cweave.w

模拟器提供了一个简单易用的平台

line by putting a |'%'| just before the last character.

@<Print warning message...@>=
{
  printf("\n! Line had to be broken (output l. %d):\n",out_line);
@.Line had to be broken@>
  term_write(out_buf+1, out_ptr-out_buf-1);
  new_line; mark_harmless;
  flush_buffer(out_ptr-1,1,1); return;
}

@ Here is a macro that outputs a section number in decimal notation.
The number to be converted by |out_section| is known to be less than
|def_flag|, so it cannot have more than five decimal digits.  If
the section is changed, we output `\.{\\*}' just after the number.

@c
void
out_section(n)
sixteen_bits n;
{
  char s[6];
  sprintf(s,"%d",n); out_str(s);
  if(changed_section[n]) out_str ("\\*");
@.\\*@>
}

@ The |out_name| procedure is used to output an identifier or index
entry, enclosing it in braces.

@c
void
out_name(p,quote_xalpha)
name_pointer p;
boolean quote_xalpha;
{
  char *k, *k_end=(p+1)->byte_start; /* pointers into |byte_mem| */
  out('{');
  for (k=p->byte_start; k<k_end; k++) {
    if (isxalpha(*k) && quote_xalpha) out('\\');
@.\\\$@>
@.\\\_@>
    out(*k);
  }
  out('}');
}

@* Routines that copy \TEX/ material.
During phase two, we use subroutines |copy_limbo|, |copy_TeX|, and
|copy_comment| in place of the analogous |skip_limbo|, |skip_TeX|, and
|skip_comment| that were used in phase one. (Well, |copy_comment|
was actually written in such a way that it functions as |skip_comment|
in phase one.)

The |copy_limbo| routine, for example, takes \TEX/ material that is not
part of any section and transcribes it almost verbatim to the output file.
The use of `\.{@@}' signs is severely restricted in such material:
`\.{@@@@}' pairs are replaced by singletons; `\.{@@l}' and `\.{@@q}' and
`\.{@@s}' are interpreted.

@c
void
copy_limbo()
{
  char c;
  while (1) {
    if (loc>limit && (finish_line(), get_line()==0)) return;
    *(limit+1)='@@';
    while (*loc!='@@') out(*(loc++));
    if (loc++<=limit) {
      c=*loc++;
      if (ccode[(eight_bits)c]==new_section) break;
      switch (ccode[(eight_bits)c]) {
        case translit_code: out_str("\\ATL"); break;
@.\\ATL@>
        case '@@': out('@@'); break;
        case noop: skip_restricted(); break;
        case format_code: if (get_next()==identifier) get_next();
          if (loc>=limit) get_line(); /* avoid blank lines in output */
          break; /* the operands of \.{@@s} are ignored on this pass */
        default: err_print("! Double @@ should be used in limbo");
@.Double @@ should be used...@>
        out('@@');
      }
    }
  }
}

@ The |copy_TeX| routine processes the \TEX/ code at the beginning of a
section; for example, the words you are now reading were copied in this
way. It returns the next control code or `\.{\v}' found in the input.
We don't copy spaces or tab marks into the beginning of a line. This
makes the test for empty lines in |finish_line| work.

@ @f copy_TeX TeX
@c
eight_bits
copy_TeX()
{
  char c; /* current character being copied */
  while (1) {
    if (loc>limit && (finish_line(), get_line()==0)) return(new_section);
    *(limit+1)='@@';
    while ((c=*(loc++))!='|' && c!='@@') {
      out(c);
      if (out_ptr==out_buf+1 && (xisspace(c))) out_ptr--;
    }
    if (c=='|') return('|');
    if (loc<=limit) return(ccode[(eight_bits)*(loc++)]);
  }
}

@ The |copy_comment| function issues a warning if more braces are opened than
closed, and in the case of a more serious error it supplies enough
braces to keep \TEX/ from complaining about unbalanced braces.
Instead of copying the \TEX/ material
into the output buffer, this function copies it into the token memory
(in phase two only).
The abbreviation |app_tok(t)| is used to append token |t| to the current
token list, and it also makes sure that it is possible to append at least
one further token without overflow.

