Hi All,
I don't know exactly what's the technical name of what I'm doing, but look at the following code:
float *pFloat;
float **pFloat2;
float (*pFloatPar)[3];
int main()
{
pFloat;
pFloat2;
pFloat
}
I've declared pFloatPar as a double array in a way that the compiler knows about the second dimension size and then I can write pFloatPar[2][1].
However, the actual code completion doesn't recognize the declaration (see the attached screenshot). My assumption is that CC cannot correctly parse the parenthesis...
I have put your code in our parsertest project, and the result looks below:
--------------M-a-i-n--L-o-g--------------
000001. ParserDummy() : ParserBase() : Instantiation of Parser object.
000002. InitTokenizer() : m_Filename='test.h', m_FileSize=108.
000003. Init() : m_Filename='test.h'
000004. test.h
000005. Parse() : Parsing 'test.h'
000006. DoParse() : Loop:m_Str='', token='float'
000007. DoParse() : Loop:m_Str='float ', token='*'
000008. DoParse() : Loop:m_Str='float ', token='pFloat'
000009. DoAddToken() : Created token='pFloat', file_idx=1, line=1, ticket=
000010. GetActualTokenType() : Searching within m_Str='float'
000011. GetActualTokenType() : Compensated m_Str='float'
000012. GetActualTokenType() : Found 'float'
000013. DoAddToken() : Prepending ''
000014. DoAddToken() : Added/updated token 'pFloat' (0), kind 'variable', type 'float*', actual 'float'. Parent is (-1)
000015. DoParse() : Loop:m_Str='float', token=';'
000016. DoParse() : Loop:m_Str='', token='float'
000017. DoParse() : Loop:m_Str='float ', token='*'
000018. DoParse() : Loop:m_Str='float ', token='*'
000019. DoParse() : Loop:m_Str='float ', token='pFloat2'
000020. DoAddToken() : Created token='pFloat2', file_idx=1, line=2, ticket=
000021. GetActualTokenType() : Searching within m_Str='float'
000022. GetActualTokenType() : Compensated m_Str='float'
000023. GetActualTokenType() : Found 'float'
000024. DoAddToken() : Prepending ''
000025. DoAddToken() : Added/updated token 'pFloat2' (1), kind 'variable', type 'float**', actual 'float'. Parent is (-1)
000026. DoParse() : Loop:m_Str='float', token=';'
000027. DoParse() : Loop:m_Str='', token='float'
000028. ReadParentheses(): (* pFloatPar), line=3
000029. DoParse() : Loop:m_Str='', token=';'
000030. DoParse() : Loop:m_Str='', token='int'
000031. DoParse() : Loop:m_Str='int ', token='main'
000032. ReadParentheses(): (), line=5
000033. HandleFunction() : Adding function 'main': m_Str='int '
000034. HandleFunction() : name='main', args='()', peek='{'
000035. HandleFunction() : !(Ctor/Dtor) 'main', m_Str='int ', localParent='<none>'
000036. HandleFunction() : Adding function 'main', ': m_Str='int ', enc_ns='nil'.
000037. HandleFunction() : Add token name='main', args='()', return type='int '
000038. GetBaseArgs() : args='()'.
000039. GetBaseArgs() : baseArgs='()'.
000040. DoAddToken() : Created token='main', file_idx=1, line=5, ticket=
000041. GetActualTokenType() : Searching within m_Str='int'
000042. GetActualTokenType() : Compensated m_Str='int'
000043. GetActualTokenType() : Found 'int'
000044. DoAddToken() : Prepending ''
000045. DoAddToken() : Added/updated token 'main' (2), kind 'function', type 'int', actual 'int'. Parent is (-1)
000046. ParserDummy() : ~ParserBase() : Destruction of Parser object.
--------------T-r-e-e--L-o-g--------------
000047. float* pFloat [1,0]
000048. float** pFloat2 [2,0]
000049. main() [5,5]
--------------L-i-s-t--L-o-g--------------
000050. variable float* pFloat [1,0]
000051. variable float** pFloat2 [2,0]
000052. function int main() [5,5]
Yes. The parser just skip the "[3]".
