Overview.bs

<h1>CSS Text Module Level 3</h1>

<pre class='metadata'>
Shortname: css-text
Level: 3
Status: ED
Work Status: Refining
Deadline: 2014-09-20
Group: csswg
ED: http://dev.w3.org/csswg/css-text-3/
TR: http://www.w3.org/TR/css-text-3/
Previous Version: http://www.w3.org/TR/2013/WD-css-text-3-20131010/
Previous Version: http://www.w3.org/TR/2012/WD-css3-text-20121113/
!Issue Tracking: <a href="http://www.w3.org/Style/CSS/Tracker/products/10">http://www.w3.org/Style/CSS/Tracker/products/10</a>
Test Suite: http://test.csswg.org/suites/css3-text/nightly-unstable/
Editor: Elika J. Etemad / fantasai, Invited Expert, http://fantasai.inkedblade.net/contact
Editor: Koji Ishii, Invited Expert, kojiishi@gluesoft.co.jp
Abstract: This CSS3 module defines properties for text manipulation and specifies their processing model. It covers line breaking, justification and alignment, white space handling, and text transformation.
At Risk: the ''full-width'' value of 'text-transform'
At Risk: the &lt;length> values of the 'tab-size' property
At Risk: the 'text-justify' property
At Risk: the percentage values of 'word-spacing'
At Risk: the 'hanging-punctuation' property
At Risk: the ''pre-wrap-auto'' value of the 'white-space' property
</pre>

  <style type="text/css">
    img { vertical-align: middle; }
    span[lang] { font-size: 125%; line-height: 1; vertical-align: middle;}

    /* Bidi & spaces example */
    .egbidiwsaA,.egbidiwsbB,.egbidiwsaB,.egbidiwsbC
      { white-space:pre;font-size:80%;font-family:monospace; vertical-align:2px; margin:1px }
    .egbidiwsaA { background:lime;padding:2px; }
    .egbidiwsbB { border:2px solid blue }
    .egbidiwsaB { background:yellow;border:2px dotted white }
    .egbidiwsbC { border:2px dotted red }

	.hyphens-ex {
		border: thin solid black;
		display: inline-block;
		padding: 4pt;
	}

    /* Start Letter-spacing Tutorial */
    .ls-ex {
      font-size: 200%;
      margin-left: 1em;
      margin-right: 1em;
    }
    .ls-fixed-width {
      width: 10em;
    }
    .color-box { background: rgb(224, 203, 82); }
    .bad { color: red; }
    .good { color: green; }
    /* End Letter-spacing Tutorial */

    .char { border: 1px dotted gray; }
    .quarter { font-size: 25%; }
    tt[lang="ja"] { font-family: "MS Gothic", "Osaka", monospace }
    div.figure table {
      margin :auto;
    }
    .feedback {
      background: #FFEECC;
      border-color: orange;
    }
    .feedback:before {
      content: "Info Needed";
      color: #FF8800;
    }
  </style>

<h2 id="intro">
Introduction</h2>

  <p>This module describes the typesetting controls of CSS;
  that is, the features of CSS that control the translation of
  source text to formatted, line-wrapped text.
  Various CSS properties provide control over
  <a href="#transforming">case transformation</a>,
  <a href="#white-space">white space collapsing</a>,
  <a href="#white-space">text wrapping</a>,
  <a href="#line-breaking">line breaking rules</a>
  and <a href="#hyphenation">hyphenation</a>,
  <a href="#justification">alignment and justification</a>,
  <a href="#spacing">spacing</a>,
  and <a href="#edge-effects">indentation</a>.

  <div class="note">
    <p>Font selection is covered in <a href="http://www.w3.org/TR/css3-fonts/">CSS Fonts Level 3</a> [[CSS3-FONTS]].

    <p>
    <span id="decoration"></span>
    <span id="text-decoration"></span>
    <span id="line-decoration"></span>
    <span id="text-decoration-line"></span>
    <span id="text-decoration-color"></span>
    <span id="text-decoration-style"></span>
    <span id="text-decoration-skip"></span>
    <span id="text-underline-position"></span>
    <span id="emphasis-marks"></span>
    <span id="text-emphasis-style"></span>
    <span id="text-emphasis-color"></span>
    <span id="text-emphasis"></span>
    <span id="text-emphasis-position"></span>
    <span id="text-shadow"></span>
    Features for decorating text,
    such as <a href="http://www.w3.org/TR/css-text-decor-3/#line-decoration">underlines</a>,
    <a href="http://www.w3.org/TR/css-text-decor-3/#emphasis-marks">emphasis marks</a>,
    and <a href="http://www.w3.org/TR/css-text-decor-3/#text-shadow-property">shadows</a>,
    (previously part of this module)
    are covered in
    <a href="http://www.w3.org/TR/css-text-decor-3/">CSS Text Decoration Level 3</a> [[CSS3-TEXT-DECOR]].

    <p><a href="http://www.w3.org/TR/css3-writing-modes/#text-direction">Bidirectional</a> and
    <a href="http://www.w3.org/TR/css3-writing-modes/#vertical-intro">vertical</a> text
    are addressed in
    <a href="http://www.w3.org/TR/css3-writing-modes/">CSS Writing Modes Level 3</a> [[CSS3-WRITING-MODES]].
  </div>

<h3 id="placement">
Module Interactions</h3>

  <p>This module, together with [[CSS3-TEXT-DECOR]],
  replaces and extends the text-level features defined in [[!CSS21]] chapter 16.

<h3 id="values">
Values</h3>

  <p>This specification follows the
  <a href="http://www.w3.org/TR/CSS21/about.html#property-defs">CSS property
  definition conventions</a> from [[!CSS21]]. Value types not defined in
  this specification are defined in CSS Level 2 Revision 1 [[!CSS21]].
  Other CSS modules may expand the definitions of these value types: for
  example [[CSS3VAL]], when combined with this module, expands the
  definition of the <var>&lt;length&gt;</var> value type as used in this specification.</p>

  <p>In addition to the property-specific values listed in their definitions,
  all properties defined in this specification also accept the
  <a href="http://www.w3.org/TR/CSS21/cascade.html#value-def-inherit">inherit</a>
  keyword as their property value. For readability it has not been repeated
  explicitly.

<h3 id="terms">
Terminology</h3>

  <p>In addition to the terms defined below,
    other terminology and concepts used in this specification are defined
    in [[!CSS21]] and [[!CSS3-WRITING-MODES]].

<h4 id="characters">
Characters and Letters</h4>

  <p>The basic unit of typesetting is the <dfn export>character</dfn>.
  However, because writing systems are not always as simple as the basic English alphabet,
  what a <a>character</a> actually is depends on the context in which the term is used.
  For example, in Hangul (the Korean writing system),
  each square representation of a syllable
  (e.g. <span lang=ko-hang title="Hangul syllable HAN">한</span>=<span lang=ko-Latn>Han</span>)
  can be considered a <a>character</a>.
  However, the square symbol is really composed of multiple letters each representing a phoneme
  (e.g. <span lang=ko-hang title="Hangul letter HIEUH">ㅎ</span>=<span lang=ko-Latn>h</span>,
   <span lang=ko-hang title="Hangul letter HIEUH">ㅏ</span>=<span lang=ko-Latn>a</span>,
   <span lang=ko-hang title="Hangul letter HIEUH">ㄴ</span>=<span lang=ko-Latn>n</span>)
  and these also could each be considered a <a>character</a>.

  <p>A basic unit of computer text encoding, for any given encoding,
  is also called a <a>character</a>,
  and depending on the encoding,
  a single encoding <a>character</a> might correspond
  to the entire pre-composed syllabic <a>character</a> (e.g. <span lang=ko-hang title="Hangul syllable HAN">한</span>),
  to the individual phonemic <a>character</a> (e.g. <span lang=ko-hang title="Hangul letter HIEUH">ㅎ</span>),
  or to smaller units such as
  a base letterform (e.g. <span lang=ko-hang title="Hangul letter IEUNG">ㅇ</span>)
  and any combining marks that vary it (e.g. extra strokes that represent aspiration).

  <p>In turn, a single encoding <a>character</a> can be represented in the data stream as one or more bytes;
  and in programming environments one byte is sometimes also called a <a>character</a>.

  <p>Therefore the term <a>character</a> is fairly ambiguous where technical precision is required.

  <p>For text layout, we will refer to the <dfn export lt="typographic character unit|typographic character">typographic character unit</dfn>
  as the basic unit of text.
  Even within the realm of text layout,
  the relevant <a>character</a> unit depends on the operation.
  For example, line-breaking and letter-spacing will segment
  a sequence of Thai characters that include U+0E33 THAI CHARACTER SARA AM differently;
  or the behaviour of a conjunct consonant in a script such as Devanagari
  may depend on the font in use.
  So the <a>typographic character</a> represents a unit of the writing system&mdash;<!--
  -->such as a Latin alphabetic letter (including its diacritics),
  Hangul syllable,
  Chinese ideographic character,
  Myanmar syllable cluster&mdash;<!--
  -->that is indivisible with respect to a particular typographic operation
  (line-breaking, first-letter effects, tracking, justification, vertical arrangement, etc.).

  <a href="http://www.unicode.org/reports/tr29/">Unicode Standard Annex #29: Text Segmentation</a>
  defines a unit called the <dfn>grapheme cluster</dfn>
  which approximates the <a>typographic character</a>.
  A UA must use the <em>extended grapheme cluster</em>
  (not <em>legacy grapheme cluster</em>), as defined in [[!UAX29]],
  as the basis for its <a>typographic character unit</a>.
  However, the UA should tailor the definitions
  as required by typographic tradition
  since the default rules are not always appropriate or ideal--
  and is expected to tailor them differently
  depending on the operation as needed.

  <p class="note">
  The rules for such tailorings are out of scope for CSS.
<!--
  however W3C currently maintains a wiki page
  where some known tailorings are collected.
-->

  <div class="example">
    The following are some examples of <a>typographic character unit</a> tailorings
    required by standard typesetting practice:
    
    <ul>
      <li>
        <p>In some scripts such as Myanmar or Devanagari,
        the <a>typographic character unit</a> for both justification and line-breaking
        is an entire syllable,
        which can include more than one [[!UAX29]] <a>grapheme cluster</a>.

      <li>
        <p>In other scripts such as Thai or Lao,
        even though for line-breaking the <a>typographic character</a>
        matches Unicode’s default <a>grapheme clusters</a>,
        for letter-spacing the relevant unit
        is <em>less</em> than a [[!UAX29]] <a>grapheme cluster</a>,
        and may require decomposition or other substitutions
        before spacing can be inserted.

        <p>For instance,
        to properly letter-space the Thai word คำ (U+0E04 + U+0E33),
        the U+0E33 needs to be decomposed into U+0E4D + U+0E32,
        and then the extra letter-space inserted before the U+0E32: คํ า.

        <p>A slightly more complex example is น้ำ (U+0E19 + U+0E49 + U+0E33).
        In this case, normal Thai shaping will first decompose the U+0E33 into U+0E4D + U+0E32
        and then swap the U+0E4D with the U+0E49, giving U+0E19 + U+0E4D + U+0E49 + U+0E32.
        As before the extra letter-space is then inserted before the U+0E32: นํ้ า.

      <li>
        <p>Vertical typesetting [[!CSS3-WRITING-MODES]] can also require tailoring.
        For example, when typesetting ''text-orientation/upright'' text,
        Tibetan tsek and shad marks are kept with the preceding grapheme cluster,
        rather than treated as an independent <a>typographic character unit</a>.
    </ul>
  </div>

  <p>A <dfn export>typographic letter unit</dfn> or <dfn>letter</dfn> for the purpose of this specification
  is a <a>typographic character unit</a> belonging to one of the Letter or Number general
  categories in Unicode. [[!UAX44]]
  See <a href="#character-properties">Character Properties</a>
  for how to determine the Unicode properties of a <a>typographic character unit</a>.

  <p>The rendering characteristics of a <a>typographic character unit</a> divided
  by an element boundary is undefined:
  it may be rendered as belonging to either side of the boundary,
  or as some approximation of belonging to both.
  Authors are forewarned that dividing <a>grapheme clusters</a>
  by element boundaries may give inconsistent or undesired results.

<h4 id="languages">
Languages and Typesetting</h4>

  <p class="note">
    Many typographic effects vary by linguistic context.
    In CSS, language-specific typographic tailorings
    are only applied when the content language is known (declared).

  <p><strong class="advisement">
    Authors should language-tag their content accurately for the best typographic behavior.
  </strong>

  <p>The <dfn>content language</dfn> of an element is the (human) language
    the element is declared to be in, according to the rules of the
    <a href="http://www.w3.org/TR/CSS21/conform.html#doclanguage">document language</a>.
    For example, the rules for determining the <a>content language</a> of an HTML
    element use the <code>lang</code> attribute and are defined in [[HTML5]],
    and the rules for determining the <a>content language</a> of an XML element use
    the <code>xml:lang</code> attribute and are
    <a href="http://www.w3.org/TR/REC-xml/#sec-lang-tag">defined</a> in [[XML10]].
    Note that it is possible for the <a>content language</a> of an element
    to be unknown.

