Support anonymous blocks for new Oracle dialect (#53)

rathboma · web-flow · commit 334ffa587e00 · 2022-06-22T01:06:14.000Z
diff --git a/README.md b/README.md
@@ -55,6 +55,7 @@ This way you have sure is a valid query before trying to identify the types.
 * ALTER_TRIGGER
 * ALTER_FUNCTION
 * ALTER_INDEX
+* ANON_BLOCK (Oracle Database only)
 * UNKNOWN (only available if strict mode is disabled)
 
 ## Execution types
@@ -64,6 +65,7 @@ Execution types allow to know what is the query behavior
 * `LISTING:` is when the query list the data
 * `MODIFICATION:` is when the query modificate the database somehow (structure or data)
 * `INFORMATION:` is show some data information such as a profile data
+* `ANON_BLOCK: ` is for an anonymous block query which may contain multiple statements of unknown type (Oracle Database only)
 * `UNKNOWN`: (only available if strict mode is disabled)
 
 ## Installation
@@ -112,7 +114,7 @@ console.log(statements);
 1. `input (string)`: the whole SQL script text to be processed
 1. `options (object)`: allow to set different configurations
     1. `strict (bool)`: allow disable strict mode which will ignore unknown types *(default=true)*
-    2. `dialect (string)`: Specify your database dialect, values: `generic`, `mysql`, `psql`, `sqlite` and `mssql`. *(default=generic)*
+    2. `dialect (string)`: Specify your database dialect, values: `generic`, `mysql`, `oracle`, `psql`, `sqlite` and `mssql`. *(default=generic)*
 
 ## Contributing
 
diff --git a/src/defines.ts b/src/defines.ts
@@ -1,4 +1,4 @@
-export const DIALECTS = ['mssql', 'sqlite', 'mysql', 'psql', 'generic'] as const;
+export const DIALECTS = ['mssql', 'sqlite', 'mysql', 'oracle', 'psql', 'generic'] as const;
 export type Dialect = typeof DIALECTS[number];
 export type StatementType =
   | 'INSERT'
@@ -27,9 +27,10 @@ export type StatementType =
   | 'ALTER_TRIGGER'
   | 'ALTER_FUNCTION'
   | 'ALTER_INDEX'
+  | 'ANON_BLOCK'
   | 'UNKNOWN';
 
-export type ExecutionType = 'LISTING' | 'MODIFICATION' | 'UNKNOWN';
+export type ExecutionType = 'LISTING' | 'MODIFICATION' | 'ANON_BLOCK' | 'UNKNOWN';
 
 export interface IdentifyOptions {
   strict?: boolean;
diff --git a/src/parser.ts b/src/parser.ts
@@ -51,15 +51,17 @@ export const EXECUTION_TYPES: Record<StatementType, ExecutionType> = {
   ALTER_FUNCTION: 'MODIFICATION',
   ALTER_INDEX: 'MODIFICATION',
   UNKNOWN: 'UNKNOWN',
+  ANON_BLOCK: 'ANON_BLOCK',
 };
 
-const statementsWithEnds = ['CREATE_TRIGGER', 'CREATE_FUNCTION'];
+const statementsWithEnds = ['CREATE_TRIGGER', 'CREATE_FUNCTION', 'ANON_BLOCK'];
 const blockOpeners: Record<Dialect, string[]> = {
   generic: ['BEGIN', 'CASE'],
   psql: ['BEGIN', 'CASE', 'LOOP', 'IF'],
   mysql: ['BEGIN', 'CASE', 'LOOP', 'IF'],
   mssql: ['BEGIN', 'CASE'],
   sqlite: ['BEGIN', 'CASE'],
+  oracle: ['DECLARE', 'BEGIN', 'CASE'],
 };
 
 interface ParseOptions {
@@ -250,6 +252,12 @@ function createStatementParserByToken(token: Token, options: ParseOptions): Stat
         return createDeleteStatementParser(options);
       case 'TRUNCATE':
         return createTruncateStatementParser(options);
+      case 'DECLARE':
+      case 'BEGIN':
+        if (options.dialect === 'oracle') {
+          return createBlockStatementParser(options);
+        }
+      // eslint-disable-next-line no-fallthrough
       default:
         break;
     }
@@ -285,6 +293,32 @@ function createSelectStatementParser(options: ParseOptions) {
   return stateMachineStatementParser(statement, steps, options);
 }
 
