2
\$\begingroup\$

I've attempted to remedy the issues outlined in the answer to my previous question. I've added several hundred blank lines to better illustrate the grouping of functions and generally make things look less dense. I've trimmed all the lines to 80 columns except for one 83 character line in ppnarg.h which was inside the original author's comment, so I chose not to alter that. I've added forward declarations for all the static functions inside the .c files so all the static "helper functions" can be placed below the non-static function that uses them, so the implementation can be presented in a more top down fashion overall. I've added a description of every API function next to its declaration in the .h file, and comments in the .c files explaining design decisions that are important for understanding the implementation.

github

README.md

omitted for size

Makefile

CFLAGS= -std=c99 -g -Wall -Wpedantic -Wextra -Wno-unused-function -Wno-unused-parameter -Wno-switch -Wno-return-type -Wunused-variable
CFLAGS+= $(cflags)

test : pc11test
    ./$<

pc11test : pc11object.o pc11parser.o pc11io.o pc11test.o
    $(CC) $(CFLAGS) -o $@ $^ $(LDLIBS)

pc11object.o : pc11object.[ch]
pc11parser.o : pc11parser.[ch] pc11object.h
pc11io.o     : pc11io.[ch]     pc11object.h pc11parser.h
pc11test.o   : pc11test.[ch]   pc11object.h pc11parser.h pc11io.h


clean :
    rm *.o pc11test.exe

count :
    wc -l -c -L pc11*[ch] ppnarg.h
    cloc pc11*[ch] ppnarg.h

ppnarg.h

omitted for size

pc11object.h

#define PC11OBJECT_H
#include <stdlib.h>
#include <stdio.h>
#if ! PPNARG_H
  #include "ppnarg.h"
#endif


/* Variant subtypes of object,
   and signatures for function object functions */

#define IS_THE_TARGET_OF_THE_HIDDEN_POINTER_  *
typedef union    object  IS_THE_TARGET_OF_THE_HIDDEN_POINTER_  object;

typedef object   integer;
typedef object   list;
typedef object   symbol;
typedef object   string;
typedef object   boolean;

typedef object   suspension;
typedef object   parser;
typedef object   operator;
typedef operator predicate;
typedef operator binoperator;

typedef object   fSuspension( object env );
typedef object   fParser( object env, list input );
typedef object   fOperator( object env, object input );
typedef boolean  fPredicate( object env, object input );
typedef object   fBinOperator( object left, object right );


typedef enum {
  INVALID,
  INT,
  LIST,
  SYMBOL,
  STRING,
  VOID,
  SUSPENSION,
  PARSER,
  OPERATOR,
  END_TAGS
} tag;

enum object_symbol_codes {
  T,
  END_OBJECT_SYMBOLS
};


struct integer {
  tag t;
  int i;
};

struct list {
  tag t;
  object first, rest;
};

struct symbol {
  tag t;
  int code;
  const char *printname;
  object data;
};

struct string {
  tag t;
  char *str;
  int disposable;
};

struct void_ {
  tag t;
  void *pointer;
};

struct suspension {
  tag t;
  object env;
  fSuspension *f;
  const char *printname;
};

struct parser {
  tag t;
  object env;
  fParser *f;
  const char *printname;
};

struct operator {
  tag t;
  object env;
  fOperator *f;
  const char *printname;
};

struct header {
  int mark;
  object next;
  int forward;
};

union object {
  tag t;
  struct integer Int;
  struct list List;
  struct symbol Symbol;
  struct string String;
  struct void_ Void;
  struct suspension Suspension;
  struct parser Parser;
  struct operator Operator;
  struct header Header;
};



/* Global true/false objects. */

extern object NIL_; /* .t == INVALID */
extern symbol T_;



/* Determine if object is non-NULL and non-NIL.
   Will also convert a boolean T_ or NIL_ to an integer 1 or 0.
 */

static int
valid( object it ){
  return  it
      &&  it->t <  END_TAGS
      &&  it->t != INVALID;
}


/* Constructors */


integer    Int( int i );

boolean    Boolean( int b );

string     String( char *str, int disposable );

object     Void( void *pointer );


/* List of one element */

list       one( object it );


/* Join two elements togther. If rest is a list or NIL_, result is a list. */

list       cons( object first, object rest );


/* Join N elements together in a list */

#define LIST(...) \
  reduce( cons, PP_NARG(__VA_ARGS__), (object[]){ __VA_ARGS__ } )


/* Macros capture printnames automatically for these constructors */

#define    Symbol( n ) \
           Symbol_( n, #n, NIL_ )
symbol     Symbol_( int code, const char *printname, object data );

#define    Suspension( env, f ) \
           Suspension_( env, f, __func__ )
suspension Suspension_( object env, fSuspension *f, const char *printname );

#define    Parser( env, f ) \
           Parser_( env, f, __func__ )
parser     Parser_( object env, fParser *f, const char *printname );

#define    Operator( env, f ) \
           Operator_( env, f, #f )
operator   Operator_( object env, fOperator *f, const char *printname );



/* Printing */


/* Print list with dot notation or any object */

void    print( object a );


/* Print list with list notation or any object */

void    print_list( object a );



/* Functions over lists */


/* car */

object  first( list it );


/* cdr */

list    rest( list it );


/* Length of list */

int     length( list ls );


/* Force n elements from the front of (lazy?) list */

list    take( int n,
          list it );


/* Skip ahead n elements in (lazy?) list */

list    drop( int n,
          list it );


/* Index a (lazy?) list */

object  nth( int n,
         list it );


/* Apply operator to (lazy?) object */

object  apply( operator op,
           object it );


/* Produce lazy lists */

list    infinite( object mother );

list    chars_from_str( char *str );

list    chars_from_file( FILE *file );


/* Lazy list adapters */

list    ucs4_from_utf8( list o );

list    utf8_from_ucs4( list o );




/* Maps and folds */


/* Transform each element of list with operator; yield new list. */

list    map( operator op,
         list it );


/* Fold right-to-left over list with f */

object  collapse( fBinOperator *f,
          list it );


/* Fold right-to-left over array of objects with f */

object  reduce( fBinOperator *f,
        int n,
        object *po );




/* Comparisons and Association Lists (Environments) */


/* Compare for equality. For symbols, just compare codes. */

boolean eq( object a,
        object b );


/* Call eq, but avoid the need to allocate a Symbol object */

boolean eq_symbol( int code,
           object b );


/* Return copy of start sharing end */

list    append( list start,
        list end );


/* Prepend n (key . value) pairs to tail */

list    env( list tail,
         int n, ... );


/* Return value associated with key */

object  assoc( object key,
           list env );


/* Call assoc, but avoid the need to allocate a Symbol object */

object  assoc_symbol( int code,
              list env );




/* Conversions */


/* Copy integers and strings into *str. modifies caller supplied pointer */

void    fill_string( char **str,
             list it );


/* Convert integers and strings from list into a string */

string  to_string( list ls );


/* Dynamically create a symbol object corresponding to printname s.
   Scans the list of allocations linearly to find a matching printname.
   Failing that, it allocates a new symbol code from the space [-2,-inf). */

symbol  symbol_from_string( string s );




/* That one lone function without a category to group it in. */


/* Report (an analogue of) memory usage.
   By current measure, an allocation is 64 bytes,
   ie. 2x 32 byte union objects. */

int count_allocations( void );

pc11object.c

#define _BSD_SOURCE
#include "pc11object.h"
#include <stdarg.h>
#include <string.h>

static void print_listn( object a );
static int leading_ones( object byte );
static int mask_off( object byte, int m );

static fSuspension  force_first;
static fSuspension  force_rest;
static fSuspension  force_apply;
fSuspension infinite;
static fSuspension  force_chars_from_string;
static fSuspension  force_chars_from_file;
static fSuspension  force_ucs4_from_utf8;
static fSuspension  force_utf8_from_ucs4;

fBinOperator map;
fBinOperator eq;
fBinOperator append;
fBinOperator assoc;