<h2 id="transforming">
  Transforming Text</h2>

  <h3 id="text-transform-property">
    <span id="caps-prop"></span>
    <span id="text-transform"></span>
    Case Transforms: the 'text-transform' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>text-transform</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>none | capitalize | uppercase | lowercase | full-width
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>none</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>all elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>as specified</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property transforms text for styling purposes.
      (It has no effect on the underlying content.)
      Values have the following meanings:</p>

    <dl dfn-for=text-transform dfn-type=value>
      <dt><dfn>none</dfn></dt>
        <dd>No effects.</dd>
      <dt><dfn>capitalize</dfn></dt>
        <dd>Puts the first <a>typographic letter unit</a> of each word in titlecase;
          other characters are unaffected.</dd>
      <dt><dfn>uppercase</dfn></dt>
        <dd>Puts all lettersin uppercase.
      <dt><dfn>lowercase</dfn></dt>
        <dd>Puts all letters in lowercase.</dd>
      <dt><dfn>full-width</dfn></dt>
        <dd>Puts all <a>typographic character units</a> in fullwidth form.
          If a character does not have a corresponding fullwidth form,
          it is left as is.
          This value is typically used to typeset Latin letters and digits
          as if they were ideographic characters.
    </dl>

    <p>For ''capitalize'', what constitutes a “word“ is UA-dependent;
      [[!UAX29]] is suggested (but not required) for determining such word
      boundaries. Authors should not expect ''capitalize'' to follow
      language-specific titlecasing conventions (such as skipping articles
      in English).

    <div class="example">
      <p>The following example converts the ASCII characters
        used in abbreviations in Japanese text to their fullwidth variants
        so that they lay out and line break like ideographs:
      <pre>abbr:lang(ja) { text-transform: full-width; }</pre>
    </div>

    <p class="note">
      Note that, as defined in <a href="#order">Text Processing Order of Operations</a>,
      transforming text affects line-breaking and other formatting operations.

    <p>
      The UA must use the full case mappings for Unicode
      characters, including any conditional casing rules, as defined in
      Default Case Algorithm section of The Unicode Standard [[!UNICODE]].
      If (and only if) the <a>content language</a>
      of the element is, according to the rules of the
      <a href="http://www.w3.org/TR/CSS21/conform.html#doclanguage">document language</a>,
      known,
      then any appropriate language-specific rules must be applied as well.
      These minimally include, but are not limited to, the language-specific
      rules in Unicode's
      <a href="http://www.unicode.org/Public/UNIDATA/SpecialCasing.txt">SpecialCasing.txt</a>.

    <div class="example">
      <p>For example, in Turkish there are two &ldquo;i&rdquo;s, one with
        a dot&mdash;&ldquo;İ&rdquo; and &ldquo;i&rdquo;&mdash; and one
        without&mdash;&ldquo;I&rdquo; and &ldquo;ı&rdquo;. Thus the usual
        case mappings between &ldquo;I&rdquo; and &ldquo;i&rdquo; are
        replaced with a different set of mappings to their respective
        undotted/dotted counterparts, which do not exist in English. This
        mapping must only take effect if the <a>content language</a> is Turkish
        (or another Turkic language that uses Turkish casing rules);
        in other languages, the usual mapping of &ldquo;I&rdquo;
        and &ldquo;i&rdquo; is required. This rule is thus conditionally
        defined in Unicode's SpecialCasing.txt file.
    </div>

    <p>The definition of fullwidth and halfwidth forms can be found on the
      Unicode consortium web site at [[!UAX11]].
      The mapping to fullwidth form is defined by taking code points with
      the <code>&lt;wide&gt;</code> or the <code>&lt;narrow&gt;</code> tag
      in their <code>Decomposition_Mapping</code> in [[!UAX44]].
      For the <code>&lt;narrow&gt;</code> tag,
      the mapping is from the code point to the decomposition (minus <code>&lt;narrow&gt;</code> tag),
      and for the <code>&lt;wide&gt;</code> tag,
      the mapping is from the decomposition (minus the <code>&lt;wide&gt;</code> tag)
      back to the original code point.

    <p>Text transformation happens after <a href="#white-space-rules">white
      space processing</a>, which means that ''full-width'' only transforms
      U+0020 spaces to U+3000 within <a>preserved</a> white space.

    <p class="note">
      A future level of CSS may introduce the ability to create custom mapping
      tables for less common text transforms, such as by an ''@text-transform''
      rule similar to ''@counter-style'' from [[CSS-COUNTER-STYLES-3]].


<h2 id="white-space-property">
  <span id="white-space-collapsing"></span><span id='text-wrap'></span>
  White Space and Wrapping: the 'white-space' property</h2>

  <table class="propdef">
      <tr>
        <th>Name:</th>
        <td><dfn>white-space</dfn></td>
      </tr>
      <tr>
        <th><a href="#values">Value</a>:
        <td>normal | pre | nowrap | pre-wrap | pre-wrap-auto | pre-line</td>
      </tr>
      <tr>
        <th>Initial:</th>
        <td>normal
      </tr>
      <tr>
        <th>Applies to:</th>
        <td>all elements</td>
      </tr>
      <tr>
        <th>Inherited:</th>
        <td>yes</td>
      </tr>
      <tr>
        <th>Percentages:</th>
        <td>N/A</td>
      </tr>
      <tr>
        <th>Media:</th>
        <td>visual</td>
      </tr>
      <tr>
        <th>Computed&#160;value:</th>
        <td>as specified
      </tr>
      <tr>
        <th>Animatable:
        <td>no
      <tr>
        <th>Canonical order:
        <td>N/A
  </table>

  <p>This property specifies two things:
  <ul>
    <li>whether and how <a>white space</a> inside the element is collapsed
    <li>whether lines may <a>wrap</a> at unforced <a>soft wrap opportunities</a>
  </ul>

  <p>Values have the following meanings, which must be interpreted
    according to
    the <a href="#white-space-rules">White Space Processing</a> and
    <a href="#line-breaking">Line Breaking</a> rules:</p>

  <dl dfn-for=white-space dfn-type=value>
    <dt><dfn>normal</dfn></dt>
    <dd>This value directs user agents to collapse sequences of <a>white space</a>
      into a single character (or <a href="#line-break-transform">in some
      cases</a>, no character).
      Lines may wrap at allowed <a>soft wrap opportunities</a>,
      as determined by the line-breaking rules in effect,
      in order to minimize inline-axis overflow.
    <dt><dfn>pre</dfn></dt>
    <dd>This value prevents user agents from collapsing sequences of <a>white space</a>.
      <a>Segment breaks</a> such as line feeds and carriage returns are preserved as <a>forced line breaks</a>.
      Lines only break at <a>forced line breaks</a>;
      content that does not fit within the block container overflows it.
    <dt><dfn>nowrap</dfn>
    <dd>Like ''white-space/normal'', this value collapses <a>white space</a>;
      but like ''pre'', it does not allow wrapping.
    <dt><dfn>pre-wrap</dfn></dt>
    <dd>Like ''pre'', this value preserves <a>white space</a>;
      but like ''white-space/normal'', it allows wrapping.
    <dt><dfn>pre-wrap-auto</dfn></dt>
    <dd>Like ''pre-wrap'', this value preserves <a>white space</a> and allows wrapping.
      However, some of the behavior is made UA dependent
      (See <a href="#white-space-rules">The White Space Processing Rules</a>
      for details).
      This flexibility allows UAs to match platform conventions,
      in particular with respect to multi-line editable text fields,
      and when preserved <a>white space</a> at the end a line
      would cause the line to overflow.
    <dt><dfn>pre-line</dfn></dt>
    <dd>Like ''white-space/normal'', this value collapses consecutive spaces and allows wrapping,
      but preserves <a>segment breaks</a> in the source as <a>forced line breaks</a>.
  </dl>

  Issue: Suggestions of a better name for ''pre-wrap-auto'' are welcome.

  Note: ''pre-wrap-auto'' is at risk.

  <p>The following informative table summarizes the behavior of various
    'white-space' values:</p>

  <table class="data">
    <colgroup class="header"></colgroup>
    <colgroup span=3></colgroup>
    <thead>
      <tr>
        <th></th>
        <th>New Lines</th>
        <th>Spaces and Tabs</th>
        <th>Text Wrapping</th>
      </tr>
    </thead>
    <tbody>
      <tr>
        <th>''white-space/normal''</th>
        <td>Collapse</td>
        <td>Collapse</td>
        <td>Wrap</td>
      </tr>
      <tr>
        <th>''pre''</th>
        <td>Preserve</td>
        <td>Preserve</td>
        <td>No wrap</td>
      </tr>
      <tr>
        <th>''nowrap''</th>
        <td>Collapse</td>
        <td>Collapse</td>
        <td>No wrap</td>
      </tr>
      <tr>
        <th>''pre-wrap''</th>
        <td>Preserve</td>
        <td>Preserve</td>
        <td>Wrap</td>
      </tr>
      <tr>
        <th>''pre-wrap-auto''</th>
        <td>Preserve</td>
        <td>Preserve</td>
        <td>Wrap</td>
      </tr>
      <tr>
        <th>''pre-line''</th>
        <td>Preserve</td>
        <td>Collapse</td>
        <td>Wrap</td>
    </tbody>
  </table>

  <p>See <a href="#white-space-processing">White Space Processing Rules</a>
    for details on how white space collapses. An informative summary of
    collapsing (''white-space/normal'' and ''nowrap'') is presented below:
    <ul>
      <li>A sequence of segment breaks and other <a>white space</a> between two
        Chinese, Japanese, or Yi characters collapses into nothing.
      <li>A zero width space before or after a white space sequence
        containing a segment break causes the entire sequence of <a>white space</a>
        to collapse into a zero width space.
      <li>Otherwise, consecutive <a>white space</a> collapses into a single space.
    </ul>

  <p>See <a href="#line-breaking">Line Breaking</a>
    for details on wrapping behavior.

<h2 id="white-space-processing">
  White Space Processing Details</h2>

  <p>The source text of a document often contains formatting
    that is not relevant to the final rendering: for example,
    <a href="http://rhodesmill.org/brandon/2012/one-sentence-per-line/"
    >breaking the source into segments</a> (lines) for ease of editing
    or adding white space characters such as tabs and spaces to indent the source code.
    CSS white space processing allows the author to control interpretation of such formatting:
    to preserve or collapse it away when rendering the document.
    White space processing in CSS interprets white space characters only for rendering:
    it has no effect on the underlying document data.

  <p>White space processing in CSS is controlled with the 'white-space' property.

  <p id="segment-normalization">
    CSS does not define document segmentation rules. Segments can be
    separated by a particular newline sequence (such as a line feed or
    CRLF pair), or delimited by some other mechanism, such as the SGML
    <code>RECORD-START</code> and <code>RECORD-END</code> tokens.
    For CSS processing, each document language&ndash;defined segment break,
    CRLF sequence (U+000D U+000A), carriage return (U+000D), and line feed (U+000A)
    in the text is treated as a <dfn export>segment break</dfn>,
    which is then interpreted for rendering as specified by the 'white-space' property.

  <p class="note">Note that a document parser might
    not only normalize any <a>segment breaks</a>,
    but also collapse other space characters or
    otherwise process white space according to markup rules.
    Because CSS processing occurs <em>after</em> the parsing stage,
    it is not possible to restore these characters for styling.
    Therefore, some of the behavior specified below
    can be affected by these limitations and
    may be user agent dependent.</p>

  <p class="note">Note that anonymous blocks consisting entirely of
    <a>collapsible</a> <a>white space</a> are removed from the rendering tree.
    Thus any such <a>white space</a> surrounding a block-level element is collapsed away.
    See [[CSS21]] section
    <a href="http://www.w3.org/TR/CSS21/visuren.html#anonymous">9.2.2.1</a></p>

  <p>
    Control characters (<a>Unicode category</a> <code>Cc</code>)
    other than tab (U+0009), line feed (U+000A), form feed (U+000C), and carriage return (U+000D)
    must be rendered as a visible glyph
    and otherwise treated as any other character
    of the Other Symbols (<code>So</code>) <a>general category</a> and Common <a lt="Unicode script">script</a>.
    The UA may use a glyph provided by a font specifically for the control character,
    substitute the glyphs provided for the corresponding symbol in the Control Pictures block,
    generate a visual representation of its codepoint value,
    or use some other method to provide an appropriate visible glyph.
    As required by [[!UNICODE]],
    unsupported <code>Default_ignorable</code> characters must be ignored for rendering.

  <h3 id="white-space-rules">
    The White Space Processing Rules</h3>

    <p>White space processing in CSS affects only
      the <dfn export lt="white space|white space characters| document white space|document white space characters">document white space characters</dfn>:
      spaces (U+0020), tabs (U+0009), and <a href="#white-space-processing">segment breaks</a>.

    <p class="note">
      Note that the set of characters considered <a>document white space</a> (part of the document content)
      and that considered syntactic white space (part of the CSS syntax)
      are not necessarily identical.
      However, since both include spaces (U+0020), tabs (U+0009), line feeds (U+000A), and carriage returns (U+000D)
      most authors won't notice any differences.

