Merge pull request #15 from optimism-java/yaml-parser

feat: yaml parse error and add ssz test cases
optimism-java · Nov 2, 2024 · 8a62f1d · 8a62f1d
2 parents fb9c95f + 5a577df
commit 8a62f1d
Show file tree

Hide file tree

Showing 3 changed files with 105 additions and 17 deletions.
diff --git a/src/spec_tests/ssz_static/root.zig b/src/spec_tests/ssz_static/root.zig
@@ -62,13 +62,13 @@ const CommonUnion = union {
     // AttestationData: types.AttestationData,
     // AttesterSlashing: types.AttesterSlashing,
     // BeaconBlock: types.BeaconBlock,
-    // BeaconBlockHeader: types.BeaconBlockHeader,
+    BeaconBlockHeader: types.BeaconBlockHeader,
     Checkpoint: types.Checkpoint,
     // Deposit: types.Deposit,
-    // DepositData: types.DepositData,
-    // DepositMessage: types.DepositMessage,
-    // Eth1Block: types.Eth1Block,
-    // Eth1Data: types.Eth1Data,
+    DepositData: types.DepositData,
+    DepositMessage: types.DepositMessage,
+    Eth1Block: types.Eth1Block,
+    Eth1Data: types.Eth1Data,
     Fork: types.Fork,
     ForkData: types.ForkData,
     // HistoricalBatch: types.HistoricalBatch,
@@ -77,9 +77,9 @@ const CommonUnion = union {
     // ProposerSlashing: types.ProposerSlashing,
     // SignedBeaconBlock: types.SignedBeaconBlock,
     // SignedBeaconBlockHeader: types.SignedBeaconBlockHeader,
-    // SigningData: types.SigningData,
+    SigningData: types.SigningData,
     SyncAggregatorSelectionData: types.SyncAggregatorSelectionData, // only exist in altair or later
-    // Validator: types.Validator,
+    Validator: types.Validator,
     VoluntaryExit: types.VoluntaryExit,
 };
 

diff --git a/src/yaml/parse.zig b/src/yaml/parse.zig
@@ -1,7 +1,3 @@
-//! The code bellow is essentially a port of https://github.com/kubkon/zig-yaml
-//! to the most recent version of zig with a couple of stylistic changes and support for
-//! json yaml.
-
 const std = @import("std");
 const assert = std.debug.assert;
 const log = std.log.scoped(.parse);
@@ -386,16 +382,17 @@ const Parser = struct {
 
         log.debug("(doc) begin {s}@{d}", .{ @tagName(self.tree.tokens[node.base.start].id), node.base.start });
         // json format yaml
-        const is_one_line: bool = if (self.eatToken(.flow_map_start, &.{})) |doc_pos| is_one_line: {
+
+        const is_json_format: bool = if (self.eatToken(.flow_map_start, &.{})) |doc_pos| is_one_line: {
             if (self.getCol(doc_pos) > 0) return error.MalformedYaml;
             if (self.eatToken(.tag, &.{ .new_line, .comment })) |_| {
                 node.directive = try self.expectToken(.literal, &.{ .new_line, .comment });
             }
             break :is_one_line true;
         } else false;
 
-        if (is_one_line) {
-            return self.one_line_doc(node);
+        if (is_json_format) {
+            return self.handle_json_format(node);
         }
 
         // Parse header
@@ -443,9 +440,9 @@ const Parser = struct {
         return &node.base;
     }
 
