ruby-web 1.1.1

data/lib/formatter.rb

@@ -0,0 +1,314 @@
+class NullFormatter
+  def initialize(writer=nil)
+    if not writer
+      writer = NullWriter.new
+    end
+    @writer = writer
+  end
+  def end_paragraph(blankline) end
+  def add_line_break() end
+  #def add_hor_rule(*args, **kw) end
+  def add_hor_rule() end
+  def add_label_data(format, counter, blankline=nil) end
+  def add_flowing_data(data) end
+  def add_literal_data(data) end
+  def flush_softspace() end
+  def push_alignment(align) end
+  def pop_alignment() end
+  def push_font(size, i, b, tt) end
+  def pop_font() end
+  def push_margin(margin) end
+  def pop_margin() end
+  def set_spacing(spacing) end
+  def push_style(*styles) end
+  def pop_style(n=1) end
+  def assert_line_data(flag=1) end
+end
+
+class AbstractFormatter < NullFormatter
+
+  def initialize(writer)
+    @writer = writer
+    @align = nil
+    @align_stack = []
+    @font_stack = []
+    @margin_stack = []
+    @spacing = nil
+    @style_stack = []
+    @nospace = true
+    @softspace = false
+    @para_end = true
+    @parskip = 0
+    @hard_break = true
+    @have_label = false
+  end
+
+  def end_paragraph(blankline)
+    if not @hard_break
+      @writer.send_line_break
+      @have_label = false
+    end
+    if (@parskip < blankline) and !@have_label
+      @writer.send_paragraph(blankline - @parskip)
+      @parskip = blankline
+      @have_label = 0
+    end
+    @hard_break = true
+    @nospace = true
+    @para_end = true
+    @softspace = false
+  end
+
+  def add_line_break
+    if not (@hard_break or @para_end)
+      @writer.send_line_break
+      @have_label = false
+      @parskip = 0
+    end
+    @hard_break = true
+    @nospace = true
+    @softspace = false
+  end
+
+  def add_hor_rule
+    if not @hard_break
+      @writer.send_line_break
+    end
+    @writer.send_hor_rule #(args, kw)
+    @hard_break = true
+    @nospace = true
+    @have_label = false
+    @para_end = false
+    @softspace = false
+    @parskip = 0
+  end
+
+  def add_label_data(format, counter, blankline = nil)
+    if @have_label or !@hard_break
+      @writer.send_line_break
+    end
+    if not @para_end
+      @writer.send_paragraph(1)  #((blankline and 1) or 0)
+    end
+    if format.kind_of? String
+      #@writer.send_label_data(format_counter(format, counter))
+    else
+      @writer.send_label_data(format)
+    end
+    @nospace = @have_label = @hard_break = @para_end = 1
+    @softspace = @parskip = 0
+  end
+
+  def add_flowing_data(data)
+    return if not data
+    prespace = (/^\s/ =~ data)
+    postspace = (/\s$/ =~ data)
+    data = data.split.join(" ")
+    if @nospace and data.length < 1
+      return
+    elsif prespace or @softspace
+      if data.length == 0
+        if not @nospace
+          @softspace = true
+          @parskip = 0
+        end
+        return
+      end
+      if not @nospace
+        data = ' ' + data
+      end
+    end
+    @hard_break = @nospace = @para_end = false
+    @have_label = false
+    @parskip = 0
+    @softspace = postspace
+    @writer.send_flowing_data(data)
+  end
+
+  def add_literal_data(data)
+    return if not data
+    if @softspace
+      @writer.send_flowing_data("_")
+    end
+    @hard_break = (data[-1,1] == "\n")
+    @nospace = @para_end = @softspace = false
+    @have_label = false
+    @parskip = 0
+    @writer.send_literal_data(data)
+  end
+
+  def flush_softspace
+    if @softspace
+      @hard_break = false
+      @para_end = false
+      @have_label = false
+      @softspace = false
+      @parskip = 0
+      @nospace = true
+      @writer.send_flowing_data(' ')
+    end
+  end
+
+  def push_font(size, i, b, tt)
+    if @softspace
+      @hard_break = false
+      @para_end = false