+function createBlockStatementParser(options: ParseOptions) {
+  const statement = createInitialStatement();
+  statement.type = 'ANON_BLOCK';
+
+  const steps: Step[] = [
+    // Select
+    {
+      preCanGoToNext: () => false,
+      validation: {
+        acceptTokens: [
+          { type: 'keyword', value: 'DECLARE' },
+          { type: 'keyword', value: 'BEGIN' },
+        ],
+      },
+      add: (token) => {
+        if (statement.start < 0) {
+          statement.start = token.start;
+        }
+      },
+      postCanGoToNext: () => true,
+    },
+  ];
+
+  return stateMachineStatementParser(statement, steps, options);
+}
+
 function createInsertStatementParser(options: ParseOptions) {
   const statement = createInitialStatement();
 
@@ -540,6 +574,9 @@ function stateMachineStatementParser(
   let prevToken: Token;
   let prevPrevToken: Token;
 
+  let lastBlockOpener: Token;
+  let anonBlockStarted = false;
+
   let openBlocks = 0;
 
   /* eslint arrow-body-style: 0, no-extra-parens: 0 */
@@ -600,11 +637,27 @@ function stateMachineStatementParser(
       if (
         token.type === 'keyword' &&
         blockOpeners[dialect].includes(token.value) &&
-        prevPrevToken.value.toUpperCase() !== 'END'
+        prevPrevToken?.value.toUpperCase() !== 'END'
       ) {
+        if (
+          dialect === 'oracle' &&
+          lastBlockOpener?.value === 'DECLARE' &&
+          token.value.toUpperCase() === 'BEGIN'
+        ) {
+          // don't open a new block!
+          setPrevToken(token);
+          lastBlockOpener = token;
+          return;
+        }
         openBlocks++;
+        lastBlockOpener = token;
         setPrevToken(token);
-        return;
+        if (statement.type === 'ANON_BLOCK' && !anonBlockStarted) {
+          anonBlockStarted = true;
+          // don't return
+        } else {
+          return;
+        }
       }
 
       if (
@@ -614,7 +667,7 @@ function stateMachineStatementParser(
         statement.parameters.push(token.value);
       }
 
-      if (statement.type) {
+      if (statement.type && statement.start >= 0) {
         // statement has already been identified
         // just wait until end of the statement
         return;
diff --git a/src/tokenizer.ts b/src/tokenizer.ts
@@ -25,6 +25,9 @@ const KEYWORDS = [
   'WITH',
   'AS',
   'MATERIALIZED',
+  'BEGIN',
+  'DECLARE',
+  'CASE',
 ];
 
 const INDIVIDUALS: Record<string, Token['type']> = {
diff --git a/test/identifier/multiple-statement.spec.ts b/test/identifier/multiple-statement.spec.ts
@@ -127,6 +127,104 @@ describe('identifier', () => {
       expect(actual).to.eql(expected);
     });
 
