initial commit into repository0.0

1 files changed, 594 insertions, 0 deletions

diff --git a/lib/sitecing_parser.ll b/lib/sitecing_parser.ll
new file mode 100644
index 0000000..6cb78f3
--- a/dev/null
+++ b/lib/sitecing_parser.ll
@@ -0,0 +1,594 @@
+%{
+ /*
+  * XXX: I have a strong feeling that this parser should be completely rewritten.
+  */
+#include <iostream>
+#include <fstream>
+#include <cassert>
+#include <stdexcept>
+using namespace std;
+#include "sitecing/sitecing_util.h"
+#include "sitecing/sitecing_exception.h"
+using namespace sitecing;
+#define sitecing_parser_flexlexer_once
+#include "sitecing/sitecing_parser.h"
+#include "sitecing/sitecing_enflesher.h"
+#undef yyFlexLexer
+#define yyFlexLexer sitecing_parserFlexLexer
+%}
+%x SLASHSTAR_COMMENT SLASHSLASH_COMMENT STRING
+%x CODELINE CLASSLINE DECLLINE IMPLLINE DECLBLOCK IMPLBLOCK VARLINE VARINIT
+%x IMPORTLINE IMPORTCOMPONENT
+%x IMPORTTYPELINE IMPORTTYPECOMPONENT
+%x DERIVELINE DERIVECOMPONENT
+%x CONSTRUCTOR DESTRUCTOR CODEMETHODLINE CODEMETHODARGS
+%x CODEMETHODBLOCK INLINE METHODLINE METHODARGS METHODBLOCK CODEBLOCK OUTPUTBLOCK
+%option 8bit c++ verbose noyywrap yyclass="sitecing_parser" prefix="sitecing_parser" stack yylineno
+
+WHITESPACE	[ \t]
+ID		[A-Za-z_][A-Za-z0-9_]*
+NOIDCHAR	[^A-Za-z0-9_]
+
+%%
+
+<INITIAL>{
+    ^\%\%class{WHITESPACE}+	{
+	// TODO: signal error if we already have class name acquired from source.
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments|modus_operandi::flag_devour_whitespace));
+	BEGIN(CLASSLINE);
+    }
+    ^\%\%decl{WHITESPACE}+	{
+	modi.push_front(modus_operandi(0));
+	anchor();
+	BEGIN(DECLLINE);
+    }
+    ^\%\%impl{WHITESPACE}+	{
+	modi.push_front(modus_operandi(0));
+	anchor();
+	BEGIN(IMPLLINE);
+    }
+    \<\%decl\>			{
+	modi.push_front(modus_operandi(0));
+	anchor();
+	BEGIN(DECLBLOCK);
+    }
+    \<\%impl\>			{
+	modi.push_front(modus_operandi(0));
+	anchor();
+	BEGIN(IMPLBLOCK);
+    }
+    ^\%\%var{WHITESPACE}+	{
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	anchor();
+	BEGIN(VARLINE);
+    }
+    ^\%\%import{WHITESPACE}+	{
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	BEGIN(IMPORTLINE);
+    }
+    ^\%\%import_type{WHITESPACE}+ {
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	BEGIN(IMPORTTYPELINE);
+    }
+    ^\%\%derive{WHITESPACE}+	{
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	BEGIN(DERIVELINE);
+    }
+    \<\%constructor\>	{
+	modi.push_front(modus_operandi());
+	anchor();
+	BEGIN(CONSTRUCTOR);
+    }
+    \<\%destructor\>	{
+	modi.push_front(modus_operandi());
+	anchor();
+	BEGIN(DESTRUCTOR);
+    }
+    \<\%codemethod{WHITESPACE}+	{
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	anchor();
+	BEGIN(CODEMETHODLINE);
+    }
+    \<\%method{WHITESPACE}+ {
+	modi.push_front(modus_operandi(modus_operandi::flag_devour_comments));
+	anchor();
+	BEGIN(METHODLINE);
+    }
+    <<EOF>>	{
+	assert(modi.size()==1);
+	M().modify(modus_operandi::modus_preop);
+	LexerOutput(";",1);
+	return 0;
+    }
+}
+<<EOF>>	throw preprocessor_error(CODEPOINT,"unexpected end of file",lineno());
+
+<CODEBLOCK,CODEMETHODBLOCK>{
+    "<%output>"	{
+	anchor();
+	yy_push_state(OUTPUTBLOCK);
+    }
+}
+
+<METHODLINE>{
+    {WHITESPACE}+	{
+			    modus_operandi& m = modi.front();
+			    if(!m.output.empty()) {
+				if(!m._lastid.empty()) {
+				    if(!m._type.empty()) m._type += ' ';
+				    m._type += m._lastid;
+				}
+				m._lastid = m.output;
+				m.output.clear();
+			    }
+			}
+    \*	{
+		    modus_operandi& m = modi.front();
+		    ECHO;
+		    if(!m._lastid.empty()) {
+			if(!m._type.empty()) m._type += ' ';
+			m._type += m._lastid;
+		    }
+		    m._lastid = m.output;
+		    m.output.clear();
+		}
+    \(	{
+	modus_operandi& m = modi.front();
+	if(m.output.empty()) {
+	    m._name=m._lastid;
+	}else{
+	    if(!m._lastid.empty()) { // XXX: lastid, I believe should never be emtpy...
