Simtel MSDOS 1992 September/Simtel20_Sept92.cdr/msdos/ddjmag/ddj8905.arc/TAWK.ASC

Text File | 1989-05-12 | 22KB | 712 lines

_TAWK, A Simple Interpreter in C++_
by Bruce Eckel


[LISTING ONE]


// FIELD.HXX: used by csascii class to build a single field.
// Fields are collected by csascii to create a record.
// by Bruce Eckel,
#include <stream.hxx>

class field { // one field in a comma-separated ASCII record
  istream * input; // where to get the data
  char * data;
  int length, fsize;
  int end_of_file;  // flag to indicate the end of file happened
  void getfield();  // recursive function to read in a field;
           // treats data, length & input as globals
  int infield; // flag used by getfield() to determine whether
        // it's inside a quoted field
 public:
  field(istream & instream);
  ~field();
  friend ostream& operator<<(ostream &s, field & f) {
    s << f.data;
    return s;
  }
  int eof() { return end_of_file; }  // to check for end
  int size() { return fsize;}
  int last_length() {return length; }
  char * string() { return data; }
};


[LISTING TWO]

// FIELD.CXX: definitions for class field
// A "recursive descent" scanning scheme is used because field
// length is always unknown.
// by Bruce Eckel
#include "field.hxx"

field::field(istream & instream) {
  input = &instream;
  length = 0;
  end_of_file = 0; // set flag to say "we're not at the end"
  infield = 0; // set flag to say "we're not inside a field"
  data = (char *)0; // to show no memory has been allocated
  getfield();  // recursively get characters until end of field
}

field::~field() {
  delete data;  // if no memory has been allocated,
  // data = (char *)0 so this will have no effect.
}

// A Comma-separated ASCII field is contained in quotes to allow
// commas within the field; these quotes must be stripped out
void field::getfield() {
  char c;
  // This happens when DEscending:
  if((input->get(c)).eof() ) {
    end_of_file++;  // just say we reached the end...
    return;
  }
  else  // watch out for the Unix vs. DOS LF/CR problem here:
    if (((c != ',') || infield) && (c != '\n')) {
      if ( (c != '"') && (c != '\r')) // watch for quotes or CR
        length++;  // no quotes -- count this character
      else {
        if ( c == '"')
          infield = !infield;  // if we weren't inside a field
          // and a quote was encountered, we are now inside
          // a field.  If we were inside a field and a quote
          // was found, we're out of the field.
        c = 0; // a quote or CR; mark it so it isn't included
      }
      getfield();  // recursively get characters in field
      // after returning from function call, we jump past
      // the following "else" part to finish the recursion
    }
    else { // This happens once, when the terminator is found:
      fsize = length;  // remember how long the string is
      data = new char[length + 1]; // space for null terminator
      data[length] = '\0';  // highest index is "length"
        // when you allocate an array of length + 1
      length--;  // notice we don't insert the delimiter
      // Now the first "if" statement evaluates to TRUE and
      // the function rises back up.
      return;
    }
  // This happens when Ascending:
  if ( c ) // if it wasn't a quote or CR,
    data[length--] = c;  // put chars in as we rise back up...
}


[LISTING THREE]

// CSASCII.HXX: class to manipulate comma-separated ASCII
// database files.
//by Bruce Eckel
#include <stream.hxx>
#include "field.hxx"

class csascii {  // manipulates comma-separated ascii files,
// generated by most database management systems (generated and
// used by the BASIC programming language).  Each field
// is separated by a comma; records are separated by newlines.
  int fieldcount;
  field ** data; // an array to hold the entire record
  istream * datafile; // file with comma separated ASCII input
  int readrecord(); // private function to read a record
 public:
  csascii( char * filename );  // Open file, get first record
  ~csascii(); // destructor
  int next(); // get next record, return 0 when EOF
  field & operator[](int index); // select a field
  int number_of_fields() { return fieldcount; }
};



[LISTING FOUR]