/* Helper macro for constructor functions. */

#define OBJECT(...) new_( (union object[]){{ __VA_ARGS__ }} )


/* Flags controlling print(). */

static int print_innards = 1;
static int print_chars = 1;
static int print_codes = 0;


/* Define simple objects T_ and NIL_, the components of our boolean type. */

static union object nil_object = { .t=INVALID };
object NIL_ = & nil_object;
object T_ = 1 + (union object[]){ {.Header={1}},
                      {.Symbol={SYMBOL, T, "T", & nil_object}} };


/* Allocation function is defined at the end of this file with
   its file scoped data protected from the vast majority of
   other functions here. */

static object new_( object prototype );



integer
Int( int i ){
  return  OBJECT( .Int = { INT, i } );
}

boolean
Boolean( int b ){
  return  b  ? T_  : NIL_;
}

string
String( char *str, int disposable ){
  return  OBJECT( .String = { STRING, str, disposable } );
}

object
Void( void *pointer ){
  return  OBJECT( .Void = { VOID, pointer } );
}

list
one( object it ){
  return  cons( it, NIL_ );
}

list
cons( object first, object rest ){
  return  OBJECT( .List = { LIST, first, rest } );
}

symbol
Symbol_( int code, const char *printname, object data ){
  return  OBJECT( .Symbol = { SYMBOL, code, printname, data } );
}

suspension
Suspension_( object env, fSuspension *f, const char *printname ){
  return  OBJECT( .Suspension = { SUSPENSION, env, f, printname } );
}

parser
Parser_( object env, fParser *f, const char *printname ){
  return  OBJECT( .Parser = { PARSER, env, f, printname } );
}

operator
Operator_( object env, fOperator *f, const char *printname ){
  return  OBJECT( .Operator = { OPERATOR, env, f, printname } );
}



void
print( object a ){
  switch(  a  ? a->t  : 0  ){
  default: printf( "() " ); break;
  case INT: printf( print_chars  ? "'%c' "  : "%d ", a->Int.i ); break;
  case LIST: printf( "(" ), print( a->List.first ), printf( "." ),
                            print( a->List.rest ), printf( ")" ); break;
  case SUSPENSION: printf( "...(%s) ", a->Suspension.printname ); break;
  case PARSER: printf( "Parser(%s", a->Parser.printname ),
               (print_innards & ! a[-1].Header.forward) &&
                 (printf( ", " ), print( a->Parser.env ),0),
               printf( ") " ); break;
  case OPERATOR: printf( "Oper(%s", a->Operator.printname ),
                 printf( ", " ), print( a->Operator.env ),
                 printf( ") " ); break;
  case STRING: printf( "\"%s\" ", a->String.str ); break;
  case SYMBOL: if(  print_codes  )
                 printf( "%d:%s ", a->Symbol.code, a->Symbol.printname );
               else
         printf( "%s ", a->Symbol.printname );
               break;
  case VOID: printf( "VOID " ); break;
  }
}


void
print_list( object a ){
  switch(  a  ? a->t  : 0  ){
  default: print( a ); break;
  case LIST: printf( "(" ), print_list( first( a ) ),
                            print_listn( rest( a ) ), printf( ") " ); break;
  }
}

static void
print_listn( object a ){
  if(  ! valid( a )  ) return;
  switch(  a->t  ){
  default: print( a ); break;
  case LIST: print_list( first( a ) ),
             print_listn( rest( a ) ); break;
  }
}


/* force_() executes a suspension function to instantiate and yield 
   a value. It may unwrap many layers of suspended operations to shake
   off any laziness at the front of a list or resolve a lazy calculation
   down to its result.
   
   In order to simulate the feature of lazy evaluation that a lazy
   list will manifest its elements "in place", the resulting object
   from force_() must be overwritten over the representation of the 
   suspension object to provide the illusion that the list magically
   manifests for all handles to that part of the list.
   
   Consequently, force_() is declared static to this file and it is
   exclusively used in the stereotyped form:

     *it = *force_( it );

   Functions outside of this module requiring the forced execution
   of a potential suspension must use side effect of take() or drop().
   Eg. drop( 1, it ) will transform a suspended calculation into its
   actual resulting value. If it is a lazy list, this will manifest 
   the list node with a new suspension as the rest().
 */

static object
force_( object it ){
  if(  it->t != SUSPENSION  ) return  it;
  return  force_( it->Suspension.f( it->Suspension.env ) );
}


object first( list it ){
  if(  it->t == SUSPENSION  ) return  Suspension( it, force_first );
  if(  it->t != LIST  ) return  NIL_;
  return  it->List.first;
}

static object force_first ( object it ){
  *it = *force_( it );
  return  first( it );
}


object rest( list it ){
  if(  it->t == SUSPENSION  ) return  Suspension( it, force_rest );
  if(  it->t != LIST  ) return  NIL_;
  return  it->List.rest;
}

static object force_rest ( object it ){
  *it = *force_( it );
  return  rest( it );
}


int
length( list ls ){
  return  valid( ls )  ?  valid( first( ls ) ) + length( rest( ls ) ) : 0;
}



list
take( int n, list it ){
  if(  n == 0  ) return  NIL_;
  *it = *force_( it );
  if(  ! valid( it )  ) return  NIL_;
  return  cons( first( it ), take( n-1, rest( it ) ) );
}


list
drop( int n, list it ){
  if(  n == 0  ) return  it;
  *it = *force_( it );
  if(  ! valid( it )  ) return  NIL_;
  return  drop( n-1, rest( it ) );
}


object nth( int n, list it ){
  return  first( take( 1, drop( n-1, it ) ) );
}


object
apply( operator op, object it ){
  if(  it->t == SUSPENSION  )
    return  Suspension( cons( op, it ), force_apply );
  return  op->Operator.f( op->Operator.env, it );
}

static object
force_apply( list env ){
  operator op = first( env );
  object it = rest( env );
  *it = *force_( it );
  return  apply( op, it );
}


list
infinite( object mother ){
  return cons( mother, Suspension( mother, infinite ) );
}


list
chars_from_str( char *str ){
  if(  ! str  ) return  NIL_;
  return  Suspension( String( str, 0 ), force_chars_from_string );
}

static list
force_chars_from_string( string s ){
  char *str = s->String.str;
  if(  ! *str  ) return  one( Symbol( EOF ) );
  return  cons( Int( *str ),
        Suspension( String( str+1, 0 ), force_chars_from_string ) );
}


list
chars_from_file( FILE *file ){
  if(  ! file  ) return  NIL_;
  return  Suspension( Void( file ), force_chars_from_file );
}

static list
force_chars_from_file( object file ){
  FILE *f = file->Void.pointer;
  int c = fgetc( f );
  if(  c == EOF  ) return  one( Symbol( EOF ) );
  return  cons( Int( c ), Suspension( file, force_chars_from_file ) );
}