+		if(!m._type.empty()) m._type += ' ';
+		m._type += m._lastid;
+	    }
+	    m._name = m.output;
+	    m.output.clear();
+	}
+	ECHO;
+	BEGIN(METHODARGS);
+    }
+}
+<METHODARGS>{
+    \%\>	{
+	modus_operandi& m = modi.front();
+	m._args = m.output;
+	m.output.clear();
+	anchor();
+	BEGIN(METHODBLOCK);
+    }
+}
+
+<INITIAL,METHODBLOCK,OUTPUTBLOCK>{
+    \<\%{WHITESPACE}+	{
+	M().modify(modus_operandi::modus_postop);
+	anchor();
+	LexerOutput("(",1);
+	yy_push_state(INLINE);
+    }
+    ^\%{WHITESPACE}	{
+	M().modify(modus_operandi::modus_code);
+	anchor();
+	yy_push_state(CODELINE);
+    }
+    \<\%code\>	{
+	M().modify(modus_operandi::modus_code);
+	anchor();
+	yy_push_state(CODEBLOCK);
+    }
+    "</%output>" {
+	if(YY_START!=OUTPUTBLOCK) throw preprocessor_error(CODEPOINT,"unexpected tag",lineno());
+	M().modify(modus_operandi::modus_code);
+	anchor();
+	yy_pop_state();
+    }
+}
+
+<INLINE>\%\>	LexerOutput(")",1); M().modus=modus_operandi::modus_preop; yy_pop_state();
+<CODELINE>\n	yy_pop_state();
+
+<CODEMETHODLINE>{
+    {WHITESPACE}+	{
+			    modus_operandi& m = modi.front();
+			    if(!m.output.empty()) {
+				if(!m._lastid.empty()) {
+				    if(!m._type.empty()) m._type += ' ';
+				    m._type += m._lastid;
+				}
+				m._lastid = m.output;
+				m.output.clear();
+			    }
+			}
+    \*	{
+		    modus_operandi& m = modi.front();
+		    ECHO;
+		    if(!m._lastid.empty()) {
+			if(!m._type.empty()) m._type += ' ';
+			m._type += m._lastid;
+		    }
+		    m._lastid = m.output;
+		    m.output.clear();
+		}
+    \(	{
+	modus_operandi& m = modi.front();
+	if(m.output.empty()) {
+	    m._name=m._lastid;
+	}else{
+	    if(!m._lastid.empty()) { // XXX: lastid, I believe should never be emtpy...
