input.c

NullSofts criptable install system2.28源代码

      while (iscmd(c));
      unget(in, c, &cpos);
    }
    /*
     * Now match the command against the list of available
     * ones.
     */
    ret.type = tok_cmd;
    ret.text = ustrdup(rs.text);
    match_kw(&ret);
    sfree(rs.text);
    return ret;
  } else if (c == '{')
  {                             /* tok_lbrace */
    ret.type = tok_lbrace;
    return ret;
  } else if (c == '}')
  {                             /* tok_rbrace */
    ret.type = tok_rbrace;
    return ret;
  } else
  {                             /* tok_word */
    /*
     * Read a word: the longest possible contiguous sequence of
     * things other than whitespace, backslash, braces and
     * hyphen. A hyphen terminates the word but is returned as
     * part of it; everything else is pushed back for the next
     * token. The `aux' field contains TRUE if the word ends in
     * a hyphen.
     */
    ret.aux = FALSE;            /* assumed for now */
    while (1)
    {
      if (iswhite(c) || c == '{' || c == '}' || c == '\\' || c == EOF)
      {
        /* Put back the character that caused termination */
        unget(in, c, &cpos);
        break;
      } else
      {
        rdadd(&rs, (wchar_t)c);
        if (c == '-')
        {
          ret.aux = TRUE;
          break;                /* hyphen terminates word */
        }
      }
      c = get(in, &cpos);
    }
    ret.type = tok_word;
    ret.text = ustrdup(rs.text);
    sfree(rs.text);
    return ret;
  }
}

/*
 * Determine whether the next input character is an open brace (for
 * telling code paragraphs from paragraphs which merely start with
 * code).
 */
int isbrace(input * in)
{
  int c;
  filepos cpos;

  c = get(in, &cpos);
  unget(in, c, &cpos);
  return (c == '{');
}

/*
 * Read the rest of a line that starts `\c'. Including nothing at
 * all (tok_word with empty text).
 */
token get_codepar_token(input * in)
{
  int c;
  token ret;
  rdstring rs = { 0, 0, NULL };
  filepos cpos;

  ret.type = tok_word;
  c = get(in, &cpos);           /* expect (and discard) one space */
  ret.pos = cpos;
  if (c == ' ')
  {
    c = get(in, &cpos);
    ret.pos = cpos;
  }
  while (!isnl(c) && c != EOF)
  {
    int c2 = c;
    c = get(in, &cpos);
    /* Discard \r just before \n. */
    if (c2 != 13 || !isnl(c))
      rdadd(&rs, (wchar_t)c2);
  }
  unget(in, c, &cpos);
  ret.text = ustrdup(rs.text);
  sfree(rs.text);
  return ret;
}

/*
 * Adds a new word to a linked list
 */
static word *addword(word newword, word *** hptrptr)
{
  word *mnewword;
  if (!hptrptr)
    return NULL;
  mnewword = mknew(word);
  *mnewword = newword;          /* structure copy */
  mnewword->next = NULL;
  **hptrptr = mnewword;
  *hptrptr = &mnewword->next;
  return mnewword;
}

/*
 * Adds a new paragraph to a linked list
 */
static paragraph *addpara(paragraph newpara, paragraph *** hptrptr)
{
  paragraph *mnewpara = mknew(paragraph);
  *mnewpara = newpara;          /* structure copy */
  mnewpara->next = NULL;
  **hptrptr = mnewpara;
  *hptrptr = &mnewpara->next;
  return mnewpara;
}

/*
 * Destructor before token is reassigned; should catch most memory
 * leaks
 */
#define dtor(t) ( sfree(t.text) )

/*
 * Reads a single file (ie until get() returns EOF)
 */
static void read_file(paragraph *** ret, input * in, indexdata * idx)
{
  token t;
  paragraph par;
  word wd, **whptr, **idximplicit;
  tree234 *macros;
  wchar_t utext[2], *wdtext;
  int style, spcstyle;
  int already;
  int iswhite, seenwhite;
  int type;
  struct stack_item {
    enum {
      stack_nop = 0,            /* do nothing (for error recovery) */
      stack_ualt = 1,           /* \u alternative */
      stack_style = 2,          /* \e, \c, \cw */
      stack_idx = 4,            /* \I, \i, \ii */
      stack_hyper = 8,          /* \W */
      stack_quote = 16,         /* \q */
    } type;
    word **whptr;               /* to restore from \u alternatives */
    word **idximplicit;         /* to restore from \u alternatives */
  } *sitem;
  stack parsestk;
  word *indexword=NULL, *uword=NULL, *iword=NULL;
  word *idxwordlist;
  rdstring indexstr;
  int index_downcase=0, index_visible=0, indexing=0;
  const rdstring nullrs = { 0, 0, NULL };
  wchar_t uchr;

  t.text = NULL;
  macros = newtree234(macrocmp);
  already = FALSE;