/* UCS4 <=> UTF8 */

list
ucs4_from_utf8( list input ){
  if(  ! input  ) return  NIL_;
  return  Suspension( input, force_ucs4_from_utf8 );
}

list
utf8_from_ucs4( list input ){
  if(  ! input  ) return  NIL_;
  return  Suspension( input, force_utf8_from_ucs4 );
}

static list
force_ucs4_from_utf8( list input ){
  *input = *force_( input );
  object byte;
  byte = first( input ), input = rest( input );
  if(  !valid(byte)  ) return  NIL_;
  if(  eq_symbol( EOF, byte )  ) return  input;
  int ones = leading_ones( byte );
  int bits = mask_off( byte, ones );
  int n = ones;
  while(  n-- > 1  ){
    *input = *force_( input );
    byte = first( input ), input = rest( input );
    if(  eq_symbol( EOF, byte )  ) return  input;
    bits = ( bits << 6 ) | ( byte->Int.i & 0x3f );
  }
  if(  bits < ((int[]){0,0,0x80,0x800,0x10000,0x110000,0x4000000})[ ones ]  )
    fprintf( stderr, "Overlength encoding in utf8 char.\n" );
  return  cons( Int( bits ), Suspension( input, force_ucs4_from_utf8 ) );
}

static list
force_utf8_from_ucs4( list input ){
  *input = *force_( input );
  object code = first( input );
  if(  eq_symbol( EOF, code )  ) return  input;
  int x = code->Int.i;
  object next = Suspension( drop( 1, input ), force_utf8_from_ucs4 );
  if(  x <= 0x7f  )
    return  cons( code, next );
  if(  x <= 0x7ff  )
    return  LIST( Int( (x >> 6)   | 0xc0 ),
                  Int( (x & 0x3f) | 0x80 ), next );
  if(  x <= 0xffff )
    return  LIST( Int(   (x >> 12)         | 0xe0 ),
                  Int( ( (x >> 6) & 0x3f ) | 0x80 ),
              Int( (  x       & 0x3f ) | 0x80 ), next );
  if(  x <= 0x10ffff  )
    return  LIST( Int(   (x >> 18)          | 0xf0 ),
              Int( ( (x >> 12) & 0x3f ) | 0x80 ),
                  Int( ( (x >> 6)  & 0x3f ) | 0x80 ),
              Int( (  x        & 0x3f ) | 0x80 ), next );
  if(  x <= 0x3ffffff  )
    return  LIST( Int(   (x >> 24)          | 0xf8 ),
              Int( ( (x >> 18) & 0x3f ) | 0x80 ),
              Int( ( (x >> 12) & 0x3f ) | 0x80 ),
                  Int( ( (x >> 6)  & 0x3f ) | 0x80 ),
              Int( (  x        & 0x3f ) | 0x80 ), next );
  if(  x <= 0x3fffffff  )
    return  LIST( Int(   (x >> 30)          | 0xfc ),
              Int( ( (x >> 24) & 0x3f ) | 0x80 ),
              Int( ( (x >> 18) & 0x3f ) | 0x80 ),
              Int( ( (x >> 12) & 0x3f ) | 0x80 ),
                  Int( ( (x >> 6)  & 0x3f ) | 0x80 ),
              Int( (  x        & 0x3f ) | 0x80 ), next );
  fprintf( stderr, "Invalid unicode code point in ucs4 char.\n" );
  return  next;
}

static int
leading_ones( object byte ){
  if(  byte->t != INT  ) return  0;
  int x = byte->Int.i;
  return  x&0200 ? x&0100 ? x&040 ? x&020 ? x&010 ? x&4 ? 6
                                                        : 5
                                                  : 4
                                          : 3
                                  : 2
                          : 1
                 : 0;
}

static int
mask_off( object byte, int m ){
  if(  byte->t != INT  ) return  0;
  int x = byte->Int.i;
  return  x & (m  ? (1<<(8-m))-1  : -1);
}




list
map( operator op, list it ){
  if(  ! valid( it )  ) return  it;
  return  cons( apply( op, first( it ) ),
        map( op, rest( it ) ) );
}

object
collapse( fBinOperator *f, list it ){
  if(  !valid( it )  ) return  it;
  object right = collapse( f, rest( it ) );
  if(  !valid( right )  ) return  first( it );
  return  f( first( it ), right );
}

object
reduce( fBinOperator *f, int n, object *po ){
  return  n==1  ? *po  : f( *po, reduce( f, n-1, po+1 ) );
}



boolean
eq( object a, object b ){
  return  Boolean(
            !valid( a ) && !valid( b )  ? 1  :
            !valid( a ) || !valid( b )  ? 0  :
            a->t != b->t                ? 0  :
            a->t == SYMBOL              ? a->Symbol.code == b->Symbol.code  :
            !memcmp( a, b, sizeof *a )  ? 1  : 0
          );
}

boolean
eq_symbol( int code, object b ){
  return  eq( (union object[]){ {.Symbol = {SYMBOL, code, "", 0} } }, b );
}


list
append( list start, list end ){
  if(  ! valid( start )  ) return  end;
  return  cons( first( start ), append( rest( start ), end ) );
}


list
env( list tail, int n, ... ){
  va_list v;
  va_start( v, n );
  list r = tail;
  while( n-- ){
    object a = va_arg( v, object );
    object b = va_arg( v, object );
    r = cons( cons( a, b ), r );
  }
  va_end( v );
  return  r;
}


object
assoc( object key, list b ){
  if(  !valid( b )  ) return  NIL_;
  object pair = first( b );
  if(  valid( eq( key, first( pair ) ) )  )
    return  rest( pair );
  else
    return  assoc( key, rest( b ) );
}

object
assoc_symbol( int code, list b ){
  return  assoc( (union object[]){ {.Symbol = {SYMBOL, code, "", 0}} }, b );
}



static int
string_length( object it ){
  switch(  it  ? it->t  : 0  ){
  default: return  0;
  case INT: return  1;
  case STRING: return  strlen( it->String.str );
  case LIST: return  string_length( first( it ) )
                   + string_length( rest( it ) );
  }
}


void
fill_string( char **str, list it ){
  switch(  it  ? it->t  : 0  ){
  default: return;
  case INT:
    *(*str)++ = it->Int.i;
    return;
  case STRING:
    strcpy( *str, it->String.str );
    *str += strlen( it->String.str );
    return;
  case LIST:
    fill_string( str, first( it ) );
    fill_string( str, rest( it ) );
    return;
  }
}


string
to_string( list ls ){
  char *str = calloc( 1 + string_length( ls ), 1 );
  string s = OBJECT( .String = { STRING, str, 1 } );
  fill_string( &str, ls );
  return  s;
}


/* The following functions are isolated to the bottom of this file
   so that their static variables are protected from all other
   functions in this file. 
 */

/* Allocation of objects */

static list allocation_list = NULL;

static object
new_( object prototype ){
  object record = calloc( 2, sizeof *record );
  if(  record  ){
    record[0] = (union object){ .Header = { 0, allocation_list } };
    allocation_list = record;
    record[1] = *prototype;
  }
  return  record + 1;
}


/* Construction of dynamic symbols */

static int next_symbol_code = -2;

symbol
symbol_from_string( string s ){
  list ls = allocation_list;
  while(  ls != NULL && valid( ls + 1 )  ){
    if(  ls[1].t == SYMBOL
    &&  strcmp( ls[1].Symbol.printname, s->String.str ) == 0  ){
      return  ls + 1;
    }
    ls = ls[0].Header.next;
  }
  return  Symbol_( next_symbol_code--, strdup( s->String.str ), NIL_ );
}


int
count_allocations( void ){
  list ls = allocation_list;
  int n = 0;
  while(  ls != NULL && valid( ls + 1 )  ){
    ++n;
    ls = ls->Header.next;
  }
  return  n;
}

pc11parser.h

#define PC11PARSER_H
#if  ! PC11OBJECT_H
  #include "pc11object.h"
#endif

enum parser_symbol_codes {
  VALUE = END_OBJECT_SYMBOLS,
  OK,
  FAIL,
  SATISFY_PRED,
  EITHER_P,
  EITHER_Q,
  SEQUENCE_P,
  SEQUENCE_Q,
  SEQUENCE_OP,
  BIND_P,
  BIND_OP,
  INTO_P,
  INTO_ID,
  INTO_Q,
  REGEX_ATOM,
  PROBE_P,
  PROBE_MODE,
  EBNF_SEQ,
  EBNF_ANY,
  EBNF_EPSILON,
  EBNF_MAYBE,
  EBNF_MANY,
  END_PARSER_SYMBOLS
};


