grammar: add column constraints

grammar.js

@@ -25,19 +25,45 @@ module.exports = grammar({
         optional($._if_not_exists),
         $.table_reference,
         $.column_definitions
+        // TODO: INHERITS
+        // TODO: PARTITION BY
+        // TODO: USING
+        // TODO: WITH
+        // TODO: ON COMMIT
+        // TODO: TABLESPACE
       ),
 
     column_definitions: ($) =>
       seq(
         "(",
-        optional(commaSepRepeat1(choice($.column_definition /*$.constraint*/))),
+        optional(
+          commaSepRepeat1(choice($.column_definition /*, $.table_constraint*/))
+        ),
         ")"
       ),
 
     column_definition: ($) =>
-      seq(field("name", $.identifier), field("datatype", $._type)),
+      seq(
+        field("name", $.identifier),
+        field("datatype", $._type),
+        repeat($.column_constraint)
+      ),
 
-    //constraint: $ => seq(),
+    column_constraint: ($) =>
+      seq(
+        optional(seq($.keyword_constraint, field("name", $.identifier))),
+        choice(
+          $._not_null,
+          $.keyword_null,
+          seq($.keyword_default, $._expression)
+          // TODO: CHECK
+          // TODO: GENERATED
+          // TODO: UNIQUE
+          // TODO: PRIMARY KEY
+          // TODO: FOREIGN KEY
+        )
+        // TODO: DEFERRABLE
+      ),
 
     table_reference: ($) =>
       seq(
@@ -45,8 +71,24 @@ module.exports = grammar({
         field("name", $.identifier)
       ),
 
+    _expression: ($) =>
+      choice(
+        $.literal
+        // TODO: add more types
+      ),
+
+    literal: ($) =>
+      choice(
+        $._number,
+        $._literal_string,
+        $.keyword_true,
+        $.keyword_false,
+        $.keyword_null
+      ),
+
     // keywords
     _if_not_exists: ($) => seq($.keyword_if, $.keyword_not, $.keyword_exists),
+    _not_null: ($) => seq($.keyword_not, $.keyword_null),
 
     keyword_create: (_) => mkKeyword("create"),
     keyword_table: (_) => mkKeyword("table"),
@@ -55,6 +97,11 @@ module.exports = grammar({
     keyword_if: (_) => mkKeyword("if"),
     keyword_not: (_) => mkKeyword("not"),
     keyword_exists: (_) => mkKeyword("exists"),
+    keyword_null: (_) => mkKeyword("null"),
+    keyword_constraint: (_) => mkKeyword("constraint"),
+    keyword_default: (_) => mkKeyword("default"),
+    keyword_true: (_) => mkKeyword("true"),
+    keyword_false: (_) => mkKeyword("false"),
 
     // References: https://www.postgresql.org/docs/15/datatype.html
     _type: ($) =>
@@ -177,10 +224,8 @@ module.exports = grammar({
         seq(mkKeyword("timestamp"), $._with_time_zone)
       ),
 
-    _without_time_zone: ($) => seq($._keyword_without, $._keyword_time_zone),
+    _without_time_zone: ($) => seq(mkKeyword("without"), $._keyword_time_zone),
-    _with_time_zone: ($) => seq($._keyword_with, $._keyword_time_zone),
+    _with_time_zone: ($) => seq(mkKeyword("with"), $._keyword_time_zone),
-    _keyword_without: (_) => mkKeyword("without"),
-    _keyword_with: (_) => mkKeyword("with"),
     _keyword_time_zone: (_) => seq(mkKeyword("time"), mkKeyword("zone")),
 
     // References: https://www.postgresql.org/docs/15/datatype-uuid.html
@@ -198,9 +243,10 @@ module.exports = grammar({
     // https://stackoverflow.com/questions/13014947/regex-to-match-a-c-style-multiline-comment
     marginalia: (_) => seq("/*", /[^*]*\*+(?:[^/*][^*]*\*+)*/, "/"),
 
+    _literal_string: ($) => choice(seq("'", /[^']*/, "'")),
     _number: (_) => /\d+/,
 
-    identifier: ($) => choice($._identifier, seq('"', $._identifier, '"')),
+    identifier: ($) => choice($._identifier, seq('"', /[^"]+/, '"')),
 