  /*
   * Loop on each paragraph.
   */
  while (1)
  {
    int start_cmd = c__invalid;
    par.words = NULL;
    par.keyword = NULL;
    whptr = &par.words;

    /*
     * Get a token.
     */
    if (!already)
    {
      dtor(t), t = get_token(in);
    }
    already = FALSE;
    if (t.type == tok_eof)
      break;

    /*
     * Parse code paragraphs separately.
     */
    if (t.type == tok_cmd && t.cmd == c_c && !isbrace(in))
    {
      par.type = para_Code;
      par.fpos = t.pos;
      while (1)
      {
        dtor(t), t = get_codepar_token(in);
        wd.type = word_WeakCode;
        wd.breaks = FALSE;      /* shouldn't need this... */
        wd.text = ustrdup(t.text);
        wd.alt = NULL;
        wd.fpos = t.pos;
        addword(wd, &whptr);
        dtor(t), t = get_token(in);
        if (t.type == tok_white)
        {
          /*
           * The newline after a code-paragraph line
           */
          dtor(t), t = get_token(in);
        }
        if (t.type == tok_eop || t.type == tok_eof)
          break;
        else if (t.type != tok_cmd || t.cmd != c_c)
        {
          error(err_brokencodepara, &t.pos);
          addpara(par, ret);
          while (t.type != tok_eop)     /* error recovery: */
            dtor(t), t = get_token(in); /* eat rest of paragraph */
          goto codeparabroken;  /* ick, but such is life */
        }
      }
      addpara(par, ret);
    codeparabroken:
      continue;
    }

    while (t.type == tok_cmd && macrolookup(macros, in, t.text, &t.pos))
    {
      dtor(t), t = get_token(in);
    }


    /*
     * This token begins a paragraph. See if it's one of the
     * special commands that define a paragraph type.
     *
     * (note that \# is special in a way, and \nocite takes no
     * text)
     */
    par.type = para_Normal;
    if (t.type == tok_cmd)
    {
      int needkw=0;
      int is_macro = FALSE;

      par.fpos = t.pos;
      switch (t.cmd)
      {
      default:
        needkw = -1;
        break;
      case c__invalid:
        error(err_badparatype, t.text, &t.pos);
        needkw = 4;
        break;
      case c__comment:
        if (isbrace(in))
          break;                /* `\#{': isn't a comment para */
        do
        {
          dtor(t), t = get_token(in);
        }
        while (t.type != tok_eop && t.type != tok_eof);
        continue;               /* next paragraph */
        /*
         * `needkw' values:
         *
         *   1 -- exactly one keyword
         *   2 -- at least one keyword
         *   4 -- any number of keywords including zero
         *   8 -- at least one keyword and then nothing else
         *  16 -- nothing at all! no keywords, no body
         *  32 -- no keywords at all
         */
      case c_A:
        needkw = 2;
        par.type = para_Appendix;
        break;
      case c_B:
        needkw = 2;
        par.type = para_Biblio;
        break;
      case c_BR:
        needkw = 1;
        par.type = para_BR;
        start_cmd = c_BR;
        break;
      case c_C:
        needkw = 2;
        par.type = para_Chapter;
        break;
      case c_H:
        needkw = 2;
        par.type = para_Heading;
        par.aux = 0;
        break;
      case c_IM:
        needkw = 2;
        par.type = para_IM;
        start_cmd = c_IM;
        break;
      case c_S:
        needkw = 2;
        par.type = para_Subsect;
        par.aux = t.aux;
        break;
      case c_U:
        needkw = 32;
        par.type = para_UnnumberedChapter;
        break;
        /* For \b and \n the keyword is optional */
      case c_b:
        needkw = 4;
        par.type = para_Bullet;
        break;
      case c_n:
        needkw = 4;
        par.type = para_NumberedList;
        break;
      case c_cfg:
        needkw = 8;
        par.type = para_Config;
        start_cmd = c_cfg;
        break;
      case c_copyright:
        needkw = 32;
        par.type = para_Copyright;
        break;
      case c_define:
        is_macro = TRUE;
        needkw = 1;
        break;
        /* For \nocite the keyword is _everything_ */
      case c_nocite:
        needkw = 8;
        par.type = para_NoCite;
        break;
      case c_preamble:
        needkw = 32;
        par.type = para_Preamble;
        break;
      case c_rule:
        needkw = 16;
        par.type = para_Rule;
        break;
      case c_title:
        needkw = 32;
        par.type = para_Title;
        break;
      case c_versionid:
        needkw = 32;
        par.type = para_VersionID;
        break;
      }

      if (needkw > 0)
      {
        rdstring rs = { 0, 0, NULL };
        int nkeys = 0;
        filepos fp;

