From fce0f4e95f65e06a44c1f37253163f5094cbf7eb Mon Sep 17 00:00:00 2001
From: Sasha Lopoukhine <superlopuh@gmail.com>
Date: Tue, 2 Jul 2024 16:49:15 +0100
Subject: [PATCH] add --xdsl-anonymize option mutually exclusive with
 --mlir-anonymize (#2)

Sometimes the names carry semantic meaning, so it's useful to check that
they are preserved.
---
 filecheckize/main.py      | 19 +++++++++++++++----
 tests/filecheck/mlir.mlir | 23 +++++++++++++++++++++++
 2 files changed, 38 insertions(+), 4 deletions(-)

diff --git a/filecheckize/main.py b/filecheckize/main.py
index ad8b181..6c5714e 100644
--- a/filecheckize/main.py
+++ b/filecheckize/main.py
@@ -18,11 +18,17 @@ def main():
         nargs="?",
         help="Input file to read. Defaults to standard input.",
     )
-    parser.add_argument(
+    group = parser.add_mutually_exclusive_group()
+    group.add_argument(
         "--mlir-anonymize",
         action="store_true",
         help="Anonymize MLIR SSA value names and basic block names.",
     )
+    group.add_argument(
+        "--xdsl-anonymize",
+        action="store_true",
+        help="Anonymize MLIR unnamed SSA values and basic block names.",
+    )
     parser.add_argument(
         "--check-empty-lines",
         action="store_true",
@@ -49,6 +55,7 @@ def main():
     args = parser.parse_args(sys.argv[1:])
 
     comment_line = re.compile(rf"^\s*{re.escape(args.comments_prefix)}.*$")
+    unnamed_ssa_value = re.compile(r"%([\d]+)")
     ssa_value_name = re.compile(r"%([\d]+|[\w$._-][\w\d$._-]*)(:|#[\d]*)?")
     basic_block_name = re.compile(r"\^([\d]+|[\w$._-][\w\d$._-]*)")
 
@@ -73,9 +80,13 @@ def main():
             if re.match(comment_line, line):
                 continue
 
-        if args.mlir_anonymize:
-            # Anonymize SSA value names
-            line = re.sub(ssa_value_name, r"%{{.*}}", line)
+        if args.mlir_anonymize or args.xdsl_anonymize:
+            if args.mlir_anonymize:
+                # Anonymize SSA value names
+                line = re.sub(ssa_value_name, r"%{{.*}}", line)
+            elif args.xdsl_anonymize:
+                # Anonymize unnamed SSA values
+                line = re.sub(unnamed_ssa_value, r"%{{.*}}", line)
 
             # Anonymize basic blocks names
             line = re.sub(basic_block_name, r"^{{.*}}", line)
diff --git a/tests/filecheck/mlir.mlir b/tests/filecheck/mlir.mlir
index 5c2e1c8..b739c86 100644
--- a/tests/filecheck/mlir.mlir
+++ b/tests/filecheck/mlir.mlir
@@ -1,26 +1,49 @@
 // RUN: filecheckize %s --strip-comments | filecheck %s --match-full-lines --check-prefix STRIP
 // RUN: filecheckize %s --strip-comments --check-empty-lines | filecheck %s --check-prefix WITH-EMPTY --match-full-lines
 // RUN: filecheckize %s --strip-comments --mlir-anonymize | filecheck %s --check-prefix ANONYMIZE --match-full-lines
+// RUN: filecheckize %s --strip-comments --xdsl-anonymize | filecheck %s --check-prefix XDSL-ANONYMIZE --match-full-lines
 
 func.func @arg_rec(%0 : !test.type<"int">) -> !test.type<"int"> {
     %1 = func.call @arg_rec(%0) : (!test.type<"int">) -> !test.type<"int">
+    %name = arith.constant : i32
+    %other_name = arith.constant : i32
+    %2 = arith.addi %name, %other_name : i32
     func.return %1 : !test.type<"int">
 }
 
 // STRIP:      // CHECK:      func.func @arg_rec(%0 : !test.type<"int">) -> !test.type<"int"> {
 // STRIP-NEXT: // CHECK-NEXT:     %1 = func.call @arg_rec(%0) : (!test.type<"int">) -> !test.type<"int">
+// STRIP-NEXT: // CHECK-NEXT:     %name = arith.constant : i32
+// STRIP-NEXT: // CHECK-NEXT:     %other_name = arith.constant : i32
+// STRIP-NEXT: // CHECK-NEXT:     %2 = arith.addi %name, %other_name : i32
 // STRIP-NEXT: // CHECK-NEXT:     func.return %1 : !test.type<"int">
 // STRIP-NEXT: // CHECK-NEXT: }
 
 // WITH-EMPTY:       // CHECK-EMPTY:
 // WITH-EMPTY-NEXT:  // CHECK-NEXT:  func.func @arg_rec(%0 : !test.type<"int">) -> !test.type<"int"> {
 // WITH-EMPTY-NEXT:  // CHECK-NEXT:      %1 = func.call @arg_rec(%0) : (!test.type<"int">) -> !test.type<"int">
+// WITH-EMPTY-NEXT:  // CHECK-NEXT:      %name = arith.constant : i32
+// WITH-EMPTY-NEXT:  // CHECK-NEXT:      %other_name = arith.constant : i32
+// WITH-EMPTY-NEXT:  // CHECK-NEXT:      %2 = arith.addi %name, %other_name : i32
 // WITH-EMPTY-NEXT:  // CHECK-NEXT:      func.return %1 : !test.type<"int">
 // WITH-EMPTY-NEXT:  // CHECK-NEXT:  }
 // WITH-EMPTY-NEXT:  // CHECK-EMPTY:
 // WITH-EMPTY-NEXT:  // CHECK-EMPTY:
+// WITH-EMPTY-NEXT:  // CHECK-EMPTY:
+// WITH-EMPTY-NEXT:  // CHECK-EMPTY:
 
 // ANONYMIZE:       // CHECK:       func.func @arg_rec(%{{.*}} : !test.type<"int">) -> !test.type<"int"> {
 // ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = func.call @arg_rec(%{{.*}}) : (!test.type<"int">) -> !test.type<"int">
+// ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = arith.constant : i32
+// ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = arith.constant : i32
+// ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = arith.addi %{{.*}}, %{{.*}} : i32
 // ANONYMIZE-NEXT:  // CHECK-NEXT:      func.return %{{.*}} : !test.type<"int">
 // ANONYMIZE-NEXT:  // CHECK-NEXT:  }
+
+// XDSL-ANONYMIZE:       // CHECK:       func.func @arg_rec(%{{.*}} : !test.type<"int">) -> !test.type<"int"> {
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = func.call @arg_rec(%{{.*}}) : (!test.type<"int">) -> !test.type<"int">
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:      %name = arith.constant : i32
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:      %other_name = arith.constant : i32
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:      %{{.*}} = arith.addi %name, %other_name : i32
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:      func.return %{{.*}} : !test.type<"int">
+// XDSL-ANONYMIZE-NEXT:  // CHECK-NEXT:  }