    <h4 id="white-space-phase-1">Phase I: Collapsing and Transformation</h4>

    <p>For each inline (including anonymous inlines;
      see [[CSS21]] section <a href="http://www.w3.org/TR/CSS21/visuren.html#anonymous">9.2.2.1</a>)
      within an inline
      formatting context, white space characters are handled as follows,
      ignoring bidi formatting characters as if they were not there:</p>

    <ul>
    <li id="collapse"><p>If 'white-space' is set to
      ''white-space/normal'', ''nowrap'', or ''pre-line'',
      white space characters are considered <dfn export lt="collapsible white space|collapsible">collapsible</dfn>
      and are processed by performing the following steps:</p>
      <ol>
        <li>All spaces and tabs immediately preceding or following a segment
          break are removed.</li>
        <li><a>Segment breaks</a> are transformed for
          rendering according to the <a href="#line-break-transform">segment break transformation rules</a>.
        </li>
        <li>Every tab is converted to a space (U+0020).</li>
        <li>Any space immediately following another collapsible space&mdash;even
          one outside the boundary of the inline containing that space,
          provided both spaces are within the same inline formatting
          context&mdash;is collapsed to have zero advance width. (It is
          invisible, but retains its <a>soft wrap opportunity</a>, if any.)</li>
      </ol>
    </li>
    <li><p>If 'white-space' is set to ''pre-wrap'',
      any sequence of spaces is preserved.
      This is achieved by treating it as a sequence of non-breaking spaces,
      except that a <a>soft wrap opportunity</a> exists
      between each of the spaces of this preserved sequence,
      as well as at the beginning
      and at the end of the sequence.</p></li>
    <li><p>If 'white-space' is set to ''pre-wrap-auto'',
      any sequence of spaces is preserved.
      This is achieved by treating it as a sequence of non-breaking spaces.
      However, unlike with ''pre-wrap'',
      whether <a>soft wrap opportunities</a> exist in this sequence
      is user-agent defined.
      These UA-specific rules may be context dependent.</p></li>
    </ul>

    <p>Then, the entire block is rendered. Inlines are laid out, taking bidi
      reordering into account, and <a>wrapping</a> as specified by the
      'white-space' property.</p>

    <div class="example" id="egbidiwscollapse">
      <p>The following example illustrates
        the interaction of white-space collapsing and bidirectionality.
        Consider the following markup fragment, taking special note of spaces
        (with varied backgrounds and borders for emphasis and identification):

      <pre><code>&lt;ltr&gt;A<span class="egbidiwsaA">&#160;</span>&lt;rtl&gt;<span class="egbidiwsbB">&#160;</span>B<span class="egbidiwsaB">&#160;</span>&lt;/rtl&gt;<span class="egbidiwsbC">&#160;</span>C&lt;/ltr&gt;</code></pre>

      <p>where the <code>&lt;ltr&gt;</code> element represents a left-to-right embedding
        and the <code>&lt;rtl&gt;</code> element represents a right-to-left embedding.
        If the 'white-space' property is set to ''white-space/normal'',
        the white-space processing model will result in the following:

      <ul style="line-height:1.3">
        <li>The space before the B (<span class="egbidiwsbB">&#160;</span>)
          will collapse with the space after the A (<span class="egbidiwsaA">&#160;</span>).
        <li>The space before the C (<span class="egbidiwsbC">&#160;</span>)
          will collapse with the space after the B (<span class="egbidiwsaB">&#160;</span>).
      </ul>

      <p>This will leave two spaces,
        one after the A in the left-to-right embedding level,
        and one after the B in the right-to-left embedding level.
        The text will then be ordered according to the Unicode bidirectional algorithm,
        with the end result being:

      <pre>A<span class="egbidiwsaA">&#160;</span><span class="egbidiwsaB">&#160;</span>BC</pre>

      <p>Note that there will be two spaces between A and B,
        and none between B and C.
        This is best avoided by putting spaces outside the element
        instead of just inside the opening and closing tags and, where practical,
        by relying on implicit bidirectionality instead of explicit embedding levels.
    </div>

    <h4 id="line-break-transform">
      Segment Break Transformation Rules</h4>

      <p>When 'white-space' is ''pre'', ''pre-wrap'', ''pre-wrap-auto'', or ''pre-line'',
      <a>segment breaks</a> are not <a>collapsible</a>
      and are instead transformed into a preserved line feed (U+000A).

      <p>For other values of 'white-space', <a>segment breaks</a> are <a>collapsible</a>,
        and are either transformed into a space (U+0020) or removed
        depending on the context before and after the break:</p>

      <ul>
        <li>If the character immediately before or immediately after the segment
          break is the zero-width space character (U+200B), then the break
          is removed, leaving behind the zero-width space.
        <li>Otherwise, if the <a>East Asian Width property</a> [[!UAX11]] of both
          the character before and after the line feed is <code>F</code>, <code>W</code>, or <code>H</code> (not <code>A</code>),
          and neither side is Hangul, then the segment break is removed.
        <li>Otherwise, the segment break is converted to a space (U+0020).
      </ul>

      <p class="note">Note that the white space processing rules have already
        removed any tabs and spaces after the segment break before these checks
        take place.</p>

      <p class="feedback issue">Comments on how well this would work in practice would
        be very much appreciated, particularly from people who work with
        Thai and similar scripts.
        Note that browser implementations do not currently follow these rules
        (although IE does in some cases transform the break).</p>

    <h4 id="white-space-phase-2">Phase II: Trimming and Positioning</h4>

    <p>As each line is laid out,</p>
    <ol>
      <li>A sequence of collapsible spaces at the beginning of a line is
        removed.
      <li>Each tab is rendered as a horizontal shift
        that lines up the start edge of the next glyph with the next <a>tab stop</a>.
        <dfn lt="tab stop">Tab stops</dfn> occur at points that are multiples of the tab size
        from the block's starting content edge.
        The tab size is given by the 'tab-size' property.
      <li>A sequence of <a>collapsible</a> spaces at the end of a line is removed.
      <li>If spaces or tabs at the end of a line are non-collapsible but
         have 'white-space' set to ''pre-wrap-auto'' the UA may visually
         collapse their character advance widths.
    </ol>

    <p>White space that was not removed or collapsed during the white space
      processing steps is called <dfn>preserved</dfn> white space.</p>

  <h3 id="tab-size-property">
    Tab Character Size: the 'tab-size' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>tab-size</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>&lt;integer&gt; | &lt;length&gt;</td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>8</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>block containers</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>the specified integer or length made absolute</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>as <a href="http://www.w3.org/TR/css3-transitions/#animtype-length">length</a>
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property determines the tab size used to render preserved tab characters (U+0009).
      Integers represent the measure as multiples of the space character's advance width (U+0020).
      Negative values are not allowed.

<h2 id="line-breaking">
  Line Breaking and Word Boundaries</h2>

  <p>When inline-level content is laid out into lines, it is broken across line boxes.
    Such a break is called a <dfn>line break</dfn>.
    When a line is broken due to explicit line-breaking controls,
    or due to the start or end of a block,
    it is a <dfn>forced line break</dfn>.
    When a line is broken due to content <dfn lt="wrapping|wrap">wrapping</dfn>
    (i.e. when the UA creates unforced line breaks in order to fit the content within the measure),
    it is a <dfn>soft wrap break</dfn>.
    The process of breaking inline-level content into lines is called <dfn lt="line-breaking-process">line breaking</dfn>.

  <p>Wrapping is only performed at an allowed break point, called a <dfn export>soft wrap opportunity</dfn>.

  <p>In most writing systems,
    in the absence of hyphenation a <a>soft wrap opportunity</a> occurs only at word boundaries.
    Many such systems use spaces or punctuation to explicitly separate words,
    and <a>soft wrap opportunities</a> can be identified by these characters.
    Scripts such as Thai, Lao, and Khmer, however,
    do not use spaces or punctuation to separate words.
    Although the zero width space (U+200B) can be used as an explicit word delimiter in these scripts,
    this practice is not common.
    As a result, a lexical resource is needed to correctly identify <a>soft wrap opportunities</a> in such texts.

  <p>In several other writing systems, (including Chinese, Japanese, Yi, and sometimes also Korean)
    a <a>soft wrap opportunity</a> is based on syllable boundaries, not word boundaries.
    In these systems a line can break anywhere <em>except</em> between certain character combinations.
    Additionally the level of strictness in these restrictions can vary with the typesetting style.</p>

  <p>CSS does not fully define where <a>soft wrap opportunities</a> occur;
    however some controls are provided to distinguish common variations.

  <div class="note">
    <p>Further information on line breaking conventions can be found in
      [[JLREQ]] and [[JIS4051]] for Japanese,
      [[ZHMARK]] for Chinese, and
      in [[!UAX14]] for all scripts in Unicode.

    <p class="feedback issue">Any guidance for appropriate references here would be
      much appreciated.
  </div>

<h3 id="line-break-details">
Line Breaking Details</h3>

  <p>When determining <a>line breaks</a>:
  <ul>
    <li>Regardless of the 'white-space' value,
      lines always break at each <a>preserved</a> forced break character:
      for all values, line-breaking behavior defined for
      the BK, CR, LF, CM, NL, and SG line breaking classes in [[!UAX14]]
      must be honored.
    <li>When 'white-space' allows wrapping,
      line breaking behavior defined for the WJ, ZW, and GL line-breaking classes in [[!UAX14]]
      must be honored.
    <li>UAs that allow wrapping at punctuation other than spaces should prioritize breakpoints.
      For example, if breaks after slashes are given a lower priority than spaces,
      the sequence "check /etc" will never break between the "/" and the "e".
      As long as care is taken to avoid such awkward breaks, allowing breaks at
      appropriate punctuation other than spaces is recommended, as it results
      in more even-looking margins, particularly in narrow measures.
      The UA may use the width of the containing block, the text's language,
      and other factors in assigning priorities.
    <li>Out-of-flow elements do not introduce a <a>forced line break</a>
      or <a>soft wrap opportunity</a> in the flow.
    <li>The line breaking behavior of a replaced element or other atomic inline
      is equivalent to that of the Object Replacement Character (U+FFFC)
      and introduces a <a>soft wrap opportunity</a> both before and after itself.
      For Web-compatibility, this rule take precedence over <code>WJ</code> and <code>GL</code> handling;
      in terms of [[!UAX14]],
      this shifts the <code>CB</code> rule (<a href="http://www.unicode.org/reports/tr14/#LB20"><code>LB20</code></a>)
      immediately above the <code>WJ</code> and <code>GL</code> rules (<a href="http://www.unicode.org/reports/tr14/#LB11"><code>LB11</code></a>/<a href="http://www.unicode.org/reports/tr14/#LB12"><code>LB12</code></a>).
    <li>For <a>soft wrap opportunities</a> created by characters that disappear at the line break (e.g. U+0020 SPACE),
      properties on the element containing that character control the line breaking at that opportunity.
      For <a>soft wrap opportunities</a> defined by the boundary between two characters,
      the properties on the element containing the boundary control breaking.
      <!-- http://lists.w3.org/Archives/Public/www-style/2008Dec/0043.html -->
    <li>For <a>soft wrap opportunities</a> before the first or after the last character of a box,
      the break occurs immediately before/after the box (at its margin edge)
      rather than breaking the box between its content edge and the content.
    <li>Line breaking in/around Ruby is defined in <a href="http://www.w3.org/TR/css-ruby-1/#line-breaks">CSS Ruby</a> [[!CSS3RUBY]].
  </ul>

  <h3 id="word-break-property">
    Breaking Rules for Letters: the 'word-break' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>word-break</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>normal | keep-all | break-all</td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>normal</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>all elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>specified value</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property specifies <a>soft wrap opportunities</a> between letters,
      i.e. where it is “normal” and permissible to break lines of text.
      It does not affect rules governing the <a>soft wrap opportunities</a>
      created by spaces and punctuation.
      (See 'line-break' for controls affecting punctuation.)

    <div class="example">
      <p>For example, in some styles of CJK typesetting, English words are allowed
        to break between any two letters, rather than only at spaces or hyphenation points;
        this can be enabled with ''word-break:break-all''.

      <div class="figure">
        <img src="images/break-all.png" alt="A snippet of Japanese text with English in it. The word 'caption' is broken into 'capt' and 'ion' across two lines.">
        <p class="caption">An example of English text embedded in Japanese
          being broken at an arbitrary point in the word.
      </div>

      <p>As another example, Korean has two styles of line-breaking:
        between any two Korean syllables (''word-break: normal'')
        or, like English, mainly at spaces (''word-break: keep-all'').
      <pre>각 줄의 마지막에 한글이 올 때 줄 나눔 기 /* break between syllables */
<!--    -->준을 “글자” 또는 “어절” 단위로 한다.</pre>
      <pre>각 줄의 마지막에 한글이 올 때 줄 나눔   /* break only at spaces */
<!--    -->기준을 “글자” 또는 “어절” 단위로 한다.</pre>
    </div>

    <p class="note">
      To enable additional break opportunities only in the case of overflow,
      see 'overflow-wrap'.

    <p>Values have the following meanings:</p>

    <dl dfn-for=word-break dfn-type=value>
      <dt><dfn>normal</dfn></dt>
      <dd>Words break according to their customary rules,
       as described <a href="#line-breaking">above</a>.
       Korean, which commonly exhibits two different behaviors,
       allows breaks between any two consecutive Hangul/Hanja.

      <dt><dfn>break-all</dfn></dt>
      <dd>Breaking is allowed within “words”:
        in addition to ''word-break/normal'' <a>soft wrap opportunities</a>,
        lines may break between any two <a>typographic letter units</a>
        (except where forbidden by the 'line-break' property).
        Hyphenation is not applied. This option is used mostly in a context where
        the text consists predominantly of CJK characters with only short non-CJK excerpts,
        and it is desired that the text be better distributed on each line.</dd>

      <dt><dfn>keep-all</dfn></dt>
      <dd>Breaking is forbidden within “words”:
        implicit <a>soft wrap opportunities</a> between <a>typographic letter units</a> are suppressed,
        i.e. breaks are prohibited between pairs of letters
        (regardless of 'line-break' settings)
        except where opportunities exist due to dictionary-based breaking.
        Otherwise this option is equivalent to ''word-break/normal''.
        In this style, sequences of CJK characters do not break.
        <p class=note>This is the other common behavior for Korean (which uses spaces between words),
        and is also useful for mixed-script text where CJK snippets are mixed
        into another language that uses spaces for separation.</dd>
    </dl>

    <p>Symbols that line-break the same way as letters of a particular category
    are affected the same way as those letters.

