CSS3 module: text

Abstract

This document presents a set of text formatting properties for CSS3. Many of these properties already existed in CSS2 [CSS2]. Many of the new properties have been added to address basic requirements in international text layout, particularly for East Asian and bidirectional text.

Status of This Document

This document is a working draft of the CSS working group which is part of the Style activity. It contains a proposal for features to be included in CSS level 3.

This is the second last call for comments. All comments in response to the previous call have been processed: see the disposition of comments. It is expected that a Candidate Recommendation will follow shortly. The deadline for comments on this draft is 5 March 2003.

Feedback is very much welcome. Comments can be sent directly to the editor, but the mailing list www-style@w3.org (see instructions) is also open and is preferred for discussion of this and other drafts in the Style area. The mailing list is archived.

This document has been produced as a combined effort of the W3C Internationalization Activity, and the Style Activity. It also includes extensive contribution made by members of the XSL Working Group (members only). Finally, some of the proposal surfaced first in the Scalable Vector Graphics (SVG) 1.0 Specification [SVG1.0]. The text has been duplicated in this document to reflect which properties and specification should eventually be referenced in CSS itself.

This working draft may be updated, replaced or rendered obsolete by other W3C documents at any time. It is inappropriate to use W3C Working Drafts as reference material or to cite them as other than "work in progress". Its publication does not imply endorsement by the W3C membership or the CSS Working Group (members only).

Patent disclosures relevant to CSS may be found on the Working Group's public patent disclosure page.

To find the latest version of this working draft, please follow the "Latest version" link above, or visit the list of W3C Technical Reports.

Contents

• 1. Dependencies on other modules
• 2. Introduction
• 3. Text layout
  • 3.1. Text layout introduction
  • 3.2. Setting the inline-progression and block-progression: the 'writing-mode' and 'direction' properties
  • 3.3. Glyph orientation within a text run: the 'glyph-orientation-vertical' and 'glyph-orientation-horizontal' properties
  • 3.4. Embedding and override: the 'unicode-bidi' property
  • 3.5. Script character classification: the 'text-script' property
• 4. Text alignment and justification
  • 4.1. Text alignment: the 'text-align' property
  • 4.2. Justification: the 'text-justify' property
  • 4.3. Last line alignment: the 'text-align-last' property
  • 4.4. Minimum and maximum font size: the 'min-font-size' and 'max-font-size' property
  • 4.5. Additional compression: The 'text-justify-trim' property
  • 4.6. Kashida effect: the 'text-kashida-space' property'
• 5. Text indentation
  • 5.1. Summary
  • 5.2. First Line indentation: the 'text-first-indent' property
  • 5.3. Block Line indentation: the 'text-block-indent' property
  • 5.4. All Lines indentation: the 'text-indent' shorthand property
• 6. Line breaking
  • 6.1. Types of line breaking
  • 6.2. Line breaking: the 'line-break' property
  • 6.3. Word breaking: the 'word-break-cjk', 'word-break-inside' properties and the shorthand 'word-break' property
• 7. Text Wrapping, White-space Control and Text Overflow
  • 7.1. Text wrapping: the 'wrap-option' property
  • 7.2. White-space control: the 'linefeed-treatment', 'white-space-treatment', 'all-space-treatment' properties and the 'white-space' shorthand property
  • 7.3. Text overflow: the 'text-overflow-mode', 'text-overflow-ellipsis' properties and the shorthand 'text-overflow' property
• 8. Text spacing
  • 8.1. Letter spacing: the 'letter-spacing' property
  • 8.2. Word spacing: the 'word-spacing' property
  • 8.3. Punctuation trimming: the 'punctuation-trim' property
  • 8.4. Adding space: the 'text-autospace' property
  • 8.5. Text kerning: the 'kerning-mode' and 'kerning-pair-threshold' property
• 9. Text decoration
  • 9.1. Introduction
  • 9.2. Text decoration style: the 'text-underline-style', 'text-line-through-style' and 'text-overline-style' properties
  • 9.3. Text decoration color: the 'text-underline-width', 'text-line-through-width' and 'text-overline-width' properties
  • 9.4. Text decoration color: the 'text-underline-color', 'text-line-through-color' and 'text-overline-color' properties
  • 9.5. Text decoration mode: the 'text-underline-mode', 'text-line-through-mode' and 'text-overline-mode' properties
  • 9.6. Other text decoration simple properties: 'text-underline-position' and 'text-blink'
  • 9.7. Text decoration shorthand properties: 'text-underline', 'text-line-through', 'text-overline-mode' and 'text-decoration'
  • 9.8. Text shadows: the 'text-shadow' property
• 10. Document grid
  • 10.1. What is document grid?
  • 10.2. Line grid mode: the 'line-grid-mode' property
  • 10.3. Line grid progression: the 'line-grid-progression' property
  • 10.4. Line grid: the 'line-grid' shorthand property
• 11. Miscellaneous text formatting
  • 11.1. Capitalization: the 'text-transform' property
  • 11.2. Hanging punctuation: the 'hanging-punctuation' property
  • 11.3. Combining text: the 'text-combine' property'
• 12. Properties index
• 13. Profiles
• 14. Glossary
• Appendix A: Vertical Layout Effect on CSS Properties
• Acknowledgments
• References
  • Normative references
  • Other references

1. Dependencies on other modules

This CSS3 module depends on the following other CSS3 modules:

• Fonts
• Line
• Syntax and grammar
• Values and unit

It has non-normative (informative) references to the following other CSS3 modules:

• Multi-column layout
• Ruby

2. Introduction

In both CSS1 and CSS2, text formatting has been limited to simple effects like for example: text decoration, text alignment and letter spacing. However, international typography contains types of formatting that could not be achieved without using special workarounds or graphics.