     _identifier: (_) => /([a-zA-Z_][0-9a-zA-Z_]*)/,
   },

src/grammar.json

@@ -181,6 +181,69 @@
             "type": "SYMBOL",
             "name": "_type"
           }
+        },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "SYMBOL",
+            "name": "column_constraint"
+          }
+        }
+      ]
+    },
+    "column_constraint": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "SEQ",
+              "members": [
+                {
+                  "type": "SYMBOL",
+                  "name": "keyword_constraint"
+                },
+                {
+                  "type": "FIELD",
+                  "name": "name",
+                  "content": {
+                    "type": "SYMBOL",
+                    "name": "identifier"
+                  }
+                }
+              ]
+            },
+            {
+              "type": "BLANK"
+            }
+          ]
+        },
+        {
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "SYMBOL",
+              "name": "_not_null"
+            },
+            {
+              "type": "SYMBOL",
+              "name": "keyword_null"
+            },
+            {
+              "type": "SEQ",
+              "members": [
+                {
+                  "type": "SYMBOL",
+                  "name": "keyword_default"
+                },
+                {
+                  "type": "SYMBOL",
+                  "name": "_expression"
+                }
+              ]
+            }
+          ]
         }
       ]
     },
@@ -222,6 +285,40 @@
         }
       ]
     },
+    "_expression": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "literal"
+        }
+      ]
+    },
+    "literal": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "_number"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "_literal_string"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "keyword_true"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "keyword_false"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "keyword_null"
+        }
+      ]
+    },
     "_if_not_exists": {
       "type": "SEQ",
       "members": [
@@ -239,6 +336,19 @@
         }
       ]
     },
+    "_not_null": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "keyword_not"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "keyword_null"
+        }
+      ]
+    },
     "keyword_create": {
       "type": "PATTERN",
       "value": "create|CREATE"
@@ -276,6 +386,26 @@
       "type": "PATTERN",
       "value": "exists|EXISTS"
     },
+    "keyword_null": {
+      "type": "PATTERN",
+      "value": "null|NULL"
+    },
+    "keyword_constraint": {
+      "type": "PATTERN",
+      "value": "constraint|CONSTRAINT"
+    },
+    "keyword_default": {
+      "type": "PATTERN",
+      "value": "default|DEFAULT"
+    },
+    "keyword_true": {
+      "type": "PATTERN",
+      "value": "true|TRUE"
+    },
+    "keyword_false": {
+      "type": "PATTERN",
+      "value": "false|FALSE"
+    },
     "_type": {
       "type": "CHOICE",
       "members": [
@@ -938,8 +1068,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "SYMBOL",
+          "type": "PATTERN",
-          "name": "_keyword_without"
+          "value": "without|WITHOUT"
         },
         {
           "type": "SYMBOL",
@@ -951,8 +1081,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "SYMBOL",
+          "type": "PATTERN",
-          "name": "_keyword_with"
+          "value": "with|WITH"
         },
         {
           "type": "SYMBOL",
@@ -960,14 +1090,6 @@
         }
       ]
     },
-    "_keyword_without": {
-      "type": "PATTERN",
-      "value": "without|WITHOUT"
-    },
-    "_keyword_with": {
-      "type": "PATTERN",
-      "value": "with|WITH"
-    },
     "_keyword_time_zone": {
       "type": "SEQ",
       "members": [
@@ -1027,6 +1149,28 @@
         }
       ]
     },
+    "_literal_string": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "SEQ",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "'"
+            },
+            {
+              "type": "PATTERN",
+              "value": "[^']*"
+            },
+            {
+              "type": "STRING",
+              "value": "'"
+            }
+          ]
+        }
+      ]
+    },
     "_number": {
       "type": "PATTERN",
       "value": "\\d+"
@@ -1046,8 +1190,8 @@
               "value": "\""
             },
             {
-              "type": "SYMBOL",
+              "type": "PATTERN",
-              "name": "_identifier"
+              "value": "[^\"]+"
             },
             {
               "type": "STRING",

