == .eof) error.UnexpectedEnd else error.UnexpectedToken;
+        self.advance();
+
+        if (self.cur.kind != .word) return error.UnexpectedEnd;
+        const rawHigh = rowmod.Value.parseCli(self.cur.text) catch return error.InvalidValue;
+        const high = try rawHigh.dupe(self.allocator);
+        self.advance();
+
+        const bounds = try self.allocator.alloc(Predicate, 2);
+        bounds[0] = .{ .gte = .{ .field = field, .value = low } };
+        bounds[1] = .{ .lte = .{ .field = field, .value = high } };
+        return .{ .and_ = bounds };
+    }
+};
+
+/// Parses a full `table select` predicate expression, e.g.
+/// `city eq s:warsaw and (age gte i:30 or not age lt i:10)`. The returned
+/// Parsed owns every slice/pointer the Predicate tree references - `text`
+/// itself does not need to outlive the call.
+pub fn parse(backing: Allocator, text: []const u8) ParseError!Parsed {
+    var arena = std.heap.ArenaAllocator.init(backing);
+    errdefer arena.deinit();
+    const allocator = arena.allocator();
+
+    var p = Parser{ .lexer = .{ .src = text }, .allocator = allocator, .cur = undefined };
+    p.cur = p.lexer.next();
+
+    const pred = try p.parseOr();
+    if (p.cur.kind != .eof) return error.UnexpectedToken;
+
+    return .{ .arena = arena, .predicate = pred };
+}
+
+fn evalStr(allocator: Allocator, text: []const u8, row: *const rowmod.Row) !bool {
+    var parsed = try parse(allocator, text);
+    defer parsed.deinit();
+    return parsed.predicate.evaluate(row);
+}
+
+test "parses a single condition" {
+    const allocator = std.testing.allocator;
+    var row = rowmod.Row.init(allocator);
+    defer row.deinit();
+    try row.set("city", .{ .text = "warsaw" });
+
+    try std.testing.expect(try evalStr(allocator, "city eq s:warsaw", &row));
+    try std.testing.expect(!try evalStr(allocator, "city eq s:krakow", &row));
+}
+
+test "and binds tighter than or" {
+    const allocator = std.testing.allocator;
+    var row = rowmod.Row.init(allocator);
+    defer row.deinit();
+    try row.set("a", .{ .int = 1 });
+    try row.set("b", .{ .int = 2 });
+    try row.set("c", .{ .int = 99 });
+
+    // a eq 1 or b eq 2 and c eq 3  ==  (a eq 1) or ((b eq 2) and (c eq 3))
+    // b eq 2 is true but c eq 3 is false, so the and-branch is false; the
+    // overall result must still be true because of the leading `a eq 1`.
+    try std.testing.expect(try evalStr(allocator, "a eq i:1 or b eq i:2 and c eq i:3", &row));
+
+    var row2 = rowmod.Row.init(allocator);
+    defer row2.deinit();
+    try row2.set("a", .{ .int = 0 });
+    try row2.set("b", .{ .int = 2 });
+    try row2.set("c", .{ .int = 99 });
+    try std.testing.expect(!try evalStr(allocator, "a eq i:1 or b eq i:2 and c eq i:3", &row2));
+}
+
+test "parentheses override default precedence" {
+    const allocator = std.testing.allocator;
+    var row = rowmod.Row.init(allocator);
+    defer row.deinit();
+    try row.set("a", .{ .int = 0 });
+    try row.set("b", .{ .int = 2 });
+    try row.set("c", .{ .int = 5 });
+
+    // (a eq 1 or b eq 2) and c eq 5 - forces the or to evaluate before and.
+    try std.testing.expect(try evalStr(allocator, "(a eq i:1 or b eq i:2) and c eq i:5", &row));
+    try std.testing.expect(!try evalStr(allocator, "(a eq i:1 or b eq i:2) and c eq i:6", &row));
+}
+
+test "not negates the following term" {
+    const allocator = std.testing.allocator;
+    var row = rowmod.Row.init(allocator);
+    defer row.deinit();
+    try row.set("age", .{ .int = 20 });
+
+    try std.testing.expect(try evalStr(allocator, "not age eq i:10", &row));
+    try std.testing.expect(!try evalStr(allocator, "not age eq i:20", &row));
+    try std.testing.expect(!try evalStr(allocator, "not (age eq i:10 or age eq i:20)", &row));
+}
+
+test "nested parentheses and multiple not" {
+    const allocator = std.testing.allocator;
+    var row = rowmod.Row.init(allocator);
+    defer row.deinit();
+    try row.set("x", .{ .int = 5 });
+
+    try std.testing.expect(try evalStr(allocator, "not not x eq i:5", &row));
+    try std.testing.expect(try evalStr(allocator, "((x eq i:5))", &row));
+}
+