-    fn one_line_doc(self: *Parser, node: *Node.Doc) ParseError!*Node {
+    fn handle_json_format(self: *Parser, node: *Node.Doc) ParseError!*Node {
         // Parse value
-        node.value = try self.value();
+        node.value = try self.json_format_map();
         if (node.value == null) {
             self.token_it.seekBy(-1);
         }
@@ -499,7 +496,6 @@ const Parser = struct {
             if (self.getCol(key_pos) < col) {
                 break;
             }
-
             const key = self.token_it.next() orelse return error.UnexpectedEof;
             switch (key.id) {
                 .literal => {},
@@ -513,7 +509,88 @@ const Parser = struct {
                     continue;
                 },
             }
+            log.debug("(map) key {s}@{d}", .{ self.tree.getRaw(key_pos, key_pos), key_pos });
+
+            // Separator
+            _ = try self.expectToken(.map_value_ind, &.{ .new_line, .comment });
+
+            // Parse value
+            const val = try self.value();
+            errdefer if (val) |v| {
+                v.deinit(self.allocator);
+            };
+
+            if (val) |v| {
+                if (self.getCol(v.start) < self.getCol(key_pos)) {
+                    return error.MalformedYaml;
+                }
+                if (v.cast(Node.Value)) |_| {
+                    if (self.getCol(v.start) == self.getCol(key_pos)) {
+                        return error.MalformedYaml;
+                    }
+                }
+            }
+
+            try node.values.append(self.allocator, .{
+                .key = key_pos,
+                .value = val,
+            });
+        }
+
+        node.base.end = self.token_it.pos - 1;
+
+        log.debug("(map) end {s}@{d}", .{ @tagName(self.tree.tokens[node.base.end].id), node.base.end });
 
+        return &node.base;
+    }
+
+    fn json_format_map(self: *Parser) ParseError!*Node {
+        const node = try self.allocator.create(Node.Map);
+        errdefer self.allocator.destroy(node);
+        node.* = .{};
+        node.base.tree = self.tree;
+        node.base.start = self.token_it.pos;
+        errdefer {
+            for (node.values.items) |entry| {
+                if (entry.value) |val| {
+                    val.deinit(self.allocator);
+                }
+            }
+            node.values.deinit(self.allocator);
+        }
+
+        log.debug("(map) begin {s}@{d}", .{ @tagName(self.tree.tokens[node.base.start].id), node.base.start });
+
+        const col = self.getCol(node.base.start);
+        var currentNewLine = self.getLine(node.base.start);
+
+        while (true) {
+            self.eatCommentsAndSpace(&.{});
+
+            // Parse key
+            const key_pos = self.token_it.pos;
+            if (self.getCol(key_pos) < col) {
+                // multiline cases, see validator.txt in test
+                const newLine = self.getLine(key_pos);
+                if (newLine > currentNewLine) {
+                    currentNewLine = newLine;
+                } else {
+                    break;
+                }
+            }
+            const key = self.token_it.next() orelse return error.UnexpectedEof;
+            switch (key.id) {
+                .literal => {},
+                .doc_start, .doc_end, .eof => {
+                    self.token_it.seekBy(-1);
+                    break;
+                },
+                else => {
+                    // TODO key not being a literal
+                    // return error.Unhandled;
+                    continue;
+                },
+            }
             log.debug("(map) key {s}@{d}", .{ self.tree.getRaw(key_pos, key_pos), key_pos });
 
             // Separator

diff --git a/src/yaml/yaml.zig b/src/yaml/yaml.zig
@@ -32,6 +32,7 @@ pub const Value = union(enum) {
     string: []const u8,
     list: List,
     map: Map,
+    bool: bool,
 
     pub fn asInt(self: Value) !i64 {
         if (self != .int) return error.TypeMismatch;
@@ -53,6 +54,14 @@ pub const Value = union(enum) {
         return self.string;
     }
 
+    pub fn asBool(self: Value) !bool {
+        if (self != .string) return error.TypeMismatch;
+        const str = self.string;
+        if (std.mem.eql(u8, str, "true")) return true;
+        if (std.mem.eql(u8, str, "false")) return false;
+        return error.TypeMismatch;
+    }
+
     pub fn asList(self: Value) !List {
         if (self == .list) {
             return self.list;
@@ -97,6 +106,7 @@ pub const Value = union(enum) {
             .uint => |uint| return writer.print("{}", .{uint}),
             .float => |float| return writer.print("{d}", .{float}),
             .string => |string| return writer.print("{s}", .{string}),
+            .bool => |data| return writer.print("{any}", .{data}),
             .list => |list| {
                 const len = list.len;
                 if (len == 0) return;
@@ -427,6 +437,7 @@ pub const Yaml = struct {
             },
             .void => error.TypeMismatch,
             .optional => unreachable,
+            .bool => try value.asBool(),
             else => error.Unimplemented,
         };
     }