@d app_tok(c) {if (tok_ptr+2>tok_mem_end) overflow("token"); *(tok_ptr++)=c;}

@<Predec...@>=
int copy_comment();

@ @c
int copy_comment(is_long_comment,bal) /* copies \TEX/ code in comments */
boolean is_long_comment; /* is this a traditional \CEE/ comment? */
int bal; /* brace balance */
{
  char c; /* current character being copied */
  while (1) {
    if (loc>limit) {
      if (is_long_comment) {
        if (get_line()==0) {
          err_print("! Input ended in mid-comment");
@.Input ended in mid-comment@>
          loc=buffer+1; goto done;
        }
      }
      else {
        if (bal>1) err_print("! Missing } in comment");
@.Missing \} in comment@>
        goto done;
      }
    }
    c=*(loc++);
    if (c=='|') return(bal);
    if (is_long_comment) @<Check for end of comment@>;
    if (phase==2) {
      if (ishigh(c)) app_tok(quoted_char);
      app_tok(c);
    }
    @<Copy special things when |c=='@@', '\\'|@>;
    if (c=='{') bal++;
    else if (c=='}') {
      if(bal>1) bal--;
      else {err_print("! Extra } in comment");
@.Extra \} in comment@>
        if (phase==2) tok_ptr--;
      }
    }
  }
done:@<Clear |bal| and |return|@>;
}

@ @<Check for end of comment@>=
if (c=='*' && *loc=='/') {
  loc++;
  if (bal>1) err_print("! Missing } in comment");
@.Missing \} in comment@>
  goto done;
}

@ @<Copy special things when |c=='@@'...@>=
if (c=='@@') {
  if (*(loc++)!='@@') {
    err_print("! Illegal use of @@ in comment");
@.Illegal use of @@...@>
    loc-=2; if (phase==2) *(tok_ptr-1)=' '; goto done;
  }
}
else if (c=='\\' && *loc!='@@')
  if (phase==2) app_tok(*(loc++)) else loc++;

@ We output
enough right braces to keep \TEX/ happy.

@<Clear |bal|...@>=
if (phase==2) while (bal-- >0) app_tok('}');
return(0);

@** Parsing.
The most intricate part of \.{CWEAVE} is its mechanism for converting
\CEE/-like code into \TEX/ code, and we might as well plunge into this
aspect of the program now. A ``bottom up'' approach is used to parse the
\CEE/-like material, since \.{CWEAVE} must deal with fragmentary
constructions whose overall ``part of speech'' is not known.

At the lowest level, the input is represented as a sequence of entities
that we shall call {\it scraps}, where each scrap of information consists
of two parts, its {\it category} and its {\it translation}. The category
is essentially a syntactic class, and the translation is a token list that
represents \TEX/ code. Rules of syntax and semantics tell us how to
combine adjacent scraps into larger ones, and if we are lucky an entire
\CEE/ text that starts out as hundreds of small scraps will join
together into one gigantic scrap whose translation is the desired \TEX/
code. If we are unlucky, we will be left with several scraps that don't
combine; their translations will simply be output, one by one.

The combination rules are given as context-sensitive productions that are
applied from left to right. Suppose that we are currently working on the
sequence of scraps $s_1\,s_2\ldots s_n$. We try first to find the longest
production that applies to an initial substring $s_1\,s_2\ldots\,$; but if
no such productions exist, we try to find the longest production
applicable to the next substring $s_2\,s_3\ldots\,$; and if that fails, we
try to match $s_3\,s_4\ldots\,$, etc.

A production applies if the category codes have a given pattern. For
example, one of the productions (see rule~3) is
$$\hbox{|exp| }\left\{\matrix{\hbox{|binop|}\cr\hbox{|ubinop|}}\right\}
\hbox{ |exp| }\RA\hbox{ |exp|}$$
and it means that three consecutive scraps whose respective categories are
|exp|, |binop| (or |ubinop|),
and |exp| are converted to one scrap whose category
is |exp|.  The translations of the original
scraps are simply concatenated.  The case of
$$\hbox{|exp| |comma| |exp| $\RA$ |exp|} \hskip4emE_1C\,\\{opt}9\,E_2$$
(rule 4) is only slightly more complicated:
Here the resulting |exp| translation
consists not only of the three original translations, but also of the
tokens |opt| and 9 between the translations of the
|comma| and the following |exp|.
In the \TEX/ file, this will specify an optional line break after the
comma, with penalty 90.

At each opportunity the longest possible production is applied.  For
example, if the current sequence of scraps is |int_like| |cast|
|lbrace|, rule 31 is applied; but if the sequence is |int_like| |cast|
followed by anything other than |lbrace|, rule 32 takes effect.

Translation rules such as `$E_1C\,\\{opt}9\,E_2$' above use subscripts
to distinguish between translations of scraps whose categories have the
same initial letter; these subscripts are assigned from left to right.

@ Here is a list of the category codes that scraps can have.
(A few others, like |int_like|, have already been defined; the
|cat_name| array contains a complete list.)

