Merge pull request #2 from hydradatabase/jd/incorrect-results-in-table

fix bug using with tables

.github/workflows/build_and_test.yml

@@ -1,10 +1,18 @@
 name: pg_stringtheory Build and Test
 on:
   push:
-  pull_request:
+    paths:
+      - '.github/**'
+      - 'src/**'
+      - 'expected/**'
+      - 'sql/**'
+      - '*.control'
+      - '*.sql'
+      - Makefile
 jobs:
   build-and-test:
     strategy:
+      fail-fast: false
       matrix:
         os: [ubuntu-latest]
         version: [REL_14_STABLE, REL_15_STABLE, REL_16_STABLE]

.gitignore

@@ -2,5 +2,5 @@
 *.o
 .clangd
 
-results
-
+/results
+/regression.*

expected/equality.out

@@ -1,4 +1,4 @@
-CREATE EXTENSION stringtheory;
+CREATE EXTENSION IF NOT EXISTS stringtheory;
 -- no match
 SELECT stringtheory.equals('hello', 'world');
  equals 
@@ -27,3 +27,25 @@ SELECT stringtheory.equals('123456', '12345');
  f
 (1 row)
 
+-- test equality in a CTE
+WITH a AS (SELECT md5(generate_series(1, 1000)::text) b)
+SELECT count(*) FROM a
+WHERE stringtheory.equals(b, md5('123'));
+ count 
+-------
+     1
+(1 row)
+
+-- test strstr in a table
+CREATE TEMPORARY TABLE stringtheory_test
+(a text);
+INSERT INTO stringtheory_test
+SELECT md5(generate_series(1,1000)::text);
+SELECT COUNT(*) FROM stringtheory_test
+WHERE stringtheory.equals(a, md5('123'));
+ count 
+-------
+     1
+(1 row)
+
+DROP TABLE stringtheory_test;

expected/strstr.out

@@ -1,5 +1,5 @@
-CREATE EXTENSION stringtheory;
-ERROR:  extension "stringtheory" already exists
+CREATE EXTENSION IF NOT EXISTS stringtheory;
+NOTICE:  extension "stringtheory" already exists, skipping
 -- no match
 SELECT stringtheory.strstr('hello', 'world');
  strstr 
@@ -42,3 +42,25 @@ SELECT stringtheory.strstr('ello', 'hello world');
      -1
 (1 row)
 
+-- test strstr in a CTE
+WITH a AS (SELECT md5(generate_series(1, 1000)::text) b)
+SELECT COUNT(*) FROM a
+WHERE stringtheory.strstr(b, '00') >= 0;
+ count 
+-------
+   114
+(1 row)
+
+-- test strstr in a table
+CREATE TEMPORARY TABLE stringtheory_test
+(a text);
+INSERT INTO stringtheory_test
+SELECT md5(generate_series(1,1000)::text);
+SELECT COUNT(*) FROM stringtheory_test
+WHERE stringtheory.strstr(a, '00') >= 0;
+ count 
+-------
+   114
+(1 row)
+
+DROP TABLE stringtheory_test;

sql/equality.sql

@@ -1,4 +1,4 @@
-CREATE EXTENSION stringtheory;
+CREATE EXTENSION IF NOT EXISTS stringtheory;
 
 -- no match
 SELECT stringtheory.equals('hello', 'world');
@@ -11,3 +11,20 @@ SELECT stringtheory.equals('1234567890123456', '1234567890123456');
 
 -- no match when partial
 SELECT stringtheory.equals('123456', '12345');
+
+-- test equality in a CTE
+WITH a AS (SELECT md5(generate_series(1, 1000)::text) b)
+SELECT count(*) FROM a
+WHERE stringtheory.equals(b, md5('123'));
+
+-- test strstr in a table
+CREATE TEMPORARY TABLE stringtheory_test
+(a text);
+
+INSERT INTO stringtheory_test
+SELECT md5(generate_series(1,1000)::text);
+
+SELECT COUNT(*) FROM stringtheory_test
+WHERE stringtheory.equals(a, md5('123'));
+
+DROP TABLE stringtheory_test;

sql/strstr.sql

@@ -1,4 +1,4 @@
-CREATE EXTENSION stringtheory;
+CREATE EXTENSION IF NOT EXISTS stringtheory;
 
 -- no match
 SELECT stringtheory.strstr('hello', 'world');
@@ -17,3 +17,20 @@ SELECT stringtheory.strstr('hello world', 'ello');
 
 -- haystack in needle not found
 SELECT stringtheory.strstr('ello', 'hello world');
+
+-- test strstr in a CTE
+WITH a AS (SELECT md5(generate_series(1, 1000)::text) b)
+SELECT COUNT(*) FROM a
+WHERE stringtheory.strstr(b, '00') >= 0;
+
+-- test strstr in a table
+CREATE TEMPORARY TABLE stringtheory_test
+(a text);
+
+INSERT INTO stringtheory_test
+SELECT md5(generate_series(1,1000)::text);
+
+SELECT COUNT(*) FROM stringtheory_test
+WHERE stringtheory.strstr(a, '00') >= 0;
+
+DROP TABLE stringtheory_test;

src/text.cpp

@@ -2,6 +2,8 @@ extern "C" {
 #include "postgres.h"
 
 #include "fmgr.h"
+#include "mb/pg_wchar.h"
+
 #if PG_VERSION_NUM >= 160000
 #include "varatt.h"
 #endif
@@ -73,7 +75,8 @@ Datum pg_strstr(PG_FUNCTION_ARGS) {
   }
 
   /* Get the results from the simd functions. */
-  size_t ret = fast_strstr(VARDATA(left), len_left, VARDATA(right), len_right);
+  size_t ret =
+      fast_strstr(VARDATA_ANY(left), len_left, VARDATA_ANY(right), len_right);
 
   PG_RETURN_INT32(ret);
 }
@@ -98,7 +101,8 @@ Datum pg_equals(PG_FUNCTION_ARGS) {
   }
 
   /* Get the results from the simd functions. */
-  size_t ret = fast_strstr(VARDATA(left), len_left, VARDATA(right), len_right);
+  size_t ret =
+      fast_strstr(VARDATA_ANY(left), len_left, VARDATA_ANY(right), len_right);
 
   /* If the result is 0, strings are equal. */
   PG_RETURN_BOOL(ret == 0);