parser.cc

Go to the documentation of this file.

/* Copyright (C) 2000, 2001 Stefan Buehler <sbuehler@uni-bremen.de>
 
   This program is free software; you can redistribute it and/or modify it
   under the terms of the GNU General Public License as published by the
   Free Software Foundation; either version 2, or (at your option) any
   later version.
 
   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.
 
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
   USA. */
 
#include "arts.h"
#include <map>
#include "messages.h"
#include "exceptions.h"
#include "file.h"
#include "auto_wsv.h"
#include "methods.h"
#include "parser.h"
#include "wsv_aux.h"
 
void SourceText::AppendFile(const String& name) 
{
  mSfLine.push_back(mText.nelem());
  mSfName.push_back(name);
 
  read_text_from_file(mText, name);    
}
 
void SourceText::AdvanceChar() 
{
  if ( mColumn < mText[mLine].nelem()-1 )
    {
      ++mColumn;
    }
  else
    {
      mLineBreak = true;
      do
        {
          if (mLine>=mText.nelem()-1)
            {
              throw Eot( "",
                         this->File(),
                         this->Line(),
                         this->Column() ); 
            }
          else
            {
              ++mLine;
              mColumn = 0;
            }
        }
      while ( 1 > mText[mLine].nelem() ); // Skip empty lines.
    }
}
 
 
void SourceText::AdvanceLine() 
{
  mLineBreak = true;
  mColumn = 0;
  do
    {
      if (mLine>=mText.nelem()-1)
        {
          throw Eot( "",
                     this->File(),
                     this->Line(),
                     this->Column() ); 
        }
      else
        {
          ++mLine;
        }
    }
  while ( 1 > mText[mLine].nelem() ); // Skip empty lines.
}
 
 
const String& SourceText::File()
{
  Index i    = 0;
  bool   stop = false;
 
  while ( i<mSfLine.nelem()-1 && !stop )
    {
      if (mLine>=mSfLine[i+1]) ++i;
      else                     stop = true;
    }
 
  return mSfName[i];
}
 
 
Index SourceText::Line()
{
  Index i    = 0;
  bool   stop = false;
 
  while ( i<mSfLine.nelem()-1 && !stop )
    {
      if (mLine>=mSfLine[i+1]) ++i;
      else                     stop = true;
    }
 
  return mLine - mSfLine[i] + 1; 
}
 
 
void SourceText::Init()
{
  mLine   = 0;
  mColumn = 0;
    
  if ( 1 > mText.nelem() )
    {
      throw Eot( "Empty text!",
                 this->File(),
                 this->Line(),
                 this->Column() ); 
    }
  else
    {
      // Skip empty lines:
      while ( 1 > mText[mLine].nelem() )
        {
          if (mLine>=mText.nelem()-1)
            {
              throw Eot( "",
                         this->File(),
                         this->Line(),
                         this->Column() ); 
            }
          else
            {
              mLineBreak = true;
              ++mLine;
            }
        }
    }
}
 
 
ostream& operator << (ostream& os, const SourceText& text)
{
  for (Index i=0; i<text.mText.nelem();++i)
    cout << i
         << "(" << text.mText[i].nelem() << ")"
         << ": " << text.mText[i] << '\n';
  return(os);
}
 
 
bool is_whitespace(const char c)
{
  switch (c) 
    {
    case ' ':
    case '\r':
    case '\t':
    case '#':
      return true;
      break;
    default:
      return false;
      break;
    }
}
 
void eat_whitespace(SourceText& text)
{
  char dummy;
 
  while (is_whitespace(dummy=text.Current())) 
    {
      switch (dummy)
        {
        case ' ':
        case '\r':
        case '\t':
          text.AdvanceChar();
          break;
        case '#':
          text.AdvanceLine();
          break;
        default:
          {
            ostringstream os;
            os << "Expected whitespace, but got `" << dummy << "'.";
            throw UnexpectedChar( os.str(),
                                  text.File(),
                                  text.Line(),
                                  text.Column() ); 
            break;
          }
        }
    }
}
 
 
void read_name(String& name, SourceText& text)
{
  bool stop = false;
  name = "";
 
  while (!stop) 
    {
      char dummy = text.Current();
 
      if ( isalnum(dummy) || '_'==dummy )
        {
          name += dummy;
          // AdvanceChar sets LineBreak if a line break occured.
          text.LineBreak() = false;
          text.AdvanceChar();
          if ( text.LineBreak() ) stop = true;
        }
      else
        {
          stop = true;
        }
    }
 
