CComment

On Sun, Feb 20, 2005 at 01:56:05PM +0100, Bernhard Fischer wrote:
>On Sat, Feb 19, 2005 at 08:05:59PM -0500, Stefan Seefeld wrote:
>>Bernhard Fischer wrote:
>>>Hi Stefan,
>>>
[snip]
>>to check in a new python frontend that calls the Cxx parser in C mode
>>and as soon as that is confirmed to work retire the ctool backend.
>>Let me know if you are interested in testing it, or want to help in
>>other ways.
>
>Yes, i'm interrested in testing it.
>Currently i am perusing linux/drivers/infiniband/ulp/sdp as my primary
>testcase (available in recent kernels).
>Another testcase is here: http://members.aon.at/berny_f/synopsis/
>>
>>>Attached is a quick C style comment parser. Before applying it, one
>>>should make the intermittent asterisk optional, though.....
>>
>>That looks good !
>>You are right about the asterisk. In fact, I would add the restriction
>>that the asterisks have to be aligned, i.e.
>>
>>/*
>> * ...
>> */
>>
>>would count, but
>>
>>/* comment showing that
>>   *foobar is the dereferenced foobar.
>>*/
>>
>>would not. The asterisk being optional should be expressed using
>
>Indeed.

Attached is a revised patch to add CComment which handles the tiny
testcase (http://members.aon.at/berny_f/synopsis/) proper for me.
>
>>the usual 'Parameter' idiom used everywhere else in the synopsis
>>frontend.
>
>Ok, i'll try to use that.

>From a short glance, i don't quite see how i'd do that. Could you
elaborate or point me to an example, please?

>
>>Would you mind adjusting your patch ?

I've got a question regarding the implementation of CPrevious.
Previous is recognized by means of the '<' character at pos 0 in the
comment text. CPrevious OTOH does not have the '<' character but rather
is a comment which is positioned after some code (See struct bar in
foo.h contained in the tiny testcase referenced above).

Would i implement CPrevious via
o putting a '<' into the comment when CComment finds code before the
  comment? This may be considered a rough hack but may allow me to
  simply reuse Previous instead of introducing a new CPrevious..
o matching only ^/[\*]+... as CComment and introducing a CPrevious which
  associates ^(^/[\*]+)... to the respective code.

_______________________________________________
Synopsis-devel mailing list
[email protected]
http://lists.fresco.org/cgi-bin/listinfo/synopsis-devel

diff -X /usr/src/cvs/excl -rdup Synopsis.oorig/Synopsis/Processors/Comments/Comments.py Synopsis/Synopsis/Processors/Comments/Comments.py
--- Synopsis.oorig/Synopsis/Processors/Comments/Comments.py	2004-11-18 16:34:16.000000000 +0100
+++ Synopsis/Synopsis/Processors/Comments/Comments.py	2005-03-01 14:34:15.000000000 +0100
@@ -84,6 +84,49 @@ class SSDComments(CommentProcessor):
 
       return string.join(self.re_ssd.findall(str),'\n')
 
+class CComments (CommentProcessor):
+   """A class that formats C /* */ style comments"""
+
+   __re_c = r"/[\*]+[ \t]*(?P<text>.*)(?P<lines>(\n[ \t]*.*)*?)(\n[ \t]*)?[\*]+/"
+   # match lines with and without asterisk.
+   # preserve space after asterisk so empty lines are formatted correctly.
+   __re_line = r"\n[ \t]*([ \t]*[\*]+(?=[ \t\n]))*(?P<text>.*)"
+
+   def __init__(self):
+      "Compiles the regular expressions"
+
+      if self.debug:
+          print 'CComment Processor requested.\n'
+      self.re_c = re.compile(CComments.__re_c)
+      self.re_line = re.compile(CComments.__re_line)
+
+   def visit_comments(self, decl):
+      "Calls process_comment on all comments"
+
+      map(self.process_comment, decl.comments())
+
+   def process_comment(self, comment):
+      """Finds comments in the C format. The format is  /* ... */.
+      It has to cater for all five line forms: "/* ...", " * ...", " ...",
+      " */" and the one-line "/* ... */".
+      """
+
+      text = comment.text()
+      text_list = []
+      mo = self.re_c.search(text)
+      while mo:
+         text_list.append(mo.group('text'))
+         lines = mo.group('lines')
+         if lines:
+            mol = self.re_line.search(lines)
+            while mol:
+               text_list.append(mol.group('text'))
+               mol = self.re_line.search(lines, mol.end())
+         mo = self.re_c.search(text, mo.end())
+      text = string.join(text_list,'\n')
+      comment.set_text(text)
+
+
 class JavaComments (CommentProcessor):
    """A class that formats java /** style comments"""