+    describe('identifying statements with anonymous blocks', () => {
+      it('should work in strict mode', () => {
+        const actual = identify(
+          `
+          DECLARE
+            PK_NAME VARCHAR(200);
+
+          BEGIN
+            EXECUTE IMMEDIATE ('CREATE SEQUENCE "untitled_table8_seq"');
+
+          SELECT
+            cols.column_name INTO PK_NAME
+          FROM
+            all_constraints cons,
+            all_cons_columns cols
+          WHERE
+            cons.constraint_type = 'P'
+            AND cons.constraint_name = cols.constraint_name
+            AND cons.owner = cols.owner
+            AND cols.table_name = 'untitled_table8';
+
+          execute immediate (
+            'create or replace trigger "untitled_table8_autoinc_trg"  BEFORE INSERT on "untitled_table8"  for each row  declare  checking number := 1;  begin    if (:new."' || PK_NAME || '" is null) then      while checking >= 1 loop        select "untitled_table8_seq".nextval into :new."' || PK_NAME || '" from dual;        select count("' || PK_NAME || '") into checking from "untitled_table8"        where "' || PK_NAME || '" = :new."' || PK_NAME || '";      end loop;    end if;  end;'
+          );
+
+          END;
+          `,
+          { dialect: 'oracle', strict: true },
+        );
+        const expected = [
+          {
+            end: 1043,
+            executionType: 'ANON_BLOCK',
+            parameters: [],
+            start: 11,
+            text: 'DECLARE\n            PK_NAME VARCHAR(200);\n\n          BEGIN\n            EXECUTE IMMEDIATE (\'CREATE SEQUENCE "untitled_table8_seq"\');\n\n          SELECT\n            cols.column_name INTO PK_NAME\n          FROM\n            all_constraints cons,\n            all_cons_columns cols\n          WHERE\n            cons.constraint_type = \'P\'\n            AND cons.constraint_name = cols.constraint_name\n            AND cons.owner = cols.owner\n            AND cols.table_name = \'untitled_table8\';\n\n          execute immediate (\n            \'create or replace trigger "untitled_table8_autoinc_trg"  BEFORE INSERT on "untitled_table8"  for each row  declare  checking number := 1;  begin    if (:new."\' || PK_NAME || \'" is null) then      while checking >= 1 loop        select "untitled_table8_seq".nextval into :new."\' || PK_NAME || \'" from dual;        select count("\' || PK_NAME || \'") into checking from "untitled_table8"        where "\' || PK_NAME || \'" = :new."\' || PK_NAME || \'";      end loop;    end if;  end;\'\n          );\n\n          END;',
+            type: 'ANON_BLOCK',
+          },
+        ];
+        expect(actual).to.eql(expected);
+      });
+
+      it('should identify a create table then a block', () => {
+        const actual = identify(
+          `
+          create table
+            "untitled_table8" (
+              "id" integer not null primary key,
+              "created_at" varchar(255) not null
+            );
+
+          DECLARE
+            PK_NAME VARCHAR(200);
+
+          BEGIN
+            EXECUTE IMMEDIATE ('CREATE SEQUENCE "untitled_table8_seq"');
+
+          SELECT
+            cols.column_name INTO PK_NAME
+          FROM
+            all_constraints cons,
+            all_cons_columns cols
+          WHERE
+            cons.constraint_type = 'P'
+            AND cons.constraint_name = cols.constraint_name
+            AND cons.owner = cols.owner
+            AND cols.table_name = 'untitled_table8';
+
+          execute immediate (
+            'create or replace trigger "untitled_table8_autoinc_trg"  BEFORE INSERT on "untitled_table8"  for each row  declare  checking number := 1;  begin    if (:new."' || PK_NAME || '" is null) then      while checking >= 1 loop        select "untitled_table8_seq".nextval into :new."' || PK_NAME || '" from dual;        select count("' || PK_NAME || '") into checking from "untitled_table8"        where "' || PK_NAME || '" = :new."' || PK_NAME || '";      end loop;    end if;  end;'
+          );
+
+          END;
+          `,
+          { dialect: 'oracle', strict: false },
+        );
+        const expected = [
+          {
+            end: 167,
+            executionType: 'MODIFICATION',
+            parameters: [],
+            start: 11,
+            text: 'create table\n            "untitled_table8" (\n              "id" integer not null primary key,\n              "created_at" varchar(255) not null\n            );',
+            type: 'CREATE_TABLE',
+          },
+          {
+            end: 1212,
+            executionType: 'ANON_BLOCK',
+            parameters: [],
+            start: 180,
+            text: 'DECLARE\n            PK_NAME VARCHAR(200);\n\n          BEGIN\n            EXECUTE IMMEDIATE (\'CREATE SEQUENCE "untitled_table8_seq"\');\n\n          SELECT\n            cols.column_name INTO PK_NAME\n          FROM\n            all_constraints cons,\n            all_cons_columns cols\n          WHERE\n            cons.constraint_type = \'P\'\n            AND cons.constraint_name = cols.constraint_name\n            AND cons.owner = cols.owner\n            AND cols.table_name = \'untitled_table8\';\n\n          execute immediate (\n            \'create or replace trigger "untitled_table8_autoinc_trg"  BEFORE INSERT on "untitled_table8"  for each row  declare  checking number := 1;  begin    if (:new."\' || PK_NAME || \'" is null) then      while checking >= 1 loop        select "untitled_table8_seq".nextval into :new."\' || PK_NAME || \'" from dual;        select count("\' || PK_NAME || \'") into checking from "untitled_table8"        where "\' || PK_NAME || \'" = :new."\' || PK_NAME || \'";      end loop;    end if;  end;\'\n          );\n\n          END;',
+            type: 'ANON_BLOCK',
+          },
+        ];
+        expect(actual).to.eql(expected);
+      });
+    });
+
     describe('identifying multiple statements with CTEs', () => {
       it('should able to detect queries with a CTE in middle query', () => {
         const actual = identify(
diff --git a/test/index.spec.ts b/test/index.spec.ts
@@ -4,7 +4,7 @@ import { expect } from 'chai';
 describe('identify', () => {
   it('should throw error for invalid dialect', () => {
     expect(() => identify('SELECT * FROM foo', { dialect: 'invalid' as Dialect })).to.throw(
-      'Unknown dialect. Allowed values: mssql, sqlite, mysql, psql, generic',
+      'Unknown dialect. Allowed values: mssql, sqlite, mysql, oracle, psql, generic',
     );
   });
 
diff --git a/test/parser/oracle.spec.ts b/test/parser/oracle.spec.ts