The proposed fix is try to peek the token after the right parenthesis, and if it is a "[", then we can guess that is is a second level pointer.
Quote from: ollydbg on November 15, 2011, 10:17:44 AM
Yes. The parser just skip the "[3]".
Even if the parser does so, the result should be
float *pFloatPar; and the parser should treat it as a pointer to a float. This would, of course, be wrong, but it seems the parser is just dropping this symbol completely.
Quote from: ollydbg on November 15, 2011, 10:17:44 AM
The proposed fix is try to peek the token after the right parenthesis, and if it is a "[", then we can guess that is is a second level pointer.
That sounds good to me, however I'm not that familiar with the CC plugin to try a patch, sorry... ;-)
Quote from: daniloz on November 15, 2011, 02:20:40 PM
Quote from: ollydbg on November 15, 2011, 10:17:44 AM
The proposed fix is try to peek the token after the right parenthesis, and if it is a "[", then we can guess that is is a second level pointer.
That sounds good to me, however I'm not that familiar with the CC plugin to try a patch, sorry... ;-)
I think I will take some time to fix it. :D
Quote from: ollydbg on November 15, 2011, 02:47:40 PM
I think I will take some time to fix it. :D
That'd be great... Thanks! 8)
This is a dirty hack on this
Index: E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/parserthread.cpp
===================================================================
--- E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/parserthread.cpp (revision 7561)
+++ E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/parserthread.cpp (working copy)
@@ -506,11 +506,11 @@
m_LastToken.Clear();
m_LastUnnamedTokenName.Clear();
- // Notice: clears the queue "m_EncounteredTypeNamespaces"
+ // Notice: clear the queue "m_EncounteredTypeNamespaces"
while (!m_EncounteredTypeNamespaces.empty())
m_EncounteredTypeNamespaces.pop();
- // Notice: clears the queue "m_EncounteredNamespaces"
+ // Notice: clear the queue "m_EncounteredNamespaces"
while (!m_EncounteredNamespaces.empty())
m_EncounteredNamespaces.pop();
@@ -942,7 +942,8 @@
int pos = peek.find(ParserConsts::ptr);
if (pos != wxNOT_FOUND)
{
- if (m_Tokenizer.PeekToken().GetChar(0) == ParserConsts::opbracket_chr)
+ peek = m_Tokenizer.PeekToken();
+ if (peek.GetChar(0) == ParserConsts::opbracket_chr)
{
arg.Trim(true).RemoveLast();
//wxString token = arg.Mid(pos+1,)
@@ -953,6 +954,18 @@
HandleFunction(arg);
}
}
+ else if (peek.GetChar(0) == ParserConsts::semicolon_chr)
+ {
+ // we meet such mode:
+ // AAA (*BBB)[ccc];
+ // The "[ccc]" was skipped by Tokenizer, so we meet the semicolon
+ m_Str << token<<_T("*");
+ //strip the "*" from arg.
+ arg.Trim(true).RemoveLast();
+ arg.Remove(0, pos+1); //remove the heading "*"
+ if (m_Options.handleVars)
+ DoAddToken(tkVariable, arg, m_Tokenizer.GetLineNumber());
+ }
}
else // wxString arg = m_Tokenizer.GetToken(); // eat args ()
m_Str = token + arg;
@@ -2646,6 +2659,7 @@
else if (tmp==ParserConsts::gt)
{
--nestLvl;
+ m_TemplateArgument.Trim();
m_TemplateArgument << tmp;
}
else if (tmp==ParserConsts::semicolon)
@@ -2658,7 +2672,7 @@
else if (tmp.IsEmpty())
break;
else
- m_TemplateArgument << tmp;
+ m_TemplateArgument << tmp <<_T(" ");
if (nestLvl <= 0)
break;
}
Well, the log is below:
