label.y

早期freebsd实现

/* -*- C++ -*-
   Copyright (C) 1989, 1990, 1991, 1992 Free Software Foundation, Inc.
     Written by James Clark ([e-mail])

This file is part of groff.

groff 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.

groff 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 groff; see the file COPYING.  If not, write to the Free Software
Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */

%{

#include "refer.h"
#include "refid.h"
#include "ref.h"
#include "token.h"

int yylex();
void yyerror(const char *);
int yyparse();

static const char *format_serial(char c, int n);

struct label_info {
  int start;
  int length;
  int count;
  int total;
  label_info(const string &);
};

label_info *lookup_label(const string &label);

struct expression {
  enum {
    // Does the tentative label depend on the reference?
    CONTAINS_VARIABLE = 01, 
    CONTAINS_STAR = 02,
    CONTAINS_FORMAT = 04,
    CONTAINS_AT = 010
  };
  virtual ~expression() { }
  virtual void evaluate(int, const reference &, string &,
			substring_position &) = 0;
  virtual unsigned analyze() { return 0; }
};

class at_expr : public expression {
public:
  at_expr() { }
  void evaluate(int, const reference &, string &, substring_position &);
  unsigned analyze() { return CONTAINS_VARIABLE|CONTAINS_AT; }
};

class format_expr : public expression {
  char type;
  int width;
  int first_number;
public:
  format_expr(char c, int w = 0, int f = 1)
    : type(c), width(w), first_number(f) { }
  void evaluate(int, const reference &, string &, substring_position &);
  unsigned analyze() { return CONTAINS_FORMAT; }
};

class field_expr : public expression {
  int number;
  char name;
public:
  field_expr(char nm, int num) : name(nm), number(num) { }
  void evaluate(int, const reference &, string &, substring_position &);
  unsigned analyze() { return CONTAINS_VARIABLE; }
};

class literal_expr : public expression {
  string s;
public:
  literal_expr(const char *ptr, int len) : s(ptr, len) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class unary_expr : public expression {
protected:
  expression *expr;
public:
  unary_expr(expression *e) : expr(e) { }
  ~unary_expr() { delete expr; }
  void evaluate(int, const reference &, string &, substring_position &) = 0;
  unsigned analyze() { return expr ? expr->analyze() : 0; }
};

// This caches the analysis of an expression.

class analyzed_expr : public unary_expr {
  unsigned flags;
public:
  analyzed_expr(expression *);
  void evaluate(int, const reference &, string &, substring_position &);
  unsigned analyze() { return flags; }
};

class star_expr : public unary_expr {
public:
  star_expr(expression *e) : unary_expr(e) { }
  void evaluate(int, const reference &, string &, substring_position &);
  unsigned analyze() {
    return ((expr ? (expr->analyze() & ~CONTAINS_VARIABLE) : 0)
	    | CONTAINS_STAR);
  }
};

typedef void map_t(const char *, const char *, string &);

class map_expr : public unary_expr {
  map_t *func;
public:
  map_expr(expression *e, map_t *f) : unary_expr(e), func(f) { }
  void evaluate(int, const reference &, string &, substring_position &);
};
  
typedef const char *extractor_t(const char *, const char *, const char **);

class extractor_expr : public unary_expr {
  int part;
  extractor_t *func;
public:
  enum { BEFORE = +1, MATCH = 0, AFTER = -1 };
  extractor_expr(expression *e, extractor_t *f, int pt)
    : unary_expr(e), func(f), part(pt) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class truncate_expr : public unary_expr {
  int n;
public:
  truncate_expr(expression *e, int i) : n(i), unary_expr(e) { } 
  void evaluate(int, const reference &, string &, substring_position &);
};

class separator_expr : public unary_expr {
public:
  separator_expr(expression *e) : unary_expr(e) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class binary_expr : public expression {
protected:
  expression *expr1;
  expression *expr2;
public:
  binary_expr(expression *e1, expression *e2) : expr1(e1), expr2(e2) { }
  ~binary_expr() { delete expr1; delete expr2; }
  void evaluate(int, const reference &, string &, substring_position &) = 0;
  unsigned analyze() {
    return (expr1 ? expr1->analyze() : 0) | (expr2 ? expr2->analyze() : 0);
  }
};

class alternative_expr : public binary_expr {
public:
  alternative_expr(expression *e1, expression *e2) : binary_expr(e1, e2) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class list_expr : public binary_expr {
public:
  list_expr(expression *e1, expression *e2) : binary_expr(e1, e2) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class substitute_expr : public binary_expr {
public:
  substitute_expr(expression *e1, expression *e2) : binary_expr(e1, e2) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

