configure_file: update \@ escape logic

Update meson logic to only handle matched pairs of \@ when
escaping variables in 'meson' and 'cmake@' style configure_file
invocations.

The previous approach, using a regex to handle a single  \@
to escape a variable has undesirable side effects,
including mangling valid perl that needs to be configured.

Closes: mesonbuild#7165

mesonbuild/utils/universal.py

@@ -1182,24 +1182,21 @@ def do_replacement(regex: T.Pattern[str], line: str,
                    variable_format: Literal['meson', 'cmake', 'cmake@'],
                    confdata: T.Union[T.Dict[str, T.Tuple[str, T.Optional[str]]], 'ConfigurationData']) -> T.Tuple[str, T.Set[str]]:
     missing_variables: T.Set[str] = set()
-    if variable_format == 'cmake':
-        start_tag = '${'
-        backslash_tag = '\\${'
-    else:
-        start_tag = '@'
-        backslash_tag = '\\@'
 
     def variable_replace(match: T.Match[str]) -> str:
-        # Pairs of escape characters before '@' or '\@'
+        # Pairs of escape characters before '@', '\@', '${' or '\${'
         if match.group(0).endswith('\\'):
             num_escapes = match.end(0) - match.start(0)
             return '\\' * (num_escapes // 2)
-        # Single escape character and '@'
-        elif match.group(0) == backslash_tag:
-            return start_tag
-        # Template variable to be replaced
+        # Handle cmake escaped \${} tags
+        elif variable_format == 'cmake' and match.group(0) == '\\${':
+            return '${'
+        # \@escaped\@ variables
+        elif match.groupdict().get('escaped') is not None:
+            return match.group('escaped')[1:-2]+'@'
         else:
-            varname = match.group(1)
+            # Template variable to be replaced
+            varname = match.group('variable')
             var_str = ''
             if varname in confdata:
                 var, _ = confdata.get(varname)
@@ -1280,11 +1277,23 @@ def get_cmake_define(line: str, confdata: 'ConfigurationData') -> str:
 
 def get_variable_regex(variable_format: Literal['meson', 'cmake', 'cmake@'] = 'meson') -> T.Pattern[str]:
     # Only allow (a-z, A-Z, 0-9, _, -) as valid characters for a define
-    # Also allow escaping '@' with '\@'
     if variable_format in {'meson', 'cmake@'}:
-        regex = re.compile(r'(?:\\\\)+(?=\\?@)|\\@|@([-a-zA-Z0-9_]+)@')
+        # Also allow escaping pairs of '@' with '\@'
+        regex = re.compile(r'''
+            (?:\\\\)+(?=\\?@)  # Matches multiple backslashes followed by an @ symbol
+            |                  # OR
+            @(?P<variable>[-a-zA-Z0-9_]+)@  # Matches a variable enclosed in @ symbols and captures the variable name
+            |                  # OR
+            (?P<escaped>\\@[-a-zA-Z0-9_]+\\@)  # Matches an escaped variable enclosed in @ symbols
+        ''', re.VERBOSE)
     else:
-        regex = re.compile(r'(?:\\\\)+(?=\\?\$)|\\\${|\${([-a-zA-Z0-9_]+)}')
+        regex = re.compile(r'''
+            (?:\\\\)+(?=\\?\$)  # Matches multiple backslashes followed by a dollar sign
+            |                  # OR
+            \\\${              # Matches a backslash followed by a dollar sign and an opening curly brace
+            |                  # OR
+            \${(?P<variable>[-a-zA-Z0-9_]+)}  # Matches a variable enclosed in curly braces and captures the variable name
+        ''', re.VERBOSE)
     return regex
 
 def do_conf_str(src: str, data: T.List[str], confdata: 'ConfigurationData',

test cases/common/14 configure file/config6.h.in

@@ -1,19 +1,16 @@
 /* No escape */
 #define MESSAGE1 "@var1@"
 
-/* Single escape means no replace */
-#define MESSAGE2 "\@var1@"
+/* Escaped whole variable */
+#define MESSAGE2 "\\@var1\\@"
 
 /* Replace pairs of escapes before '@' or '\@' with escape characters
  * (note we have to double number of pairs due to C string escaping)
  */
 #define MESSAGE3 "\\\\@var1@"
 
-/* Pairs of escapes and then single escape to avoid replace */
-#define MESSAGE4 "\\\\\@var1@"
-
-/* Check escaped variable does not overlap following variable */
-#define MESSAGE5 "\@var1@var2@"
+/* Pairs of escapes and then an escaped variable */
+#define MESSAGE4 "\\\\\@var1\@"
 
 /* Check escape character outside variables */
-#define MESSAGE6 "\\ @ \@ \\\\@ \\\\\@"
+#define MESSAGE5 "\\ @ \\\\@"

test cases/common/14 configure file/dummy-perl.pl.in

@@ -0,0 +1,8 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+
+my @array = qw(1 2);
+use Data::Dumper;
+say Dumper(\@array);

test cases/common/14 configure file/meson.build

@@ -297,6 +297,15 @@ configure_file(
   configuration: conf
 )
 
+# Do not mangle (e.g.) perl files when input legitimately contains '\@'
+configure_file(
+    input: 'dummy-perl.pl.in',
+    output: 'dummy-perl.pl',
+    configuration: conf,
+)
+run_command('cmp', files('dummy-perl.pl.in'),
+    join_paths(meson.current_build_dir(), 'dummy-perl.pl'), check: true)
+
 test('configure-file', test_file)
 
 # Dictionaries

test cases/common/14 configure file/prog6.c

@@ -1,11 +1,17 @@
 #include <string.h>
 #include <config6.h>
+#include <stdio.h>
 
 int main(void) {
+    fprintf(stderr, "MESSAGE1 %s\n", MESSAGE1);
+    fprintf(stderr, "MESSAGE2 %s\n", MESSAGE2);
+    fprintf(stderr, "MESSAGE3 %s\n", MESSAGE3);
+    fprintf(stderr, "MESSAGE4 %s\n", MESSAGE4);
+    fprintf(stderr, "MESSAGE5 %s\n", MESSAGE5);
+
     return strcmp(MESSAGE1, "foo")
         || strcmp(MESSAGE2, "@var1@")
         || strcmp(MESSAGE3, "\\foo")
         || strcmp(MESSAGE4, "\\@var1@")
-        || strcmp(MESSAGE5, "@var1bar")
-        || strcmp(MESSAGE6, "\\ @ @ \\@ \\@");
+        || strcmp(MESSAGE5, "\\ @ \\@");
 }

test cases/common/14 configure file/subdir/meson.build

@@ -35,4 +35,4 @@ configure_file(
   input: '../test.py.in',
   output: 'no_write_conflict.txt',
   configuration: conf
-)
+)