     <div class="example">
      <p>Here's a mixed-script sample text:
      <pre>这是一些汉字, and some Latin,<!--
        --> &#x0648; &#x06A9;&#x0645;&#x06CC; &#x0646;&#x0648;&#x0634;&#x062A;&#x0646; &#x0639;&#x0631;&#x0628;&#x06CC;, <!--
        -->และตัวอย่างการเขียนภาษาไทย.</pre>
      <p>The break-points are determined as follows (indicated by &lsquo;&middot;&rsquo;):
      <dl>
        <dt>''word-break: normal''</dt>
        <dd>
      <pre>这·是·一·些·汉·字,·and·some·Latin,<!--
        -->·&#x0648;·&#x06A9;&#x0645;&#x06CC;·&#x0646;&#x0648;&#x0634;&#x062A;&#x0646;·&#x0639;&#x0631;&#x0628;&#x06CC;,·<!--
        -->และ·ตัวอย่าง·การเขียน·ภาษาไทย.</pre>
        <dt>''word-break: break-all''</dt>
        <dd>
      <pre>这·是·一·些·汉·字,·a·n·d·s·o·m·e·L·a·t·i·n,<!--
        -->·&#x0648;·&#xFB90;·&#xFEE4;·&#xFEF0;·&#xFEE7;·&#xFEEE;·&#xFEB7;·&#xFE98;·&#xFEE6;·&#xFECB;·&#xFEAE;·&#xFE91;·&#xFEF0;,·<!--
        -->แ·ล·ะ·ตั·ว·อ·ย่·า·ง·ก·า·ร·เ·ขี·ย·น·ภ·า·ษ·า·ไ·ท·ย.</pre>
        <dt>''word-break: keep-all''</dt>
        <dd>
      <pre>这是一些汉字,·and·some·Latin,<!--
        -->·&#x0648;·&#x06A9;&#x0645;&#x06CC;·&#x0646;&#x0648;&#x0634;&#x062A;&#x0646;·&#x0639;&#x0631;&#x0628;&#x06CC;,·<!--
        -->แและ·ตัวอย่าง·การเขียน·ภาษาไทย.</pre>
      </dl>
    </div>

    <p>When shaping scripts such as Arabic are allowed to break within words
      due to ''break-all'', the characters must still be shaped as if the
      word were not broken.</p>

  <h3 id="line-break-property">
    Breaking Rules for Punctuation: the 'line-break' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>line-break</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>auto | loose | normal | strict</td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>auto</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>all elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>specified value</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property specifies the strictness of line-breaking rules applied
      within an element:
      particularly how <a>wrapping</a> interacts with punctuation and symbols.
      Values have the following meanings:</p>

    <dl dfn-for=line-break dfn-type=value>
      <dt><dfn>auto</dfn></dt>
      <dd>The UA determines the set of line-breaking restrictions to use,
        and it may vary the restrictions based on the length of the line; e.g.,
        use a less restrictive set of line-break rules for short lines.
      <dt><dfn>loose</dfn></dt>
      <dd>Breaks text using the least restrictive set of line-breaking
        rules. Typically used for short lines, such as in newspapers.</dd>
      <dt><dfn>normal</dfn></dt>
      <dd>Breaks text using the most common set of line-breaking rules.</dd>
      <dt><dfn>strict</dfn></dt>
      <dd>Breaks text using the most stringent set of line-breaking
        rules.</dd>
    </dl>

    <p class="feedback issue">
      The rules here are following guidelines from <a href="http://www.w3.org/TR/klreq/#line-break">KLREQ</a> for Korean,
      which don't allow the Chinese/Japanese-specific breaks.
      However, the resulting behavior could use some review and feedback to make sure they are correct,
      particularly when “word basis” breaking is used (''word-break: keep-all'') in Korean.

    <p>CSS distinguishes between three levels of strictness in the rules for
      text wrapping.
      The precise set of rules in effect for each level is up to the UA
      and should follow language conventions.
      However, this specification does require that:</p>

    <ul>
      <li>
        The following breaks are forbidden in ''strict'' line breaking
        and allowed in ''line-break/normal'' and ''loose'':
        <ul>
          <li>breaks before Japanese small kana or the Katakana-Hiragana prolonged sound mark:
            i.e. characters with the Unicode Line Break property <code>CJ</code>.
            (See <a href="http://www.unicode.org/Public/UNIDATA/LineBreak.txt">LineBreak.txt</a> in [[!UNICODE]].)
        </ul>
      <li>
        The following breaks are allowed for ''line-break/normal'' and ''loose'' line breaking
        if the <a>content language</a> is Chinese or Japanese,
        and are otherwise forbidden:
        <ul>
          <li>breaks before hyphens:<br>
            &#x2010;&nbsp;U+2010,  &#x2013;&nbsp;U+2013,  &#x301C;&nbsp;U+301C,
            &#x30A0;&nbsp;U+30A0
        </ul>
      <li>
        The following breaks are forbidden for ''line-break/normal'' and ''strict'' line breaking
        and allowed in ''loose'':
        <ul>
          <li>breaks before iteration marks:<br>
            &#x3005;&nbsp;U+3005,  &#x303B;&nbsp;U+303B,  &#x309D;&nbsp;U+309D,
            &#x309E;&nbsp;U+309E,  &#x30FD;&nbsp;U+30FD,  &#x30FE;&nbsp;U+30FE
          <li>breaks between inseparable characters
            such as &#x2025;&nbsp;U+2025,  &#x2026;&nbsp;U+2026
            i.e. characters with the Unicode Line Break property <code>IN</code>.
            (See <a href="http://www.unicode.org/Public/UNIDATA/LineBreak.txt">LineBreak.txt</a> in [[!UNICODE]].)
        </ul>
      <li>
        The following breaks are allowed for ''loose''
        if the <a>content language</a> is Chinese or Japanese
        and are otherwise forbidden:
        <ul>
          <li>breaks before certain centered punctuation marks:<br>
            &#x30FB;&nbsp;U+30FB,
            &#xFF1A;&nbsp;U+FF1A,  &#xFF1B;&nbsp;U+FF1B,  &#xFF65;&nbsp;U+FF65,
            &#x203C;&nbsp;U+203C,
            &#x2047;&nbsp;U+2047,  &#x2048;&nbsp;U+2048,  &#x2049;&nbsp;U+2049,
            &#xFF01;&nbsp;U+FF01,  &#xFF1F;&nbsp;U+FF1F
          <li>breaks before suffixes:<br>
            Characters with the Unicode Line Break property <code>PO</code>
            and the <a>East Asian Width property</a> [[!UAX11]] <code>A</code>, <code>F</code>, or <code>W</code>.
          <li>breaks after prefixes:<br>
            Characters with the Unicode Line Break property <code>PR</code>
            and the <a>East Asian Width property</a> [[!UAX11]] <code>A</code>, <code>F</code>, or <code>W</code>.
        </ul>
    </ul>

    <p class="note">In the recommended list above, no distinction is made among the levels of
      strictness in non-CJK text: only CJK codepoints are affected, unless
      the text is marked as Chinese or Japanese, in which case some additional
      common codepoints are affected. However a future level of CSS may add
      behaviors affecting non-CJK text.

    <p class="note">The CSSWG recognizes that in a future edition of the
      specification finer control over line breaking may be necessary to
      satisfy high-end publishing requirements.

<h2 id="hyphenation">Breaking Within Words</h2>

    <p><dfn id=hyphenate lt="hyphenation|hyphenate">Hyphenation</dfn>
      allows the controlled splitting of words
      to improve the layout of paragraphs,
      typically splitting words at syllabic or morphemic boundaries
      and visually indicating the split (usually by inserting a hyphen, U+2010).
      In some cases, hyphenation may also alter the spelling of a word.
      Regardless, hyphenation is a rendering effect only:
      it must have no effect on the underlying document content
      or on text selection or searching.

    <p>Hyphenation occurs when the line breaks at a valid <dfn>hyphenation opportunity</dfn>,
      which creates a <a>soft wrap opportunity</a> within the word.
      In CSS it is controlled with the 'hyphens' property.
      CSS Text Level 3 does not define the exact rules for hyphenation,
      however UAs are strongly encouraged to optimize their line-breaking implementation
      to choose good break points and appropriate hyphenation points.
      Hyphenation opportunities <em>are</em> considered when calculating
      <a lt="min-content size">min-content intrinsic sizes</a>.
    <!-- This is because it allows tables to hyphenate instead of overflowing,
         which is particularly important in long-word languages like German.
         https://bugzilla.mozilla.org/show_bug.cgi?id=418975 -->

    <p>CSS also provides the 'overflow-wrap' property, which can allow
      arbitrary breaking within words when the text would otherwise overflow
      its container.

  <h3 id="hyphens-property">Hyphenation Control: the 'hyphens' property</h3>

    <table class=propdef>
      <tr>
        <td>Name:
        <td><dfn>hyphens</dfn>
      <tr>
        <th><a href="#values">Value:</a>
        <td>none | manual | auto
      <tr>
        <th>Initial:
        <td>manual
      <tr>
        <th>Applies to:
        <td>all elements
      <tr>
        <th>Inherited:
        <td>yes
      <tr>
        <th>Percentages:
        <td>N/A
      <tr>
        <th>Media:
        <td>visual
      <tr>
        <th>Computed&nbsp;value:
        <td>specified value
      <tr>
        <th>Animatable:
        <td>no
      <tr>
        <th>Canonical order:
        <td>N/A
    </table>

    <p>This property controls whether hyphenation is allowed to create more
      <a>soft wrap opportunities</a> within a line of text.
      Values have the following meanings:

    <dl dfn-for=hyphens dfn-type=value>
      <dt><dfn>none</dfn>
        <dd>Words are not hyphenated, even if characters inside
          the word explicitly define <a>hyphenation opportunities</a>.
      <dt><dfn>manual</dfn>
        <dd>Words are only hyphenated where there are characters inside the word
          that explicitly suggest <a>hyphenation opportunities</a>.
        <div class="example">
        <p>In Unicode, U+00AD is a conditional "soft hyphen" and U+2010 is an
          unconditional hyphen. Unicode Standard Annex #14 describes the
          <a href="http://unicode.org/reports/tr14/#SoftHyphen">role of soft hyphens in</a>
          Unicode line breaking. [[!UAX14]]
          In HTML, &amp;shy; represents the soft hyphen character,
          which suggests a hyphenation opportunity.
          <pre>ex&amp;shy;ample</pre>
        </div>
      <dt><dfn>auto</dfn>
        <dd>Words may be broken at <a>hyphenation opportunities</a>
          determined automatically by a language-appropriate hyphenation resource
          in addition to those indicated explicitly by a conditional hyphen.
          Automatic <a>hyphenation opportunities</a> within a word must be ignored
          if the word contains a conditional hyphen (&amp;shy; or U+00AD),
          in favor of the conditional hyphen(s).
          However, if, even after breaking at such opportunities,
          a portion of that word is is still too long to fit on one line,
          an automatic hyphenation opportunity may be used.
    </dl>

    <p>Correct automatic hyphenation requires a hyphenation resource
      appropriate to the language of the text being broken.
      The UA is therefore only required to automatically hyphenate text
      for which the <a>content language</a> is known
      and for which it has an appropriate hyphenation resource.

    <p class="advisement">
      Authors should correctly tag their content’s <a lt="content language">language</a>
      (e.g. using the HTML ''lang'' attribute)
      in order to obtain correct automatic hyphenation.
      UAs may refuse to automatically hyphenate untagged content
      regardless of the 'hyphens' property value.

    <p>When shaping scripts such as Arabic are allowed to break within words
      due to hyphenation, the characters must still be shaped as if the word
      were not broken.</p>

    <div class="example">
      <p>For example, if the Uyghur word &ldquo;داميدى&rdquo;
      were hyphenated, it would appear as
      <img src="images/uyghur-hyphenate-joined.png"
           alt="[isolated DAL + isolated ALEF + initial MEEM + medial YEH + hyphen + line-break + final DAL + isolated ALEF MAKSURA]">
      not as
      <img src="images/uyghur-hyphenate-unjoined.png"
           alt="[isolated DAL + isolated ALEF + initial MEEM + final YEH + hyphen + line-break + isolated DAL + isolated ALEF MAKSURA]">
    </div>

  <h3 id="overflow-wrap-property">
    Overflow Wrapping: the 'overflow-wrap'/'word-wrap' property</h3>

<pre class='propdef'>
Name: overflow-wrap, word-wrap
Value: normal | break-word
Initial: normal
Applies to: all elements
Inherited: yes
Percentages: N/A
Media: visual
Computed value: specified value
Animatable: no
Canonical order: N/A
</pre>

    <p>This property specifies whether the UA may arbitrarily break within a word
      to prevent overflow when an otherwise-unbreakable string is too
      long to fit within the line box. It only has an effect when
      'white-space' allows <a>wrapping</a>. Possible values:</p>

    <dl dfn-for=overflow-wrap dfn-type=value>
    <dt><dfn>normal</dfn></dt>
    <dd>Lines may break only at allowed break points. However, the restrictions
      introduced by ''word-break: keep-all'' may be relaxed to match ''word-break: normal''
      if there are no otherwise-acceptable break points in the line.</dd>
    <dt><dfn>break-word</dfn></dt>
    <dd>An otherwise unbreakable sequence of <a>characters</a>
      may be broken at an arbitrary point if
      there are no otherwise-acceptable break points in the line.
      Shaping characters are still shaped as if the word were not
      broken, and grapheme clusters must stay together as one unit.
      No hyphenation character is inserted at the break point.
    </dl>

    <p><a>Soft wrap opportunities</a> introduced by ''overflow-wrap: break-word''
       are not considered when calculating <a lt="min-content size">min-content intrinsic sizes</a>.