/* Parse the input using parser p. */

list    parse( parser p,
           list input );


/* Check result from parse(). */

int     is_ok( list result );

int     not_ok( list result );


/* Return OK or FAIL result. */

parser  succeeds( list result );

parser  fails( list errormsg );


/* Emit debugging output from p.
   Print on ok iff mode&1; print not ok iff mode&2. */

parser  probe( parser p,
           int mode );


/* The basic (leaf) parser. */

parser  satisfy( predicate pred );


/* Simple parsers built with satisfy(). */

parser  alpha( void );
parser  upper( void );
parser  lower( void );
parser  digit( void );
parser  literal( object example );
parser  chr( int c );
parser  str( char *s );
parser  anyof( char *s );
parser  noneof( char *s );

/* Accept any single element off the input list. */

parser  item( void );



/* Choice ("OR" branches) */


/* Combine 2 parsers into a choice. */

parser  either( parser p,
        parser q );


/* Combine N parsers into a choice. */

#define ANY(...)                    \
  reduce( either,                   \
          PP_NARG(__VA_ARGS__),     \
          (object[]){ __VA_ARGS__ } )



/* Sequence ("AND" branches) */


/* Combine 2 parsers into a sequence,
   using op to merge the value portions of results. */

parser  sequence( parser p,
          parser q,
          binoperator op );


/* Sequence 2 parsers but drop result from first. */

parser  xthen( parser x,
           parser q );


/* Sequence 2 parsers but drop result from second. */

parser   thenx( parser p,
        parser x );


/* Sequence 2 parsers and concatenate results. */

parser   then( parser p,
           parser q );


/* Sequence N parsers and concatenate results. */

#define SEQ(...)                    \
  reduce( then,                     \
          PP_NARG(__VA_ARGS__),     \
          (object[]){ __VA_ARGS__ } )


/* Sequence 2 parsers, but pass result from first as a
   (id.value) pair in second's env. */

parser  into( parser p,
          object id,
          parser q );



/* Repetitions */


/* Accept 0 or 1 successful results from p. */

parser  maybe( parser p );


/* Accept 0 or more successful results from p. */

parser  many( parser p );


/* Accept 1 or more successful results from p. */

parser  some( parser p );



/* Transform of values */


/* Process succesful result from p
   by transforming the value portion with op. */

parser  bind( parser p,
          operator op );



/* Building recursive parsers */


/* Create an empty parser, useful for building loops.
   A forward declaration of a parser. */

parser  forward( void );



/* Compilers */


/* Compile a regular expression into a parser. */
// E->T ('|' T)*
// T->F*
// F->A ('*' | '+' | '?')?
// A->'.' | '('E')' | C
// C->S|L|P
// S->'\' ('.' | '|' | '(' | ')' | '[' | ']' | '/' )
// L->'[' '^'? ']'? [^]]* ']'
// P->Plain char

parser  regex( char *re );


/* Compile a block of EBNF definitions into a list of
   (symbol.parser) pairs. */
// D->N '=' E ';'
// N->name
// E->T ('|' T)*
// T->F*
// F->R | N | '[' E ']' | '{' E '}' | '(' E ')' | '/' regex '/'
// R->'"' [^"]* '"' | "'" [^']* "'"

list    ebnf( char *productions,
          list supplements,
          list handlers );

pc11parser.c

#include "pc11parser.h"
#include <ctype.h>
#include <string.h>

static fParser    success;
static fParser    fail;

static fParser    parse_satisfy;
static fPredicate is_upper;
static fPredicate is_alpha;
static fPredicate is_lower;
static fPredicate is_digit;
static fPredicate is_literal;
static fPredicate is_range;
static fPredicate is_anyof;
static fPredicate is_noneof;
static fPredicate always_true;

static fParser    parse_either;
fBinOperator  either;

static fParser    parse_sequence;

static fBinOperator concat;
fBinOperator  then;

static fBinOperator left;
static fBinOperator right;
fBinOperator  xthen;
fBinOperator  thenx;

static fParser    parse_bind;

static fParser    parse_into;

static fParser    parse_probe;

static fOperator  apply_meta;
static fOperator  on_dot;
static fOperator  on_chr;
static fOperator  on_meta;
static fOperator  on_class;
static fOperator  on_term;
static fOperator  on_expr;

static fOperator  stringify;
static fOperator  symbolize;
static fOperator  encapsulate;

static fOperator  make_matcher;
static fOperator  make_sequence;
static fOperator  make_any;
static fOperator  make_maybe;
static fOperator  make_many;

static fOperator  define_forward;
static fOperator  compile_bnf;
static fOperator  compile_rhs;
static fOperator  define_parser;
static fOperator  wrap_handler;


/* Execute a parser upon an input stream by invoking its function,
   supplying its env. */

list
parse( parser p, list input ){
  if(  !valid( p ) || !valid( input ) || p->t != PARSER  )
    return  cons( Symbol(FAIL),
          cons( String("parse() validity check failed",0),
            input ) ); 
  return  p->Parser.f( p->Parser.env, input );
}

/*
  The result structure from a parser is either
    ( OK . ( <value> . <remaining input ) )
  or
    ( FAIL . ( <error message> . <remaining input> ) )
*/

static object
success( object result, list input ){
  return  cons( Symbol(OK),
        cons( result,
              input ) );
}

static object
fail( object errormsg, list input ){
  return  cons( Symbol(FAIL),
        cons( errormsg,
              input ) );
}


int
is_ok( list result ){
  return  valid( eq_symbol( OK, first( result ) ) );
}

int
not_ok( list result ){
  return  ! is_ok( result );
}


parser
succeeds( list result ){
  return  Parser( result, success );
}

parser
fails( list errormsg ){
  return  Parser( errormsg, fail );
}


/* For all of the parsers after this point, the associated parse_*()
   function should be considered the "lambda" or "closure" function for
   the constructed parser object.
   C, of course, doesn't have lambdas. Hence these closely associated
   functions are close by and have related names.
   These parse_* functions receive an association list of (symbol.value)
   pairs in their env parameter, and they extract their needed values
   using assoc_symbol().
*/

/* The satisfy(pred) parser is the basis for all "leaf" parsers.
   Importantly, it forces the first element off of the (lazy?) input
   list. Therefore, all other functions that operate upon this result
   of this parser need not fuss with suspensions at all. */

parser
satisfy( predicate pred ){
  return  Parser( env( NIL_, 1, Symbol(SATISFY_PRED), pred ), parse_satisfy );
}

static list
parse_satisfy( object env, list input ){
  predicate pred = assoc_symbol( SATISFY_PRED, env );
  drop( 1, input );
  object item = first( input );
  if(  ! valid( item )  ) return  fail( String( "empty input", 0 ),
                    input );
  return  valid( apply( pred, item ) )
            ? success( item,
               rest( input ) )
            : fail( LIST( String( "predicate not satisfied", 0 ), pred, NIL_ ),
            input );
}


parser item( void ){
  return  satisfy( Operator( NIL_, always_true ) );
}

boolean
always_true( object v, object it ){
  return  T_;
}


parser
alpha( void ){
  return  satisfy( Operator( NIL_, is_alpha ) );
}

static boolean
is_alpha( object v, object it ){
  return  Boolean( it->t == INT && isalpha( it->Int.i ) );
}


