feat: add literal aliases (#11)

corpus/spec.txt

@@ -1105,8 +1105,9 @@ block: ># comment
 --------------------------------------------------------------------------------
 
 (ERROR
-  (flow_node
+  (block_mapping_pair
-    (plain_scalar)))
+    key: (flow_node
+      (plain_scalar))))
 
 ================================================================================
 Comment without whitespace after doublequoted scalar [INVALID]
@@ -2129,8 +2130,9 @@ a:
 --------------------------------------------------------------------------------
 
 (ERROR
-  (flow_node
+  (block_mapping_pair
-    (plain_scalar)))
+    key: (flow_node
+      (plain_scalar))))
 
 ================================================================================
 Invalid tag [INVALID]
@@ -2310,9 +2312,11 @@ Mapping starting at --- line [INVALID]
 
 --------------------------------------------------------------------------------
 
-(ERROR
+(stream
-  (flow_node
+  (document
-    (plain_scalar)))
+    (flow_node
+      (plain_scalar)))
+  (MISSING _eof))
 
 ================================================================================
 Mapping with anchor on document start line [INVALID]
@@ -3581,8 +3585,9 @@ key: - a
 --------------------------------------------------------------------------------
 
 (ERROR
-  (flow_node
+  (block_mapping_pair
-    (plain_scalar)))
+    key: (flow_node
+      (plain_scalar))))
 
 ================================================================================
 Sequence With Same Indentation as Parent Mapping
@@ -8576,10 +8581,8 @@ c"
           key: (flow_node
             (plain_scalar)))
         (ERROR
-          (flow_node
+          (plain_scalar)
-            (plain_scalar))
+          (plain_scalar))))))
-          (flow_node
-            (plain_scalar)))))))
 
 ================================================================================
 Wrong indented sequence item [INVALID]

grammar.js

@@ -18,7 +18,8 @@ module.exports = grammar({
     $._r_blk_key_bgn,  $._br_blk_key_bgn, $._b_blk_key_bgn, // ?
     $._r_blk_val_bgn,  $._br_blk_val_bgn, $._b_blk_val_bgn, // :
     $._r_blk_imp_bgn,                                       // : (implicit)
-    $._r_blk_str_bgn,  $._br_blk_str_bgn,                   // | >
+    $._r_blk_lit_bgn,  $._br_blk_lit_bgn,                   // |
+    $._r_blk_fld_bgn,  $._br_blk_fld_bgn,                   // >
                        $._br_blk_str_ctn,                   // block scalar content
     $._r_flw_seq_bgn,  $._br_flw_seq_bgn, $._b_flw_seq_bgn, // [
     $._r_flw_seq_end,  $._br_flw_seq_end,                   // ]
@@ -229,8 +230,8 @@ module.exports = grammar({
     _r_blk_str_val: $ => choice($._r_blk_str, seq($._r_prp, choice($._r_blk_str, $._br_blk_str))),
     _br_blk_str_val: $ => choice($._br_blk_str, seq($._br_prp, choice($._r_blk_str, $._br_blk_str))),
 
-    _r_blk_str: $ => seq($._r_blk_str_bgn, repeat($._br_blk_str_ctn), $._bl),
+    _r_blk_str: $ => seq(choice($._r_blk_lit_bgn, $._r_blk_fld_bgn), repeat($._br_blk_str_ctn), $._bl),
-    _br_blk_str: $ => seq($._br_blk_str_bgn, repeat($._br_blk_str_ctn), $._bl),
+    _br_blk_str: $ => seq(choice($._br_blk_lit_bgn, $._br_blk_fld_bgn), repeat($._br_blk_str_ctn), $._bl),
 
     // flow value in block
 
@@ -435,7 +436,7 @@ module.exports = grammar({
   },
 });
 
-module.exports = global_alias(module.exports, {
+module.exports = global_alias(global_alias(module.exports, {
   ..._("yaml_directive", "_s_dir_yml"),
   ..._("yaml_version", "_r_dir_yml_ver"),
   ..._("tag_directive", "_s_dir_tag"),
@@ -479,6 +480,27 @@ module.exports = global_alias(module.exports, {
   ..._("escape_sequence", "_r_dqt_esc_nwl", "_br_dqt_esc_nwl",
                           "_r_dqt_esc_seq", "_br_dqt_esc_seq",
                           "_r_sqt_esc_sqt", "_br_sqt_esc_sqt"),
+}), {
+  ..._("---", "_s_drs_end"),
+  ..._("...", "_s_doc_end"),
+  ..._("-", "_r_blk_seq_bgn", "_br_blk_seq_bgn", "_b_blk_seq_bgn"),
+  ..._("?", "_r_blk_key_bgn", "_br_blk_key_bgn", "_b_blk_key_bgn"),
+  ..._(":", "_r_blk_val_bgn", "_br_blk_val_bgn", "_b_blk_val_bgn"),
+  ..._(":", "_r_blk_imp_bgn"),
+  ..._("|", "_r_blk_lit_bgn", "_br_blk_lit_bgn"),
+  ..._(">", "_r_blk_fld_bgn", "_br_blk_fld_bgn"),
+  ..._("[", "_r_flw_seq_bgn", "_br_flw_seq_bgn", "_b_flw_seq_bgn"),
+  ..._("]", "_r_flw_seq_end", "_br_flw_seq_end"),
+  ..._("{", "_r_flw_map_bgn", "_br_flw_map_bgn", "_b_flw_map_bgn"),
+  ..._("}", "_r_flw_map_end", "_br_flw_map_end"),
+  ..._(",", "_r_flw_sep_bgn", "_br_flw_sep_bgn"),
+  ..._("?", "_r_flw_key_bgn", "_br_flw_key_bgn"),
+  ..._(":", "_r_flw_jsv_bgn", "_br_flw_jsv_bgn"),
+  ..._(":", "_r_flw_njv_bgn", "_br_flw_njv_bgn"),
+  ..._("\"", "_r_dqt_str_bgn", "_br_dqt_str_bgn", "_b_dqt_str_bgn"),
+  ..._("\"", "_r_dqt_str_end", "_br_dqt_str_end"),
+  ..._("'", "_r_sqt_str_bgn", "_br_sqt_str_bgn", "_b_sqt_str_bgn"),
+  ..._("'", "_r_sqt_str_end", "_br_sqt_str_end"),
 });
 