src/node-types.json

@@ -25,6 +25,48 @@
       ]
     }
   },
+  {
+    "type": "column_constraint",
+    "named": true,
+    "fields": {
+      "name": {
+        "multiple": false,
+        "required": false,
+        "types": [
+          {
+            "type": "identifier",
+            "named": true
+          }
+        ]
+      }
+    },
+    "children": {
+      "multiple": true,
+      "required": true,
+      "types": [
+        {
+          "type": "keyword_constraint",
+          "named": true
+        },
+        {
+          "type": "keyword_default",
+          "named": true
+        },
+        {
+          "type": "keyword_not",
+          "named": true
+        },
+        {
+          "type": "keyword_null",
+          "named": true
+        },
+        {
+          "type": "literal",
+          "named": true
+        }
+      ]
+    }
+  },
   {
     "type": "column_definition",
     "named": true,
@@ -137,6 +179,16 @@
           }
         ]
       }
+    },
+    "children": {
+      "multiple": true,
+      "required": false,
+      "types": [
+        {
+          "type": "column_constraint",
+          "named": true
+        }
+      ]
     }
   },
   {
@@ -282,6 +334,29 @@
     "named": true,
     "fields": {}
   },
+  {
+    "type": "literal",
+    "named": true,
+    "fields": {},
+    "children": {
+      "multiple": false,
+      "required": false,
+      "types": [
+        {
+          "type": "keyword_false",
+          "named": true
+        },
+        {
+          "type": "keyword_null",
+          "named": true
+        },
+        {
+          "type": "keyword_true",
+          "named": true
+        }
+      ]
+    }
+  },
   {
     "type": "marginalia",
     "named": true,
@@ -409,6 +484,10 @@
     "type": "\"",
     "named": false
   },
+  {
+    "type": "'",
+    "named": false
+  },
   {
     "type": "(",
     "named": false
@@ -457,6 +536,10 @@
     "type": "keyword_bytea",
     "named": true
   },
+  {
+    "type": "keyword_constraint",
+    "named": true
+  },
   {
     "type": "keyword_create",
     "named": true
@@ -473,10 +556,18 @@
     "type": "keyword_decimal",
     "named": true
   },
+  {
+    "type": "keyword_default",
+    "named": true
+  },
   {
     "type": "keyword_exists",
     "named": true
   },
+  {
+    "type": "keyword_false",
+    "named": true
+  },
   {
     "type": "keyword_if",
     "named": true
@@ -501,6 +592,10 @@
     "type": "keyword_not",
     "named": true
   },
+  {
+    "type": "keyword_null",
+    "named": true
+  },
   {
     "type": "keyword_numeric",
     "named": true
@@ -529,6 +624,10 @@
     "type": "keyword_text",
     "named": true
   },
+  {
+    "type": "keyword_true",
+    "named": true
+  },
   {
     "type": "keyword_unlogged",
     "named": true
@@ -540,9 +639,5 @@
   {
     "type": "keyword_xml",
     "named": true
-  },
-  {
-    "type": "literal",
-    "named": true
   }
 ]

test/corpus/create/table.txt

@@ -167,3 +167,113 @@ create table foo (
     )
   )
 )
+
+================================================================================
+Create a table with column constraints
+================================================================================
+create table foo (
+  c1  text  not null,
+  c2  text  null,
+  c2  text  not null default 'hello'
+);
+--------------------------------------------------------------------------------
+
+(source_file
+  (statement
+    (create_table
+      (keyword_create)
+      (keyword_table)
+      (table_reference
+        name: (identifier)
+      )
+      (column_definitions
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_not)
+            (keyword_null)
+          )
+        )
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_null)
+          )
+        )
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_not)
+            (keyword_null)
+          )
+          (column_constraint
+            (keyword_default)
+            (literal)
+          )
+        )
+      )
+    )
+  )
+)
+
+================================================================================
+Create a table with named column constraints
+================================================================================
+create table foo (
+  c1  text  constraint strong_c1 not null,
+  c2  text  constraint weak_c2 null,
+  c3  text  constraint "c3 with power" not null constraint "c2 set default hello" default 'hello'
+);
+--------------------------------------------------------------------------------
+
+(source_file
+  (statement
+    (create_table
+      (keyword_create)
+      (keyword_table)
+      (table_reference
+        name: (identifier)
+      )
+      (column_definitions
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_constraint)
+            name: (identifier)
+            (keyword_not)
+            (keyword_null)
+          )
+        )
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_constraint)
+            name: (identifier)
+            (keyword_null)
+          )
+        )
+        (column_definition
+          name: (identifier)
+          datatype: (keyword_text)
+          (column_constraint
+            (keyword_constraint)
+            name: (identifier)
+            (keyword_not)
+            (keyword_null)
+          )
+          (column_constraint
+            (keyword_constraint)
+            name: (identifier)
+            (keyword_default)
+            (literal)
+          )
+        )
+      )
+    )
+  )
+)