parser
upper( void ){
  return  satisfy( Operator( NIL_, is_upper ) );
}

static boolean
is_upper( object v, object it ){
  return  Boolean( it->t == INT && isupper( it->Int.i ) );
}


parser
lower( void ){
  return  satisfy( Operator( NIL_, is_lower ) );
}

static boolean
is_lower( object v, object it ){
  return  Boolean( it->t == INT && islower( it->Int.i ) );
}


parser
digit( void ){
  return  satisfy( Operator( NIL_, is_digit ) );
}

static boolean
is_digit( object v, object it ){
  return  Boolean( it->t == INT && isdigit( it->Int.i ) );
}


parser
literal( object example ){
  return  satisfy( Operator( example, is_literal ) );
}

static boolean
is_literal( object example, object it ){
  return  eq( example, it );
}


parser
chr( int c ){
  return  literal( Int( c ) );
}


parser
str( char *s ){
  return  !*s  ? succeeds( NIL_ )
               : !s[1]  ? chr( *s )
                        : then( chr( *s ), str( s+1 ) );
}



parser
range( int lo, int hi ){
  return  satisfy( Operator( cons( Int( lo ), Int( hi ) ), is_range ) );
}

static boolean
is_range( object bounds, object it ){
  int lo = first( bounds )->Int.i,
      hi = rest( bounds )->Int.i;
  return  Boolean( it->t == INT && lo <= it->Int.i && it->Int.i <= hi );
}


parser
anyof( char *s ){
  return  satisfy( Operator( String( s, 0 ), is_anyof ) );
}

static boolean
is_anyof( object set, object it ){
  return  Boolean( it->t == INT && strchr( set->String.str, it->Int.i ) );
}


parser
noneof( char *s ){
  return  satisfy( Operator( String( s, 0 ), is_noneof ) );
}

static boolean
is_noneof( object set, object it ){
  return  Boolean( it->t == INT && ! strchr( set->String.str, it->Int.i ) );
}


/* The choice combinator. Result is success if either p or q succeed.
   Short circuits q if p was successful. Not lazy. */


parser
either( parser p, parser q ){
  return  Parser( env( NIL_, 2,
               Symbol(EITHER_Q), q,
                       Symbol(EITHER_P), p ),
                  parse_either );
}

static object
parse_either( object env, list input ){
  parser p = assoc_symbol( EITHER_P, env );
  object result = parse( p, input );
  if(  is_ok( result )  ) return  result;
  parser q = assoc_symbol( EITHER_Q, env );
  return  parse( q, input );
}


/* Sequence 2 parsers and join the 2 results using a binary operator.
   By parameterizing this "joining" operator, this parser supports
   then(), thenx() and xthen() while being completely agnostic as to
   how joining might or might not be done.
 */


parser
sequence( parser p, parser q, binoperator op ){
  return  Parser( env( NIL_, 3,
                       Symbol(SEQUENCE_OP), op,
                       Symbol(SEQUENCE_Q), q,
                       Symbol(SEQUENCE_P), p ),
                  parse_sequence );
}

static object
parse_sequence( object env, list input ){
  parser p = assoc_symbol( SEQUENCE_P, env );
  object p_result = parse( p, input );
  if(  not_ok( p_result )  ) return  p_result;

  parser q = assoc_symbol( SEQUENCE_Q, env );
  list remainder = rest( rest( p_result ) );
  object q_result = parse( q, remainder );
  if(  not_ok( q_result )  ){
    object q_error = first( rest( q_result ) );
    object q_remainder = rest( rest( q_result ) );
    return  fail( LIST( q_error, String( "after", 0),
            first( rest( p_result ) ), NIL_ ),
          q_remainder );
  }

  binoperator op = assoc_symbol( SEQUENCE_OP, env );
  return  success( op->Operator.f( first( rest( p_result ) ),
                                   first( rest( q_result ) ) ),
                   rest( rest( q_result ) ) );
}


parser
then( parser p, parser q ){
  return  sequence( p, q, Operator( NIL_, concat ) );
}

parser
xthen( parser x, parser q ){
  return  sequence( x, q, Operator( NIL_, right ) );
}

parser
thenx( parser p, parser x ){
  return  sequence( p, x, Operator( NIL_, left ) );
}


/* Some hacking and heuristics to massage 2 objects together into a list,
   taking care if either is already a list */

static object
concat( object l, object r ){
  if(  ! valid( l )  ) return  r;
  if(  r->t == LIST
    && valid( eq_symbol( VALUE, first( first( r ) ) ) )
    && ! valid( rest( r ) )
    && ! valid( rest( first( r ) ) )  )
    return  l;
  switch(  l->t  ){
  case LIST: return  cons( first( l ), concat( rest( l ), r ) );
  default: return  cons( l, r );
  }
}

static object
right( object l, object r ){
  return  r;
}

static object
left( object l, object r ){
  return  l;
}



/* Sequence parsers p and q, but define the value portion of the result of p 
   (if successful) as (id.value) in the env of q.
 */

parser
into( parser p, object id, parser q ){
  return  Parser( env( NIL_, 3,
                       Symbol(INTO_P), p,
                       Symbol(INTO_ID), id,
                       Symbol(INTO_Q), q ),
                  parse_into );
}

static object
parse_into( object v, list input ){
  parser p = assoc_symbol( INTO_P, v );
  object p_result = parse( p, input );
  if(  not_ok( p_result )  ) return  p_result;
  object id = assoc_symbol( INTO_ID, v );
  parser q = assoc_symbol( INTO_Q, v );
  object q_result = q->Parser.f( env( q->Parser.env, 1,
                                      id, first( rest( p_result ) ) ),
                         rest( rest( p_result ) ) );
  if(  not_ok( q_result )  ){
    object q_error = first( rest( q_result ) );
    object q_remainder = rest( rest( q_result ) );
    return  fail( LIST( q_error, String( "after", 0),
            first( rest( p_result ) ), NIL_ ),
          q_remainder );
  }
  return  q_result;
}



/* If the parser p succeeds, great! return its result.
   If not, who cares?! call it a success, but give a nothing value.
   If this parser is composed using then(), the merging of values will
   simply ignore this nothing value. It just disappears.
   
   If you bind() this parser to an operator, the operator can test
   if valid( input ) to tell whether p succeeded (and yielded a value)
   or not (which yielded NIL).
 */

parser
maybe( parser p ){
  return  either( p, succeeds( NIL_ ) );
}


/* Uses a forward() to build an infinite sequence of maybe(p). */

parser
many( parser p ){
  parser q = forward();
  *q = *maybe( then( p, q ) );
  return  q;
}


parser
some( parser p ){
  return  then( p, many( p ) );
}



/* Bind transforms a succesful result from the child parser
   through the operator. The operator's environment is supplemented
   with the environment passed to bind itself.
 */

parser
bind( parser p, operator op ){
  return  Parser( env( NIL_, 2,
                       Symbol(BIND_P), p,
                       Symbol(BIND_OP), op ),
                  parse_bind );
}

static object
parse_bind( object env, list input ){
  parser p = assoc_symbol( BIND_P, env );
  operator op = assoc_symbol( BIND_OP, env );
  object result = parse( p, input );
  if(  not_ok( result )  ) return  result;
  object payload = rest( result ),
         value = first( payload ),
         remainder = rest( payload );
  return  success( apply( (union object[]){{.Operator={
    OPERATOR,
    append(op->Operator.env, env),
    op->Operator.f,
    op->Operator.printname
  }}}, value ), remainder );
}



