feat: support load data in mysql

taozhi8833998 · taozhi8833998 · commit 705fca03f6d2 · 2025-06-25T08:53:21.000+08:00
diff --git a/pegjs/mysql.pegjs b/pegjs/mysql.pegjs
@@ -447,6 +447,7 @@ cmd_stmt
   / grant_stmt
   / explain_stmt
   / transaction_stmt
+  / load_data_stmt
 
 create_stmt
   = create_table_stmt
@@ -2091,6 +2092,69 @@ transaction_stmt
     }
   }
 
+load_data_field
+  = k:('FIELDS'i / 'COLUMNS'i) __ t:('TERMINATED'i __ 'BY'i __ ident_without_kw_type)? __ en:(('OPTIONALLY'i)? __ 'ENCLOSED'i __ 'BY'i __ ident_without_kw_type)? __ es:('ESCAPED'i __ 'BY'i __ ident_without_kw_type)? {
+    if (t) t[4].prefix = 'TERMINATED BY'
+    if (en) en[6].prefix = `${en[0] && en[0].toUpperCase() === 'OPTIONALLY' ? 'OPTIONALLY ' : ''}ENCLOSED BY`
+    if (es) es[4].prefix = 'ESCAPED BY'
+    return {
+      keyword: k,
+      terminated: t && t[4],
+      enclosed: en && en[6],
+      escaped: es && es[4]
+    }
+  }
+
+load_data_line_starting
+  = k:('STARTING'i / 'TERMINATED'i) __ 'BY'i __ s:ident_without_kw_type {
+    s.prefix = `${k.toUpperCase()} BY`
+    return {
+      type: k.toLowerCase(),
+      [k.toLowerCase()]: s
+    }
+  }
+load_data_line
+  = k:'LINES'i __ s:load_data_line_starting? __ t:load_data_line_starting? {
+    if (s && t && s.type === t.type) throw new Error('LINES cannot be specified twice')
+    if (s) Reflect.deleteProperty(s, 'type')
+    if (t) Reflect.deleteProperty(t, 'type')
+    return {
+      keyword: k,
+      ...(s || {}),
+      ...(t || {})
+    }
+  }
+
+load_data_stmt
+  = 'LOAD'i __ 'DATA'i __ lc:('LOW_PRIORITY'i / 'CONCURRENT'i)? __ lo:('LOCAL'i)? __
+  'INFILE'i __ file:ident_without_kw_type __ ri:replace_insert? __
+  'INTO'i __ 'TABLE'i __ table:table_name __
+  pa:insert_partition? __
+  cs:(create_option_character_set_kw __ ident_without_kw_type)? __
+  fields:load_data_field? __
+  lines:load_data_line? __
+  ig:(KW_IGNORE __ literal_numeric __ ('LINES'i / 'ROWS'i))? __
+  co:column_clause? __
+  set:(KW_SET __ set_list)? {
+    return {
+      type: 'load_data',
+      mode: lc,
+      local: lo,
+      file: file,
+      replace_ignore: ri,
+      table: table,
+      partition: pa,
+      character_set: cs,
+      fields: fields,
+      lines: lines,
+      ignore: ig && {
+        count: ig[2],
+        suffix: ig[4]
+      },
+      column: co,
+      set: set && set[2]
+    }
+  }
 priv_type_table
   =  p:(KW_ALL / KW_ALTER / KW_CREATE __ 'VIEW'i / KW_CREATE / KW_DELETE / KW_DROP / 'GRANT'i __ 'OPTION'i / KW_INDEX / KW_INSERT / KW_REFERENCES / KW_SELECT / KW_SHOW __ KW_VIEW / KW_TRIGGER / KW_UPDATE) {
     return {
@@ -3372,7 +3436,7 @@ double_quoted_ident
   }
 
