[pypy-commit] cffi cpy-extension: Remove comments.

[arigo]

Author: Armin Rigo <arigo at tunes.org>
Branch: cpy-extension
Changeset: r330:7caa8ac321f2
Date: 2012-06-14 13:16 +0200
http://bitbucket.org/cffi/cffi/changeset/7caa8ac321f2/

Log:	Remove comments.

diff --git a/cffi/cparser.py b/cffi/cparser.py
--- a/cffi/cparser.py
+++ b/cffi/cparser.py
@@ -2,6 +2,7 @@
 from . import api, model
 import pycparser, weakref, re
 
+_r_comment = re.compile(r"/\*.*?\*/|//.*?$", re.DOTALL | re.MULTILINE)
 _r_partial_enum = re.compile(r"\.\.\.\s*\}")
 _r_enum_dotdotdot = re.compile(r"__dotdotdot\d+__$")
 _parser_cache = None
@@ -13,12 +14,19 @@
     return _parser_cache
 
 def _preprocess(csource):
+    # Remove comments
+    csource = _r_comment.sub(' ', csource)
+    # Replace "...}" with "__dotdotdotNUM__}".  This construction should
+    # occur only at the end of enums; at the end of structs we have "...;}"
+    # and at the end of vararg functions "...);"
     matches = list(_r_partial_enum.finditer(csource))
     for number, match in enumerate(reversed(matches)):
         p = match.start()
         assert csource[p:p+3] == '...'
         csource = '%s __dotdotdot%d__ %s' % (csource[:p], number,
                                              csource[p+3:])
+    # Replace all remaining "..." with the same name, "__dotdotdot__",
+    # which is declared with a typedef for the purpose of C parsing.
     return csource.replace('...', ' __dotdotdot__ ')
 
 class Parser(object):
diff --git a/testing/test_parsing.py b/testing/test_parsing.py
--- a/testing/test_parsing.py
+++ b/testing/test_parsing.py
@@ -122,3 +122,20 @@
     type = ffi._parser.parse_type("fn_t")
     BType = ffi._get_cached_btype(type)
     assert BType == '<func (<pointer to <int>>), <int>, False>'
+
+def test_remove_comments():
+    ffi = FFI(backend=FakeBackend())
+    ffi.cdef("""
+        double /*comment here*/ sin   // blah blah
+        /* multi-
+           line-
+           //comment */  (
+        // foo
+        double // bar      /* <- ignored, because it's in a comment itself
+        x, double/*several*//*comment*/y) /*on the same line*/
+        ;
+    """)
+    m = ffi.rawload("m")
+    func = m.sin
+    assert func.name == 'sin'
+    assert func.BType == '<func (<double>, <double>), <double>, False>'