class ternary_expr : public expression {
protected:
  expression *expr1;
  expression *expr2;
  expression *expr3;
public:
  ternary_expr(expression *e1, expression *e2, expression *e3)
    : expr1(e1), expr2(e2), expr3(e3) { }
  ~ternary_expr() { delete expr1; delete expr2; delete expr3; }
  void evaluate(int, const reference &, string &, substring_position &) = 0;
  unsigned analyze() {
    return ((expr1 ? expr1->analyze() : 0)
	    | (expr2 ? expr2->analyze() : 0)
	    | (expr3 ? expr3->analyze() : 0));
  }
};

class conditional_expr : public ternary_expr {
public:
  conditional_expr(expression *e1, expression *e2, expression *e3)
    : ternary_expr(e1, e2, e3) { }
  void evaluate(int, const reference &, string &, substring_position &);
};

static expression *parsed_label = 0;
static expression *parsed_date_label = 0;
static expression *parsed_short_label = 0;

static expression *parse_result;

string literals;

%}

%union {
  int num;
  expression *expr;
  struct { int ndigits; int val; } dig;
  struct { int start; int len; } str;
}

/* uppercase or lowercase letter */
%token <num> TOKEN_LETTER
/* literal characters */
%token <str> TOKEN_LITERAL
/* digit */
%token <num> TOKEN_DIGIT

%type <expr> conditional
%type <expr> alternative
%type <expr> list
%type <expr> string
%type <expr> substitute
%type <expr> optional_conditional
%type <num> number
%type <dig> digits
%type <num> optional_number
%type <num> flag

%%

expr:
	optional_conditional
		{ parse_result = ($1 ? new analyzed_expr($1) : 0); }
	;

conditional:
	alternative
		{ $$ = $1; }
	| alternative '?' optional_conditional ':' conditional
		{ $$ = new conditional_expr($1, $3, $5); }
	;

optional_conditional:
	/* empty */
		{ $$ = 0; }
	| conditional
		{ $$ = $1; }
	;

alternative:
	list
		{ $$ = $1; }
	| alternative '|' list
		{ $$ = new alternative_expr($1, $3); }
	| alternative '&' list
		{ $$ = new conditional_expr($1, $3, 0); }
	;	

list:
	substitute
		{ $$ = $1; }
	| list substitute
		{ $$ = new list_expr($1, $2); }
	;

substitute:
	string
		{ $$ = $1; }
	| substitute '~' string
		{ $$ = new substitute_expr($1, $3); }
	;

string:
	'@'
		{ $$ = new at_expr; }
	| TOKEN_LITERAL
		{
		  $$ = new literal_expr(literals.contents() + $1.start,
					$1.len);
		}
	| TOKEN_LETTER
		{ $$ = new field_expr($1, 0); }
	| TOKEN_LETTER number
		{ $$ = new field_expr($1, $2 - 1); }
	| '%' TOKEN_LETTER
		{
		  switch ($2) {
		  case 'I':
		  case 'i':
		  case 'A':
		  case 'a':
		    $$ = new format_expr($2);
		    break;
		  default:
		    command_error("unrecognized format `%1'", char($2));
		    $$ = new format_expr('a');
		    break;
		  }
		}
	
	| '%' digits
		{
		  $$ = new format_expr('0', $2.ndigits, $2.val);
		}
	| string '.' flag TOKEN_LETTER optional_number
		{
		  switch ($4) {
		  case 'l':
		    $$ = new map_expr($1, lowercase);
		    break;
		  case 'u':
		    $$ = new map_expr($1, uppercase);
		    break;
		  case 'c':
		    $$ = new map_expr($1, capitalize);
		    break;
		  case 'r':
		    $$ = new map_expr($1, reverse_name);
		    break;
		  case 'a':
		    $$ = new map_expr($1, abbreviate_name);
		    break;
		  case 'y':
		    $$ = new extractor_expr($1, find_year, $3);
		    break;
		  case 'n':
		    $$ = new extractor_expr($1, find_last_name, $3);
		    break;
		  default:
		    $$ = $1;
		    command_error("unknown function `%1'", char($4));
		    break;
		  }
		}

	| string '+' number
		{ $$ = new truncate_expr($1, $3); }
	| string '-' number
		{ $$ = new truncate_expr($1, -$3); }
	| string '*'
		{ $$ = new star_expr($1); }
	| '(' optional_conditional ')'
		{ $$ = $2; }
	| '<' optional_conditional '>'
		{ $$ = new separator_expr($2); }
	;

optional_number:
	/* empty */
		{ $$ = -1; }
	| number
		{ $$ = $1; }
	;

number:
	TOKEN_DIGIT
		{ $$ = $1; }
	| number TOKEN_DIGIT
		{ $$ = $1*10 + $2; }
	;

digits:
	TOKEN_DIGIT
		{ $$.ndigits = 1; $$.val = $1; }
	| digits TOKEN_DIGIT
		{ $$.ndigits = $1.ndigits + 1; $$.val = $1.val*10 + $2; }
	;
	
      
flag:
	/* empty */
		{ $$ = 0; }
	| '+'
		{ $$ = 1; }
	| '-'
		{ $$ = -1; }
	;