// CSASCII.CXX: function definitions for comma-separated
// ascii database manipulation class
// by Bruce Eckel,
#include "csascii.hxx"

int csascii::readrecord() {
  for (int fieldnum = 0; fieldnum < fieldcount; fieldnum++ ) {
    data[fieldnum] = new field(*datafile);
    if (data[fieldnum]->eof()) return 0;
  }
  return 1;
}

csascii::csascii( char * filename ) {
  char c;
  fieldcount = 0;
  int quote = 0;
  // first, determine the number of fields in a record:
  {
    // See text for dangers of opening files this way:
    istream infile(new filebuf->open(filename, input));
    while(infile.get(c), c != '\n') {
      // keep track of being inside a quoted string:
      if (c == '"') quote = !quote;
      // fields are delimited by unquoted commas:
      if ( c == ',' && !quote)
        fieldcount++;
    }
  }  // infile goes out of scope; file closed
  fieldcount++; // last field terminated by newline, not comma
  // an array of field pointers:
  data = new field * [ fieldcount ];
  // re-open at start; dynamically allocate so it isn't scoped:
  datafile = new istream(new filebuf->open(filename, input));
  readrecord();
}

csascii::~csascii() {
  delete data;
  delete datafile; // calls istream destructor to close file
}

int csascii::next() {
  for (int i = 0; i < fieldcount; i++ )
    delete data[i];  // free all the data storage
  return readrecord(); // 0 when end of file
}

field & csascii::operator[](int index) {
  if (index >= fieldcount) {
    cerr << "index too large for number of fields in record\n";
    exit(1);
  }
  return *(data[index]);
}


[LISTING FIVE]

// LOOKUP.CXX: simple use of csascii to find name in a database
// by Bruce Eckel,
#include "csascii.hxx"
#include <string.h>

main(int argc, char ** argv) {
  if (argc < 2) {
    cerr << "usage: lookup lastname\n";
    exit(1);
  }
  // This puts the database file in the root directory:
  csascii file("\\ppquick.asc"); // create object & open file
  int found = 0;  // indicates one record was found
  do {
    if (strcmp(file[0].string(),argv[1]) == 0) {
      found++;  // found one.  File is sorted, so if we stop
      // finding them, quit instead of wasting time.
      cout << chr(27) << "[2J"; // ANSI clear screen
      for (int i = 0; i < file.number_of_fields(); i++)
        cout << file[i] << "\n";
      cout << chr(27) << "[7m" << "press any key" <<
        chr(27) << "[0m";
      if( getch() == 27) break;
    }  else if (found) exit(0);  // quit if that was the last
  } while (file.next());
}


[LISTING SIX]


// PARSE.HXX: class to parse a tawk script file.  Creates
// a structure which can be used at run-time to "execute"
// the tawk script.
// by Bruce Eckel,
#include <stream.hxx>

// types of tokens the scanner can find:
enum tokentype {
  fieldnumber, string, if_, else_, endif_, phase_change
};

// preamble and conclusion of the tawk script are only executed
// once, while main is executed once for every data record
enum phase { preamble, tmain, conclusion};

class token {
  tokentype ttype;
  union {  // an "anonymous union"
    int fieldnum;  // if type is a fieldnumber
    unsigned char * literal; // if type is a string
  };
  int if_level;  // if this is an if_, then_, or else_
  // private functions:
  void get_token();  // recursive descent scanner
  // Functions to help in scanning:
  void getnext(char & c); // used by get_token();
  unsigned char get_value(char delimiter, char * msg);
  void dumpline(); // for @! comments
  void error(char * msg = "", char * msg2 = "");
 public:
  token(istream & input);
  ~token();
  friend ostream & operator<<(ostream &s, token &t);
  int field_number() { return fieldnum; }
  int token_type() { return ttype; }
  int nesting_level() { return if_level;}
};

// The following is called a "container class," since its sole
// purpose is to hold a list of objects (tokens, in this case):
class parse_array {
  token ** tokenarray; // an array of token pointers
  istream * parse_stream;
  int token_count;
  int end; // the size of the array
  phase p_section; // of the program (preamble, etc.)
  void build_array(); // another recursive function
 public:
  parse_array(istream & input);
  ~parse_array();
  int size() { return end; } // how big is it?
  token & operator[](int index); // select a token
  phase section() { return p_section; }
};


[LISTING SEVEN]

// PARSE.CXX: class parse function definitions
// by Bruce Eckel,
#include "csascii.hxx"
#include "parse.hxx"
#include <ctype.h>
#include <stdlib.h>

