html.c

将HTML转换为TXT文件的程序

void
OrderedList::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<OL" << attributes << ">" << separator;
  if (items.get()) foreach(*items, os, separator);
  os << "</OL>" << separator;
}

void
UnorderedList::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<UL" << attributes << ">" << separator;
  if (items.get()) foreach(*items, os, separator);
  os << "</UL>" << separator;
}

void
Dir::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<DIR" << attributes << ">" << separator;
  if (items.get()) foreach(*items, os, separator);
  os << "</DIR>" << separator;
}

void
Menu::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<MENU" << attributes << ">" << separator;
  if (items.get()) foreach(*items, os, separator);
  os << "</MENU>" << separator;
}

void
ListNormalItem::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<LI" <<attributes << ">" << separator;
  if (flow.get()) foreach(*flow, os, separator);
  os << "</LI>" << separator;
}

void
ListBlockItem::unparse(ostream &os, ostream_manipulator separator) const
{
  if (block.get()) block->unparse(os, separator);
}

/* ------------------------------------------------------------------------- */

void
DefinitionList::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<DL" << attributes << ">" << separator;
  if (items.get()) foreach(*items, os, separator);
  os << "</DL>" << separator;
}

void
TermName::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<DT>" << separator;
  if (flow.get()) foreach(*flow, os, separator);
  os << "</DT>" << separator;
}

void
TermDefinition::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<DD>" << separator;
  if (flow.get()) foreach(*flow, os, separator);
  os << "</DD>" << separator;
}

/* ------------------------------------------------------------------------- */

void
Table::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<TABLE" << attributes << ">" << separator;
  if (caption.get()) caption->unparse(os, separator);
  if (rows.get()) foreach(*rows, os, separator);
  os << "</TABLE>" << separator;
}

void
NoBreak::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<NOBR>" << separator;
  if (content.get()) foreach(*content, os, separator);
  os << "</NOBR>" << separator;
}

void
Font::unparse(ostream &os, ostream_manipulator separator) const
{
  const char *tag = (
    attribute == HTMLParser::TT     ? "TT" :
    attribute == HTMLParser::I      ? "I" :
    attribute == HTMLParser::B      ? "B" :
    attribute == HTMLParser::U      ? "U" :
    attribute == HTMLParser::STRIKE ? "STRIKE" :
    attribute == HTMLParser::BIG    ? "BIG" :
    attribute == HTMLParser::SMALL  ? "SMALL" :
    attribute == HTMLParser::SUB    ? "SUB" :
    attribute == HTMLParser::SUP    ? "SUP" :
    "???"
  );

  os << "<" << tag << ">" << separator;
  if (texts.get()) foreach(*texts, os, separator);
  os << "</" << tag << ">" << separator;
}

void
Phrase::unparse(ostream &os, ostream_manipulator separator) const
{
  const char *tag = (
    attribute == HTMLParser::EM     ? "EM" :
    attribute == HTMLParser::STRONG ? "STRONG" :
    attribute == HTMLParser::DFN    ? "DFN" :
    attribute == HTMLParser::CODE   ? "CODE" :
    attribute == HTMLParser::SAMP   ? "SAMP" :
    attribute == HTMLParser::KBD    ? "KBD" :
    attribute == HTMLParser::VAR    ? "VAR" :
    attribute == HTMLParser::CITE   ? "CITE" :
    "???"
  );

  os << "<" << tag << ">" << separator;
  if (texts.get()) foreach(*texts, os, separator);
  os << "</" << tag << ">" << separator;
}

void
Anchor::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<A" << attributes << ">" << separator;
  if (texts.get()) foreach(*texts, os, separator);
  os << "</A>" << separator;
}

void
BaseFont::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<BASEFONT" << attributes << ">" << separator;
}

void
LineBreak::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<BR" << attributes << ">" << separator;
}

void
Map::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<MAP" << attributes << ">" << separator;
  if (areas.get()) {
    const list<auto_ptr<list<TagAttribute> > >           &al(*areas);
    list<auto_ptr<list<TagAttribute> > >::const_iterator i;
    for (i = al.begin(); i != al.end(); ++i) {
      os << "<AREA" << *i << ">" << separator;
    }
  }
  os << "</MAP>" << separator;
}

void
Font2::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<FONT" << attributes << ">" << separator;
  if (elements.get()) foreach(*elements, os, separator);
  os << "</FONT>" << separator;
}

void
TableRow::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<TR" << attributes << ">" << separator;
  if (cells.get()) foreach(*cells, os, separator);
  os << "</TR>" << separator;
}

void
TableCell::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<TD" << attributes << ">" << separator;
  if (content.get()) foreach(*content, os, separator);
  os << "</TD>" << separator;
}

void
TableHeadingCell::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<TH" << attributes << ">" << separator;
  if (content.get()) foreach(*content, os, separator);
  os << "</TH>" << separator;
}

void
Caption::unparse(ostream &os, ostream_manipulator separator) const
{
  os << "<CAPTION" << attributes << ">" << separator;
  if (texts.get()) foreach(*texts, os, separator);
  os << "</CAPTION>" << separator;
}

/* ------------------------------------------------------------------------- */

string
get_attribute(
  const list<TagAttribute> *as,
  const char               *name,
  const char               *dflt
)
{
  if (as) {
    list<TagAttribute>::const_iterator i;
    for (i = as->begin(); i != as->end(); ++i) {
      if (cmp_nocase((*i).first, name) == 0) return string((*i).second);
    }
  }
  return string(dflt);
}

int
get_attribute(
  const list<TagAttribute> *as,
  const char               *name,
  int                      dflt
)
{
  if (as) {
    list<TagAttribute>::const_iterator i;
    for (i = as->begin(); i != as->end(); ++i) {
      if (cmp_nocase((*i).first, name) == 0) return atoi((*i).second.c_str());
    }
  }
  return dflt;
}

int
get_attribute(
  const list<TagAttribute> *as,
  const char               *name,
  int                      dflt,
  const char               *s1,
  int                      v1,
  ...
)
{
  if (as) {
    list<TagAttribute>::const_iterator i;
    for (i = as->begin(); i != as->end(); ++i) {
      if (cmp_nocase((*i).first, name) == 0) {
        const char *s = s1;
        int        v = v1;

        va_list va;
        va_start(va, v1);
        for (;;) {
          if (cmp_nocase(s, (*i).second) == 0) break;
          s = va_arg(va, const char *);
          if (!s) { v = dflt; break; }
          v = va_arg(va, int);
        }
        va_end(va);
        return v;
      }
    }
  }
  return dflt;
}

int
get_attribute(
  const list<TagAttribute> *as,
  const char               *name,   // Attribute name
  const char               *dflt1,  // If attribute not specified
  int                      dflt2,   // If string value does not match s1, ...
  const char               *s1,
  int                      v1,
  ...
)
{
  if (as) {
    list<TagAttribute>::const_iterator i;
    for (i = as->begin(); i != as->end(); ++i) {
      if (cmp_nocase((*i).first, name) == 0) {
	dflt1 = (*i).second.c_str();
        break;
      }
    }
  }

  if (!dflt1) return dflt2;

  const char *s = s1;
  int        v = v1;

  va_list va;
  va_start(va, v1);
  for (;;) {
    if (cmp_nocase(s, dflt1) == 0) break;
    s = va_arg(va, const char *);
    if (!s) break;
    v = va_arg(va, int);
  }
  va_end(va);

  return s ? v : dflt2;
}

/* ------------------------------------------------------------------------- */