htmlparser.y

将HTML转换为TXT文件的程序

    $$ = bq;
  }
  | FORM body_content END_FORM {
    Form *f = new Form;
    f->attributes.reset($1);
    f->content.reset($2);
    $$ = f;
  }
  | HR {
    HorizontalRule *h = new HorizontalRule;
    h->attributes.reset($1);
    $$ = h;
  }
  | TABLE opt_caption table_rows END_TABLE {
    Table *t = new Table;
    t->attributes.reset($1);
    t->caption.reset($2);
    t->rows.reset($3);
    $$ = t;
  }
  ;

list:
  OL { ++list_nesting; } list_content END_OL {
    OrderedList *ol = new OrderedList;
    ol->attributes.reset($1);
    ol->items.reset($3);
    ol->nesting = --list_nesting;
    $$ = ol;
  }
  | UL { ++list_nesting; } list_content END_UL {
    UnorderedList *ul = new UnorderedList;
    ul->attributes.reset($1);
    ul->items.reset($3);
    ul->nesting = --list_nesting;
    $$ = ul;
  }
  | DIR { ++list_nesting; } list_content END_DIR {
    Dir *d = new Dir;
    d->attributes.reset($1);
    d->items.reset($3);
    d->nesting = --list_nesting;
    $$ = d;
  }
  | MENU { ++list_nesting; } list_content END_MENU {
    Menu *m = new Menu;
    m->attributes.reset($1);
    m->items.reset($3);
    m->nesting = --list_nesting;
    $$ = m;
  }
  ;

list_content:
  /* empty */ {
    $$ = 0;
  }
  | list_content error {
    $$ = $1;
  }
  | list_content list_item {
    $$ = $1 ? $1 : new list<auto_ptr<ListItem> >;
    $$->push_back(auto_ptr<ListItem>($2));
  }
  ;

list_item:
  LI opt_flow opt_END_LI {
    ListNormalItem *lni = new ListNormalItem;
    lni->attributes.reset($1);
    lni->flow.reset($2);
    $$ = lni;
  }
  | block {   /* EXTENSION: Handle a "block" in a list as an indented block. */
    ListBlockItem *lbi = new ListBlockItem;
    lbi->block.reset($1);
    $$ = lbi;
  }
  | texts {              /* EXTENSION: Treat "texts" in a list as an "<LI>". */
    ListNormalItem *lni = new ListNormalItem;
    lni->flow.reset($1);
    $$ = lni;
  }
  ;

definition_list:
                                           /* EXTENSION: Allow nested <DL>s. */
                                             /* EXTENSION: "</DL>" optional. */
  DL opt_flow opt_error definition_list opt_END_DL {
    delete $1;
    delete $2; /* Kludge */
    $$ = $4;
  }
                                 /* EXTENSION: Accept a "preamble" in the DL */
  | DL opt_flow opt_error definition_list_content END_DL {
    DefinitionList *dl = new DefinitionList;
    dl->attributes.reset($1);
    dl->preamble.reset($2);
    dl->items.reset($4);
    $$ = dl;
  }
  ;

definition_list_content:
  /* empty */ {
    $$ = 0;
  }
  | definition_list_content {
    $$ = $1;
  }
  | definition_list_content term_name {
    $$ = $1 ? $1 : new list<auto_ptr<DefinitionListItem> >;
    $$->push_back(auto_ptr<DefinitionListItem>($2));
  }
  | definition_list_content term_definition {
    $$ = $1 ? $1 : new list<auto_ptr<DefinitionListItem> >;
    $$->push_back(auto_ptr<DefinitionListItem>($2));
  }
  ;

term_name:
  DT opt_flow opt_error {      /* EXTENSION: Allow "flow" instead of "texts" */
    delete $1;
    $$ = new TermName;
    $$->flow.reset($2);
  }
  | DT opt_flow END_DT opt_P opt_error {/* EXTENSION: Ignore <P> after </DT> */
    delete $1;
    delete $4;
    $$ = new TermName;
    $$->flow.reset($2);
  }
  ;