// The following are "file static," which means no one outside
// this file can know about them.  This is the meaning when a
// global variable is declared "static."
static istream * tokenstream;
static int length; // to remember size of string
static int line_number = 1;  // line counting for errors
static int if_counter = 0; // monitors "if" statement nesting
static phase program_section = preamble;  // ... until @main
static int end_of_file = 0; // zero means not end of file

token::token(istream & input) {
  // initialize values and start the descent
  tokenstream = &input;
  length = 0;
  get_token();  // recursively get characters to end of token
}

token::~token() { // delete heap if any has been allocated:
  if (ttype == string)
    delete literal;
}

void token::error(char * msg, char * msg2) {
  cerr << "token error on line " << line_number << ": " <<
      msg << " " << msg2 << "\n";
  exit(1);
}

ostream & operator<<(ostream &s, token &t) {
  switch (t.ttype) {
    case string:
      s << (char *)t.literal;
      break;
    case fieldnumber: // only for testing
      s << " fieldnumber: " << t.fieldnum << "\n";
  }
  return s;
}

// Get a character from the tokenstream, checking for
// end-of-file and newlines
void token::getnext(char & c) {
  if(end_of_file)
    error("attempt to read after @end statement\n",
       "missing @conclusion ?");
  if((tokenstream->get(c)).eof() )
    error("@end statement missing");
  if (c == '\n')
    line_number++; // keep track of the line count
}

// See text for description of tokens
void token::get_token() {
  char c;
  // This happens when DEscending:
  getnext(c);
  if ( c == '@') {
    if (length == 0) { // length 0 means start of token
      getnext(c);
      switch(c) {
        case '!': // comment line
          dumpline(); // dump the comment
          get_token(); // get a real token
          break;
        case 'p' : case 'P' : // preamble statement
          if ( program_section != preamble )
            error("only one preamble allowed");
          dumpline(); // just for looks, ignore it
          get_token(); // get a real token
          break;
        case 'm' : case 'M' : // start of main loop
          dumpline(); // toss rest of line
          program_section = tmain;
          ttype = phase_change;
          return; // very simple token
        case 'c' : case 'C' : // start conclusion
          dumpline();
          program_section = conclusion;
          ttype = phase_change;
          return; // very simple token
        case 'e' : case 'E': // end statement
          end_of_file++;  // set flag
          ttype = fieldnumber; // so destructor doesn't
                    // delete free store for this token.
          if (if_counter)
            error("unclosed 'if' statement(s)");
          return;
        case '(' :
          if ( program_section == preamble ||
            program_section == conclusion )
           error("@() not allowed in preamble or conclusion");
          fieldnum = get_value(')',"@()");
          ttype = fieldnumber;
          // This is a complete token, so quit
          return;
        case '<' :
          c = get_value('>',"@<>");
          length++;
          get_token(); // get more...
          break;
        case '?' : // beginning of an "if" statement
          if ( program_section == preamble ||
            program_section == conclusion )
           error("@? not allowed in preamble or conclusion");
          fieldnum = get_value('@',"@?@");
          ttype = if_;
          getnext(c);  // just eat the colon
          if(c != ':')
            error("@? must be followed by @: (then)");
          if_level = ++if_counter;  // for nesting
          return;
        case '~' : // the "else" part of an "if" statement
          ttype = else_;
          if_level = if_counter;
          return;
        case '.' : // "endif" terminator of an "if" statement
          ttype = endif_;
          if_level = if_counter--;
          if(if_counter < 0)
            error("incorrect nesting of if-then-else clauses");
          return;
        case '@' : // two '@' in a row mean print an '@'
          length++;  // just leave '@' as the value of c
          get_token();
          break;
        default:
          error("'@' must be followed by:",
          "'(', '<', '?',':','~','.','p','m','c' or '@'");
      }
    } else { // an '@' in the middle of a string; terminate
      // the string.  Putback() is part of the stream class.
      // It is only safe to put one character back on the input
      tokenstream->putback(c); // to be used by the next token
      // allocate space, put the null in and return up the stack
      literal = new unsigned char[length + 1]; // space for '\0'
      literal[length--] = '\0'; // string delimiter
      ttype = string; // what kind of token this is
      return; // back up the stack
    }
  } else { // not an '@', must be plain text
    length++;
    get_token();
  }
  // This occurs on the "tail" of the recursion:
  literal[length--] = c;  // put chars in as we rise back up...
}