@d exp 1 /* denotes an expression, including perhaps a single identifier */
@d unop 2 /* denotes a unary operator */
@d binop 3 /* denotes a binary operator */
@d ubinop 4
  /* denotes an operator that can be unary or binary, depending on context */
@d cast 5 /* denotes a cast */
@d question 6 /* denotes a question mark and possibly the expressions flanking it */
@d lbrace 7 /* denotes a left brace */
@d rbrace 8 /* denotes a right brace */
@d decl_head 9 /* denotes an incomplete declaration */
@d comma 10 /* denotes a comma */
@d lpar 11 /* denotes a left parenthesis or left bracket */
@d rpar 12 /* denotes a right parenthesis or right bracket */
@d prelangle 13 /* denotes `$<$' before we know what it is */
@d prerangle 14 /* denotes `$>$' before we know what it is */
@d langle 15 /* denotes `$<$' when it's used as angle bracket in a template */
@d colcol 18 /* denotes `::' */
@d base 19 /* denotes a colon that introduces a base specifier */
@d decl 20 /* denotes a complete declaration */
@d struct_head 21 /* denotes the beginning of a structure specifier */
@d stmt 23 /* denotes a complete statement */
@d function 24 /* denotes a complete function */
@d fn_decl 25 /* denotes a function declarator */
@d semi 27 /* denotes a semicolon */
@d colon 28 /* denotes a colon */
@d tag 29 /* denotes a statement label */
@d if_head 30 /* denotes the beginning of a compound conditional */
@d else_head 31 /* denotes a prefix for a compound statement */
@d if_clause 32 /* pending \.{if} together with a condition */
@d lproc 35 /* begins a preprocessor command */
@d rproc 36 /* ends a preprocessor command */
@d insert 37 /* a scrap that gets combined with its neighbor */
@d section_scrap 38 /* section name */
@d dead 39 /* scrap that won't combine */
@d ftemplate 59 /* \\{make\_pair} */
@d new_exp 60 /* \&{new} and a following type identifier */
@d begin_arg 61 /* \.{@@[} */
@d end_arg 62 /* \.{@@]} */

@<Glo...@>=
char cat_name[256][12];
eight_bits cat_index;

@ @<Set in...@>=
    for (cat_index=0;cat_index<255;cat_index++)
      strcpy(cat_name[cat_index],"UNKNOWN");
@.UNKNOWN@>
    strcpy(cat_name[exp],"exp");
    strcpy(cat_name[unop],"unop");
    strcpy(cat_name[binop],"binop");
    strcpy(cat_name[ubinop],"ubinop");
    strcpy(cat_name[cast],"cast");
    strcpy(cat_name[question],"?");
    strcpy(cat_name[lbrace],"{"@q}@>);
    strcpy(cat_name[rbrace],@q{@>"}");
    strcpy(cat_name[decl_head],"decl_head");
    strcpy(cat_name[comma],",");
    strcpy(cat_name[lpar],"(");
    strcpy(cat_name[rpar],")");
    strcpy(cat_name[prelangle],"<");
    strcpy(cat_name[prerangle],">");
    strcpy(cat_name[langle],"\\<");
    strcpy(cat_name[colcol],"::");
    strcpy(cat_name[base],"\\:");
    strcpy(cat_name[decl],"decl");
    strcpy(cat_name[struct_head],"struct_head");
    strcpy(cat_name[alfop],"alfop");
    strcpy(cat_name[stmt],"stmt");
    strcpy(cat_name[function],"function");
    strcpy(cat_name[fn_decl],"fn_decl");
    strcpy(cat_name[else_like],"else_like");
    strcpy(cat_name[semi],";");
    strcpy(cat_name[colon],":");
    strcpy(cat_name[tag],"tag");
    strcpy(cat_name[if_head],"if_head");
    strcpy(cat_name[else_head],"else_head");
    strcpy(cat_name[if_clause],"if()");
    strcpy(cat_name[lproc],"#{"@q}@>);
    strcpy(cat_name[rproc],@q{@>"#}");
    strcpy(cat_name[insert],"insert");
    strcpy(cat_name[section_scrap],"section");
    strcpy(cat_name[dead],"@@d");
    strcpy(cat_name[public_like],"public");
    strcpy(cat_name[operator_like],"operator");
    strcpy(cat_name[new_like],"new");
    strcpy(cat_name[catch_like],"catch");
    strcpy(cat_name[for_like],"for");
    strcpy(cat_name[do_like],"do");
    strcpy(cat_name[if_like],"if");
    strcpy(cat_name[delete_like],"delete");
    strcpy(cat_name[raw_ubin],"ubinop?");
    strcpy(cat_name[const_like],"const");
    strcpy(cat_name[raw_int],"raw");
    strcpy(cat_name[int_like],"int");
    strcpy(cat_name[case_like],"case");
    strcpy(cat_name[sizeof_like],"sizeof");
    strcpy(cat_name[struct_like],"struct");
    strcpy(cat_name[typedef_like],"typedef");
    strcpy(cat_name[define_like],"define");
    strcpy(cat_name[template_like],"template");
    strcpy(cat_name[ftemplate],"ftemplate");
    strcpy(cat_name[new_exp],"new_exp");
    strcpy(cat_name[begin_arg],"@@["@q]@>);
    strcpy(