/* Construct a forwarding parser to aid building of loops.
   This parser can be composed with other parsers.
   Later, the higher level composed parser can be copied over this object 
   to create the point of recursion in the parser graph.
   Remembers the fact that it was created as a forward
   by storing a flag in the hidden allocation record for the parser.
   This flag is not altered by overwriting the parser's normal union object.
 */

parser
forward( void ){
  parser p = Parser( 0, 0 );
  p[-1].Header.forward = 1;
  return  p;
}





parser
probe( parser p, int mode ){
  return  Parser( env( NIL_, 2,
               Symbol(PROBE_MODE), Int( mode ),
               Symbol(PROBE_P), p ),
          parse_probe );
}

static object
parse_probe( object env, object input ){
  parser p = assoc_symbol( PROBE_P, env );
  int mode = assoc_symbol( PROBE_MODE, env )->Int.i;
  object result = parse( p, input );
  if(  is_ok( result ) && mode&1  )
    print( result ), puts("");
  else if(  not_ok( result ) && mode&2  )
    print_list( result ), puts("");
  return  result;
}



/* Regex compiler */

static parser regex_grammar( void );
static parser regex_parser;

parser
regex( char *re ){
  if(  !regex_parser  ) regex_parser = regex_grammar();
  object result = parse( regex_parser, chars_from_str( re ) );
  if(  not_ok( result )  ) return  result;
  return  first( rest( result ) );
}

#define META     "*+?"
#define SPECIAL  META ".|()[]/"

static parser
regex_grammar( void ){
  parser dot       = bind( chr('.'), Operator( NIL_, on_dot ) );
  parser meta      = anyof( META );
  parser escape    = xthen( chr('\\'), anyof( SPECIAL "\\" ) );
  parser class     = xthen( chr('['),
                thenx( SEQ( maybe( chr('^') ),
                            maybe( chr(']') ),
                                        many( noneof( "]" ) ) ),
                       chr(']') ) );
  parser character = ANY( bind( escape, Operator( NIL_, on_chr ) ),
              bind( class, Operator( NIL_, on_class ) ),
              bind( noneof( SPECIAL ), Operator( NIL_, on_chr ) ) );
  parser expr      = forward();
  {
    parser atom    = ANY( dot,
                          xthen( chr('('), thenx( expr, chr(')') ) ),
                          character );
    parser factor  = into( atom, Symbol(REGEX_ATOM),
                           bind( maybe( meta ),
                                 Operator( NIL_, on_meta ) ) );
    parser term    = bind( many( factor ),
                           Operator( NIL_, on_term ) );
    *expr  = *bind( then( term, many( xthen( chr('|'), term ) ) ),
                    Operator( NIL_, on_expr ) );
  }
  return  expr;
}


/* syntax directed compilation to parser */

static parser
apply_meta( parser a, object it ){
  switch(  it->Int.i  ){
  default:  return  a;
  case '*': return  many( a );
  case '+': return  some( a );
  case '?': return  maybe( a );
  }
}

static parser
on_dot( object v, object it ){
  return  item();
}

static parser
on_chr( object v, object it ){
  return  literal( it );
}

static parser
on_meta( object v, object it ){
  parser atom = assoc_symbol( REGEX_ATOM, v );
  if(  it->t == LIST 
    && valid( eq_symbol( VALUE, first( first( it ) ) ) )
    && ! valid( rest( it ) )
    && ! valid( rest( rest( it ) ) )  )
    return  atom;
  return  apply_meta( atom, it );
}

static parser
on_class( object v, object it ){
  if(  first( it )->Int.i == '^'  )
    return  satisfy( Operator( to_string( rest( it ) ), is_noneof ) );
  return  satisfy( Operator( to_string( it ), is_anyof ) );
}

static parser
on_term( object v, object it ){
  if(  ! valid( it )  ) return  NIL_;
  if(  it->t == LIST  &&  ! valid( rest( it ) )  ) it = first( it ); 
  if(  it->t == PARSER  ) return  it;
  return  collapse( then, it );
}

static parser
on_expr( object v, object it ){
  if(  it->t == LIST  &&  ! valid( rest( it ) )  ) it = first( it );
  if(  it->t == PARSER  ) return  it;
  return  collapse( either, it );
}



/* EBNF compiler */

static parser ebnf_grammar( void );

/* Compile a block of EBNF definitions into an association list
   of (symbol.parser) pairs.
   Accepts an association list of supplemental parsers for any syntactic
   constructs that are easier to build outside of the EBNF syntax.
   Accepts an association list of operators to bind the results of any
   named parser from the EBNF block or the supplements.
 */

list
ebnf( char *productions, list supplements, list handlers ){
  static parser ebnf_parser;
  if(  !ebnf_parser  ) ebnf_parser = ebnf_grammar();

  object result = parse( ebnf_parser, chars_from_str( productions ) );
  if(  not_ok( result )  ) return  result;

  object payload = first( rest( result ) );
  list defs = append( payload,
              env( supplements, 1,
               Symbol(EBNF_EPSILON), succeeds(NIL_) ) );
  list forwards = map( Operator( NIL_, define_forward ), defs );
  list parsers = map( Operator( forwards, compile_rhs ), defs );
  list final = map( Operator( forwards, define_parser ), parsers );
  map( Operator( forwards, wrap_handler ), handlers );
  return  final;
}

static parser
ebnf_grammar( void ){
  if(  !regex_parser  ) regex_parser = regex_grammar();
  parser spaces = many( anyof( " \t\n" ) );
  parser defining_symbol = thenx( chr( '=' ), spaces );
  parser choice_symbol = thenx( chr( '|' ), spaces );
  parser terminating_symbol = thenx( chr( ';' ), spaces );
  parser name = some( either( anyof( "-_" ), alpha() ) );
  parser identifier = thenx( name, spaces );
  parser terminal =
    bind( 
      thenx( either( thenx( xthen( chr( '"'), many( noneof("\"") ) ),
                chr( '"') ),
                     thenx( xthen( chr('\''), many( noneof( "'") ) ),
                chr('\'') ) ),
             spaces ),
      Operator( NIL_, make_matcher ) );
  parser symb = bind( identifier, Operator( NIL_, symbolize ) );
  parser nonterminal = symb;
  parser expr = forward();
  {
    parser factor = ANY( terminal,
             nonterminal,
             bind( xthen( then( chr( '[' ), spaces ),
                      thenx( expr,
                         then( chr( ']' ), spaces ) ) ),
                   Operator( NIL_, make_maybe ) ),
             bind( xthen( then( chr( '{' ), spaces ),
                                      thenx( expr,
                         then( chr( '}' ), spaces ) ) ),
                               Operator( NIL_, make_many ) ),
                         bind( xthen( then( chr( '(' ), spaces ),
                      thenx( expr,
                             then( chr( ')' ), spaces ) ) ),
                               Operator( NIL_, encapsulate ) ),
                         bind( xthen( chr( '/' ),
                                      thenx( regex_parser, chr( '/' ) ) ),
                               Operator( NIL_, encapsulate ) ) );
    parser term = bind( many( factor ), Operator( NIL_, make_sequence ) );
    *expr = *bind( then( term, many( xthen( choice_symbol, term ) ) ),
           Operator( NIL_, make_any ) );
  };
  parser definition = bind( then( symb,
                xthen( defining_symbol,
                   thenx( expr, terminating_symbol ) ) ),
                Operator( NIL_, encapsulate) );
  return  some( definition );
}


/* helpers */

static string
stringify( object env, object input ){
  return  to_string( input );
}

static symbol
symbolize( object env, object input ){
  return  symbol_from_string( to_string( input ) );
}

static list
encapsulate( object env, object input ){
  return  one( input );
}


/* syntax directed translation to list form */

static parser
make_matcher( object env, object input ){
  return  str( to_string( input )->String.str );
}

static list
make_sequence( object env, object input ){
  if(  length( input ) == 0  ) return  Symbol(EBNF_EPSILON);
  if(  length( input ) < 2  ) return  input;
  return  one( cons( Symbol(EBNF_SEQ), input ) );
}

static list
make_any( object env, object input ){
  if(  length( input ) < 2  ) return  input;
  return  one( cons( Symbol(EBNF_ANY), input ) );
}

static list
make_maybe( object env, object input ){
  return  one( cons( Symbol(EBNF_MAYBE), input ) );
}

static list
make_many( object env, object input ){
  return  one( cons( Symbol(EBNF_MANY), input ) );
}