  //  cout << "Name: " << name << '\n';
}
 
void assertain_character(char c, SourceText& text)
{
  if ( c != text.Current() )
    {
      ostringstream os;
      os << "Expected `" << c << "', but got `" << text.Current() << "'.";
      throw UnexpectedChar( os.str(),
                            text.File(),
                            text.Line(),
                            text.Column() ); 
    }
  
  text.AdvanceChar();
}
 
void parse_String(String& res, SourceText& text)
{
  bool stop = false;
  res = "";
 
  text.LineBreak() = false;
  assertain_character('"',text);
  if ( text.LineBreak() )
    throw IllegalLinebreak( "Line break before end of String.",
                            text.File(),
                            text.Line(),
                            text.Column() ); 
 
  while (!stop) 
    {
      char dummy = text.Current();
      if ( dummy != '"' )
        {
          res += dummy;
          text.AdvanceChar();
 
          if ( text.LineBreak() )
            throw IllegalLinebreak( "Line break before end of String.",
                                    text.File(),
                                    text.Line(),
                                    text.Column() ); 
        }
      else
        {
          stop = true;
          text.AdvanceChar();
        }
    }
}
 
void read_integer(String& res, SourceText& text)
{
  bool stop = false;
  res = "";
  char dummy;
  text.LineBreak() = false;
 
  dummy = text.Current();
  if ( '+' == dummy || '-' == dummy )
    {
      res += dummy;
      text.AdvanceChar();
      if ( text.LineBreak() )
        throw IllegalLinebreak( "Line break after sign.",
                                text.File(),
                                text.Line(),
                                text.Column() ); 
    }
 
  if (!isdigit(text.Current()))
    {
      ostringstream os;
      os << "Expected digit, but got `" << text.Current() << "'.";
      throw UnexpectedChar(os.str(),
                           text.File(),
                           text.Line(),
                           text.Column());
    }
 
  while (!stop) 
    {
      char dummy = text.Current();
      if ( isdigit(dummy) )
        {
          res += dummy;
          text.AdvanceChar();
          if ( text.LineBreak() ) stop = true;
        }
      else
        {
          stop = true;
        }
    }
}
 
void read_numeric(String& res, SourceText& text)
{
  bool stop;
  res = "";
  char dummy;
  text.LineBreak() = false;
 
  // To make sure that there is at least one digit:
  bool found_digit = false;
 
  // Check if there is a sign:
  dummy = text.Current();
  if ( '+' == dummy || '-' == dummy )
    {
      res += dummy;
      text.AdvanceChar();
      if ( text.LineBreak() )
        throw IllegalLinebreak( "Linebreak after sign.",
                                text.File(),
                                text.Line(),
                                text.Column() ); 
    }
 
  // There could be some digits here:
  stop = false;
  while (!stop) 
    {
      char dummy = text.Current();
      if ( isdigit(dummy) )
        {
          found_digit = true;
          res += dummy;
          text.AdvanceChar();
          if ( text.LineBreak() ) return; // Line break ends scanning immediately.
        }
      else
        {
          stop = true;
        }
    }
 
  // Next there can be a decimal point
  if ( '.' == text.Current() )
    {
      res += ".";
      text.AdvanceChar();
      if ( text.LineBreak() )
        {
          if (found_digit)
            {
              // Line break ends scanning immediately, if we have
              // already found at least one digit.
              return;
            }
          else
            {
              throw IllegalLinebreak("Expected at least one digit.",
                                     text.File(),
                                     text.Line(),
                                     text.Column());
            }
        }
 
      // ... followed by optional more digits
      stop = false;
      while (!stop) 
        {
          char dummy = text.Current();
          if ( isdigit(dummy) )
            {
              found_digit = true;
              res += dummy;
              text.AdvanceChar();
              if ( text.LineBreak() ) return; // Line break ends scanning immediately.
            }
          else
            {
              stop = true;
            }
        }    
    }
 
  // At this point, we must have found at least one digit.
  if (!found_digit)
    throw ParseError("Expected at least one digit.",
                     text.File(),
                     text.Line(),
                     text.Column());
 
  // Now there could be a `e' or `E':
  dummy = text.Current();
  if ( 'e' == dummy || 'E' == dummy )
    {
      res += dummy;
      text.AdvanceChar();
      if ( text.LineBreak() )
        throw IllegalLinebreak( "Linebreak after e/E.",
                                text.File(),
                                text.Line(),
                                text.Column() );
      
      // Now there must be an integer (with optional sign)
      {
        String s;
        read_integer(s,text);
        res += s;
      }
    }
}
 
void parse_integer(Index& n, SourceText& text)
{
  String res;
  read_integer(res, text);
  istringstream is(res);
  is >> n;
}
 
void parse_numeric(Numeric& n, SourceText& text)
{
  String res;
  read_numeric(res, text);
  istringstream is(res);
  is >> n;
}
 
void parse_Stringvector(ArrayOfString& res, SourceText& text)
{
  bool first = true;            // To skip the first comma.
  res.resize(0);                        // Clear the result vector (just in case).
 