// This function is used by get_token when it encounters a @(
// or a @< to get a number until it finds "delimiter."
// If an error occurs, msg is used to notify the user what
// kind of statement it is.
unsigned char token::get_value(char delimiter, char * msg) {
  char c;
  char buf[5];
  int i = 0;
  while(getnext(c), c != delimiter) {
    if (!isdigit(c))
      error("must use only digits inside", msg);
    buf[i++] = c;
  }
  buf[i] = 0;
  return atoi(buf);
}

void token::dumpline() { // called when '@!' encountered
  char c;
  while(getnext(c), c != '\n')
    ; // just eat characters until newline
}

// Since there's no way to know how big a parse_array is
// going to be until the entire tawkfile has been tokenized,
// the recursive approach is again used:

parse_array::parse_array(istream & input) {
  parse_stream = &input;
  token_count = 0;
  p_section = program_section; // so we know at run-time
  build_array();
}

void parse_array::build_array() {
  token * tk = new token(*parse_stream);
  if( ! end_of_file && tk->token_type() != phase_change) {
    // normal token, not end of file or phase change:
    token_count++;
    // recursively get tokens until eof or phase change:
    build_array();
  } else { // end of file or phase change
    // only done once per object:
    // allocate memory and return up the stack
    tokenarray = new token * [end = token_count];
    if(token_count) token_count--; // only if non-zero
    return;
  }
  tokenarray[token_count--] = tk;  // performed on the "tail"
}


parse_array::~parse_array() {
  for (int i = 0; i < end; i++)
    delete tokenarray[i];
  delete tokenarray;
}

token & parse_array::operator[](int index) {
  if ( index >= end ) {
    cerr << "parse_array error: index " << index
      << " out of bounds\n";
    exit(1);
  }
  return *tokenarray[index];
}


[LISTING EIGHT]

// TAWK.CXX: parses a tawk script and reads an ascii file;
// generates results according to the tawk script.
// by Bruce Eckel,
#include "csascii.hxx"
#include "parse.hxx"

main (int argc, char * argv[]) {
  int screen = 0;  // flag set true if screen output desired
  if (argc < 3) {
    cerr << "usage: tawk tawkfile datafile\n" <<
        "trailing -s pages output to screen";
    exit(1);
  }
  if (argc == 4) {
    if (argv[3][0] != '-') {
      cerr << "must use '-' before trailing flag\n";
      exit(1);
    } else
    if (argv[3][1] != 's') {
      cerr << "'s' is only trailing flag allowed";
      exit(1);
    } else
      screen++; // set screen output flag true
  }
  istream tawkfile(new filebuf->open(argv[1], input));
  parse_array Apreamble(tawkfile);  // the @preamble
  parse_array Amain(tawkfile);  // the @main section
  parse_array Aconclusion(tawkfile); // the @conclusion
  csascii datafile(argv[2]); // make a comma-separated ASCII
                             // object from the second arg
  // ------ @preamble  ------
  for (int i = 0; i < Apreamble.size(); i++)
    cout << Apreamble[i]; // preamble can only contain strings
  if(screen) {
    // ANSI reverse video sequence:
    cout << chr(27) << "[7m" << "press any key" <<
      chr(27) << "[0m";
    getch();
  }
  // ------ The Central Loop (@main) -------
  do {  // for each record in the data file
    if(screen) cout << chr(27) << "[2J"; // ANSI clear screen
    for(int i = 0; i < Amain.size(); i++) {
      switch(Amain[i].token_type()) {
        case fieldnumber:
          cout << datafile[Amain[i].field_number()];
          break;
        case string:
          cout << Amain[i];
          break;
        case if_:
          int fn = Amain[i].field_number();
          if (datafile[fn].size() == 0) { // conditional false
            int level = Amain[i].nesting_level();
            // find the "else" statement on the same level:
            while ( !(Amain[i].token_type() == else_
                && Amain[i].nesting_level() == level))
                  i++;
          } // conditional true -- just continue
          break;
        case else_: // an "if" conditional was true so skip
          // all the statements in the "else" clause
          int level = Amain[i].nesting_level();
          // find the "endif" statement on the same level:
          while ( !(Amain[i].token_type() == endif_
              && Amain[i].nesting_level() == level))
                i++;
          break;
        case endif_: // after performing the "else" clause
          break; // ignore it; only used to find the end
          // of the conditional when "if" is true.
        default: // should never happen (caught in parsing)
          cerr << "unknown statement encountered at run-time\n";
          exit(1);
      }
    }
    if(screen) {
      cout << chr(27) << "[7m" <<
        "press a key (ESC quits)" << chr(27) << "[0m";
      if( getch() == 27) break;
    }
  } while (datafile.next()); // matches do { ...
  //  ------ @conclusion ------
  for ( i = 0; i < Aconclusion.size(); i++)
    cout << Aconclusion[i]; //conclusion contains only strings
}