+      @softspace = false
+      @nospace = true
+      @writer.send_flowing_data(' ')
+    end
+    if @font_stack.length > 0
+      csize, ci, cb, ctt = @font_stack[-1]
+      if size == nil
+        size = csize
+      end
+      if i == nil
+        i = ci
+      end
+      if b == nil
+        b = cb
+      end
+      if tt == nil
+        tt = ctt
+      end
+    end
+    font = [size, i, b, tt]
+    @font_stack << font
+    @writer.new_font(font)
+  end
+
+  def pop_font
+    if @font_stack.length > 0
+      @font_stack.pop
+    end
+    if @font_stack
+      font = @font_stack[-1]
+    else
+      font = nil
+    end
+    @writer.new_font(font)
+  end
+
+  def push_margin(margin)
+    @margin_stack << margin
+    fstack = @margin_stack.compact
+    if (!margin) && (fstack.length > 0)
+      margin = fstack[-1]
+    end
+    @writer.new_margin(margin, fstack.length)
+  end
+
+  def pop_margin
+    if @margin_stack
+      @margin_stack.pop
+    end
+    fstack = @margin_stack.compact
+    if fstack.length > 0
+      margin = fstack[-1]
+    else
+      margin = nil
+    end
+    @writer.new_margin(margin, fstack.length)
+  end
+
+  def set_spacing(spacing)
+    @spacing = spacing
+    @writer.new_spacing(spacing)
+  end
+
+  def assert_line_data(flag=true)
+    @nospace = !flag
+    @hard_break = !flag
+    @para_end = false
+    @have_label = false
+    @parskip = 0
+  end
+
+end
+
+class NullWriter
+  def initialize() end
+  def flush() end
+  def new_alignment(align) end
+  def new_font(font) end
+  def new_margin(margin, level) end
+  def new_spacing(spacing) end
+  def new_styles(styles) end
+  def send_paragraph(blankline) end
+  def send_line_break() end
+  def send_hor_rule() end
+  def send_label_data(data) end
+  def send_flowing_data(data) end
+  def send_literal_data(data) end
+end
+
+class DumbWriter < NullWriter
+
+  def initialize(file=STDOUT, maxcol=72)
+    @file = file
+    @maxcol = maxcol
+    super()
+    reset
+  end
+
+  def reset
+    @col = 0
+    @atbreak = false
+  end
+
+  def send_paragraph(blankline)
+    @file.write("\n" + "\n"*blankline)
+    @col = 0
+    @atbreak = false
+  end
+
+  def send_line_break
+    @file.write("\n")
+    @col = 0
+    @atbreak = false
+  end
+
+  def send_hor_rule #(*args, **kw)
+    @file.write("\n")
+    @file.write('-'*@maxcol)
+    @file.write("\n")
+    @col = 0
+    @atbreak = false
+  end
+
+  def send_literal_data(data)
+    @file.write(data)
+    i = data.rindex("\n")
+    if i
+      @col = 0
+      data = data[i+1..-1]
+      #data = string.expandtabs(data)
+      @col = @col + data.length
+      @atbreak = false
+    end
+  end
+
+  def send_flowing_data(data)
+    return if not data
+    atbreak = (@atbreak || (/^\s/ =~ data))
+    col = @col
+    maxcol = @maxcol
+    for word in data.split
+      if atbreak
+        if col + word.length >= maxcol
+          @file.write("\n")
+          col = 0
+        else
+          @file.write(' ')
+          col = col + 1
+        end
+      end
+      @file.write(word)
+      col = col + word.length
+      atbreak = true
+    end
+    @col = col
+    @atbreak = (/\s$/ =~ data)
+  end
+
+end

data/lib/html-parser.rb

@@ -0,0 +1,429 @@
+# HTML parser
+
+require 'sgml-parser'
+
+class HTMLParser < SGMLParser
+
+  def initialize(formatter, verbose=nil)
+    super(verbose)
+    @formatter = formatter
+    @savedata = nil
+    @isindex = 0
+    @title = nil
+    @base = nil
+    @anchor = nil
+    @anchorlist = []
+    @nofill = 0
+    @list_stack = []
+  end
+
+
+  def handle_data(data)
+    if @savedata
+      @savedata = @savedata + data
+    else
+      if @nofill != 0
+        @formatter.add_literal_data(data)
+      else
+        @formatter.add_flowing_data(data)