+		if(!m._type.empty()) m._type += ' ';
+		m._type += m._lastid;
+	    }
+	    m._name = m.output;
+	    m.output.clear();
+	}
+	ECHO;
+	BEGIN(CODEMETHODARGS);
+    }
+}
+<CODEMETHODARGS>{
+    \%\>	{
+	modus_operandi& m = modi.front();
+	m._args = m.output;
+	m.output.clear();
+	m.flags=0;
+	anchor();
+	BEGIN(CODEMETHODBLOCK);
+    }
+}
+
+<IMPORTLINE>{
+    {WHITESPACE}+	{ }
+    {ID}	{
+		    if(!modi.front()._name.empty())
+			throw preprocessor_error(CODEPOINT,"syntax error",lineno());
+		    modi.front()._name = yytext;
+		}
+    \=		{
+	modi.front().output.clear();
+	BEGIN(IMPORTCOMPONENT);
+    }
+}
+<IMPORTCOMPONENT>{
+    {WHITESPACE}+	{ }
+    \n	{
+	modus_operandi& m = M();
+	string::size_type t = m.output.find_first_not_of(" \t");
+	if(t!=string::npos)
+	    m.output.erase(0,t);
+	t = m.output.find_last_not_of(" \t;");
+	if(t!=string::npos)
+	    m.output.erase(t+1);
+	if(m.output[0]=='"' && m.output[m.output.length()-1]=='"') {
+	    m.output.erase(0,1);
+	    m.output.erase(m.output.length()-1);
+	}
+	string c = combine_path(component_basename,m.output);
+	member_variables.push_back(member_variable(m._type,m._name,normalize_path(c,strip_leading_slash),true));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+}
+
+<IMPORTTYPELINE>{
+    {WHITESPACE}+	{ }
+    {ID}	{
+		    if(!modi.front()._name.empty())
+			throw preprocessor_error(CODEPOINT,"syntax error",lineno());
+		    modi.front()._name = yytext;
+		}
+    \=		{
+	modi.front().output.clear();
+	BEGIN(IMPORTTYPECOMPONENT);
+    }
+}
+<IMPORTTYPECOMPONENT>{
+    {WHITESPACE}+	{ }
+    \n	{
+	modus_operandi& m = M();
+	string::size_type t = m.output.find_first_not_of(" \t");
+	if(t!=string::npos)
+	    m.output.erase(0,t);
+	t = m.output.find_last_not_of(" \t;");
+	if(t!=string::npos)
+	    m.output.erase(t+1);
+	if(m.output[0]=='"' && m.output[m.output.length()-1]=='"') {
+	    m.output.erase(0,1);
+	    m.output.erase(m.output.length()-1);
+	}
+	string c = combine_path(component_basename,m.output);
+	member_variables.push_back(member_variable(m._type,m._name,normalize_path(c,strip_leading_slash),true,true));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+}
+
+<DERIVELINE>{
+    {WHITESPACE}+	{ }
+    {ID}	{
+		    if(!modi.front()._name.empty())
+			throw preprocessor_error(CODEPOINT,"syntax_error",lineno());
+		    modi.front()._name = yytext;
+		}
+    \=		{
+	modi.front().output.clear();
+	BEGIN(DERIVECOMPONENT);
+    }
+}
+<DERIVECOMPONENT>{
+    {WHITESPACE}+	{ }
+    \n {
+	modus_operandi& m = M();
+	string::size_type t = m.output.find_first_not_of(" \t");
+	if(t!=string::npos)
+	    m.output.erase(0,t);
+	t = m.output.find_last_not_of(" \t;");
+	if(t!=string::npos)
+	    m.output.erase(t+1);
+	if(m.output[0]=='"' && m.output[m.output.length()-1]=='"') {
+	    m.output.erase(0,1);
+	    m.output.erase(m.output.length()-1);
+	}
+	string c = combine_path(component_basename,m.output);
+	ancestor_classes.push_back(ancestor_class(m._name,normalize_path(c,strip_leading_slash)));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+}
+
+<VARLINE>{
+    {WHITESPACE}+	{
+	modus_operandi& m = modi.front();
+	if(!m.output.empty()) {
+	    if(!m._lastid.empty()) {
+		if(!m._type.empty()) m._type += ' ';
+		m._type += m._lastid;
+	    }
+	    m._lastid = m.output;
+	    m.output.clear();
+	}
+    }
+    \*	{
+		    modus_operandi& m = modi.front();
+		    ECHO;
+		    if(!m._lastid.empty()) {
+			if(!m._type.empty()) m._type += ' ';
+			m._type += m._lastid;
+		    }
+		    m._lastid = m.output;
+		    m.output.clear();
+		}
+    \;|\n|\=	{
+	modus_operandi& m = modi.front();
+	if(m.output.empty()) {
+	    m._name=m._lastid;
+	}else{
+	    if(!m._lastid.empty()) { // XXX: lastid should never be emtpy, I believe?