[LISTING NINE]

# makefile for tawk.exe & lookup.exe
# Zortech C++:
CPP = ztc
# Glockenspiel C++ w/ MSC 4:
#CPP = ccxx !4

all: tawk.exe lookup.exe

tawk.exe : tawk.obj parse.obj csascii.obj field.obj
        $(CPP)  tawk.obj parse.obj csascii.obj field.obj

lookup.exe : lookup.cxx csascii.obj field.obj
        $(CPP) lookup.cxx csascii.obj field.obj

tawk.obj : tawk.cxx parse.hxx csascii.hxx field.hxx
        $(CPP)  -c tawk.cxx

parse.obj : parse.cxx parse.hxx
        $(CPP)  -c parse.cxx

csascii.obj : csascii.cxx csascii.hxx field.hxx
        $(CPP)  -c csascii.cxx

field.obj : field.cxx field.hxx
        $(CPP)  -c field.cxx



[LISTING TEN]

@! REFORM.TWK
@! A tawk script to reformat a comma-separated ASCII file
@! with 6 fields.  This creates a new CS-ASCII file with
@! fields 4 and 5 combined.
@main
"@(0)","@(1)","@(2)","@(3)","@(4)@?4@: @~@.@(5)"
@conclusion
@end



[LISTING ELEVEN]


@! WALLET.TWK
@! Tawkfile to create a tiny phone listing for a wallet
@! on a Hewlett-Packard Laserjet-compatible printer
@! From a comma-separated ASCII file generated by a DBMS
@preamble
@<27>&l5C@! approximately 10 lines per inch
@<27>(s16.66H@! small typeface, built into Laserjet
@main
@! last, first, (area code) phone1
@(0),@(1)(@(2))@?3@:@(3)
@ phone2, if it exists
@?4@:@(4)
@~@.@~@?4@:@(4)
@~
@.@.@conclusion
@<27>E @! Reset the Laserjet
@end

[EXAMPLE 1]

class tiny {
  // private stuff here (this is a comment)
  int i;
 public: // public stuff here:
  print() { // an "in-line" function
    printf("i = %d\n",i);
  }
  tiny(int j); // constructors have the class name
  ~tiny() {} // destructors use a tilde
};  // classes end with a brace and a semicolon

tiny::tiny(int j) { // non inline definition
  i = j;
}

main() {
  tiny A(2);  // implicit constructor call
  // A.i = 30; // error! private member
  A.print(); // calling a member function
  // implicit destructor call at end of scope
}





[EXAMPLE 2]


#include <stream.hxx> // cout automatically defined
main() {
  cout << "Hello, world!\n" << "I am "
       << 6 << "today!\n";
}




[EXAMPLE 3]

filebuf f1;
if (f1.open(argv[1],input) == 0) {
  cout << "cannot open " << argv[1] << "\n";
  exit(1);
}
istream infile(&f1);



[EXAMPLE 4]


"Ball","Mike","Oregon Software C++ Compiler"
"Bright","Walter","Zortech C++ Compiler"
"Carolan","John","Glockenspiel C++ Translator"
"Stroustrup","Bjarne","AT&T, C++ Creator"
"Tiemann","Michael","Free Software Foundation C++ Compiler"