/* stages of constructing the parsers from list form */

static list
define_forward( object env, object it ){
  if(  rest( it )->t == PARSER  ) return  it;
  return  cons( first( it ), forward() );
}

static parser
compile_bnf( object env, object it ){
  operator self = (union object[]){{.Operator={OPERATOR,env,compile_bnf}}};
  switch(  it->t  ){
  default:
    return  it;
  case SYMBOL: {
    object ob = assoc( it, env );
    return  valid( ob )  ? ob  : it;
  }
  case LIST:   {
    object f = first( it );
    if(  valid( eq_symbol( EBNF_SEQ, f ) )  )
      return  collapse( then,
            map( self, rest( it ) ) );
    if(  valid( eq_symbol( EBNF_ANY, f ) )  )
      return  collapse( either,
            map( self, rest( it ) ) );
    if(  valid( eq_symbol( EBNF_MANY, f ) )  )
      return  many( map( self, rest( it ) ) );
    if(  valid( eq_symbol( EBNF_MAYBE, f ) )  )
      return  maybe( map( self, rest( it ) ) );
    if(  length( it ) == 1  )
      return  compile_bnf( env, f );
    return  map( self, it );
  }
  }
}

static list
compile_rhs( object env, object it ){
  if(  rest( it )->t == PARSER  ) return  it;
  object result = cons( first( it ),
      map( (union object[]){{.Operator={OPERATOR,env,compile_bnf}}},
       rest( it ) ) );
  return  result;
}

static list
define_parser( object env, object it ){
  object lhs = assoc( first( it ), env );
  if(  valid( lhs ) && lhs->t == PARSER && lhs->Parser.f == NULL  ){
    object rhs = rest( it );
    if(  rhs->t == LIST  ) rhs = first( rhs );
    *lhs = *rhs;
  }
  return  it;
}

static list
wrap_handler( object env, object it ){
  object lhs = assoc( first( it ), env );
  if(  valid( lhs ) && lhs->t == PARSER  ){
    object op = rest( it );
    parser copy = Parser( 0, 0 );
    *copy = *lhs;
    *lhs = *bind( copy, op );
  }
  return  it;
}

pc11io.h

#define PC11IO_H
#if  ! PC11PARSER_H
  #include "pc11parser.h"
#endif

enum io_symbol_codes {
  ARGS = END_PARSER_SYMBOLS,
  END_IO_SYMBOLS
};

int pprintf( char const *fmt, ... );

int pscanf( char const *fmt, ... );

pc11io.c

code omitted for size

pc11test.h

#define PC11TEST_H
#if ! PC11IO_H
  #include "pc11io.h"
#endif

int main( void );

pc11test.c

#include <ctype.h>
#include "pc11test.h"

enum test_symbol_codes {
  TEST = END_IO_SYMBOLS,
  DIGIT,
  UPPER,
  NAME,
  NUMBER,
  EOL,
  SP,
  postal_address,
  name_part,
  street_address,
  street_name,
  zip_part,
  END_TEST_SYMBOLS
};

static int test_basics();
static int test_parsers();
static int test_regex();
static int test_ebnf();
static int test_io();

int main( void ){
  return  0
      ||  test_basics()
      ||  test_parsers()
      ||  test_regex()
      ||  test_ebnf()
      ||  test_io()
      ;
}

static fOperator to_upper;

static integer
to_upper( object env, integer it ){
  return  Int( toupper( it->Int.i ) );
}

static int
test_basics(){
  puts( __func__ );
  list ch = chars_from_str( "abcdef" );
    print( ch ), puts("");
    print_list( ch ), puts("");
  integer a = apply( Operator( NIL_, to_upper ), first( ch ) );
    print( a ), puts("");
  drop( 1, a );
    print( a ), puts("");
  drop( 6, ch );
    print( ch ), puts("");
    print_list( ch ), puts("");
  drop( 7, ch );
    print( ch ), puts("");
    print_list( ch ), puts("");
  puts("");
  list xs = infinite( Int('x') );
    print_list( xs ), puts("");
  drop( 3, xs );
    print_list( xs ), puts("");
  puts("");
  return  0;
}

static int
test_parsers(){
  puts( __func__ );
  list ch = chars_from_str( "a b c d 1 2 3 4" );
  parser p = succeeds( Int('*') );
    print_list( parse( p, ch ) ), puts("");
  parser q = fails( String("Do you want a cookie?",0) );
    print_list( parse( q, ch ) ), puts("");
  parser r = item();
    print_list( parse( r, ch ) ), puts("");
  parser s = either( alpha(), item() );
    print_list( parse( s, ch ) ), puts("");
  parser t = literal( Int('a') );
    print_list( parse( t, ch ) ), puts("");
  puts("");
  return  0;
}

static int
test_regex(){
  puts( __func__ );
  parser a = regex( "." );
  print_list( a ), puts("");
  print_list( parse( a, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( a, chars_from_str( "." ) ) ), puts("");
  print_list( parse( a, chars_from_str( "\\." ) ) ), puts("");
  puts("");

  parser b = regex( "\\." );
  print_list( b ), puts("");
  print_list( parse( b, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( b, chars_from_str( "." ) ) ), puts("");
  print_list( parse( b, chars_from_str( "\\." ) ) ), puts("");
  puts("");

  parser c = regex( "\\\\." );
  print_list( c ), puts("");
  print_list( parse( c, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( c, chars_from_str( "." ) ) ), puts("");
  print_list( parse( c, chars_from_str( "\\." ) ) ), puts("");
  print_list( parse( c, chars_from_str( "\\a" ) ) ), puts("");
  puts("");

  parser d = regex( "\\\\\\." );
  print_list( d ), puts("");
  print_list( parse( d, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( d, chars_from_str( "." ) ) ), puts("");
  print_list( parse( d, chars_from_str( "\\." ) ) ), puts("");
  print_list( parse( d, chars_from_str( "\\a" ) ) ), puts("");
  puts("");

  parser e = regex( "\\\\|a" );
  print_list( e ), puts("");
  print_list( parse( e, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( e, chars_from_str( "." ) ) ), puts("");
  print_list( parse( e, chars_from_str( "\\." ) ) ), puts("");
  print_list( parse( e, chars_from_str( "\\a" ) ) ), puts("");
  puts("");
  
  parser f = regex( "[abcd]" );
  print_list( f ), puts("");
  print_list( parse( f, chars_from_str( "a" ) ) ), puts("");
  print_list( parse( f, chars_from_str( "." ) ) ), puts("");
  puts("");

  return  0;
}

static fOperator stringify;

static string
stringify( object env, list it ){
  return  to_string( it );
}

static int
test_ebnf(){
  puts( __func__ );
  Symbol(postal_address);
  Symbol(name_part);
  Symbol(street_address);
  Symbol(street_name);
  Symbol(zip_part);