+		if(!m._type.empty()) m._type += ' ';
+		m._type += m._lastid;
+	    }
+	    m._name=m.output;
+	    m.output.clear();
+	}
+	BEGIN(VARINIT);
+	if(*yytext!='=')
+	    unput('\n');
+    }
+}
+<VARINIT>{
+    \n	{
+	modus_operandi& m = modi.front();
+	string::size_type t = m.output.find_first_not_of(" \t");
+	if(t!=string::npos)
+	    m.output.erase(0,t);
+	t = m.output.find_last_not_of(" \t;");
+	if(t!=string::npos)
+	    m.output.erase(t+1);
+	member_variables.push_back(member_variable(m._type,m._name,m.output));
+	if(!m.output.empty())
+	    have_initializers=true;
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+}
+<DECLLINE>\n	{
+    ECHO;
+    decl += modi.front().output;
+    modi.pop_front();
+    BEGIN(INITIAL);
+}
+<IMPLLINE>\n	{
+    ECHO;
+    impl += modi.front().output;
+    modi.pop_front();
+    BEGIN(INITIAL);
+}
+<CLASSLINE>\n	{
+    class_name = modi.front().output;
+    modi.pop_front();
+    BEGIN(INITIAL);
+}
+<CLASSLINE,DECLLINE,IMPLLINE,VARLINE,VARINIT,IMPORTLINE,IMPORTCOMPONENT,CODEMETHODLINE,CODEMETHODARGS,INLINE,METHODLINE,METHODARGS,DECLBLOCK,IMPLBLOCK,CONSTRUCTOR,DESTRUCTOR,CODEMETHODBLOCK,CODELINE,CODEBLOCK>{
+    "/*"	{
+	yy_push_state(SLASHSTAR_COMMENT);
+	if(!M().devour_comments()) {
+	    ECHO;
+	}
+    }
+    "//"	{
+	yy_push_state(SLASHSLASH_COMMENT);
+	if(!M().devour_comments()) {
+	    ECHO;
+	}
+    }
+    \"		{
+	yy_push_state(STRING);
+    	ECHO;
+    }
+    \'\\.\'	{
+	ECHO;
+    }
+}
+
+<INITIAL,METHODBLOCK,OUTPUTBLOCK>{
+    \"	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\\"",2);
+    \n	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\n",2);
+    \r	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\r",2);
+    \t	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\t",2);
+    \b	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\b",2);
+    \a	soft_anchor(); M().modify(modus_operandi::modus_text); LexerOutput("\\a",2);
+    .	soft_anchor(); M().modify(modus_operandi::modus_text); ECHO;
+    {WHITESPACE}+	soft_anchor(); M().modify(modus_operandi::modus_text); ECHO;
+}
+
+<DECLBLOCK,IMPLBLOCK,CONSTRUCTOR,DESTRUCTOR,CODEMETHODBLOCK,METHODBLOCK,CODEBLOCK>{
+    \<\/\%decl\>	{
+	if(YY_START!=DECLBLOCK) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	decl += modi.front().output;
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/\%impl\>	{
+	if(YY_START!=IMPLBLOCK) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	impl += modi.front().output;
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/\%constructor\>	{
+	if(YY_START!=CONSTRUCTOR) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	member_functions.push_back(member_function("","","",modi.front().output));
+	have_constructor = true;
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/\%destructor\>	{
+	if(YY_START!=DESTRUCTOR) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	member_functions.push_back(member_function("","~","",modi.front().output));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/\%codemethod\>	{
+	if(YY_START!=CODEMETHODBLOCK) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	modus_operandi& m = modi.front();
+	member_functions.push_back(member_function(m._type,m._name,m._args,m.output));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/%method\> {