 single_quoted_ident
-  = "'" chars:[^']+ "'" {
+  = "'" chars:[^']* "'" {
     return {
       type: 'single_quote_string',
       value: chars.join('')
diff --git a/src/expr.js b/src/expr.js
@@ -12,6 +12,7 @@ import { jsonExprToSQL, jsonVisitorExprToSQL } from './json'
 import { selectToSQL } from './select'
 import { showToSQL } from './show'
 import { arrayStructExprToSQL } from './array-struct'
+import { loadDataToSQL } from './load'
 import { tablesToSQL, unnestToSQL } from './tables'
 import { unionToSQL } from './union'
 import { namedWindowExprListToSQL, windowFuncToSQL } from './window'
@@ -35,6 +36,7 @@ const exprToSQLConvertFn = {
   fulltext_search   : fullTextSearchToSQL,
   function          : funcToSQL,
   lambda            : lambdaToSQL,
+  load_data         : loadDataToSQL,
   insert            : unionToSQL,
   interval          : intervalToSQL,
   json              : jsonExprToSQL,
diff --git a/src/insert.js b/src/insert.js
@@ -89,5 +89,6 @@ function insertToSQL(stmt) {
 export {
   conflictToSQL,
   insertToSQL,
+  partitionToSQL,
   valuesToSQL,
 }
diff --git a/src/load.js b/src/load.js
@@ -0,0 +1,58 @@
+import { commonOptionConnector, toUpper, hasVal, literalToSQL } from './util'
+import { partitionToSQL } from './insert'
+import { tableToSQL } from './tables'
+import { columnsToSQL } from './column'
+import { setToSQL } from './update'
+
+function loadDataFields(expr) {
+  if (!expr) return ''
+  const { keyword, terminated, enclosed, escaped } = expr
+  return [
+    toUpper(keyword),
+    literalToSQL(terminated),
+    literalToSQL(enclosed),
+    literalToSQL(escaped),
+  ].filter(hasVal).join(' ')
+}
+
+function loadDataLines(expr) {
+  if (!expr) return ''
+  const { keyword, starting, terminated } = expr
+  return [
+    toUpper(keyword),
+    literalToSQL(starting),
+    literalToSQL(terminated),
+  ].filter(hasVal).join(' ')
+}
+
+function loadDataIgnore(expr) {
+  if (!expr) return ''
+  const { count, suffix } = expr
+  return ['IGNORE', literalToSQL(count), suffix].filter(hasVal).join(' ')
+}
+function loadDataToSQL(expr) {
+  if (!expr) return ''
+  const { mode, local, file, replace_ignore, table, partition, character_set, column, fields, lines, set, ignore } = expr
+  const result = [
+    'LOAD DATA',
+    toUpper(mode),
+    toUpper(local),
+    'INFILE',
+    literalToSQL(file),
+    toUpper(replace_ignore),
+    'INTO TABLE',
+    tableToSQL(table),
+    partitionToSQL(partition),
+    commonOptionConnector('CHARACTER SET', literalToSQL, character_set),
+    loadDataFields(fields),
+    loadDataLines(lines),
+    loadDataIgnore(ignore),
+    columnsToSQL(column),
+    commonOptionConnector('SET', setToSQL, set),
+  ]
+  return result.filter(hasVal).join(' ')
+}
+
+export {
+  loadDataToSQL,
+}
diff --git a/src/sql.js b/src/sql.js
@@ -1,6 +1,6 @@
 import { unionToSQL, multipleToSQL } from './union'
 
-const supportedTypes = ['analyze', 'attach', 'select', 'deallocate', 'delete', 'exec', 'update', 'insert', 'drop', 'rename', 'truncate', 'call', 'desc', 'use', 'alter', 'set', 'create', 'lock', 'unlock', 'declare', 'show', 'replace', 'if', 'grant', 'revoke', 'proc', 'raise', 'execute', 'transaction', 'explain', 'comment']
+const supportedTypes = ['analyze', 'attach', 'select', 'deallocate', 'delete', 'exec', 'update', 'insert', 'drop', 'rename', 'truncate', 'call', 'desc', 'use', 'alter', 'set', 'create', 'lock', 'unlock', 'declare', 'show', 'replace', 'if', 'grant', 'revoke', 'proc', 'raise', 'execute', 'transaction', 'explain', 'comment', 'load_data']
 