  list parsers = ebnf(
    "postal_address = name_part street_address zip_part ;\n"
    "name_part = personal_part SP last_name SP opt_suffix_part EOL\n"
    "          | personal_part SP name_part ;\n"
    "personal_part = initial '.' | first_name ;\n"
    "street_address = house_num SP street_name opt_apt_num EOL ;\n"
    "zip_part = town_name ',' SP state_code SP zip_code EOL ;\n"
    "opt_suffix_part = 'Sr.' | 'Jr.' | roman_numeral | ;\n"
    "opt_apt_num = [ apt_num ] ;\n"
    "apt_num = NUMBER ;\n"
    "town_name = NAME ;\n"
    "state_code = UPPER UPPER ;\n"
    "zip_code = DIGIT DIGIT DIGIT DIGIT DIGIT ;\n"
    "initial = 'Mrs' | 'Mr' | 'Ms' | 'M' ;\n"
    "roman_numeral = 'I' [ 'V' | 'X' ] { 'I' } ;\n"
    "first_name = NAME ;\n"
    "last_name = NAME ;\n"
    "house_num = NUMBER ;\n"
    "street_name = NAME ;\n",
    env( NIL_, 6,
     Symbol(EOL), chr('\n'),
     Symbol(DIGIT), digit(),
     Symbol(UPPER), upper(),
     Symbol(NUMBER), some( digit() ),
         Symbol(NAME), some( alpha() ),
     Symbol(SP), many( anyof( " \t\n" ) ) ),
    env( NIL_, 2,
         Symbol(name_part), Operator( NIL_, stringify ),
         Symbol(street_name), Operator( NIL_, stringify ) )
  );

  parser start = assoc_symbol( postal_address, parsers );
  if(  valid( start ) && start->t == LIST  )
    start = first( start );
  print_list( start ), puts("\n");
  
  print_list( parse( start,
      chars_from_str( "Mr. luser droog I\n"
                      "2357 Streetname\n"
                      "Anytown, ST 00700\n" ) ) ),
    puts("");

  printf( "%d objects\n", count_allocations() );
  return  0;
}

static int
test_io(){
  pprintf( "%s:%c-%c\n", "does it work?", '*', '@' );
  return  0;
}

Size:

$ make count
wc -l -c -L pc11*[ch] ppnarg.h
  180  4442    78 pc11io.c
   13   218    36 pc11io.h
  549 13731    77 pc11object.c
  361  5955    77 pc11object.h
  818 20944    80 pc11parser.c
  214  3601    63 pc11parser.h
  202  5453    69 pc11test.c
    6    82    21 pc11test.h
   29  1018    83 ppnarg.h
 2372 55444    83 total
cloc pc11*[ch] ppnarg.h
       9 text files.
       9 unique files.                              
       0 files ignored.

github.com/AlDanial/cloc v 1.93  T=0.05 s (194.9 files/s, 51356.5 lines/s)
-------------------------------------------------------------------------------
Language                     files          blank        comment           code
-------------------------------------------------------------------------------
C                                4            316             98           1335
C/C++ Header                     5            241             99            283
-------------------------------------------------------------------------------
SUM:                             9            557            197           1618
-------------------------------------------------------------------------------

Any improvements to make to the interface, implementation, or documentation?

\$\endgroup\$
5
  • \$\begingroup\$ Why so many spaces between parts of a line of code where 1 space will do? if( it->t == LIST && ! valid( rest( it ) ) ) it = first( it ); \$\endgroup\$
    – chux
    Commented Jul 1, 2022 at 14:55
  • \$\begingroup\$ Just a few conventions I've adopted. Adding 2 spaces inside the parens of if and while and for statements helps it to look different than a function call (which I write with just one space inside the parens). A separate, but related, convention is to add extra spaces in long or complex expressions to help the operator with the lowest precedence to stand out. I tend to do this with &&, ||, and to the left of ? and :. I find it less noisy than adding extra parens, but the intent is similar: to help the reader to see how the expression will be parsed. \$\endgroup\$ Commented Jul 2, 2022 at 1:08
  • 2
    \$\begingroup\$ luser droog., manual formatting is simply not productive. If you can tailor/write a code formatter to match your style --> great. Otherwise, use an auto-formatter. \$\endgroup\$
    – chux
    Commented Jul 2, 2022 at 4:32
  • 1
    \$\begingroup\$ Agreed. Part of the very purpose of this code is to build such a formatter. I felt like I needed something this flexible to even express a style rule like: "make sure the spacing hierarchy matches the precedence hierarchy". \$\endgroup\$ Commented Jul 3, 2022 at 3:51
  • 1
    \$\begingroup\$ The vertical spacing definitely helps! \$\endgroup\$
    – pacmaninbw
    Commented Jul 3, 2022 at 13:29

1 Answer 1

3
\$\begingroup\$

Makefile improvements

At first glance, the makefile looks OK, but a few improvements can be made. First, you always override CFLAGS, although you allow to add things to be added to it via the environment variable $cflags. However, that lower case form is very non-standard, and it's more common to expect CFLAGS=... make to work. The usual solution is this:

CFLAGS ?= -g -Wall -Wpedantic -Wextra -Wno-unused-function -Wno-unused-parameter -Wno-switch -Wno-return-type -Wunused-variable
CFLAGS += -std=c99

Where in the first line, we only add those options if no CFLAGS were provided via the environment, and in the second line we unconditionally add any required flags for the build to work.

Second, targets that don't build anything but just run commands should be marked as .PHONY, so if you accidentily created a file test that has a timestamp newer than pc11test, it wouldn't prevent make test from working as expect. So add:

.PHONY: test clean count

About forward declarations

I've added forward declarations for all the static functions inside the .c files so all the static "helper functions" can be placed below the non-static function that uses them, so the implementation can be presented in a more top down fashion overall.

Personally I don't think that is very helpful. Now you have both the forward declaration and the actual definition to keep in sync. The files are long enough that you are going to use search functionality anyway to jump to functions.

Documentation

It is great that you are documenting all the functions and also having them grouped in a sensible way. However, I recommend that you write these documentation comments in Doxygen format. Doxygen is a widely used standard for documenting C and C++ code, and the Doxygen tools can then do all kinds of nice stuff for you: apart from generating documentation in navigatable PDF and HTML formats, it can also warn you when you forgot to document functions and/or function parameters.

Naming things

Especially in pc11object.h, I am a bit surprised by some of the function names, in particular when the comments above them describe those in terms that don't match the function name itself.

For example, drop() has as documentation "Skip ahead n elements". Why not call the function skip() then, or alternatively, write "Drop the first n elements" in the comments. There are more examples of this, like map() "Transform", collapse()/reduce() "Fold", env() "Prepend", and so on.

Some comments don't make sense at all to a C programmer, like first() being documented as "car". If you don't know LISP, you might think "what does this have to do with automobiles?".

Some comments are needlessly complicated, like append() being documented as "return copy of start sharing end". But it also raises questions: does this append one list to another like the function name implies, or does it create a new list that is the concatenation of two lists like the comments hint at?

Make sure the function name, while concise, conveys clearly what is is going to do, and make sure the documentation matches. I would go over all the function names and make changes where appropriate. For example, instead of having to remember whether collapse() is for lists and reduce() is for arrays of objects, why not make them fold_list() and fold_objects()?

\$\endgroup\$
3
  • 1
    \$\begingroup\$ "comments in Doxygen format. Doxygen is the de facto standard for documenting C and C++ code" --> Any support for the de facto assertion? \$\endgroup\$
    – chux
    Commented Jul 1, 2022 at 14:57
  • \$\begingroup\$ @chux-ReinstateMonica They say so on their website, so it must be true! I can't find any authorative page on the popularity of each documentation generator for C though, but in my personal experience it is the most commonly used, and just searching for "C documentation generator" or "C code documentation" lists Doxygen or sites referring to Doxygen at the top. \$\endgroup\$
    – G. Sliepen
    Commented Jul 1, 2022 at 15:26
  • \$\begingroup\$ UV for an Appeal to Popular Belief. \$\endgroup\$
    – chux
    Commented Jul 1, 2022 at 17:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.