    <p>For legacy reasons, UAs must treat 'word-wrap' as an alternate name
      for the 'overflow-wrap' property, as if it were a <a>shorthand</a> of 'overflow-wrap'.

<h2 id="justification">
  Alignment and Justification</h2>

  <p>Alignment and justification controls how inline content is distributed within a line box.

  <h3 id="text-align-property">
    Text Alignment: the 'text-align' shorthand</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>text-align</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>start | end | left | right | center |
              justify | match-parent | justify-all
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>start</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>block containers</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>specified value, except for ''match-parent'' which computes as defined below</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This <a>shorthand property</a>
      sets the 'text-align-all' and 'text-align-last' properties
      and describes how the inline-level content of a block
      is aligned along the inline axis
      if the content does not completely fill the line box.
      Values have the following meanings:</p>

    <dl dfn-for=text-align dfn-type=value>
    <dt><dfn>start</dfn></dt>
      <dd>Inline-level content is aligned to the <a spec=css-writing-modes-3>start</a> edge of the line box.
    <dt><dfn>end</dfn></dt>
      <dd>Inline-level content is aligned to the <a spec=css-writing-modes-3>end</a> edge of the line box.
    <dt><dfn>left</dfn></dt>
      <dd>Inline-level content is aligned to the
        <a href="http://www.w3.org/TR/css3-writing-modes/#line-left">line left</a>
        edge of the line box.
        (In vertical writing modes,
         this will be either the physical top or bottom,
         depending on 'text-orientation'.) [[CSS3-WRITING-MODES]]
    <dt><dfn>right</dfn></dt>
      <dd>Inline-level content is aligned to the
        <a href="http://www.w3.org/TR/css3-writing-modes/#line-right">line right</a>
        edge of the line box.
        (In vertical writing modes,
         this will be either the physical top or bottom,
         depending on 'text-orientation'.) [[CSS3-WRITING-MODES]]
    <dt><dfn>center</dfn></dt>
      <dd>Inline-level content is centered within the line box.
    <dt><dfn>justify</dfn></dt>
      <dd>Text is justified according to the method specified by the 'text-justify' property,
        in order to exactly fill the line box.
        Unless otherwise specified by 'text-align-last',
        the last line before a forced break or the end of the block is ''start''-aligned.
    <dt><dfn>justify-all</dfn></dt>
      <dd>Same as ''justify'', except 'text-align-last' is also set to ''justify'',
        forcing the last line to justify as well.
    <dt><dfn>match-parent</dfn></dt>
      <dd>This value behaves the same as ''inherit''
        (computes to its parent's computed value)
        except that an inherited ''start'' or ''end'' keyword
        is interpreted against its parent's 'direction' value
        and results in a computed value of either 'left' or 'right'.
    </dl>

    <p>All values except ''justify-all'' reset 'text-align-last' to ''text-align-last/auto''.

    <p>A block of text is a stack of
      <a href="http://www.w3.org/TR/CSS21/visuren.html#line-box">line boxes</a>.
      This property specifies how the inline-level boxes within each line box
      align with respect to the start and end sides of the line box.
      Alignment is not with respect to the
      <a href="http://www.w3.org/TR/CSS21/visuren.html#viewport">viewport</a>
      or containing block.

    <p>In the case of ''justify'', the UA may stretch or shrink any inline boxes
      by <a href="#text-justify-property">adjusting</a> their text. (See 'text-justify'.)
      If an element's white space is not <a href="#collapse">collapsible</a>,
      then the UA is not required to adjust its text for the purpose of justification
      and may instead treat the text as having no <a>justification opportunities</a>.
      If the UA chooses to adjust the text, then it must ensure
      that <a>tab stops</a> continue to line up as required by the
      <a href="#white-space-rules">white space processing rules</a>.

    <p>If (after justification, if any) the inline contents of a line box are too long to fit within it,
      then the contents are <a spec=css-writing-modes-3>start</a>-aligned:
      any content that doesn't fit overflows the line box's <a spec=css-writing-modes-3>end</a> edge.

    <p>See <a href="#bidi-linebox">Bidirectionality and line boxes</a>
      for details on how to determine the <a spec=css-writing-modes-3>start</a> and <a spec=css-writing-modes-3>end</a> edges of a line box.

  <h3 id="text-align-all-property">
    Default Text Alignment: the 'text-align-all' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:
          <td><dfn>text-align-all</dfn>
        <tr>
          <th><a href="#values">Value</a>:
          <td>start | end | left | right | center |
              justify | match-parent
        <tr>
          <th>Initial:
          <td>start
        <tr>
          <th>Applies to:
          <td>block containers
        <tr>
          <th>Inherited:
          <td>yes
        <tr>
          <th>Percentages:
          <td>N/A
        <tr>
          <th>Media:
          <td>visual
        <tr>
          <th>Computed&#160;value:
          <td>specified value
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This longhand of the 'text-align' <a>shorthand property</a>
    specifies the inline alignment of all lines of inline content in the block container,
    except for last lines overridden by a non-''text-align-last/auto'' value of 'text-align-last'.
    See 'text-align' for a full description of values.

    <p class="advisement">
    Authors should use the 'text-align' shorthand instead of this property.

  <h3 id="text-align-last-property">
    Last Line Alignment: the 'text-align-last' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>text-align-last</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>auto | start | end | left | right | center | justify
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>auto</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>block containers</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>specified value</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property describes how the last line of a block or a line
      right before a <a>forced line break</a> is aligned.
      If a line is also the first line of the block
      or the first line after a <a>forced line break</a>,
      'text-align-last' takes precedence over 'text-align-all'.

    <p>If <dfn dfn-for=text-align-last dfn-type=value>auto</dfn> is specified,
      content on the affected line is aligned per 'text-align'
      unless 'text-align' is set to ''justify'',
      in which case it is start-aligned.
      All other values have the same meanings as in 'text-align'.

  <h3 id="text-justify-property">
    Justification Method: the 'text-justify' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>text-justify</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>auto | none | inter-word | inter-character
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>auto</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>block containers and, optionally, inline elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>specified value</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

    <p>This property selects the justification method used when a line's
      alignment is set to ''justify'' (see 'text-align').
      The property applies to block containers, but the UA may (but
      is not required to) also support it on inline elements.
      It takes the following values:</p>

    <dl dfn-for=text-justify dfn-type=value>
      <dt><dfn>auto</dfn></dt>
        <dd>The UA determines the justification algorithm to follow, based
          on a balance between performance and adequate presentation quality.
          Since justification rules vary by writing system and language,
          UAs should, where possible, use a justification algorithm appropriate to the text.

          <p class="example">
          For example, the UA could use by default a justification method that is a
          simple universal compromise for all writing systems&mdash;such as
          primarily expanding <a>word separators</a>
          and between CJK <a>typographic letter units</a>
          along with secondarily expanding between Southeast Asian <a>typographic letter units</a>.
          Then, in cases where the <a>content language</a> of the paragraph is known,
          it could choose a more language-tailored justification behavior
          e.g. following [[JLREQ]] for Japanese,
          using cursive elongation for Arabic,
          using ''inter-word'' for German,
          etc.

        <div class="figure" id="fig-text-justify-cursive">
          <p>
            <img alt="Two lines of calligraphic Arabic end together due to a mix of compressed and swash forms."
                 title="Swash forms elongate the first line while a compressed contextual ligature shortens the second, allowing both to end precisely together."
                 src="images/text-justify-cursive.png"></p>
          <p class="caption">An example of cursively-justified Arabic text,
          rendered by <a href="http://www.decotype.com/">Tasmeem</a>.
          Like English, Arabic can be justified by adjusting the spacing between words,
          but in most styles it can also be justified by calligraphically elongating or compressing the letterforms themselves.
          In this example, the upper text is extended to fill the line by the use of elongated (kashida) forms and swash forms,
          while the bottom line is compressed slightly by using a stacked combination for the characters between ت and م.
          By employing traditional calligraphic techniques,
          a typesetter can justify the line while preserving flow and color,
          providing a very high quality justification effect.
          However, this is by its nature a very script-specific effect.
        </div>

        <div class="figure" id="fig-text-justify-compromise">
          <p>
            <img alt="Extra space is partly to spaces and partly among CJK and Thai letters."
                 src="images/text-justify-compromise.png"></p>
          <p class="caption">Mixed-script text with ''text-justify: auto'':
            this interpretation uses a universal-compromise justification method,
            expanding at spaces as well as between CJK and Southeast Asian letters.
            This effectively uses inter-word + inter-ideograph spacing
            for lines that have word-separators and/or CJK characters
            and falls back to inter-cluster behavior for lines that don't
            or for which the space stretches too far.
        </div>

      <dt><dfn>none</dfn></dt>
        <dd>Justification is disabled: there are no <a>justification opportunities</a> within the text.
          <div class="figure" id="fig-text-justify-none">
            <p>
              <img alt="No extra space is insesrted."
                   src="images/text-justify-none.png"></p>
            <p class="caption">Mixed-script text with ''text-justify: none''
          </div>
          <p class="note">This value is intended for use in user stylesheets
          to improve readability or for accessibility purposes.

      <dt><dfn>inter-word</dfn></dt>
        <dd>Justification adjusts spacing at <a>word separators</a> only
          (effectively varying the used 'word-spacing' on the line).
          This behavior is typical for languages that separate words using spaces,
          like English or Korean.
          <div class="figure" id="fig-text-justify-interword">
            <p>
              <img alt="Extra space is equally distributed mainly to spaces."
                   src="images/text-justify-interword.png"></p>
            <p class="caption">Mixed-script text with ''text-justify: inter-word''
          </div>
      <dt><dfn>inter-character</dfn></dt>
        <dd>Justification adjusts spacing between each pair of adjacent <a>typographic character units</a>
          (effectively varying the used 'letter-spacing' on the line).
          This value is sometimes used in East Asian systems such as Japanese.
          <div class="figure" id="fig-text-justify-distribute">
            <p>
              <img alt="Extra space is equally distributed at points between spaces and letters of all writing systems."
                   src="images/text-justify-distribute.png"></p>
            <p class="caption">Mixed-script text with ''text-justify: inter-character''
          </div>
          For legacy reasons, UAs must also support the alternate keyword <dfn>distribute</dfn>
          with the exact same meaning and behavior.
    </dl>

    <p class="advisement">
      Since optimal justification is language-sensitive,
      authors should correctly language-tag their content for the best results.

    <p>The exact justification algorithm is UA-dependent;
      however, CSS provides some general guidelines below.

<h4 id="expanding-text">
Expanding and Compressing Text</h4>

  <p>When justifying text, the user agent takes the remaining space between
    the ends of a line's contents and the edges of its line box, and
    distributes that space throughout its contents so that the contents
    exactly fill the line box.
    The user agent may alternatively distribute negative space,
    putting more content on the line than would otherwise fit under normal spacing conditions.

  <p>A <span id="expansion-opportunity"><dfn export>justification opportunity</dfn></span> is
    a point where the justification algorithm may alter spacing within the text.
    A justification opportunity can be provided by a single <a>typographic character unit</a>
    (such as a <a>word separator</a>),
    or by the juxtaposition of two <a>typographic character units</a>.
    Space distributed by justification is <em>in addition to</em>
    the spacing defined by the 'letter-spacing' or 'word-spacing' properties.
    However, when space is distributed to a <a>word separator</a> <a>justification opportunity</a>,
    it is applied under the same rules as for 'word-spacing'.
    Similarly, when space is distributed to an <a>justification opportunity</a> between
    two <a>typographic character units</a>,
    it is applied under the same rules as for 'letter-spacing'.

  <p>A justification algorithm may divide <a>justification opportunities</a> into different priority levels.
    All <a>justification opportunities</a> within a given level
    are expanded or compressed at the same priority,
    regardless of which <a>typographic character units</a> created that opportunity.
    For example, if <a>justification opportunities</a> between two Han characters
    and between two Latin letters are defined to be at the same level
    (as they are in the ''inter-character'' justification style),
    they are not treated differently because they originate from different <a>typographic character units</a>.
    It is not defined in this level
    whether or how other factors
    (such as font size, letter-spacing, glyph shape, position within the line, etc.)
    may influence the distribution of space to <a>justification opportunities</a> within the line.

  <p>The UA may enable or break optional ligatures or use other font features
  such as alternate glyphs or glyph compression
  to help justify the text under any method.
  This behavior is not controlled by this level of CSS.
  However, UAs <em>must not</em> break required ligatures
  or otherwise disable features required to correctly shape complex scripts.

<h4 id="justify-symbols">
Handling Symbols and Punctuation</h4>

    <p>When determining <a>justification opportunities</a>,
      a <a>typographic character unit</a> from the Unicode Symbols (S*) and Punctuation (P*) classes
      is generally treated the same as a <a>typographic letter unit</a> of the same script
      (or, if the character's script property is Common,
      then as a <a>typographic letter unit</a> of the dominant script).

    <p>However, by typographic tradition there may be additional rules
      controlling the justification of symbols and punctuation.
      Therefore, the UA may reassign specific characters
      or introduce additional levels of prioritization
      to handle <a>justification opportunities</a> involving symbols and punctuation.