term_definition:
  DD opt_flow opt_error {
    delete $1;
    $$ = new TermDefinition;
    $$->flow.reset($2);
  }
  | DD opt_flow END_DD opt_P opt_error {/* EXTENSION: Ignore <P> after </DD> */
    delete $1;
    delete $4;
    $$ = new TermDefinition;
    $$->flow.reset($2);
  }
  ;

flow:
  flow_ {
    $$ = new list<auto_ptr<Element> >;
    $$->push_back(auto_ptr<Element>($1));
  }
  | flow error {
    $$ = $1;
  }
  | flow flow_ {
    ($$ = $1)->push_back(auto_ptr<Element>($2));
  }
  ;

flow_:
  texts {
    if ($1->size() == 1) {
      $$ = $1->front().release();
      delete $1;
    } else {
      Paragraph *p = new Paragraph;
      p->texts.reset($1);
      $$ = p;
    }
  }
  | heading {          /* EXTENSION: Allow headings in "flow", i.e. in lists */
    $$ = $1;
  }
  | block {
    $$ = $1;
  }
  ;

preformatted:
  PRE opt_texts END_PRE {
    $$ = new Preformatted;
    $$->attributes.reset($1);
    $$->texts.reset($2);
  }
  ;

caption:
  CAPTION opt_texts END_CAPTION {
    $$ = new Caption;
    $$->attributes.reset($1);
    $$->texts.reset($2);
  }
  ;

table_rows:
  /* empty */ {
    $$ = new list<auto_ptr<TableRow> >;
  }
  | table_rows error {
    $$ = $1;
  }
  | table_rows TR table_cells opt_END_TR {
    TableRow *tr = new TableRow;
    tr->attributes.reset($2);
    tr->cells.reset($3);
    ($$ = $1)->push_back(auto_ptr<TableRow>(tr));
  }
  ;

table_cells:
  /* empty */ {
    $$ = new list<auto_ptr<TableCell> >;
  }
  | table_cells error {
    $$ = $1;
  }
  | table_cells TD body_content opt_END_TD {
    TableCell *tc = new TableCell;
    tc->attributes.reset($2);
    tc->content.reset($3);
    ($$ = $1)->push_back(auto_ptr<TableCell>(tc));
  }
  | table_cells TH body_content opt_END_TH opt_END_TD {
                            /* EXTENSION: Allow "</TD>" in place of "</TH>". */
    TableHeadingCell *thc = new TableHeadingCell;
    thc->attributes.reset($2);
    thc->content.reset($3);
    ($$ = $1)->push_back(auto_ptr<TableCell>(thc));
  }
  | table_cells INPUT {    /* EXTENSION: Ignore <INPUT> between table cells. */
    delete $2;
    $$ = $1;
  }
  ;

address:
  ADDRESS opt_texts END_ADDRESS { /* Should be "address_content"... */
    delete $1;
    $$ = new Address;
    $$->content.reset($2);
  }
  ;

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

texts:
  text {
    $$ = new list<auto_ptr<Element> >;
    $$->push_back(auto_ptr<Element>($1));
  }
  | texts text {
    ($$ = $1)->push_back(auto_ptr<Element>($2));
  }
  ;

text:
  pcdata                    opt_error { $$ = $1; }
  | font                    opt_error { $$ = $1; }
  | phrase                  opt_error { $$ = $1; }
  | special                 opt_error { $$ = $1; }
  | form                    opt_error { $$ = $1; }
  | NOBR opt_texts END_NOBR opt_error { /* EXTENSION: NS 1.1 / IE 2.0 */
    NoBreak *nb = new NoBreak;
    delete $1;
    nb->content.reset($2);
    $$ = nb;
  }
  ;