%%

/* bison defines const to be empty unless __STDC__ is defined, which it
isn't under cfront */

#ifdef const
#undef const
#endif

const char *spec_ptr;
const char *spec_end;
const char *spec_cur;

int yylex()
{
  while (spec_ptr < spec_end && csspace(*spec_ptr))
    spec_ptr++;
  spec_cur = spec_ptr;
  if (spec_ptr >= spec_end)
    return 0;
  unsigned char c = *spec_ptr++;
  if (csalpha(c)) {
    yylval.num = c;
    return TOKEN_LETTER;
  }
  if (csdigit(c)) {
    yylval.num = c - '0';
    return TOKEN_DIGIT;
  }
  if (c == '\'') {
    yylval.str.start = literals.length();
    for (; spec_ptr < spec_end; spec_ptr++) {
      if (*spec_ptr == '\'') {
	if (++spec_ptr < spec_end && *spec_ptr == '\'')
	  literals += '\'';
	else {
	  yylval.str.len = literals.length() - yylval.str.start;
	  return TOKEN_LITERAL;
	}
      }
      else
	literals += *spec_ptr;
    }
    yylval.str.len = literals.length() - yylval.str.start;
    return TOKEN_LITERAL;
  }
  return c;
}

int set_label_spec(const char *label_spec)
{
  spec_cur = spec_ptr = label_spec;
  spec_end = strchr(label_spec, '\0');
  literals.clear();
  if (yyparse())
    return 0;
  delete parsed_label;
  parsed_label = parse_result;
  return 1;
}

int set_date_label_spec(const char *label_spec)
{
  spec_cur = spec_ptr = label_spec;
  spec_end = strchr(label_spec, '\0');
  literals.clear();
  if (yyparse())
    return 0;
  delete parsed_date_label;
  parsed_date_label = parse_result;
  return 1;
}

int set_short_label_spec(const char *label_spec)
{
  spec_cur = spec_ptr = label_spec;
  spec_end = strchr(label_spec, '\0');
  literals.clear();
  if (yyparse())
    return 0;
  delete parsed_short_label;
  parsed_short_label = parse_result;
  return 1;
}

void yyerror(const char *message)
{
  if (spec_cur < spec_end)
    command_error("label specification %1 before `%2'", message, spec_cur);
  else
    command_error("label specification %1 at end of string",
		  message, spec_cur);
}

void at_expr::evaluate(int tentative, const reference &ref,
		       string &result, substring_position &)
{
  if (tentative)
    ref.canonicalize_authors(result);
  else {
    const char *end, *start = ref.get_authors(&end);
    if (start)
      result.append(start, end - start);
  }
}

void format_expr::evaluate(int tentative, const reference &ref,
			   string &result, substring_position &)
{
  if (tentative)
    return;
  const label_info *lp = ref.get_label_ptr();
  int num = lp == 0 ? ref.get_number() : lp->count;
  if (type != '0')
    result += format_serial(type, num + 1);
  else {
    const char *ptr = itoa(num + first_number);
    int pad = width - strlen(ptr);
    while (--pad >= 0)
      result += '0';
    result += ptr;
  }
}

static const char *format_serial(char c, int n)
{
  assert(n > 0);
  static char buf[128]; // more than enough.
  switch (c) {
  case 'i':
  case 'I':
    {
      char *p = buf;
      // troff uses z and w to represent 10000 and 5000 in Roman
      // numerals; I can find no historical basis for this usage
      const char *s = c == 'i' ? "zwmdclxvi" : "ZWMDCLXVI";
      if (n >= 40000)
	return itoa(n);
      while (n >= 10000) {
	*p++ = s[0];
	n -= 10000;
      }
      for (int i = 1000; i > 0; i /= 10, s += 2) {
	int m = n/i;
	n -= m*i;
	switch (m) {
	case 3:
	  *p++ = s[2];
	  /* falls through */
	case 2:
	  *p++ = s[2];
	  /* falls through */
	case 1:
	  *p++ = s[2];
	  break;
	case 4:
	  *p++ = s[2];
	  *p++ = s[1];
	  break;
	case 8:
	  *p++ = s[1];
	  *p++ = s[2];
	  *p++ = s[2];
	  *p++ = s[2];
	  break;
	case 7:
	  *p++ = s[1];
	  *p++ = s[2];
	  *p++ = s[2];
	  break;
	case 6:
	  *p++ = s[1];
	  *p++ = s[2];
	  break;
	case 5:
	  *p++ = s[1];
	  break;
	case 9:
	  *p++ = s[2];
	  *p++ = s[0];
	}
      }
      *p = 0;
      break;
    }
  case 'a':
  case 'A':
    {
      char *p = buf;
      // this is derived from troff/reg.c
      while (n > 0) {
	int d = n % 26;