Merge pull request #359 from constructive-io/fix/edge-case-literals

fix(deparser): keep empty bit strings, collision-free dollar tags and case-sensitive option strings valid

Author: pyramation

__fixtures__/generated/generated.json

@@ -21580,6 +21580,27 @@
   "misc/literal-escaping-30.sql": "SELECT * FROM XMLTABLE(XMLNAMESPACES('http://x' AS x, DEFAULT 'http://d'), '/x:r' PASSING (SELECT doc FROM t) COLUMNS a int) AS xt",
   "misc/literal-escaping-31.sql": "SELECT * FROM t, LATERAL XMLTABLE('/r' PASSING t.doc COLUMNS a int PATH 'a') xt",
   "misc/literal-escaping-32.sql": "SELECT 1_000.000_1, 1_000.5e1_0, 0x_FFFF_FFFF_FFFF_FFFF",
+  "misc/literal-edge-cases-1.sql": "SELECT X''",
+  "misc/literal-edge-cases-2.sql": "SELECT B''",
+  "misc/literal-edge-cases-3.sql": "SELECT X'', B'', X'1F', B'101'",
+  "misc/literal-edge-cases-4.sql": "DO $x$ BEGIN RAISE NOTICE '$_PGFN_$'; END $x$",
+  "misc/literal-edge-cases-5.sql": "DO $x$ BEGIN RAISE NOTICE '$$'; END $x$",
+  "misc/literal-edge-cases-6.sql": "DO $x$ BEGIN RAISE NOTICE '$$ $EOFCODE$'; END $x$",
+  "misc/literal-edge-cases-7.sql": "DO $x$ BEGIN RAISE NOTICE '$$ $EOFCODE$ $EOFCODE1$'; END $x$",
+  "misc/literal-edge-cases-8.sql": "DO $x$BEGIN NULL; END; --$$x$",
+  "misc/literal-edge-cases-9.sql": "CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$ SELECT length('$_PGFN_$') $x$",
+  "misc/literal-edge-cases-10.sql": "CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$ SELECT length('$$ $EOFCODE$') $x$",
+  "misc/literal-edge-cases-11.sql": "CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$SELECT 1 --$$x$",
+  "misc/literal-edge-cases-12.sql": "CREATE VIEW v WITH (security_barrier = 'Foo') AS SELECT 1",
+  "misc/literal-edge-cases-13.sql": "CREATE VIEW v WITH (security_barrier = 'true') AS SELECT 1",
+  "misc/literal-edge-cases-14.sql": "CREATE INDEX i ON t (a) WITH (fillfactor = 'int')",
+  "misc/literal-edge-cases-15.sql": "CREATE INDEX i ON t (a) WITH (opt = 'Foo')",
+  "misc/literal-edge-cases-16.sql": "CREATE TABLE t (a int) WITH (opt = 'Foo')",
+  "misc/literal-edge-cases-17.sql": "CREATE TABLE t (a int) WITH (opt = 'int')",
+  "misc/literal-edge-cases-18.sql": "CREATE TABLE t (a int) WITH (opt = 'select')",
+  "misc/literal-edge-cases-19.sql": "ALTER TABLE t SET (opt = 'Foo')",
+  "misc/literal-edge-cases-20.sql": "ALTER TABLE t SET (opt = 'int')",
+  "misc/literal-edge-cases-21.sql": "ALTER INDEX i SET (opt = 'Foo')",
   "misc/launchql-ext-types-1.sql": "CREATE DOMAIN attachment AS jsonb CHECK ( value ?& ARRAY['url', 'mime'] AND (value->>'url') ~ '^(https?)://[^\\s/$.?#].[^\\s]*$' )",
   "misc/launchql-ext-types-2.sql": "COMMENT ON DOMAIN attachment IS E'@name launchqlInternalTypeAttachment'",
   "misc/launchql-ext-types-3.sql": "CREATE DOMAIN email AS citext CHECK ( value ~ '^[a-zA-Z0-9.!#$%&''*+/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*$' )",

__fixtures__/kitchen-sink/misc/literal-edge-cases.sql

@@ -0,0 +1,29 @@
+-- Edge-case literals the deparser must keep valid
+-- Ref: constructive-io/constructive-planning#2175
+
+-- Empty bit-string literals
+SELECT X'';
+SELECT B'';
+SELECT X'', B'', X'1F', B'101';
+
+-- Dollar-quoted bodies that contain the default or fallback delimiter
+DO $x$ BEGIN RAISE NOTICE '$_PGFN_$'; END $x$;
+DO $x$ BEGIN RAISE NOTICE '$$'; END $x$;
+DO $x$ BEGIN RAISE NOTICE '$$ $EOFCODE$'; END $x$;
+DO $x$ BEGIN RAISE NOTICE '$$ $EOFCODE$ $EOFCODE1$'; END $x$;
+DO $x$BEGIN NULL; END; --$$x$;
+CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$ SELECT length('$_PGFN_$') $x$;
+CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$ SELECT length('$$ $EOFCODE$') $x$;
+CREATE FUNCTION f() RETURNS int LANGUAGE sql AS $x$SELECT 1 --$$x$;
+
+-- String option values must keep their case and stay strings
+CREATE VIEW v WITH (security_barrier = 'Foo') AS SELECT 1;
+CREATE VIEW v WITH (security_barrier = 'true') AS SELECT 1;
+CREATE INDEX i ON t (a) WITH (fillfactor = 'int');
+CREATE INDEX i ON t (a) WITH (opt = 'Foo');
+CREATE TABLE t (a int) WITH (opt = 'Foo');
+CREATE TABLE t (a int) WITH (opt = 'int');
+CREATE TABLE t (a int) WITH (opt = 'select');
+ALTER TABLE t SET (opt = 'Foo');
+ALTER TABLE t SET (opt = 'int');
+ALTER INDEX i SET (opt = 'Foo');