+      end
+    end
+  end
+
+  def save_bgn
+    @savedata = ''
+  end
+
+  def save_end
+    data = @savedata
+    @savedata = nil
+    data = '' if data == nil
+    if @nofill == 0
+      data = data.split.join(" ")
+    end
+    return data
+  end
+
+  def anchor_bgn(href, name, type)
+    @anchor = href
+    if @anchor
+      @anchorlist << href
+    end
+  end
+
+  def anchor_end
+    if @anchor
+      #handle_data(format "[%d]", @anchorlist.length)
+      @anchor = nil
+    end
+  end
+
+  def handle_image(src, alt, *args)
+    handle_data(alt)
+  end
+
+  def start_html(attrs) end
+  def end_html() end
+
+  def start_head(attrs) end
+  def end_head() end
+
+  def start_body(attrs) end
+  def end_body() end
+
+  def start_title(attrs)
+    save_bgn
+  end
+
+  def end_title
+    @title = save_end
+  end
+
+  def do_base(attrs)
+    for a, v in attrs
+      if a == 'href'
+        @base = v
+      end
+    end
+  end
+
+  def do_isindex(attrs)
+    @isindex = 1
+  end
+
+  def do_link(attrs)
+  end
+
+  def do_meta(attrs)
+  end
+
+  def do_nextid(attrs) # Deprecated
+  end
+
+
+  def start_h1(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h1', 0, 1, 0)
+  end
+
+  def end_h1
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def start_h2(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h2', 0, 1, 0)
+  end
+
+  def end_h2
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def start_h3(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h3', 0, 1, 0)
+  end
+
+  def end_h3
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def start_h4(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h4', 0, 1, 0)
+  end
+
+  def end_h4
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def start_h5(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h5', 0, 1, 0)
+  end
+
+  def end_h5
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def start_h6(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font('h6', 0, 1, 0)
+  end
+
+  def end_h6
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+  end
+
+  def do_p(attrs)
+    @formatter.end_paragraph(1)
+  end
+
+  def start_pre(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_font(nil, nil, nil, 1)
+    @nofill = @nofill + 1
+  end
+
+  def end_pre
+    @formatter.end_paragraph(1)
+    @formatter.pop_font()
+    @nofill = @nofill - 1
+    if @nofill < 0 then @nofill = 0 end
+  end
+
+  def start_xmp(attrs)
+    start_pre(attrs)
+    setliteral('xmp') # Tell SGML parser
+  end
+
+  def end_xmp
+    end_pre
+  end
+
+  def start_listing(attrs)
+    start_pre(attrs)
+    setliteral('listing') # Tell SGML parser
+  end
+
+  def end_listing
+    end_pre
+  end
+
+  def start_address(attrs)
+    @formatter.end_paragraph(0)
+    @formatter.push_font(nil, 1, nil, nil)
+  end
+
+  def end_address
+    @formatter.end_paragraph(0)
+    @formatter.pop_font()
+  end
+
+  def start_blockquote(attrs)
+    @formatter.end_paragraph(1)
+    @formatter.push_margin('blockquote')
+  end
+
+  def end_blockquote
+    @formatter.end_paragraph(1)
+    @formatter.pop_margin()
+  end
+
+  def start_ul(attrs)
+    @formatter.end_paragraph(0)
+    @formatter.push_margin('ul')
+    @list_stack << ['ul', '*', 0]
+  end
+
+  def end_ul
+    if @list_stack
+      @list_stack.pop
+    end
+    @formatter.end_paragraph(0)
+    @formatter.pop_margin
+  end
+
+  def do_li(attrs)
+    @formatter.end_paragraph(0)
+    if @list_stack && @list_stack.size > 0
+      dummy, label, counter = top = @list_stack[-1]
+      top[2] = counter = counter+1
+    else
+      label, counter = '*', 0
+    end
+    @formatter.add_label_data(label, counter)
+  end
+
+  def start_ol(attrs)
+    @formatter.end_paragraph(0)
+    @formatter.push_margin('ol')
+    label = '1.'
