Handle pcpp relative path quirk

Author: virtuald

cxxheaderparser/preprocessor.py

@@ -3,6 +3,7 @@
 """
 
 import io
+import re
 import os
 from os.path import relpath
 import typing
@@ -36,7 +37,6 @@ def _filter_self(fname: str, fp: typing.TextIO) -> str:
     # isn't what a typical user of cxxheaderparser would want, so we strip out
     # the line directives and any content that isn't in our original file
 
-    # pcpp always emits line directives that match whatever is passed in to it
     line_ending = f'{fname}"\n'
 
     new_output = io.StringIO()
@@ -100,6 +100,18 @@ def _preprocess_file(filename: str, content: str) -> str:
         if retain_all_content:
             return fp.read()
         else:
+            # pcpp emits the #line directive using the filename you pass in
+            # but will rewrite it if it's on the include path it uses. This
+            # is copied from pcpp:
+            abssource = os.path.abspath(filename)
+            for rewrite in pp.rewrite_paths:
+                temp = re.sub(rewrite[0], rewrite[1], abssource)
+                if temp != abssource:
+                    filename = temp
+                    if os.sep != "/":
+                        filename = filename.replace(os.sep, "/")
+                    break
+
             return _filter_self(filename, fp)
 
     return _preprocess_file

tests/test_preprocessor.py

@@ -1,3 +1,4 @@
+import os
 import pathlib
 
 from cxxheaderparser.options import ParserOptions
@@ -66,6 +67,46 @@ def test_preprocessor_omit_content(tmp_path: pathlib.Path) -> None:
     )
 
 
+def test_preprocessor_omit_content2(tmp_path: pathlib.Path) -> None:
+    """
+    Ensure that content in other headers is omitted while handling pcpp
+    relative path quirk
+    """
+    h_content = '#include "t2.h"' "\n" "int x = X;\n"
+    h2_content = "#define X 2\n" "int omitted = 1;\n"
+
+    tmp_path2 = tmp_path / "l1"
+    tmp_path2.mkdir()
+
+    with open(tmp_path2 / "t1.h", "w") as fp:
+        fp.write(h_content)
+
+    with open(tmp_path2 / "t2.h", "w") as fp:
+        fp.write(h2_content)
+
+    options = ParserOptions(
+        preprocessor=make_pcpp_preprocessor(include_paths=[str(tmp_path)])
+    )
+
+    # Weirdness happens here
+    os.chdir(tmp_path)
+    data = parse_file(tmp_path2 / "t1.h", options=options)
+
+    assert data == ParsedData(
+        namespace=NamespaceScope(
+            variables=[
+                Variable(
+                    name=PQName(segments=[NameSpecifier(name="x")]),
+                    type=Type(
+                        typename=PQName(segments=[FundamentalSpecifier(name="int")])
+                    ),
+                    value=Value(tokens=[Token(value="2")]),
+                )
+            ]
+        )
+    )
+
+
 def test_preprocessor_encoding(tmp_path: pathlib.Path) -> None:
     """Ensure we can handle alternate encodings"""
     h_content = b"// \xa9 2023 someone\n" b'#include "t2.h"' b"\n" b"int x = X;\n"