  // Make sure that the current character really is `[' and proceed.
  assertain_character('[',text);
  // There might have occured a linebreak, which is fine.
  
  eat_whitespace(text);
 
  // Read the elements of the vector (`]' means that we have
  // reached the end):
  while ( ']' != text.Current() )
    {
      String dummy;
 
      if (first)
        first = false;
      else
        {
          assertain_character(',',text);
          eat_whitespace(text);
        }
 
      parse_String(dummy, text);
      res.push_back(dummy);
      eat_whitespace(text);
    }
 
  text.AdvanceChar();
}
 
void parse_intvector(ArrayOfIndex& res, SourceText& text)
{
  bool first = true;            // To skip the first comma.
  res.resize(0);                        // Clear the result vector (just in case).
 
  // Make sure that the current character really is `[' and proceed.
  assertain_character('[',text);
  // There might have occured a linebreak, which is fine.
  
  eat_whitespace(text);
 
  // Read the elements of the vector (`]' means that we have
  // reached the end):
  while ( ']' != text.Current() )
    {
      Index dummy;
 
      if (first)
        first = false;
      else
        {
          assertain_character(',',text);
          eat_whitespace(text);
        }
 
      parse_integer(dummy, text);
      res.push_back(dummy);
      eat_whitespace(text);
    }
 
  text.AdvanceChar();
}
 
void parse_numvector(Vector& res, SourceText& text)
{
  bool first = true;            // To skip the first comma.
 
  // We need a temporary Array<Numeric>, so that we can use push_back
  // to store the values. FIXME: Need also constructor for Vector from
  // Array<Numeric>.
  Array<Numeric> tres;
 
  // Make sure that the current character really is `[' and proceed.
  assertain_character('[',text);
  // There might have occured a linebreak, which is fine.
  
  eat_whitespace(text);
 
  // Read the elements of the vector (`]' means that we have
  // reached the end):
  while ( ']' != text.Current() )
    {
      Numeric dummy;
 
      if (first)
        first = false;
      else
        {
          assertain_character(',',text);
          eat_whitespace(text);
        }
 
      parse_numeric(dummy, text);
      tres.push_back(dummy);
      eat_whitespace(text);
    }
 
  // Copy tres to res:
  res.resize(tres.nelem());
  res = tres;
 
  text.AdvanceChar();
}
 
bool parse_method(Index& id, 
                  Array<TokVal>& values,
                  ArrayOfIndex& output,
                  ArrayOfIndex& input,
                  SourceText& text,
                  const std::map<String, Index> MdMap,
                  const std::map<String, Index> WsvMap)
{
  extern const Array<WsvRecord> wsv_data;
  extern const Array<MdRecord> md_data;
  extern const ArrayOfString wsv_group_names;
 
  Index wsvid;                  // Workspace variable id, is used to
                                // access data in wsv_data.
 
  // Clear all output variables:
  id = 0;
  values.resize( 0 );
  output.resize( 0 );
  input.resize(  0 );
 
  {
    String methodname;
    read_name(methodname, text);
 
    {
      // Find method id:
      const map<String, Index>::const_iterator i = MdMap.find(methodname);
      if ( i == MdMap.end() )
        throw UnknownMethod(methodname,
                            text.File(),
                            text.Line(),
                            text.Column());
 
      id = i->second;
    }
    //    cout << "id=" << id << '\n';   
    // cout << "Method: " << md_data[id].Name() << '\n';
  }
 
  eat_whitespace(text);
 
  // For generic methods the output and input workspace variables have 
  // to be parsed (given in round brackets).
  if ( 0 < md_data[id].GOutput().nelem() + md_data[id].GInput().nelem() )
    {
      //      cout << "Generic!" << id << md_data[id].Name() << '\n';
      String wsvname;
      bool first = true;        // To skip the first comma.
 
      assertain_character('(',text);
      eat_whitespace(text);
 
      // First read all output Wsvs:
      for ( Index j=0 ; j<md_data[id].GOutput().nelem() ; ++j )
        {
          if (first)
            first = false;
          else
            {
              assertain_character(',',text);
              eat_whitespace(text);
            }
 
          read_name(wsvname, text);
 