    <p class="example">
      For example, there are traditionally no <a>justification opportunities</a>
      between consecutive
      U+2014 Em Dash ‘—’,
      U+2015 Horizontal Bar ‘―’,
      U+2026 Horizontal Ellipsis ‘…’,
      or U+2025 Two Dot Leader ‘‥’
      characters [[JLREQ]];
      thus a UA might assign these characters to a “never” prioritization level.
      As another example, certain fullwidth punctuation characters
      (such as U+301A Left White Square Bracket ‘〚’)
      are considered to contain a <a>justification opportunity</a> in Japanese.
      The UA might therefore assign these characters to a higher prioritization
      level than the opportunities between ideographic characters.

<h4 id="justify-limits">
Unexpandable Text</h4>

    <p>If the inline contents of a line cannot be stretched to the full width of the line box,
      then they must be aligned as specified by the 'text-align-last' property.
      (If 'text-align-last' is ''justify'', then
       they must be aligned as for ''center'' if 'text-justify' is ''inter-character'',
       and as ''start'' otherwise.)

<h4 id="justify-cursive">
Cursive Scripts</h4>

    <p>Justification <em>must not</em> introduce gaps between the joined <a>typographic letter units</a>
      of <a>cursive scripts</a> such as Arabic.
      If it is able, the UA <em>may</em>
      translate space distributed to <a>justification opportunities</a> within a run of such <a>typographic letter units</a>
      into some form of cursive elongation for that run.
      It otherwise <em>must</em> assume that no <a>justification opportunity</a> exists
      between any pair of <a>typographic letter units</a> in <a>cursive script</a>
      (regardless of whether they join).

    <div class="example">
      <p>The following are examples of unacceptable justification:
      <div class="figure">
        <p><img alt="" src="images/arabic-stretch-spaced.png">
        <p class="caption">Adding gaps between every pair of Arabic letters
      </div>
      <div class="figure">
        <p><img alt="" src="images/arabic-stretch-unjoined.png">
        <p class="caption">Adding gaps between every pair of unjoined Arabic letters
      </div>
    </div>

    <p>Some font designs allow for the use of the tatweel character for justification.
    A UA that performs tatweel-based justification must properly handle the rules for its use.
    Note that correct insertion of tatweel characters depends on context, including
    the letter-combinations involved, location within the word, and location of the word within the line.

<h4 id="justify-algos">
Possible Algorithms</h4>

  <p>For <a value for=text-justify>auto</a> justification, this specification does not define
  what all of the <a>justification opportunities</a> are,
  how they are prioritized, or
  when and how multiple levels of <a>justification opportunities</a> interact.
  However, it does require that

  <ul>
    <li>
      If a <a>justification opportunity</a> exists within a line,
      it must be justified.
    <li>
      Unless contraindicated by the typographic traditions of the <a>content language</a> or adjacent symbols/punctuation,
      each of the following provides a <a>justification opportunity</a>:
      <ul>
        <li><a>Word separators</a>
        <li>The boundary between a <a>typographic character unit</a> of any <a>block scripts</a> and any other <a>typographic character unit</a>
        <li>The boundary between a <a>typographic character unit</a> of any <a>clustered scripts</a> and any other <a>typographic character unit</a>
      </ul>
    <li>
      All <a>letters</a> belonging to all <a>block scripts</a> are treated the same,
      and all <a>letters</a> belonging to all <a>clustered scripts</a> are treated the same.
      For example, no distiction is made between
      the justification opportunity between a Han letter followed by another Han letter,
      vs. the justification opportunity between a Han letter followed by a Hangul letter.
  </ul>

  <p>The guidelines in this level of CSS do not describe a complete
  justification algorithm. They are merely a minimum set of requirements
  that a complete algorithm should meet. Limiting the set of requirements
  gives UAs some latitude in choosing a justification algorithm that
  meets their needs and desired balance of quality, speed, and complexity.


  <div class="example">
    <p>For instance, a basic but fast justification algorithm
    might use a simple greedy method for determining line breaks,
    then distribute leftover space.
    This algorithm could follow the guidelines by expanding word spaces first,
    expanding between letters only if the spaces between words hit a limit defined by the UA.

    <p>A more sophisticated but harder-to-implement justification algorithm
    might use a Knuth/Plass method
    where <a>justification opportunities</a> and limits were assigned weights
    and assessed with other line breaking considerations.
    This algorithm could follow the guidelines by giving more weight
    to <a>word separators</a> than letter spacing.
  </div>

  <div class="example">
    <p>A UA can also tailor its justification rules by language,
    to produce results more closely aligned to the typography of that language.
    For example, it's not defined whether expansion or compression is preferred,
    so a UA may, for example, bias towards compression for CJK languages
    but towards expansion for Western alphabetic languages.

    <p>As another example,
    3.8 Line Adjustment in [[JLREQ]] gives an example of a set of rules for
    how a text formatter can justify Japanese text.
    A UA could use this algorithm when the 'text-justify' property is ''text-justify:auto''.
    However, since the rules described in the document specifically target Japanese,
    they may produce non-optimal results
    when used to justify other languages such as English.
    The UA could adapt the rules to accommodate other scripts by, for instance,
    omitting the rule to compress half-width spaces (rule a. of 3.8.3).
    Or it could keep the rule, but only enable it when the content language is known to be Japanese.
  </div>

<h2 id="spacing">
  Spacing</h2>

  <p>CSS offers control over text spacing
  via the 'word-spacing' and 'letter-spacing' properties, which specify additional space
  around <a>word separators</a> or between <a>typographic character units</a>, respectively.
  The 'word-spacing' property can now be specified in percentages,
  making it possible to, for example, double or eliminate word spacing.

  <div class="example">
    <p>In the following example, word spacing is halved,
    but may expand if needed for text justification.
    <pre>p { word-spacing: -50%; }</pre>
  </div>

<h3 id="word-spacing-property">
Word Spacing: the 'word-spacing' property</h3>

  <table class="propdef">
    <tr>
      <th>Name:
      <td><dfn>word-spacing</dfn>
    <tr>
      <th><a href="#values">Value</a>:
      <td>[ normal | <var>&lt;length&gt;</var> | <var>&lt;percentage&gt;</var>]
    <tr>
      <th>Initial:
      <td>normal
    <tr>
      <th>Applies to:
      <td>all elements
    <tr>
      <th>Inherited:
      <td>yes
    <tr>
      <th>Percentages:
      <td>refers to width of the affected glyph
    <tr>
      <th>Media:
      <td>visual
    <tr>
      <th>Computed&#160;value:
      <td>an absolute length
    <tr>
      <th>Animatable:
      <td>as <a href="http://www.w3.org/TR/css3-transitions/#animtype-lpcalc">length, percentage, or calc</a>
    <tr>
      <th>Canonical order:
      <td>N/A
  </table>

    <p>This property specifies additional spacing
      between &ldquo;words&rdquo;.
      Missing values are assumed to be ''word-spacing:normal''.
      Values are interpreted as defined below:

  <dl dfn-for=word-spacing dfn-type=value>
    <dt><dfn>normal</dfn>
      <dd>No additional spacing is applied.
        Computes to zero.
    <dt><dfn>&lt;length&gt;</dfn>
      <dd>Specifies extra spacing <em>in addition to</em>
        the intrinsic inter-word spacing defined by the font.
    <dt><dfn>&lt;percentage&gt;</dfn>
      <dd>Specifies the additional spacing as a percentage of the affected
        character's advance width.
  </dl>

  <p>Additional spacing is applied to each <a>word separator</a>
    left in the text after the <a href="#white-space-rules">white space processing rules</a> have been applied,
    and should be applied half on each side of the character
    unless otherwise dictated by typographic tradition.
    Values may be negative, but there may be implementation-dependent limits.

  <div class="example">
    <p>The following example will make all the spaces between words in Arabic
    be rendered as zero-width, and double the width of each space in English:

    <pre>:lang(ar) { word-spacing: -100%; }
<!--  -->:lang(en) { word-spacing: 100%; }</pre>

    <p>The following example will <em>add</em> half the the width of the
      &ldquo;0&rdquo; glyph to word spacing character [[CSS3VAL]]:
    <pre>p { word-spacing: 0.5ch; }</pre>
  </div>

    <p><dfn id="word-separator" lt="word-separator character | word separator">Word-separator characters</dfn>
      are <a>typographic character units</a> whose purpose and general usage is to separate words.
      In [[UNICODE]] this includes
      the space (U+0020), the no-break space (U+00A0), the Ethiopic word space (U+1361),
      the Aegean word separators (U+10100,U+10101), the Ugaritic word divider (U+1039F),
      and the Phoenician Word Separator (U+1091F).

      If there are no word-separator characters, or if a word-separating
      character has a zero advance width (such as the zero width space U+200B)
      then the user agent must not create an additional spacing between words.

      General punctuation and fixed-width spaces (such as U+3000 and U+2000
      through U+200A) are not considered word-separator characters.</p>

  <h3 id="letter-spacing-property">
    Tracking: the 'letter-spacing' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>letter-spacing</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>normal | <var>&lt;length&gt;</var>
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>normal</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>all elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>an absolute length
        </tr>
        <tr>
          <th>Animatable:
          <td>as <a href="http://www.w3.org/TR/css3-transitions/#animtype-length">length</a>
        <tr>
          <th>Canonical order:
          <td>N/A
    </table>

  <p>This property specifies additional spacing (commonly called <dfn>tracking</dfn>)
  between adjacent <a>typographic character units</a>.
  Letter-spacing is applied after
  <a href="http://www.w3.org/TR/css3-writing-modes/#text-direction">bidi reordering</a>
  and is in addition to any 'word-spacing'.
  Depending on the justification rules in effect,
  user agents may further increase or decrease the space between <a>typographic character units</a>
  in order to <a href="#text-justify-property">justify text</a>.

  <p>Values have the following meanings:

  <dl dfn-for=letter-spacing dfn-type=value>
    <dt><dfn>normal</dfn>
      <dd>No additional spacing is applied. Computes to zero.
    <dt><dfn>&lt;length&gt;</dfn>
      <dd>Specifies <em>additional</em> spacing between <a>typographic character units</a>.
        Values may be negative, but there may be implementation-dependent limits.
  </dl>

    <p>For the purpose of 'letter-spacing', each consecutive run of atomic
      inlines (such as images and inline blocks) is treated as a single
      <a>typographic character unit</a>.

    <p>Letter-spacing must not be applied at the beginning or at the end of a line.

  <div class="example">
    <p>Because letter-spacing is not applied at the beginning or end of a line,
    text always fits flush with the edge of the block.
    <pre>
<!-- -->p    { letter-spacing: 1em; }

<!-- -->&lt;p>abc&lt;/p></pre>

    <p class="ls-ex good ls-fixed-width color-box" style="text-align: left">a&emsp;b&emsp;c</p>
    <p class="ls-ex good ls-fixed-width color-box" style="text-align: right">a&emsp;b&emsp;c</p>

    <p>UAs therefore must not append letter spacing to the right or trailing edge of a line:</p>

    <p class="ls-ex bad ls-fixed-width color-box" style="text-align: right">a&emsp;b&emsp;c&emsp;</p>
  </div>

    <p>Letter spacing between two <a>typographic character units</a> effectively “belongs”
      to the innermost element that contains the two <a>typographic character units</a>:
      the total letter spacing between two adjacent <a>typographic character units</a> (after bidi reordering)
      is specified by and rendered within
      the innermost element that <em>contains</em> the boundary between the two <a>typographic character units</a>.

  <div class="example">
    <p>A given value of 'letter-spacing' only affects the spacing
    between characters completely contained within the element for which it is specified:

    <pre>
<!-- -->p    { letter-spacing: 1em; }
<!-- -->span { letter-spacing: 2em; }

<!-- -->&lt;p>a&lt;span>bb&lt;/span>c&lt;/p></pre>

    <p class="ls-ex">a&emsp;<span class="color-box">b&emsp;&emsp;b</span>&emsp;c</p>

    <p>This also means that applying 'letter-spacing' to
    an element containing only a single character
    has no effect on the rendered result:

    <pre>
<!-- -->p    { letter-spacing: 1em; }
<!-- -->span { letter-spacing: 2em; }

<!-- -->&lt;p>a&lt;span>b&lt;/span>c&lt;/p></pre>

    <p class="ls-ex">a&emsp;<span class="color-box">b</span>&emsp;c</p>

    <p>An inline box only includes
    letter spacing between characters completely contained within that element:

    <pre>
<!-- -->p    { letter-spacing: 1em; }

<!-- -->&lt;p>a&lt;span>bb&lt;/span>c&lt;/p></pre>

    <p class="ls-ex good">a&emsp;<span class="color-box">b&emsp;b</span>&emsp;c</p>

    <p>It is incorrect to include the letter spacing on the right or trailing edge of the element:

    <p class="ls-ex bad">a&emsp;<span class="color-box">b&emsp;b&emsp;</span>c</p>

    <p>Letter spacing is inserted <strong>after</strong> RTL reordering,
    so the letter spacing applied to the inner span below has no effect,
    since after reordering the "c" doesn't end up next to "&#x5d0;":
    <pre>
<!-- -->p    { letter-spacing: 1em; }
<!-- -->span { letter-spacing: 2em; }

<!-- -->&lt;!-- abc followed by Hebrew letters alef (&#x5d0;), bet (&#x5d1;) and gimel (&#x5d2;) -->
<!-- -->&lt;!-- Reordering will display these in reverse order. --&gt;
<!-- --><bdo dir=ltr>&lt;p>ab&lt;span>c&#x5d0;&lt;/span>&#x5d1;&#x5d2;&lt;/p></bdo></pre>

    <p class="ls-ex">a&emsp;b</span>&emsp;<span class="color-box">c</span>&emsp;<span class="color-box">&#x5d0;</span>&emsp;&#x5d1;&emsp;&#x5d2;</p>
  </div>

    <p>Letter spacing ignores invisible zero-width formatting characters
      (such as those from the Unicode Cf category).
      Spacing must be added as if those characters did not exist in the document.