 function _(alias_value, ...rule_names) {
@@ -529,9 +551,11 @@ function recursive_alias(rule, alias_map, checklist) {
     case "SYMBOL":
       if (rule.name in alias_map) {
         checklist[rule.name]++;
-        return { type: "ALIAS", content: rule, named: true, value: alias_map[rule.name].name };
+        const alias = alias_map[rule.name].name;
+        return { type: "ALIAS", content: rule, named: /[a-z]/i.test(alias), value: alias };
       }
     case "BLANK":
+    case "ALIAS":
       return rule;
     default:
       throw new Error(`Unexpected rule type ${JSON.stringify(rule.type)}`);

src/node-types.json

@@ -354,6 +354,50 @@
       ]
     }
   },
+  {
+    "type": "\"",
+    "named": false
+  },
+  {
+    "type": "'",
+    "named": false
+  },
+  {
+    "type": ",",
+    "named": false
+  },
+  {
+    "type": "-",
+    "named": false
+  },
+  {
+    "type": "---",
+    "named": false
+  },
+  {
+    "type": "...",
+    "named": false
+  },
+  {
+    "type": ":",
+    "named": false
+  },
+  {
+    "type": ">",
+    "named": false
+  },
+  {
+    "type": "?",
+    "named": false
+  },
+  {
+    "type": "[",
+    "named": false
+  },
+  {
+    "type": "]",
+    "named": false
+  },
   {
     "type": "alias",
     "named": true
@@ -393,5 +437,17 @@
   {
     "type": "yaml_version",
     "named": true
+  },
+  {
+    "type": "{",
+    "named": false
+  },
+  {
+    "type": "|",
+    "named": false
+  },
+  {
+    "type": "}",
+    "named": false
   }
 ]

src/scanner.cc

@@ -17,7 +17,8 @@ enum TokenType {
   R_BLK_KEY_BGN,  BR_BLK_KEY_BGN, B_BLK_KEY_BGN,
   R_BLK_VAL_BGN,  BR_BLK_VAL_BGN, B_BLK_VAL_BGN,
   R_BLK_IMP_BGN,
-  R_BLK_STR_BGN,  BR_BLK_STR_BGN,
+  R_BLK_LIT_BGN,  BR_BLK_LIT_BGN,
+  R_BLK_FLD_BGN,  BR_BLK_FLD_BGN,
                   BR_BLK_STR_CTN,
   R_FLW_SEQ_BGN,  BR_FLW_SEQ_BGN, B_FLW_SEQ_BGN,
   R_FLW_SEQ_END,  BR_FLW_SEQ_END,
@@ -848,9 +849,12 @@ struct Scanner {
         if (is_br_dqt_esc_seq) return scn_dqt_esc_seq(lexer, BR_DQT_ESC_SEQ);
         return false;
       }
-    } else if (LKA == '|' || LKA == '>') {
+    } else if (LKA == '|') {
-      if (VLD[R_BLK_STR_BGN] && is_r) return scn_blk_str_bgn(lexer, R_BLK_STR_BGN);
+      if (VLD[R_BLK_LIT_BGN] && is_r) return scn_blk_str_bgn(lexer, R_BLK_LIT_BGN);
-      if (VLD[BR_BLK_STR_BGN] && is_br) return scn_blk_str_bgn(lexer, BR_BLK_STR_BGN);
+      if (VLD[BR_BLK_LIT_BGN] && is_br) return scn_blk_str_bgn(lexer, BR_BLK_LIT_BGN);
+    } else if (LKA == '>') {
+      if (VLD[R_BLK_FLD_BGN] && is_r) return scn_blk_str_bgn(lexer, R_BLK_FLD_BGN);
+      if (VLD[BR_BLK_FLD_BGN] && is_br) return scn_blk_str_bgn(lexer, BR_BLK_FLD_BGN);
     }
 
     bool maybe_sgl_pln_blk = (VLD[R_SGL_PLN_BLK] && is_r) || (VLD[BR_SGL_PLN_BLK] && is_br) || (VLD[B_SGL_PLN_BLK] && is_b);