+    for a, v in attrs
+      if a == 'type'
+        if v.length == 1
+          v = v + '.'
+          label = v
+        end
+      end
+    end
+    @list_stack << ['ol', label, 0]
+  end
+
+  def end_ol
+    if @list_stack
+      @list_stack.pop
+    end
+    @formatter.end_paragraph(0)
+    @formatter.pop_margin
+  end
+
+  def start_menu(attrs)
+    start_ul(attrs)
+  end
+
+  def end_menu
+    end_ul
+  end
+
+  def start_dir(attrs)
+    start_ul(attrs)
+  end
+
+  def end_dir
+    end_ul
+  end
+
+  def start_dl(attrs)
+    @formatter.end_paragraph(1)
+    @list_stack << ['dl', '', 0]
+  end
+
+  def end_dl
+    ddpop(1)
+    if @list_stack.length > 0
+      @list_stack.pop
+    end
+  end
+
+  def do_dt(attrs)
+    ddpop
+  end
+
+  def do_dd(attrs)
+    ddpop
+    @formatter.push_margin('dd')
+    @list_stack << ['dd', '', 0]
+  end
+
+  def ddpop(bl=0)
+    @formatter.end_paragraph(bl)
+    if @list_stack.length > 0
+      if @list_stack[-1][0] == 'dd'
+        @list_stack.pop
+        @formatter.pop_margin
+      end
+    end
+  end
+
+  def start_cite(attrs) start_i(attrs) end
+  def end_cite() end_i end
+
+  def start_code(attrs) start_tt(attrs) end
+  def end_code() end_tt end
+
+  def start_em(attrs) start_i(attrs) end
+  def end_em() end_i end
+
+  def start_kbd(attrs) start_tt(attrs) end
+  def end_kbd() end_tt end
+
+  def start_samp(attrs) start_tt(attrs) end
+  def end_samp() end_tt end
+
+  def start_strong(attrs) start_b(attrs) end
+  def end_strong() end_b end
+
+  def start_var(attrs) start_i(attrs) end
+  def end_var() end_i end
+
+  def start_i(attrs)
+    @formatter.push_font(nil, 1, nil, nil)
+  end
+  def end_i
+    @formatter.pop_font
+  end
+
+  def start_b(attrs)
+    @formatter.push_font(nil, nil, 1, nil)
+  end
+  def end_b
+    @formatter.pop_font
+  end
+
+  def start_tt(attrs)
+    @formatter.push_font(nil, nil, nil, 1)
+  end
+  def end_tt
+    @formatter.pop_font
+  end
+
+  def start_a(attrs)
+    href = nil
+    name = nil
+    type = nil
+    for attrname, value in attrs
+      value = value.strip
+      if attrname == 'href'
+        href = value
+      end
+      if attrname == 'name'
+        name = value
+      end
+      if attrname == 'type'
+        type = value.downcase
+      end
+    end
+    anchor_bgn(href, name, type)
+  end
+
+  def end_a
+    anchor_end
+  end
+
+  def do_br(attrs)
+    @formatter.add_line_break
+  end
+
+  def do_hr(attrs)
+    @formatter.add_hor_rule
+  end
+
+  def do_img(attrs)
+    align = nil
+    alt = '(image)'
+    ismap = nil
+    src = nil
+    width = 0
+    height = 0
+    for attrname, value in attrs
+      if attrname == 'align'
+        align = value
+      end
+      if attrname == 'alt'
+        alt = value
+      end
+      if attrname == 'ismap'
+        ismap = value
+      end
+      if attrname == 'src'
+        src = value
+      end
+      if attrname == 'width'
+        width = Integer(value)
+      end
+      if attrname == 'height'
+        height = Integer(value)
+      end
+    end
+    handle_image(src, alt, ismap, align, width, height)
+  end
+
+  def do_plaintext(attrs)
+    start_pre(attrs)
+    setnomoretags # Tell SGML parser
+  end
+
+  def unknown_starttag(tag, attrs)
+  end
+
+  def unknown_endtag(tag)
+  end
+
+end