Quote
--------------M-a-i-n--L-o-g--------------
000001. ParserDummy() : ParserBase() : Instantiation of Parser object.
000002. InitTokenizer() : m_Filename='test.h', m_FileSize=108.
000003. Init() : m_Filename='test.h'
000004. test.h
000005. Parse() : Parsing 'test.h'
000006. DoParse() : Loop:m_Str='', token='float'
000007. DoParse() : Loop:m_Str='float ', token='*'
000008. DoParse() : Loop:m_Str='float ', token='pFloat'
000009. DoAddToken() : Created token='pFloat', file_idx=1, line=1, ticket=
000010. GetActualTokenType() : Searching within m_Str='float'
000011. GetActualTokenType() : Compensated m_Str='float'
000012. GetActualTokenType() : Found 'float'
000013. DoAddToken() : Prepending ''
000014. DoAddToken() : Added/updated token 'pFloat' (0), kind 'variable', type 'float*', actual 'float'. Parent is (-1)
000015. DoParse() : Loop:m_Str='float', token=';'
000016. DoParse() : Loop:m_Str='', token='float'
000017. DoParse() : Loop:m_Str='float ', token='*'
000018. DoParse() : Loop:m_Str='float ', token='*'
000019. DoParse() : Loop:m_Str='float ', token='pFloat2'
000020. DoAddToken() : Created token='pFloat2', file_idx=1, line=2, ticket=
000021. GetActualTokenType() : Searching within m_Str='float'
000022. GetActualTokenType() : Compensated m_Str='float'
000023. GetActualTokenType() : Found 'float'
000024. DoAddToken() : Prepending ''
000025. DoAddToken() : Added/updated token 'pFloat2' (1), kind 'variable', type 'float**', actual 'float'. Parent is (-1)
000026. DoParse() : Loop:m_Str='float', token=';'
000027. DoParse() : Loop:m_Str='', token='float'
000028. ReadParentheses(): (* pFloatPar), line=3
000029. DoAddToken() : Created token=' pFloatPar', file_idx=1, line=3, ticket=
000030. GetActualTokenType() : Searching within m_Str='float*'
000031. GetActualTokenType() : Compensated m_Str='float*'
000032. GetActualTokenType() : Found 'float'
000033. DoAddToken() : Prepending ''
000034. DoAddToken() : Added/updated token ' pFloatPar' (2), kind 'variable', type 'float*', actual 'float'. Parent is (-1)
