HeaderParser.pm - parse single quoted backslash properly

Karl gave me a test script which was breaking because we did not handle single quoted escapes like '\\' and '\n' and the like. This fixes the tokenizer pattern to handle these without dieing, although we do not validate that the escape sequence itself is legal, we no longer choke on having backslash something in the single quote. This includes tests for this, as well as for constructs like #if 1 which we had no tests for previously.
Perl · Feb 13, 2025 · 15bf1b6 · 15bf1b6
1 parent 280818b
commit 15bf1b6
Show file tree

Hide file tree

Showing 2 changed files with 102 additions and 1 deletion.
diff --git a/regen/HeaderParser.pm b/regen/HeaderParser.pm
@@ -115,7 +115,7 @@ BEGIN {
             (?<literal>
                 (?<define> defined\(\w+\) )
             |   (?<func>   \w+\s*\(\s*\w+(?:\s*,\s*\w+)*\s*\) )
-            |   (?<const>  (?:0x[a-fA-F0-9]+|\d+[LU]*|'.') )
+            |   (?<const>  (?:0x[a-fA-F0-9]+|\d+[LU]*|'.'|'\\.') )
             |   (?<sym>    \w+ )
             )
         |   (?<op> $binop_pat | $unop_pat )

diff --git a/t/porting/header_parser.t b/t/porting/header_parser.t
@@ -55,6 +55,11 @@ $hp->parse_text(<<~'EOF');
     #error #pragma
     #error #include
     #error #define
+    #if 1
+    #if CH == 'x' || CH == '\\'
+    #define CH_IS_IT
+    #endif
+    #endif
     EOF
 my $normal= $hp->lines_as_str();
 my $lines= $hp->lines();
@@ -353,6 +358,97 @@ is($lines_as_str,<<~'DUMP_EOF', "Simple data structure as expected") or show_tex
             "start_line_num" => 27,
             "sub_type" => "#error",
             "type" => "content"
+          }, 'HeaderLine' ),
+          bless( {
+            "cond" => [
+              [
+                1
+              ]
+            ],
+            "flat" => "#if 1",
+            "level" => 0,
+            "line" => "#if 1\n",
+            "n_lines" => 1,
+            "raw" => "#if 1\n",
+            "source" => "(buffer)",
+            "start_line_num" => 28,
+            "sub_type" => "#if",
+            "type" => "cond"
+          }, 'HeaderLine' ),
+          bless( {
+            "cond" => [
+              [
+                1
+              ],
+              [
+                "CH == '\\\\' || CH == 'x'"
+              ]
+            ],
+            "flat" => "#if CH == '\\\\' || CH == 'x'",
+            "level" => 1,
+            "line" => "# if CH == '\\\\' || CH == 'x'\n",
+            "n_lines" => 1,
+            "raw" => "#if CH == 'x' || CH == '\\\\'\n",
+            "source" => "(buffer)",
+            "start_line_num" => 29,
+            "sub_type" => "#if",
+            "type" => "cond"
+          }, 'HeaderLine' ),
+          bless( {
+            "cond" => [
+              [
+                1
+              ],
+              [
+                "CH == '\\\\' || CH == 'x'"
+              ]
+            ],
+            "flat" => "#define CH_IS_IT",
+            "level" => 2,
+            "line" => "#   define CH_IS_IT\n",
+            "n_lines" => 1,
+            "raw" => "#define CH_IS_IT\n",
+            "source" => "(buffer)",
+            "start_line_num" => 30,
+            "sub_type" => "#define",
+            "type" => "content"
+          }, 'HeaderLine' ),
+          bless( {
+            "cond" => [
+              [
+                1
+              ],
+              [
+                "CH == '\\\\' || CH == 'x'"
+              ]
+            ],
+            "flat" => "#endif",
+            "inner_lines" => 2,
+            "level" => 1,
+            "line" => "# endif\n",
+            "n_lines" => 1,
+            "raw" => "#endif\n",
+            "source" => "(buffer)",
+            "start_line_num" => 31,
+            "sub_type" => "#endif",
+            "type" => "cond"
+          }, 'HeaderLine' ),
+          bless( {
+            "cond" => [
+              [
+                1
+              ]
+            ],
+            "flat" => "#endif",
+            "inner_lines" => 4,
+            "level" => 0,
+            "line" => "#endif\n",
+            "n_lines" => 1,
+            "raw" => "#endif\n",
+            "source" => "(buffer)",
+            "start_line_num" => 32,
+            "sub_type" => "#endif",
+            "type" => "cond"
           }, 'HeaderLine' )
         ];
         DUMP_EOF
@@ -383,6 +479,11 @@ is($normal,<<~'EOF',"Normalized text as expected");
     #error #pragma
     #error #include
     #error #define
+    #if 1
+    # if CH == '\\' || CH == 'x'
+    #   define CH_IS_IT
+    # endif
+    #endif
     EOF
 
 {