Source code positions after CPP macro expansion

Question

I have the following Haskell file that contains a CPP macro.

#define FOO( x, y ) x ++ y

foo =
  FOO("a",
       "b"
       ++ "c")

bar = 1

When I call cpphs on this file, I get the following result.

#line 1 "Test.hs"


foo =
  "a" ++ "b"
       ++ "c"

bar = 1

The line directive states that the line that is following the directive is line 1 of Test.hs. Therefore, foo is still positioned at line 3 while bar is now positioned at line 7, while it has been in line 8 in the original file. It looks like all whitespace tokens after the comma and before the first non-whitespace token are dropped, which loses the original line break. I have the following questions.

Is this an expected behavior?
Is there any option that I overlooked to insert a line directive to correctly fix the location of bar?
Is there any other preprocessor that would preserve the location of bar by using a line directive or some other means?

I am not able to change the content of the file in any way.

EDIT:

To answer 3. myself, it looks like clang and gcc add additional line directives as I would have expected for cpphs as well. They yield the following output.

# 1 "<stdin>"
# 1 "<built-in>" 1
# 1 "<built-in>" 3
# 417 "<built-in>" 3
# 1 "<command line>" 1
# 1 "<built-in>" 2
# 1 "<stdin>" 2


foo =
   "a" ++ 
      
# 4 "<stdin>"
  "b"
        
# 4 "<stdin>"
  ++ "c"



bar = 1

The additional line directives correctly pin bar to line 8 of the file.

EDIT:

To answer 1. myself, as long as I do not miss some option that fixes this behavior, it looks like it is not an expected behavior.

If I use the following variant of the example module

#define FOO( x, y ) x ++ y

foo =
  FOO( "a",
      "b"
        ++ "c"
     )


bar :: String
bar = 1

and invoke ghc -cpp -pgmP cpphs -optP --cpp Test.hs I get the following error.

Test.hs:9:7: error:
    • No instance for (Num String) arising from the literal ‘1’
    • In the expression: 1
      In an equation for ‘bar’: bar = 1
  |
9 | 
  |       ^

That is, using cpphs GHC is not able to correctly display the position of the type error.

"While foo is still positioned in line 3" - I see the foo is on the line 4. — 3CEZVQ
– 3CEZVQ, Commented Nov 29, 2024 at 20:59
Sorry, the word "positioned" is misleading here. The line directive resets the line count. Ich have edited the question to clarify this aspect. — Jan Christiansen
– Jan Christiansen, Commented Nov 30, 2024 at 5:58
Why is that important? It smells like the XY-problem. A real C preprocessor expands macros into single lines and you would get "a" ++ "b" ++ "c". Moreover, it joins blank lines. So, this is unexpected in the C standard. — 3CEZVQ
– 3CEZVQ, Commented Nov 30, 2024 at 19:24
I would like to highlight identifiers in the original source file. The source code positions of these identifiers are extracted from an abstract syntax tree, therefore, I have to preprocess the file first. I was under the impression that line directives are used to correctly identify entities in the original file, therefore, I would have expected, that there is a way to correctly extract the source code position of bar in the original file from the preprocessed file. — Jan Christiansen
– Jan Christiansen, Commented Nov 30, 2024 at 20:34

3CEZVQ · Accepted Answer · 2024-11-30 20:49:09Z

1

You have to write own simple preprocessor for input files, your pre-processor will insert line directives before each line.

Input:

#define FOO( x, y ) x ++ y

foo =
  FOO("a",
       "b"
       ++ "c")

bar = 1

Output:

#line 1
#define FOO( x, y ) x ++ y
#line 2

#line 3
foo =
#line 4
  FOO("a",
#line 5
       "b"
#line 6
       ++ "c")
#line 7

#line 8
bar = 1

Then this output can be passed to cpphs. You will get what you want but quite verbose.

#line 1 "Test.hs"

#line 2 "Test.hs"

#line 3 "Test.hs"
foo =
#line 4 "Test.hs"
#line 5 "Test.hs"
  "a" ++ "b"
#line 6 "Test.hs"
       ++ "c"
#line 7 "Test.hs"

#line 8 "Test.hs"
bar = 1

answered Nov 30, 2024 at 20:49

3CEZVQ

42.9k11 gold badges92 silver badges100 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Source code positions after CPP macro expansion

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related