          {
            // Find Wsv id:
            const map<String, Index>::const_iterator i = WsvMap.find(wsvname);
            if ( i == WsvMap.end() )
              throw UnknownWsv( wsvname,
                                text.File(),
                                text.Line(),
                                text.Column() );
 
            wsvid = i->second;
          }
 
          // Check that this Wsv belongs to the correct group:
          if ( wsv_data[wsvid].Group() != md_data[id].GOutput()[j] )
            {
            throw WrongWsvGroup( wsvname+" is not "+
                        wsv_group_names[md_data[id].GOutput()[j]]+", it is "+ 
                        wsv_group_names[wsv_data[wsvid].Group()],
                                 text.File(),
                                 text.Line(),
                                 text.Column() );
            }
 
          // Add this one to the list of output workspace variables:
          output.push_back(wsvid);
          
          eat_whitespace(text);
        }
 
      // Then read all input Wsvs:
      for ( Index j=0 ; j<md_data[id].GInput().nelem() ; ++j )
        {
          if (first)
            first = false;
          else
            {
              assertain_character(',',text);
              eat_whitespace(text);
            }
 
          read_name(wsvname, text);
 
          {
            // Find Wsv id:
            const map<String, Index>::const_iterator i = WsvMap.find(wsvname);
            if ( i == WsvMap.end() )
              throw UnknownWsv( wsvname,
                                text.File(),
                                text.Line(),
                                text.Column() );
 
            wsvid = i->second;
          }
 
          // Check that this Wsv belongs to the correct group:
          if ( wsv_data[wsvid].Group() != md_data[id].GInput()[j] )
            throw WrongWsvGroup( wsvname+" is not "+
                        wsv_group_names[md_data[id].GInput()[j]]+", it is "+ 
                        wsv_group_names[wsv_data[wsvid].Group()],
                                 text.File(),
                                 text.Line(),
                                 text.Column() );
 
          // Add this one to the list of input workspace variables:
          input.push_back(wsvid);
          
          eat_whitespace(text);
        }
 
      assertain_character(')',text);
      eat_whitespace(text);
    }
 
  // Now look for the curly braces:
  assertain_character('{',text);
  eat_whitespace(text);
  
  // Now we have to deal with two different cases: Keywords with
  // parameters, or (optionally) only a parameter without a keyword
  // for methods that have only a single argument.
  //
  // We can distinguish the two cases if we check whether the current
  // character is a letter. (If the parameter is specified directly it
  // must be either a number, a +- sign or a quotation mark)
  //
  // KEYWORDS THAT START WITH A NUMBER WILL BREAK THIS CODE!!
  //
  for ( Index i=0 ; i<md_data[id].Keywords().nelem() ; ++i )
    {
      if (!isalpha(text.Current()) && 1==md_data[id].Keywords().nelem())
        {
          // Parameter specified directly, without a keyword. This is only
          // allowed for single parameter methods!
 
          // We don't have to do anything here.
        }
      else  
        {      // Look for the keywords and read the parameters:
          
          String keyname;
          read_name(keyname,text);
 
          // Is the keyname the expected keyname?
          if ( keyname != md_data[id].Keywords()[i] )
            {
              throw UnexpectedKeyword( keyname,
                                       text.File(),
                                       text.Line(),
                                       text.Column());
            }
 
          eat_whitespace(text);
 
          // Look for '='
          assertain_character('=',text);
          eat_whitespace(text);
        }
 
      // Now parse the key value. This can be:
      // String_t,    Index_t,    Numeric_t,
      // Array_String_t, Array_Index_t, Vector_t,
      switch (md_data[id].Types()[i]) 
        {
        case String_t:
          {
            String dummy;
            parse_String(dummy, text);
            values.push_back(dummy);
            break;
          }
        case Index_t:
          {
            Index n;
            parse_integer(n, text);
            values.push_back(n);
            break;
          }
        case Numeric_t:
          {
            Numeric n;
            parse_numeric(n, text);
            values.push_back(n);
            break;
          }
        case Array_String_t:
          {
            ArrayOfString dummy;
            parse_Stringvector(dummy, text);
            values.push_back(dummy);
            break;
          }
        case Array_Index_t:
          {
            ArrayOfIndex dummy;
            parse_intvector(dummy, text);
            values.push_back(dummy);
            break;
          }
        case Vector_t:
          {
            Vector dummy;
            parse_numvector(dummy, text);
            values.push_back(dummy);
            break;
          }
        default:
          throw logic_error("Impossible parameter type.");
          break;
        }
 
      eat_whitespace(text);
 
      // Check:
      //      cout << "Value: " << md_data[id].Values()[i] << '\n';
    }
 