packages/deparser/__tests__/kitchen-sink/misc-literal-edge-cases.test.ts

@@ -0,0 +1,29 @@
+
+import { FixtureTestUtils } from '../../test-utils';
+const fixtures = new FixtureTestUtils();
+
+it('misc-literal-edge-cases', async () => {
+  await fixtures.runFixtureTests([
+  "misc/literal-edge-cases-1.sql",
+  "misc/literal-edge-cases-2.sql",
+  "misc/literal-edge-cases-3.sql",
+  "misc/literal-edge-cases-4.sql",
+  "misc/literal-edge-cases-5.sql",
+  "misc/literal-edge-cases-6.sql",
+  "misc/literal-edge-cases-7.sql",
+  "misc/literal-edge-cases-8.sql",
+  "misc/literal-edge-cases-9.sql",
+  "misc/literal-edge-cases-10.sql",
+  "misc/literal-edge-cases-11.sql",
+  "misc/literal-edge-cases-12.sql",
+  "misc/literal-edge-cases-13.sql",
+  "misc/literal-edge-cases-14.sql",
+  "misc/literal-edge-cases-15.sql",
+  "misc/literal-edge-cases-16.sql",
+  "misc/literal-edge-cases-17.sql",
+  "misc/literal-edge-cases-18.sql",
+  "misc/literal-edge-cases-19.sql",
+  "misc/literal-edge-cases-20.sql",
+  "misc/literal-edge-cases-21.sql"
+]);
+});

packages/deparser/src/deparser.ts

@@ -1,4 +1,4 @@
-import { QuoteUtils } from '@pgsql/quotes';
+import { keywordKindOf, QuoteUtils } from '@pgsql/quotes';
 import { Node } from '@pgsql/types';
 import * as t from '@pgsql/types';
 
@@ -149,7 +149,7 @@ function isWrappedParseResult(obj: any): obj is { ParseResult: t.ParseResult } {
  */
 function formatBitString(bsval: string): string {
   const quote = (value: string) => value.replace(/'/g, "''");
-  if (/^x[0-9A-Fa-f]+$/.test(bsval)) {
+  if (/^x[0-9A-Fa-f]*$/.test(bsval)) {
     return `x'${bsval.substring(1)}'`;
   }
   if (bsval.startsWith('b')) {
@@ -158,6 +158,12 @@ function formatBitString(bsval: string): string {
   return `b'${quote(bsval)}'`;
 }
 
+// A dollar-quoted body ends at the first occurrence of its tag, including one
+// formed by the body's trailing characters plus the closing tag (`...$` + `$$`).
+function dollarTagConflicts(body: string, tag: string): boolean {
+  return (body + tag).indexOf(tag) < body.length;
+}
+
 const NUMERIC_LITERAL = /^[+-]?(?:0[xX](?:_?[0-9A-Fa-f])+|0[oO](?:_?[0-7])+|0[bB](?:_?[01])+|(?:\d(?:_?\d)*(?:\.(?:\d(?:_?\d)*)?)?|\.\d(?:_?\d)*)(?:[eE][+-]?\d(?:_?\d)*)?)$/;
 
 function formatInteger(ival: unknown): string {
@@ -239,10 +245,19 @@ export class Deparser implements DeparserVisitor {
    */
   private getFunctionDelimiter(body: string): string {
     const delimiter = this.options.functionDelimiter || '$$';
-    if (body.includes(delimiter)) {
-      return this.options.functionDelimiterFallback || '$EOFCODE$';
+    if (!dollarTagConflicts(body, delimiter)) {
+      return delimiter;
+    }
+    const fallback = this.options.functionDelimiterFallback || '$EOFCODE$';
+    if (!dollarTagConflicts(body, fallback)) {
+      return fallback;
+    }
+    const base = fallback.slice(0, -1);
+    let counter = 1;
+    while (dollarTagConflicts(body, `${base}${counter}$`)) {
+      counter++;
     }
-    return delimiter;
+    return `${base}${counter}$`;
   }
 
   /**
@@ -10739,7 +10754,9 @@ export class Deparser implements DeparserVisitor {
   private formatOptionValue(arg: any, argValue: any): any {
     if (arg && this.getNodeType(arg) === 'String') {
       const sval: string = this.getNodeData(arg).sval || '';
-      return /^[A-Za-z_][A-Za-z0-9_]*$/.test(sval) ? sval : QuoteUtils.escape(sval);
+      return /^[a-z_][a-z0-9_]*$/.test(sval) && keywordKindOf(sval) === 'NO_KEYWORD'
+        ? sval
+        : QuoteUtils.escape(sval);
     }
     return argValue;
   }