font:
  TT       opt_texts opt_END_TT     { delete $1; $$ = new Font(TT,     $2); }
  | I      opt_texts opt_END_I      { delete $1; $$ = new Font(I,      $2); }
  | B      opt_texts opt_END_B      { delete $1; $$ = new Font(B,      $2); }
  | U      opt_texts opt_END_U      { delete $1; $$ = new Font(U,      $2); }
  | STRIKE opt_texts opt_END_STRIKE { delete $1; $$ = new Font(STRIKE, $2); }
  | BIG    opt_texts opt_END_BIG    { delete $1; $$ = new Font(BIG,    $2); }
  | SMALL  opt_texts opt_END_SMALL  { delete $1; $$ = new Font(SMALL,  $2); }
  | SUB    opt_texts opt_END_SUB    { delete $1; $$ = new Font(SUB,    $2); }
  | SUP    opt_texts opt_END_SUP    { delete $1; $$ = new Font(SUP,    $2); }
  ;

phrase:
  EM       opt_texts opt_END_EM     { delete $1; $$ = new Phrase(EM,     $2); }
  | STRONG opt_texts opt_END_STRONG { delete $1; $$ = new Phrase(STRONG, $2); }
  | DFN    opt_texts opt_END_DFN    { delete $1; $$ = new Phrase(DFN,    $2); }
  | CODE   opt_texts opt_END_CODE   { delete $1; $$ = new Phrase(CODE,   $2); }
  | SAMP   opt_texts opt_END_SAMP   { delete $1; $$ = new Phrase(SAMP,   $2); }
  | KBD    opt_texts opt_END_KBD    { delete $1; $$ = new Phrase(KBD,    $2); }
  | VAR    opt_texts opt_END_VAR    { delete $1; $$ = new Phrase(VAR,    $2); }
  | CITE   opt_texts opt_END_CITE   { delete $1; $$ = new Phrase(CITE,   $2); }
  ;

special:
                        /* EXTENSION: Allow "flow" in <A>, not only "texts". */
                                 /* EXTENSION: Allow useless <LI> in anchor. */
                                               /* EXTENSION: "</A>" optional.*/
  A opt_LI opt_flow opt_END_A {
    delete $2;
    Anchor *a = new Anchor;
    a->attributes.reset($1);
    a->texts.reset($3);
    $$ = a;
  }
  | IMG {
    Image *i = new Image;
    i->attributes.reset($1);
    $$ = i;
  }
  | APPLET applet_content END_APPLET {
    Applet *a = new Applet;
    a->attributes.reset($1);
    a->content.reset($2);
    $$ = a;
  }
                   /* EXTENSION: "flow" in <FONT> allowed, not only "texts". */
                                           /* EXTENSION: "</FONT>" optional. */
  | FONT opt_flow opt_END_FONT {
    Font2 *f2 = new Font2;
    f2->attributes.reset($1);
    f2->elements.reset($2);
    $$ = f2;
  }
  | BASEFONT {
    BaseFont *bf = new BaseFont;
    bf->attributes.reset($1);
    $$ = bf;
  }
  | BR {
    LineBreak *lb = new LineBreak;
    lb->attributes.reset($1);
    $$ = lb;
  }
  | MAP map_content END_MAP {
    Map *m = new Map;
    m->attributes.reset($1);
    m->areas.reset($2);
    $$ = m;
  }
  ;

applet_content:
  /* empty */ {
    $$ = 0;
  }
  | applet_content text {
    $$ = $1 ? $1 : new list<auto_ptr<Element> >;
    $$->push_back(auto_ptr<Element>($2));
  }
  | applet_content PARAM {
    $$ = $1 ? $1 : new list<auto_ptr<Element> >;
    Param *p = new Param;
    p->attributes.reset($2);
    $$->push_back(auto_ptr<Element>(p));
  }
  ;

map_content:
  /* empty */ {
    $$ = 0;
  }
  | map_content error {
    $$ = $1;
  }
  | map_content AREA {
    $$ = $1 ? $1 : new list<auto_ptr<list<TagAttribute> > >;
    $$->push_back(auto_ptr<list<TagAttribute> >($2));
  }
  ;

form:
  INPUT {
    Input *i = new Input;
    i->attributes.reset($1);
    $$ = i;
  }
  | SELECT select_content END_SELECT {
    Select *s = new Select;
    s->attributes.reset($1);
    s->content.reset($2);
    $$ = s;
  }
  | TEXTAREA pcdata END_TEXTAREA {
    TextArea *ta = new TextArea;
    ta->attributes.reset($1);
    ta->pcdata.reset($2);
    $$ = ta;
  }
  ;