    <p class="example">For example, 'letter-spacing' applied to
      <code>A&amp;#x200B;B</code> is identical to <code>AB</code>,
      regardless of where any element boundaries might fall.

    <p>When the effective spacing between two characters is not zero
      (due to either <a href="#text-justify-property">justification</a>
      or a non-zero value of 'letter-spacing'),
      user agents should not apply optional ligatures.</p>

    <p class="example">
      For example, if the word “filial” is letter-spaced,
      an “fi” ligature should not be used
      as it will prevent even spacing of the text.

    <p>If it is able, the UA <em>may</em> apply letter spacing to <a>cursive scripts</a>
    by translating the total extra space to be distributed to a run of such letters
    into some form of cursive elongation for that run.
    Otherwise, if the UA cannot expand text from a <a>cursive script</a>
    without breaking its cursive connections,
    it <em>must not</em> apply spacing
    between any pair of that script's <a>typographic letter units</a> at all
    (effectively treating each word as a single <a>typographic letter unit</a>
    for the purpose of letter-spacing).
    Both cases will result in an effective spacing of zero between such letters;
    however the former will preserve the sense of stretching out the text.

    <div class="example">
      <style scoped>
        .no { color: red; }
        .ok { color: green; }
        th em { display: block; font-size: smaller; font-weight: normal; font-style: italic; }
      </style>
      <p>Below are some appropriate and inappropriate examples of spacing out Arabic text.
      <table class="data">
      <thead>
        <tr><td><img src="images/arabic-stretch-original.png" alt="">
            <td>&mdash;
            <td>Original text
      <tbody>
        <tr><td><img src="images/arabic-stretch-spaced.png" alt="">
            <td class="no">BAD
            <th>Even distribution of space between each letter.
                <em>Notice this breaks cursive joins!</em>
        <tr><td><img src="images/arabic-stretch-kashida.png" alt="">
            <td class="ok">OK
            <th>Distributing &sum;<var>letter-spacing</var> by typographically-appropriate cursive elongation.
        <tr><td><img src="images/arabic-stretch-suppressed.png" alt="">
            <td class="ok">OK
            <th>Suppressing 'letter-spacing' between Arabic letters.
                <em>Notice 'letter-spacing' is nonetheless applied to non-Arabic characters (like spaces).</em>
        <tr><td><img src="images/arabic-stretch-unjoined.png" alt="">
            <td class="no">BAD
            <th>Applying 'letter-spacing' only between non-joined letters.
                <em>This distorts typographic color and obfuscates word boundaries.</em>
      </table>
    </div>
    <div class="note">
      Note that proper cursive elongation of a text can vary depending on the
      script, typeface, language,
      location within a word, location within a line,
      implementation complexity, font capabilities,
      and calligraphic preferences.
      It is outside the scope of CSS to define rules for such elongation.
      Authors should avoid applying 'letter-spacing' to cursive scripts
      unless they are prepared to accept non-interoperable results.
    </div>

<h2 id="edge-effects">
  Edge Effects</h2>

  <p>Edge effects control
    the indentation of lines with respect to other lines in the block ('text-indent')
    and how content is measured at the start and end edges of a line ('hanging-punctuation').

  <h3 id="text-indent-property">
    First Line Indentation: the 'text-indent' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>text-indent</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>[ <span class="value-inst-length"><a href="http://www.w3.org/TR/CSS21/syndata.html#value-def-length">&lt;length&gt;</a></span>
              | <span class="value-inst-percentage"><a href="http://www.w3.org/TR/CSS21/syndata.html#value-def-percentage">&lt;percentage&gt;</a></span>
              ] && hanging? && each-line?
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>0</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>block containers</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>refers to width of containing block</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>the percentage as specified or the absolute length,
            plus any keywords as specified
        </tr>
        <tr>
          <th>Animatable:
          <td>as <a href="http://www.w3.org/TR/css3-transitions/#animtype-lpcalc">length, percentage, or calc</a>, but only if keywords match
        <tr>
          <th>Canonical order:
          <td><abbr title="follows order of property value definition">per grammar</abbr>
    </table>

    <p>This property specifies the indentation applied to lines of inline
      content in a block. The indent is treated as a margin applied to
      the start edge of the line box.

    <p>Unless otherwise specified by the ''each-line'' and/or <a value for=text-indent>hanging</a> keywords,
      only lines that are the
      <a href="http://www.w3.org/TR/CSS21/selector.html#first-line-pseudo">first formatted line</a> [[!CSS21]]
      of an element are affected.
      For example, the first line of an anonymous block box is only affected
      if it is the first child of its parent element.

    <p>Values have the following meanings:</p>

    <dl dfn-for=text-indent dfn-type=value>
      <dt><dfn>&lt;length&gt;</dfn>
      <dd>Gives the amount of the indent as an absolute length.</dd>
      <dt><dfn>&lt;percentage&gt;</dfn>
      <dd>Gives the amount of the indent as a percentage of the containing
        block's logical width.</dd>
      <dt><dfn>each-line</dfn>
      <dd>Indentation affects the first line of each block container
        and each line after a <a>forced line break</a>
        (but not lines after a <a>soft wrap break</a>).
      <dt><dfn>hanging</dfn>
      <dd>Inverts which lines are affected.</dd>
    </dl>

    <div class="example">
      <p>If 'text-align' is ''start'' and 'text-indent' is ''5em'' in
      left-to-right text with no floats present, then first line of text
      will start 5em into the block:</p>

      <pre class="output">
             Since CSS1 it has been possible to
        indent the first line of a block element
        5em by setting the 'text-indent' property&nbsp;
        to '5em'.
      </pre>
      <p>If we add the ''text-indent/hanging'' keyword,
      then the first line will start flush,
      but other lines will be indented 5em:
      <pre class="output">
        In CSS3 we can instead indent all other
             lines of the block element by 5em
             by setting the 'text-indent' property
             to 'hanging 5em'.
      </pre>
    </div>

    <div class="example">
      <p>Since the 'text-indent' property only affects the “first formatted line”,
      a line after a forced break will not be indented.
      <pre class="output">
           For example, in the middle of
        this paragraph is an equation,
        which is centered:
                         x + y = z
        The first line after the equation
        is flush (else it would look like
        we started a new paragraph).
      </pre>
      <p>However, sometimes (as in poetry or code),
      it is appropriate to indent each line
      that happens to be long enough to wrap.
      In the following example, 'text-indent'
      is given a value of ''3em hanging each-line'',
      giving the third line of the poem a hanging indent
      where it soft-wraps at the block's right boundary:
      <pre class="output">
        In a short line of text
        There need be no wrapping,
        But when we go on and on and on&nbsp;&nbsp;
           and on,
        Sometimes a soft break
        Can help us stay on the page.
      </pre>
    </div>

    <p class="note">Note that since the 'text-indent' property inherits,
      when specified on a block element, it will affect descendant
      inline-block elements.
      For this reason, it is often wise to specify 'text-indent: 0' on
      elements that are specified 'display: inline-block'.</p>

  <h3 id="hanging-punctuation-property">
    Hanging Punctuation: the 'hanging-punctuation' property</h3>

    <table class="propdef">
        <tr>
          <th>Name:</th>
          <td><dfn>hanging-punctuation</dfn></td>
        </tr>
        <tr>
          <th><a href="#values">Value</a>:
          <td>none | [ first || [ force-end | allow-end ] || last ]
          </td>
        </tr>
        <tr>
          <th>Initial:</th>
          <td>none</td>
        </tr>
        <tr>
          <th>Applies to:</th>
          <td>inline elements</td>
        </tr>
        <tr>
          <th>Inherited:</th>
          <td>yes</td>
        </tr>
        <tr>
          <th>Percentages:</th>
          <td>N/A</td>
        </tr>
        <tr>
          <th>Media:</th>
          <td>visual</td>
        </tr>
        <tr>
          <th>Computed&#160;value:</th>
          <td>as specified</td>
        </tr>
        <tr>
          <th>Animatable:
          <td>no
        <tr>
          <th>Canonical order:
          <td><abbr title="follows order of property value definition">per grammar</abbr>
    </table>

    <p>This property determines whether a punctuation mark, if one is present,
      <a>hangs</a> and may be placed outside the line box (or in the indent)
      at the start or at the end of a line of text.

    <p class="note">Note that if there is not sufficient padding on the
      block container, 'hanging-punctuation' can trigger overflow.</p>

    <p>When a punctuation mark <dfn lt="hang">hangs</dfn>, it is not considered
      when measuring the line's contents for fit, alignment, or justification.
      Depending on the line's alignment/justification, this can
      result in the mark being placed outside the line box.

    <p>Values have the following meanings:</p>

    <dl dfn-for=hanging-punctuation dfn-type=value>
      <dt><dfn>none</dfn></dt>
        <dd>No character <a>hangs</a>.</dd>
      <dt><dfn>first</dfn></dt>
        <dd>An opening bracket or quote at the start of the
          <a href="http://www.w3.org/TR/CSS21/selector.html#first-line-pseudo">first
          formatted line</a> of an element <a>hangs</a>.
          This applies to all characters in the Unicode categories Ps, Pf, Pi.</dd>
      <dt><dfn>last</dfn></dt>
        <dd>A closing bracket or quote at the end of the
          last formatted line of an element <a>hangs</a>.
          This applies to all characters in the Unicode categories Pe, Pf, Pi.</dd>
      <dt><dfn>force-end</dfn></dt>
        <dd>A <a>stop or comma</a> at the end of a line <a>hangs</a>.</dd>
      <dt><dfn>allow-end</dfn></dt>
        <dd>A <a>stop or comma</a> at the end of a line <a>hangs</a> if it
          does not otherwise fit prior to justification.</dd>
    </dl>

    <p>Non-zero inline-axis borders or padding between
      a <a>hang</a>able mark and the edge of the line prevent the mark from hanging.
      For example, a period at the end of an inline box with end padding
      does not <a>hang</a> at the end edge of a line.
      At most one punctuation character may <a>hang</a> at each edge of the line.

    <p>A <a lt=hang>hanging</a> punctuation mark is still enclosed inside its inline box
      and participates in text justification:
      its character advance width is just not measured when determining
      how much content fits on the line,
      how much the line's contents need to be expanded or compressed for justification,
      or how to position the content within the line box for text alignment.

    <p><dfn lt="stop or comma">Stops and commas</dfn> allowed to <a>hang</a> include:
    <table class="data">
      <tr><td>U+002C <td>&#x002C; <td>COMMA
      <tr><td>U+002E <td>&#x002E; <td>FULL STOP
      <tr><td>U+060C <td>&#x060C; <td>ARABIC COMMA
      <tr><td>U+06D4 <td>&#x06D4; <td>ARABIC FULL STOP
      <tr><td>U+3001 <td>&#x3001; <td>IDEOGRAPHIC COMMA
      <tr><td>U+3002 <td>&#x3002; <td>IDEOGRAPHIC FULL STOP
      <tr><td>U+FF0C <td>&#xFF0C; <td>FULLWIDTH COMMA
      <tr><td>U+FF0E <td>&#xFF0E; <td>FULLWIDTH FULL STOP
      <tr><td>U+FE50 <td>&#xFE50; <td>SMALL COMMA
      <tr><td>U+FE51 <td>&#xFE51; <td>SMALL IDEOGRAPHIC COMMA
      <tr><td>U+FE52 <td>&#xFE52; <td>SMALL FULL STOP
      <tr><td>U+FF61 <td>&#xFF61; <td>HALFWIDTH IDEOGRAPHIC FULL STOP
      <tr><td>U+FF64 <td>&#xFF64; <td>HALFWIDTH IDEOGRAPHIC COMMA
    </table>
    <p>The UA may include other characters as appropriate.

    <p class="note">The CSS Working Group would appreciate if UAs including
      other characters would <a href="#status">inform the working group</a>
      of such additions.</p>

    <div class="example">
      <p>The ''allow-end'' and ''force-end'' are two variations of
        hanging punctuation used in East Asia.</p>
      <div>
      <div class="sidefigure">
        <img src="images/hanging-punctuation-allow-end.png"
          width="202" height="51"
          alt="hanging-punctuation: allow-end">
      </div>
      <pre><code class="css"><!--
     -->p {
<!-- -->   text-align: justify;
<!-- -->   hanging-punctuation: allow-end;
<!-- -->}
<!----></code></pre>
      <div class="sidefigure">
        <img src="images/hanging-punctuation-force-end.png"
          width="202" height="51"
          alt="hanging-punctuation: force-end">
      </div>
      <pre><code class="css"><!--
     -->p {
<!-- -->   text-align: justify;
<!-- -->   hanging-punctuation: force-end;
<!-- -->}
<!----></code></pre>
      </div>
      <p>The punctuation at the end of the first line for ''allow-end''
        does not hang, because it fits without hanging.
        However, if ''force-end'' is used, it is forced to hang.
        The justification measures the line without the hanging punctuation.
        Therefore when the line is expanded, the punctuation is pushed outside the line.</p>
    </div>

<h3 id="bidi-linebox">
Bidirectionality and Line Boxes</h3>

  <p>The <a spec=css-writing-modes-3>start</a> and <a spec=css-writing-modes-3>end</a> edges of a line box
    are determined by the <a>inline base direction</a> of the line box.
    In most cases, this is given by its containing block's computed 'direction'.