+	if(YY_START!=METHODBLOCK) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	modus_operandi& m = modi.front();
+	m.modify(modus_operandi::modus_code);
+	member_functions.push_back(member_function(m._type,m._name,m._args,m.output));
+	modi.pop_front();
+	BEGIN(INITIAL);
+    }
+    \<\/%code\> {
+	if(YY_START!=CODEBLOCK) throw preprocessor_error(CODEPOINT,"tags mismatch",lineno());
+	yy_pop_state();
+    }
+    \n		ECHO;
+}
+
+<SLASHSTAR_COMMENT>{
+    "*/"	{
+	if(!M().devour_comments()) {
+	    ECHO;
+	}
+	yy_pop_state();
+	unput(' ');
+    }
+    \n	{
+	if(!M().devour_comments()) {
+	    ECHO;
+	}
+    }
+}
+<SLASHSLASH_COMMENT>{
+    \n	{
+	if(!M().devour_comments()) {
+	    ECHO;
+	}
+	yy_pop_state();
+	if(YY_START!=CODEBLOCK && YY_START!=CODEMETHODBLOCK && YY_START!=IMPLBLOCK && YY_START!=DECLBLOCK)
+	    unput('\n');
+    }
+}
+<SLASHSTAR_COMMENT,SLASHSLASH_COMMENT>.	{
+    if(!M().devour_comments()) {
+	ECHO;
+    }
+}
+<STRING>{
+    \\.	ECHO;
+    \"	ECHO; yy_pop_state();
+    .	ECHO;
+}
+
+{WHITESPACE}+	{
+    if(!(M().flags&modus_operandi::flag_devour_whitespace)) {
+	ECHO;
+    }
+}
+
+%%
+
+sitecing_parser::sitecing_parser(component_factory& f)
+    : factory(f), have_initializers(false), have_constructor(false),
+    base_class("sitecing::cgi_component"),
+    base_header("sitecing/cgi_component.h"),
+    skeleton(__SC_DEFAULT_SKELETON) {
+    }
+
+void sitecing_parser::preprocess(const string& in) {
+    ifstream ifs(in.c_str(),ios::in);
+    if(!ifs.good())
+	throw preprocessor_error(CODEPOINT,"failed to open input file");
+    input_file = in;
+    modi.push_front(modus_operandi(0));
+    switch_streams(&ifs,NULL);
+    if(yylex())
+	throw preprocessor_error(CODEPOINT,"unknown error");
+    member_functions.push_back(member_function("void","main","(int _magic,va_list _args)",M().output));
+    if(have_initializers && !have_constructor)
+	member_functions.push_back(member_function("","","",""));
+    sitecing_enflesher enflesher(*this);
+    enflesher.enflesh();
+}
+
+void sitecing_parser::LexerOutput(const char* buf,int size) {
+    assert(modi.size());
+    M().output.append(buf,size);
+}
+
+static const char *modus_transitions
+    [sitecing_parser::modus_operandi::modi]
+    [sitecing_parser::modus_operandi::modi] = {
+// To:
+// code   preop               postop                text                         From:
+ { "",    "(*(__SCIF->out))", "(*(__SCIF->out))<<", "(*(__SCIF->out))<<\"" }, // code
+ { ";",   "",                 "<<",                 "<<\""                 }, // preop
+ { NULL,  NULL,               "",                   "\""                   }, // postop
+ { "\";", "\"",               "\"<<",               ""                     }  // text
+};
+
+void sitecing_parser::modus_operandi::modify(modus_t m) {
+    const char * x = modus_transitions[modus][m];
+    assert(x);
+    output += x;
+    modus = m;
+}
+
+void sitecing_parser::soft_anchor() {
+    if(M().modus!=modus_operandi::modus_text)
+	anchor();
+}
+void sitecing_parser::anchor() {
+    if(M().modus==modus_operandi::modus_text)
+	M().modify(modus_operandi::modus_preop);
+    M().output += "\n#line ";
+    char tmp[7];
+    snprintf(tmp,sizeof(tmp),"%d",lineno());
+    M().output += tmp;
+    M().output += " \"";
+    M().output += input_file;
+    M().output += "\"\n";
+}
+/* vim:set ft=lex: */