 function checkSupported(expr) {
   const ast = expr && expr.ast ? expr.ast : expr
diff --git a/src/union.js b/src/union.js
@@ -26,6 +26,7 @@ import {
 import { execToSQL } from './exec'
 import { orderOrPartitionByToSQL } from './expr'
 import { limitToSQL } from './limit'
+import { loadDataToSQL } from './load'
 import { procToSQL } from './proc'
 import { transactionToSQL } from './transaction'
 import { showToSQL } from './show'
@@ -47,6 +48,7 @@ const typeToSQLFn = {
   update      : updateToSQL,
   if          : ifToSQL,
   insert      : insertToSQL,
+  load_data   : loadDataToSQL,
   drop        : commonCmdToSQL,
   truncate    : commonCmdToSQL,
   replace     : insertToSQL,
diff --git a/test/mysql-mariadb.spec.js b/test/mysql-mariadb.spec.js
@@ -1403,6 +1403,18 @@ describe('mysql', () => {
         expect(selectIntoToSQL({})).to.be.undefined
       })
     })
+    describe('load data', () => {
+      it('should support load data', () => {
+        let sql = `LOAD DATA LOCAL INFILE "mycsv.csv" INTO TABLE mytable FIELDS TERMINATED BY ' ^' IGNORE 1 LINES;`
+        expect(getParsedSql(sql)).to.equal(`LOAD DATA LOCAL INFILE "mycsv.csv" INTO TABLE \`mytable\` FIELDS TERMINATED BY ' ^' IGNORE 1 LINES`)
+        sql = `LOAD DATA LOW_PRIORITY LOCAL INFILE '/tmp/test.txt' REPLACE INTO TABLE test FIELDS TERMINATED BY '\t' ENCLOSED BY '' ESCAPED BY '\\' LINES TERMINATED BY '\n' STARTING BY ''`
+        expect(getParsedSql(sql)).to.equal("LOAD DATA LOW_PRIORITY LOCAL INFILE '/tmp/test.txt' REPLACE INTO TABLE `test` FIELDS TERMINATED BY '\t' ENCLOSED BY '' ESCAPED BY '\\' LINES STARTING BY '' TERMINATED BY '\n'")
+        sql = `LOAD DATA LOW_PRIORITY LOCAL INFILE '/tmp/test.txt' REPLACE INTO TABLE test FIELDS TERMINATED BY '\t' OPTIONALLY ENCLOSED BY '' ESCAPED BY '\\' IGNORE 1 LINES (column1, @var1) SET column2 = @var1/100;`
+        expect(getParsedSql(sql)).to.equal("LOAD DATA LOW_PRIORITY LOCAL INFILE '/tmp/test.txt' REPLACE INTO TABLE \`test\` FIELDS TERMINATED BY '\t' OPTIONALLY ENCLOSED BY '' ESCAPED BY '\\' IGNORE 1 LINES (`column1`, @var1) SET `column2` = @var1 / 100")
+        sql = `LOAD DATA INFILE 'file.txt' INTO TABLE t1 (column1, column2) SET column3 = CURRENT_TIMESTAMP;`
+        expect(getParsedSql(sql)).to.equal("LOAD DATA INFILE 'file.txt' INTO TABLE `t1` (`column1`, `column2`) SET `column3` = CURRENT_TIMESTAMP")
+      })
+    })
   })
 
 })

Original file line number	Diff line number	Diff line change
`@@ -89,5 +89,6 @@ function insertToSQL(stmt) {`
`89`	`89`	`export {`
`90`	`90`	`conflictToSQL,`
`91`	`91`	`insertToSQL,`
	`92`	`+ partitionToSQL,`
`92`	`93`	`valuesToSQL,`
`93`	`94`	`}`