        /* Get keywords. */
        dtor(t), t = get_token(in);
        fp = t.pos;
        while (t.type == tok_lbrace)
        {
          /* This is a keyword. */
          nkeys++;
          /* FIXME: there will be bugs if anyone specifies an
           * empty keyword (\foo{}), so trap this case. */
          while (dtor(t), t = get_token(in),
                 t.type == tok_word ||
                 t.type == tok_white ||
                 (t.type == tok_cmd && t.cmd == c__nbsp) ||
                 (t.type == tok_cmd && t.cmd == c__escaped))
          {
            if (t.type == tok_white ||
                (t.type == tok_cmd && t.cmd == c__nbsp))
              rdadd(&rs, ' ');
            else
              rdadds(&rs, t.text);
          }
          if (t.type != tok_rbrace)
          {
            error(err_kwunclosed, &t.pos);
            continue;
          }
          rdadd(&rs, 0);        /* add string terminator */
          dtor(t), t = get_token(in);   /* eat right brace */
        }

        rdadd(&rs, 0);          /* add string terminator */

        /* See whether we have the right number of keywords. */
        if ((needkw & 48) && nkeys > 0)
          error(err_kwillegal, &fp);
        if ((needkw & 11) && nkeys == 0)
          error(err_kwexpected, &fp);
        if ((needkw & 5) && nkeys > 1)
          error(err_kwtoomany, &fp);

        if (is_macro)
        {
          /*
           * Macro definition. Get the rest of the line
           * as a code-paragraph token, repeatedly until
           * there's nothing more left of it. Separate
           * with newlines.
           */
          rdstring macrotext = { 0, 0, NULL };
          while (1)
          {
            dtor(t), t = get_codepar_token(in);
            if (macrotext.pos > 0)
              rdadd(&macrotext, L'\n');
            rdadds(&macrotext, t.text);
            dtor(t), t = get_token(in);
            if (t.type == tok_eop)
              break;
          }
          macrodef(macros, rs.text, macrotext.text, fp);
          continue;             /* next paragraph */
        }

        par.keyword = rdtrim(&rs);

        /* Move to EOP in case of needkw==8 or 16 (no body) */
        if (needkw & 24)
        {
          /* We allow whitespace even when we expect no para body */
          while (t.type == tok_white)
            dtor(t), t = get_token(in);
          if (t.type != tok_eop && t.type != tok_eof &&
              (start_cmd == c__invalid ||
               t.type != tok_cmd || t.cmd != start_cmd))
          {
            error(err_bodyillegal, &t.pos);
            /* Error recovery: eat the rest of the paragraph */
            while (t.type != tok_eop && t.type != tok_eof &&
                   (start_cmd == c__invalid ||
                    t.type != tok_cmd || t.cmd != start_cmd))
              dtor(t), t = get_token(in);
          }
          if (t.type == tok_cmd)
            already = TRUE;     /* inhibit get_token at top of loop */
          addpara(par, ret);
          continue;             /* next paragraph */
        }
      }
    }

    /*
     * Now read the actual paragraph, word by word, adding to
     * the paragraph list.
     *
     * Mid-paragraph commands:
     *
     *  \K \k
     *  \c \cw
     *  \e
     *  \i \ii
     *  \I
     *  \u
     *  \W
     *  \date
     *  \\ \{ \}
     */
    parsestk = stk_new();
    style = word_Normal;
    spcstyle = word_WhiteSpace;
    indexing = FALSE;
    seenwhite = TRUE;
    while (t.type != tok_eop && t.type != tok_eof)
    {
      iswhite = FALSE;
      already = FALSE;

      /* Handle implicit paragraph breaks after \IM, \BR etc */
      if (start_cmd != c__invalid &&
          t.type == tok_cmd && t.cmd == start_cmd)
      {
        already = TRUE;         /* inhibit get_token at top of loop */
        break;
      }