 
  // Now look for the closing curly braces.
  // We have to catch Eot, because after a method description is a
  // good place to end the control file. 
  try
    {
     assertain_character('}',text);
    }
  catch (const Eot x)
    {
      //      cout << "EOT!!!!" << endl;
      return true;
    }
  return false;
}
 
void parse(Array<MRecord>& tasklist,
           SourceText& text,
           const std::map<String, Index> MdMap,
           const std::map<String, Index> WsvMap)
{
  extern const Array<MdRecord> md_data;
  bool last = false;
  // For method ids:
  Index id;             
 // For keyword parameter values:
  Array<TokVal> values;
  // Output workspace variables (for generic methods):
  ArrayOfIndex output;          
  // Input workspace variables (for generic methods):
  ArrayOfIndex input;
 
  out3 << "\nParsing:\n";
 
  //  cout << text ;
 
  eat_whitespace(text);
 
  while (!last)
    {
      last = parse_method(id,values,output,input,text,MdMap,WsvMap);
 
      // Append taks to task list:
      tasklist.push_back(MRecord(id,values,output,input));
 
      {
        // Everything in this block is just to generate some
        // informative output.  
        extern const Array<WsvRecord> wsv_data;
 
        out3 << "- " << md_data[id].Name() << "\n";
 
        for ( Index j=0 ; j<values.nelem() ; ++j )
          {
            out3 << "   " 
                 << md_data[id].Keywords()[j] << ": "
                 << values[j] << '\n';
          }
          
        // Output workspace variables for generic methods:
        if ( 0 < md_data[id].GOutput().nelem() + md_data[id].GInput().nelem() )
          {
            out3 << "   Output: ";
            for ( Index j=0 ; j<output.nelem() ; ++j )
              {
                out3 << wsv_data[output[j]].Name() << " ";
              }
            out3 << "\n";
 
            out3 << "   Input: ";
            for ( Index j=0 ; j<input.nelem() ; ++j )
              {
                out3 << wsv_data[input[j]].Name() << " ";
              }
            out3 << "\n";
          }
      }
 
      // If last is set, then we have anyway reached the end of the
      // text, so we don't have to eat whitespace.
      if (!last)
        try
          {
            eat_whitespace(text);
          }
        catch (const Eot x)
          {
            last = true;
          }
    }
}
 
void parse_main(Array<MRecord>& tasklist, SourceText& text)
{
  //  extern const Array<MdRecord> md_data;
  //  extern const Array<WsvRecord> wsv_data;
  extern const std::map<String, Index> MdMap;
  extern const std::map<String, Index> WsvMap;
 
  try 
    {
      text.Init();
      parse(tasklist,text,MdMap,WsvMap);
    }
  catch (const Eot x)
    {
      // Unexpected end of the source text:
      out0 << "Unexpected end of control script.\n";
      out0 << "File: " << x.file() << '\n';
      out0 << "Line: " << x.line() << '\n';
      exit(true);
    }
  catch (const UnexpectedChar x)
    {
      // Unexpected Character:
      out0 << "Unexpected character:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out0 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const IllegalLinebreak x)
    {
      // A line break in an illegal position:
      out0 << "Illegal Line break:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      exit(true);
    }
  catch (const UnknownMethod x)
    {
      // Method unknown:
      // [**This should give a hint on how to obtain a list of allowed 
      // methods.]
      out0 << "Unknown Method:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out3 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const UnknownWsv x)
    {
      // Workspace variable unknown:
      // [**This should give a hint on how to obtain a list of allowed 
      // Wsvs.]
      out0 << "Unknown workspace variable:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out3 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const WrongWsvGroup x)
    {
      // Workspace variable unknown:
      // [**This should give a hint on how to obtain a list of Wsvs in 
      // this group.
      out0 << "Workspace variable belongs to the wrong group:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out3 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const UnexpectedKeyword x)
    {
      // Keyword unknown:
      // [**This should give a hint on how to obtain a list of allowed 
      // keywords.]
      out0 << "Unknown keyword:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out3 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const ParseError x)
    {
      // General Parse Error (parent of all the above):
      out0 << "Parse error:\n";
      out0 << x.what()   << '\n';
      out0 << "File: "   << x.file() << '\n';
      out0 << "Line: "   << x.line() << '\n';
      out0 << "Column: " << x.column() << '\n';
      exit(true);
    }
  catch (const runtime_error x)
    {
      cout << "Runtime error: ";
      cout << x.what() << '\n';
    }
  catch (const logic_error x)
    {
      cout << "Logic error: ";
      cout << x.what() << '\n';
    }
 
}