select_content:
  option {
    $$ = new list<auto_ptr<Option> >;
    $$->push_back(auto_ptr<Option>($1));
  }
  | select_content option {
    ($$ = $1)->push_back(auto_ptr<Option>($2));
  }
  ;

option:
  OPTION pcdata opt_END_OPTION {
    $$ = new Option;
    $$->attributes.reset($1);
    $$->pcdata.reset($2);
  }
  ;

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

HX:
  H1   { $$ = new Heading; $$->level = 1; $$->attributes.reset($1); }
  | H2 { $$ = new Heading; $$->level = 2; $$->attributes.reset($1); }
  | H3 { $$ = new Heading; $$->level = 3; $$->attributes.reset($1); }
  | H4 { $$ = new Heading; $$->level = 4; $$->attributes.reset($1); }
  | H5 { $$ = new Heading; $$->level = 5; $$->attributes.reset($1); }
  | H6 { $$ = new Heading; $$->level = 6; $$->attributes.reset($1); }
  ;

END_HX:
  END_H1   { $$ = 1; }
  | END_H2 { $$ = 2; }
  | END_H3 { $$ = 3; }
  | END_H4 { $$ = 4; }
  | END_H5 { $$ = 5; }
  | END_H6 { $$ = 6; }
  ;

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

opt_pcdata:     /* empty */ { $$ = 0; } | pcdata  { $$ = $1; };
opt_caption:    /* empty */ { $$ = 0; } | caption { $$ = $1; };
opt_texts:      /* empty */ { $$ = 0; } | texts   { $$ = $1; };
opt_flow:       /* empty */ { $$ = 0; } | flow    { $$ = $1; };

opt_LI:         /* empty */ { $$ = 0; } | LI      { $$ = $1; };
opt_P:          /* empty */ { $$ = 0; } | P       { $$ = $1; };

opt_END_A:      /* empty */ | END_A;
opt_END_B:      /* empty */ | END_B;
opt_END_BIG:    /* empty */ | END_BIG;
opt_END_CENTER: /* empty */ | END_CENTER;
opt_END_CITE:   /* empty */ | END_CITE;
opt_END_CODE:   /* empty */ | END_CODE;
opt_END_DFN:    /* empty */ | END_DFN;
opt_END_DL:     /* empty */ | END_DL;
opt_END_EM:     /* empty */ | END_EM;
opt_END_FONT:   /* empty */ | END_FONT;
opt_END_I:      /* empty */ | END_I;
opt_END_KBD:    /* empty */ | END_KBD;
opt_END_LI:     /* empty */ | END_LI;
opt_END_OPTION: /* empty */ | END_OPTION;
opt_END_P:      /* empty */ | END_P;
opt_END_SAMP:   /* empty */ | END_SAMP;
opt_END_SMALL:  /* empty */ | END_SMALL;
opt_END_STRIKE: /* empty */ | END_STRIKE;
opt_END_STRONG: /* empty */ | END_STRONG;
opt_END_SUB:    /* empty */ | END_SUB;
opt_END_SUP:    /* empty */ | END_SUP;
opt_END_TD:     /* empty */ | END_TD;
opt_END_TH:     /* empty */ | END_TH;
opt_END_TITLE:  /* empty */ | END_TITLE;
opt_END_TR:     /* empty */ | END_TR;
opt_END_TT:     /* empty */ | END_TT;
opt_END_U:      /* empty */ | END_U;
opt_END_VAR:    /* empty */ | END_VAR;

opt_error:      /* empty */ | error;

%% /* } */

/*
 * Some C++ compilers (e.g. EGCS 2.91.66) have problems if all virtual
 * methods of a class are inline or pure virtual, so we define the destructor,
 * which is the only virtual method, non-inline, although it is empty.
 */

HTMLParser::~HTMLParser()
{
}

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