From 7545e153d43dd652cbb863d946d27ad046999870 Mon Sep 17 00:00:00 2001
From: Anton Lydike <me@antonlydike.de>
Date: Tue, 2 Jul 2024 18:25:47 +0100
Subject: [PATCH] fix CHECK-NEXT to consume as few matches as possible

---
 README.md                   |  1 +
 filecheck/compiler.py       |  2 --
 filecheck/finput.py         | 17 ++++++++++++-----
 filecheck/matcher.py        |  4 ++--
 tests/filecheck/simple.test |  6 ++++++
 5 files changed, 21 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index 1dc0828..d4ce483 100644
--- a/README.md
+++ b/README.md
@@ -52,6 +52,7 @@ There are some features that are left out for now (e.g. [pseudo-numeric variable
   - [ ] MLIR/xDSL integration tests
 - **UX:**
   - [ ] Good error messages: I have some error messages, but could be a lot better
+  - [ ] Error messages for malformed regexes
 - **Infrastructure:**
   - [X] Formatting: black
   - [X] Pyright: Almost passes rn
diff --git a/filecheck/compiler.py b/filecheck/compiler.py
index f05f5fa..2aa9419 100644
--- a/filecheck/compiler.py
+++ b/filecheck/compiler.py
@@ -22,8 +22,6 @@ def compile_uops(
     expr: list[str] = []
     if check.name == "NEXT":
         expr.append(r"\n?[^\n]*")
-    elif check.name == "SAME":
-        expr.append("[^\n]*")
     elif check.name == "EMPTY":
         return CHECK_EMPTY_EXPR, dict()
 
diff --git a/filecheck/finput.py b/filecheck/finput.py
index 2fcd686..3812e6f 100644
--- a/filecheck/finput.py
+++ b/filecheck/finput.py
@@ -61,15 +61,22 @@ def match(self, pattern: re.Pattern[str]) -> re.Match[str] | None:
         """
         Match (exactly from the current position)
         """
-        print(f"matching on {pattern}")
+        print(f"matching on r{repr(pattern.pattern)}")
         return pattern.match(self.content, pos=self.pos)
 
-    def find(self, pattern: re.Pattern[str]) -> re.Match[str] | None:
+    def find(
+        self, pattern: re.Pattern[str], this_line: bool = False
+    ) -> re.Match[str] | None:
         """
         Find the first occurance of a pattern, might be far away.
+
+        If this_line is given, match only until the next newline.
         """
-        print(f'searching for r"{pattern.pattern}"')
-        return pattern.search(self.content, pos=self.pos)
+        print(f"searching for r{repr(pattern.pattern)}")
+        endpos = self.content.find("\n", self.pos) if this_line else -1
+        if endpos == -1:
+            endpos = sys.maxsize
+        return pattern.search(self.content, pos=self.pos, endpos=endpos)
 
     def find_between(
         self, pattern: re.Pattern[str], start: int, end: int
@@ -77,7 +84,7 @@ def find_between(
         """
         Find the first occurance of a pattern, might be far away.
         """
-        print(f"searching for {pattern} in input[{start}:{end}]")
+        print(f"searching for r{repr(pattern.pattern)} in input[{start}:{end}]")
         return pattern.search(self.content, pos=start, endpos=end)
 
     def print_line_with_current_pos(self, pos_override: int | None = None):
diff --git a/filecheck/matcher.py b/filecheck/matcher.py
index 1baf8ed..f4fd6ab 100644
--- a/filecheck/matcher.py
+++ b/filecheck/matcher.py
@@ -60,7 +60,7 @@ def run(self) -> int:
             "NOT": self.enqueue_not,
             "EMPTY": self.check_empty,
             "NEXT": self.match_immediately,
-            "SAME": self.match_immediately,
+            "SAME": self.match_eventually,
             "LABEL": self.check_label,
             "CHECK": self.match_eventually,
         }
@@ -204,7 +204,7 @@ def match_eventually(self, op: CheckOp):
         Uses file.find
         """
         pattern, repl = compile_uops(op, self.ctx.live_variables, self.opts)
-        if match := self.file.find(pattern):
+        if match := self.file.find(pattern, op.name == "SAME"):
             print(
                 f"matched: {op.check_line_repr()}, groups={match.groups()} mapping to {repl}"
             )
diff --git a/tests/filecheck/simple.test b/tests/filecheck/simple.test
index 26abf42..0282f4a 100644
--- a/tests/filecheck/simple.test
+++ b/tests/filecheck/simple.test
@@ -17,3 +17,9 @@ SOME B C
 
 // CHECK-NOT: 404
 suffix
+
+test A val1 val2 val1 val2 val1
+// CHECK: test A
+// CHECK-SAME: val2
+// CHECK-SAME: val1
+// CHECK-SAME: val2