  <p>However if its containing block has ''unicode-bidi: plaintext'' [[!CSS3-WRITING-MODES]],
    the line box's <a>inline base direction</a> must be determined
    by the <a>inline base direction</a> of the <a>bidi paragraph</a> to which it belongs:
    that is, the <a>bidi paragraph</a> for which the line box holds content.
    An empty line box
    (i.e. one that contains no atomic inlines or
     characters other than the line-breaking character, if any),
    takes its <a>inline base direction</a> from the preceding line box (if any), or,
    if this is the first line box in the containing block,
    then from the 'direction' property of the containing block.

  <div class="example">
    <p>In the following example, assuming the <code>&lt;block&gt;</code>
    is a preformatted block (''display: block; white-space: pre'') inheriting
    ''text-align: start'', every other line is right-aligned:</p>
    <pre>&lt;block style="unicode-bidi: plaintext"&gt;
<!-- -->  Latin
<!-- -->  &#x0648;·&#x06A9;&#x0645;&#x06CC;
<!-- -->  Latin
<!-- -->  &#x0648;·&#x06A9;&#x0645;&#x06CC;
<!-- -->  Latin
<!-- -->  &#x0648;·&#x06A9;&#x0645;&#x06CC;
<!-- -->&lt;/block&gt;</pre>
  </div>

  <p class="note">Note that the inline base direction determined here
    applies to the line box itself, and not to its contents.
    It affects 'text-align', 'text-align-last', 'text-indent', and 'hanging-punctuation',
    i.e. the position and alignment of its contents with respect to its edges.
    It does not affect the formatting or ordering of its content.

  <div class="example">
    <p>In the following example:
    <pre>
<!-- -->&lt;para style="display: block; direction: rtl; unicode-bidi:plaintext">
<!-- -->&lt;quote style="unicode-bidi:plaintext">שלום!&lt;/quote>", he said.
<!-- -->&lt;/para&gt;</pre>

    <p>The result should be a left-aligned line looking like this:
    <pre>"!&#1513;&#1500;&#1493;&#1501;", he said.</pre>

    <p>The line is left-aligned
      (despite the containing block having ''direction: rtl'')
      because the containing block (the <code>&lt;para></code>) has ''unicode-bidi:plaintext'',
      and the line box belongs to a bidi paragraph that is LTR.
      This is because that paragraph's first character with a strong direction
      is the LTR "h" from "he". The RTL "שלום!" does precede the "he",
      but it sits in its own bidi-isolated paragraph that is <em>not</em>
      immediately contained by the <code>&lt;para&gt;</code>,
      and is thus irrelevant to the line box's alignment.
      From from the standpoint of the bidi paragraph immediately contained
      by the <code>&lt;para&gt;</code> containing block,
      the <code>&lt;quote&gt;</code>&rsquo;s bidi-isolated paragraph inside it is,
      by definition, just a neutral U+FFFC character,
      so the immediately-contained paragraph becomes LTR by virtue
      of the "he" following it.
  </div>

  <div class="example">
    <pre>
<!-- -->&lt;fieldset style="direction: rtl">
<!-- -->&lt;textarea style="unicode-bidi:plaintext">
<!-- -->
<!-- -->Hello!
<!-- -->
<!-- -->&lt;/textarea>
<!-- -->&lt;/fieldset></pre>

    <p>As expected, the "Hello!" should be displayed LTR
      (i.e. with the exclamation mark on the right end,
      despite the <code>&lt;textarea></code>’s ''direction:rtl'')
      and left-aligned.
      This makes the empty line following it left-aligned as well,
      which means that the caret on that line should appear at its
      left edge. The first empty line, on the other hand, should
      be right-aligned, due to the RTL direction of its containing
      paragraph, the <code>&lt;textarea></code>.
  </div>

<h2 class="no-num" id="order">Appendix A:
Text Processing Order of Operations</h2>

  <p>The following list defines the order of text operations.
  (Implementations are not bound to this order as long as the resulting layout is the same.)

  <ol>
    <li><a href="#white-space-phase-1">white space processing</a> part I (pre-wrapping)
    <li><a href="#transforming">text transformation</a>
    <li><a href="http://www.w3.org/TR/css3-writing-modes/#text-combine-horizontal">text combination</a>
    <li><a href="http://www.w3.org/TR/css3-writing-modes/#text-orientation">text orientation</a> [[!CSS3-WRITING-MODES]]
    <li><a href="#wrapping">text wrapping</a> while applying per line:
      <ol style="list-style: lower-roman">
        <li><a href="#text-indent-property">indentation</a>
        <li><a href="http://www.w3.org/TR/css3-writing-modes/#text-direction">bidirectional reordering</a> [[!CSS21]] / [[!CSS3-WRITING-MODES]]
        <li><a href="#white-space-phase-2">white space processing</a> part II
        <li><a href="http://www.w3.org/TR/css3-fonts/">font/glyph selection and positioning</a> [[!CSS21]] / [[!CSS3-FONTS]]
        <li>'letter-spacing' and 'word-spacing'
        <li><a href="#hanging-punctuation-property">hanging punctuation</a>
      </ol>
    <li><a href="#justification">justification</a> (which may affect glyph selection and/or text wrapping, looping back into that step)
    <li><a href="#text-align-property">text alignment</a>
  </ol>

<h2 class="no-num" id="default-stylesheet">
Appendix B: Default UA Stylesheet</h2>

  <p>This appendix is informative,
  and is to help UA developers to implement a default stylesheet for HTML,
  but UA developers are free to ignore or modify as appropriate.

  <div class="example">
    <pre>
<!-- -->/* make list items and option elements align together */
<!-- -->li, option { text-align: match-parent; }</pre>
  </div>
  <p class="feedback issue">If you find any issues, recommendations to add, or corrections,
    please send the information to <a href="mailto:www-style@w3.org">www-style@w3.org</a>
    with <kbd>[css-text]</kbd> in the subject line.</p>

<h2 class="no-num" id="script-groups">
Appendix C: Scripts and Spacing</h2>

  <p><em>This appendix is normative.</em></p>

  <p>Typographic behavior varies somewhat by language, but varies drastically by writing system.
  	This appendix categorizes some common <a lt="Unicode script">scripts</a> in Unicode 6.0
    according to their justification and spacing behavior.
  	Category descriptions are descriptive, not prescriptive;
  	the determining factor is the prioritization of <a>justification opportunities</a>.

  <dl>
  	<dt><dfn>block scripts</dfn></dt>
  	<dd>CJK and by extension all Wide characters (see [[!UAX11]].)
  	  The following <a>Unicode scripts</a> are included:
      Bopomofo, Han, Hangul, Hiragana, Katakana, and Yi.
  	  Characters of the <a>East Asian Width property</a> <code>W</code> and <code>F</code> are also included,
  	  but <code>A</code> characters are included only if the <a>content language</a> is Chinese, Korean, or Japanese.
  	<dt><dfn>clustered scripts</dfn></dt>
  	<dd>Clustered scripts have discrete units
      and break only at word boundaries,
      but do not use visible word separators.
      They prioritize stretching spaces,
      but comfortably admit inter-character spacing for justification.
  	  The clustered scripts include, but are not limited to, the following <a>Unicode scripts</a>:
      Khmer,
      Lao,
      Myanmar,
      New Tai Lue,
      Tai Le,
      Tai Tham,
      Tai Viet,
      Thai
  	<dt><dfn lt="cursive script">cursive scripts</dfn>
  	<dd>Cursive scripts do not admit gaps between their letters for either justification or 'letter-spacing'.
      The following <a>Unicode scripts</a> are included:
      Arabic,
      Mandaic,
      Mongolian,
      N'Ko,
      Phags Pa,
      Syriac
  </dl>
  <p>User agents should update this list as they update their Unicode support
    to handle as-yet-unencoded cursive scripts in future versions of Unicode,
    and are encouraged to ask the CSSWG to update this spec accordingly.

  <p class="issue">Should block and cluster scripts be merged?
  They have different tolerances for space-justification vs inter-character justification,
  but both admit both.

<h2 id="character-properties" class="no-num">Appendix D.
Characters and Properties</h2>

  <p>Unicode defines four codepoint-level properties that are referenced
    in CSS typesetting:
  <dl export>
    <dt><dfn lt="Unicode East Asian Width|East Asian Width property"><a href="http://www.unicode.org/reports/tr11/#Definitions">East Asian width property</a></dfn>
    <dd>Defined in [[!UAX11]] and given as the <code>East_Asian_Width</code> property
      in the Unicode Character Database [[!UAX44]].
    <dt><dfn lt="Unicode General Category|Unicode category|General Category"><a href="http://www.unicode.org/reports/tr44/#General_Category_Values">general category</a></dfn>
    <dd>Defined in [[!UAX44]] and given as the <code>General_Category</code> property
      in the Unicode Character Database [[!UAX44]].
    <dt><dfn lt="Unicode Script|Script property"><a href="http://www.unicode.org/reports/tr24/#Values">script property</a></dfn>
    <dd>Defined in [[!UAX24]] and given as the <code>Script</code> property
      in the Unicode Character Database [[!UAX44]].
      (UAs must include any ScriptExtensions.txt assignments in this mapping.)
    <dt><a href="http://www.unicode.org/reports/tr50/">Vertical Orientation</a>
    <dd>Defined in [[!UTR50]] as the Vertical_Orientation property
      and given in the UTR50 data file.
  </dl>

  <p>Unicode defines properties for individual codepoints, but sometimes
    it is necessary to determine the properties of a <a>typographic character unit</a>.
    For the purposes of CSS Text,
    the properties of a <a>typographic character unit</a> are given by
    the base character of its first <a>grapheme cluster</a>—except in two cases:
  <ul>
    <li><a>Grapheme clusters</a> formed with an Enclosing Mark (<code>Me</code>) of the Common script
    are considered to be Other Symbols (<code>So</code>) in the Common script.
    They are assumed to have the same Unicode properties as the Replacement Character U+FFFD.
    <li>Grapheme clusters</a> formed with a Space Separator (<code>Zs</code>) as the base
    are considered to be Modifier Symbols (<code>Sk</code>).
    They are assumed to have the same East Asian Width property as the base,
    but take their other properties from the first combining character in the sequence.
  </ul>

<h2 class="no-num" id="acknowledgements">
Acknowledgements</h2>

  <p>This specification would not have been possible without the help from:
  Ayman Aldahleh, Bert Bos, Tantek Çelik, James Clark, Stephen Deach, John Daggett,
  Martin Dürst,
  Laurie Anna Edlund, Ben Errez, Yaniv Feinberg, Arye Gittelman, Ian
  Hickson, Martin Heijdra, Richard Ishida, Masayasu Ishikawa,
  Michael Jochimsen, Eric LeVine, Ambrose Li, Håkon Wium Lie, Chris Lilley,
  Ken Lunde, Nat McCully, IM Mincheol, Shinyu Murakami, Paul Nelson, Chris Pratley, Marcin Sawicki,
  Arnold Schrijver, Rahul Sonnad, Alan Stearns, Michel Suignard, Takao Suzuki,
  Frank Tang, Chris Thrasher, Etan Wexler, Chris Wilson, Masafumi Yabe
  and Steve Zilles.

<h2 class="no-num" id="changes">
Changes</h2>

  <h3 class="no-num" id="recent-changes">
  Changes from the <a href="http://www.w3.org/TR/2013/WD-css-text-3-20131010/">October
  2013 CSS3 Text <abbr title="Last Call Working Draft">LCWD</abbr></a></h3>

  <ul>
    <li>Made 'text-align' a shorthand of 'text-align-last' and the new 'text-align-all' property.
    <li>Removed dependence of 'text-align-last' on ''text-align: justify'',
        since the problem it solves is now solved by the shorthanding relationship.
  </ul>

  <h3 class="no-num" id="changes-2013">
  Changes from the <a href="http://www.w3.org/TR/2013/WD-css-text-3-20131010/">November
  2012 CSS3 Text <abbr title="Working Draft">WD</abbr></a></h3>

  <p>Major changes include:
  <ul>
    <li>Removed ''inter-ideograph'', ''inter-cluster'', and ''kashida'' values of 'text-justify'.
    <li>Changed ''text-justify: inter-word'' to prohibit justification between letters.
    <li>Dropped minimum and maximum values in 'letter-spacing' and 'word-spacing'.
    <li>Removed <var>&lt;string&gt;</var> value of 'text-align'.
    <li>Changed the character list of 'line-break'.
    <li>Changed orders of <a href="#white-space-rules">white space processing</a>,
        <a href="http://www.w3.org/TR/css3-writing-modes/#text-combine-horizontal">text combination</a> and
        <a href="http://www.w3.org/TR/css3-writing-modes/#text-orientation">text orientation</a>,
        'letter-spacing', and 'word-spacing'
        of the <a href="#order">Text Processing Order of Operations</a>.
    <li>Changed the definition of <a>word-separator characters</a>.
    <li>Changed to allow 'letter-spacing' to justify.
    <li>Changed the computed value of ''letter-spacing:normal'' value of 'letter-spacing' and 'word-spacing'.
    <li>Changed the grammar rule of 'text-indent'.
  </ul>

  <p>Significant details updated:
  <ul>
    <li>Clarified ''match-parent'' value of 'text-align'.
    <li>Clarified ''text-justify::none'' value of 'text-justify'.
    <li>Clarified <a href="#expanding-text">expanding and compressing text</a> for 'text-justify'.
    <li>Clarified the behavior of 'letter-spacing'.
    <li>Clarified the behavior of 'text-align' for too long lines.
    <li>Clarified the interaction of 'word-break' and 'line-break'.
  </ul>