000035. DoParse() : Loop:m_Str='float*', token=';'
000036. DoParse() : Loop:m_Str='', token='int'
000037. DoParse() : Loop:m_Str='int ', token='main'
000038. ReadParentheses(): (), line=5
000039. HandleFunction() : Adding function 'main': m_Str='int '
000040. HandleFunction() : name='main', args='()', peek='{'
000041. HandleFunction() : !(Ctor/Dtor) 'main', m_Str='int ', localParent='<none>'
000042. HandleFunction() : Adding function 'main', ': m_Str='int ', enc_ns='nil'.
000043. HandleFunction() : Add token name='main', args='()', return type='int '
000044. GetBaseArgs() : args='()'.
000045. GetBaseArgs() : baseArgs='()'.
000046. DoAddToken() : Created token='main', file_idx=1, line=5, ticket=
000047. GetActualTokenType() : Searching within m_Str='int'
000048. GetActualTokenType() : Compensated m_Str='int'
000049. GetActualTokenType() : Found 'int'
000050. DoAddToken() : Prepending ''
000051. DoAddToken() : Added/updated token 'main' (3), kind 'function', type 'int', actual 'int'. Parent is (-1)
000052. ParserDummy() : ~ParserBase() : Destruction of Parser object.
--------------T-r-e-e--L-o-g--------------
000053. float* pFloat [1,0]
000054. float** pFloat2 [2,0]
000055. float* pFloatPar [3,0]
000056. main() [5,5]
--------------L-i-s-t--L-o-g--------------
000057. variable float* pFloat [1,0]
000058. variable float** pFloat2 [2,0]
000059. variable float* pFloatPar [3,0]
000060. function int main() [5,5]
...I got another one (a vector) not working properly for me:
#include <vector>
class AClass
{
public:
AClass() {;}
~AClass() {;}
void MyMethod() {;}
};
class AnotherClass
{
public:
AnotherClass() {;}
~AnotherClass() {;}
inline const std::vector<AClass*> GetAClassVec()
{
return mvpAClass;
}
private:
std::vector<AClass*> mvpAClass;
};
int main()
{
AnotherClass a;
a. // does not show GetAClassVec
}
Look, it seems the function does not parse correctly. See the screen shot below
(http://i683.photobucket.com/albums/vv194/ollydbg_cb/2011-11-18221725.png)
OK, we arrived in this if clause.
else if (peek==ParserConsts::dcolon)
{
if (m_Str.IsEmpty())
m_EncounteredTypeNamespaces.push(token); // it's a type's namespace
else
m_EncounteredNamespaces.push(token);
m_Tokenizer.GetToken(); // eat ::
}
Here, I just parse the third line:
class AnotherClass
{
inline const std::vector<AClass*> GetAClassVec() {;}
};
Now:
token = "std"
peek = "::"
m_Str = "const "
The logic is:
We put the "std" as the m_EncounteredNamespaces,
but the correct way is: "std" should be put in m_EncounteredTypeNamespaces.
One possible way is: we should skip the "const" (do not put "const" into m_Str, as m_Str is like a returned type stack), so that when we peek a "::", we see the m_Str is empty, and "std" will put into m_EncounteredTypeNamespaces.
OK, can you understand my explanation???
Quote from: ollydbg on November 18, 2011, 03:39:46 PM
We put the "std" as the m_EncounteredNamespaces,
but the correct way is: "std" should be put in m_EncounteredTypeNamespaces.
OK, can you understand my explanation???
Yes, but I think removing const wouldn't be wise because you wanna know, don't you.
What about doing something like (untested, blind meta-code):
else if (peek==ParserConsts::dcolon)
{
if ( m_Str.IsEmpty()
|| m_Str.Trim(true).Trim(false).IsSameAs(ParserConsts::const)
|| m_Str.Trim(true).Trim(false).IsSameAs(ParserConsts::volatile) ) // what else?!
m_EncounteredTypeNamespaces.push(token); // it's a type's namespace
else
m_EncounteredNamespaces.push(token);
m_Tokenizer.GetToken(); // eat ::
}
Quote from: MortenMacFly on November 18, 2011, 09:23:30 PM
What about doing something like (untested, blind meta-code):
(...answering myself...:)
Raises immediately the question, what about such things then:
inline const mynamespace::anothernamespace::std::vector<AClass*> GetAClassVec() {;}
???
Quote from: MortenMacFly on November 18, 2011, 09:28:58 PM
Quote from: MortenMacFly on November 18, 2011, 09:23:30 PM
What about doing something like (untested, blind meta-code):
(...answering myself...:)
Raises immediately the question, what about such things then:
inline const mynamespace::anothernamespace::std::vector<AClass*> GetAClassVec() {;}
It works OK, because the next time we enter the if clause, the have the same m_Str == const.
else if (peek==ParserConsts::dcolon)
{
if ( m_Str.IsEmpty()
|| m_Str.Trim(true).Trim(false).IsSameAs(ParserConsts::kw_const)
|| m_Str.Trim(true).Trim(false).IsSameAs(ParserConsts::kw_volatile) ) // what else?!
m_EncounteredTypeNamespaces.push(token); // it's a type's namespace
else
m_EncounteredNamespaces.push(token);
m_Tokenizer.GetToken(); // eat ::
}