Along with already existing text-related properties, this document presents a number of new CSS properties to represent such formatting. The features this proposal covers include two of the most important features for East Asian typography: vertical text and layout grid.

There are a number of illustrations in this document for which the following legend is used:

- wide-cell glyph (e.g. Han) which is the n-th glyph in the text run,

- narrow-cell glyph (e.g. Latin) which is the n-th glyph in the text run,

- connected glyph (e.g. Arabic) which is the n-th glyph in the text run.

Many typographical properties in East Asian typography depends on the fact that a character is typically rendered as either a wide or narrow glyph. All characters described by the Unicode Standard [UNICODE] can be categorized by a width property. This is covered by the Unicode Standard Annex [UAX-11].

The orientation which the above symbols assume in the diagrams corresponds to the orientation that the glyphs they represent are intended to assume when rendered in the UA (user agent). Spacing between these glyphs in the diagrams is usually symbolic, unless intentionally changed to make a point.

Furthermore, all properties, in addition to the noted values, take 'initial' and 'inherit'. These values are not repeated in property value enumerations.

This module uses extensively the 'before', 'after', 'start' and 'end' notation to specify the four edges of a box relative to its text advance direction, independently of its absolute positioning in terms of 'top', 'bottom', 'left' and 'right' (corresponding respectively to the 'before', 'after', 'start' and 'end' positions in a typical Western text layout). This notation is also used extensively in [XSL1.0] for the same purpose.

The term 'Latin' is used frequently in this document to designate behavior shared among popular writing scripts in Europe and America based on the Latin, Greek and Cyrillic scripts.

Finally, in this document, requirements are expressed using the key words "MUST", "MUST NOT", "REQUIRED", "SHALL" and "SHALL NOT". Recommendations are expressed using the key words "SHOULD", "SHOULD NOT" and "RECOMMENDED". "MAY" and "OPTIONAL" are used to indicate optional features or behavior. These keywords are used in accordance with [RFC2119]. For legibility these keywords are used in lowercase form.

3. Text layout

3.1. Text layout introduction

This section describes the text layout features supported by CSS, which includes support for various international writing directions, such as left-to-right (e.g., Latin scripts), right-to-left (e.g., Hebrew or Arabic), bidirectional (e.g., mixing Latin with Arabic) and vertical (e.g., Asian scripts).

The 'writing-mode' property determines an inline-progression and a block (line to line) progression. For example, Latin scripts are typically written left to right and top to bottom. The 'direction' property, already defined in CSS2, is a subset of the 'writing-mode' property as it only determines an inline-progression. The 'direction' property may still be used when no block-progression change is desired.

The glyph orientation determines the orientation of the rendered visual shape of characters relative to the inline-progression.

Within a line, the adjustment to the current text position is based on the current glyph orientation relative to the inline-progression, the metrics of the glyph just rendered, kerning tables in the font, and the current values of various attributes and properties, such as the spacing properties.

Bidirectionality introduces another level of complexity in text layout, as in many combinations of 'writing-mode' and glyph orientation values, the proper directionality and ordering of text are determined the Unicode Standard ([UNICODE] bidirectional algorithm. CSS relies on that algorithm to achieve proper bidirectional text rendering and possible reordering. Furthermore, using the 'unicode-bidi' property, the user agent can influence the bidirectional algorithm by allowing new embedding levels and direction overrides.

Note: The Unicode Standard ([UNICODE], section 3.12) defines a bidirectional algorithm determining the directionality for bidirectional text. The display ordering of bidirectional text depends upon the directional properties of the characters in the text.

 The HTML 4.01 specification ([HTML401], section 8.2) defines bidirectionality behavior for HTML elements. Conforming HTML user agents may therefore ignore the 'direction' and 'unicode-bidi' properties in author and user style sheets. The style sheet rules that would achieve the bidirectionality behavior specified in HTML 4.01 are given in the sample style sheet. The HTML 4.01 specification also contains more information on bidirectionality issues.

Note: HTML 4.01 does not cover the more general case described by the 'writing-mode' property.

3.2. Setting the inline-progression and block-progression: the 'writing-mode' and 'direction' properties

The 'writing-mode' property specifies whether the inline-progression shall be left-to-right, right-to-left, or top-to-bottom. This property also changes the 'direction' property for the element. For more on bidirectional text, see the section about Embedding and override.

Note: Even when the inline-progression is left-to-right or right-to-left, some or all of the content within a given element might advance in the opposite direction because of the Unicode [UNICODE] bidirectional algorithm or because of explicit text advance overrides due to this property or 'direction' and 'unicode-bidi'.

lr-tb | lr

Sets the inline-progression to left-to-right, and the block-progression to top-to-bottom, as is common in the Latin-based documents. For most characters, the current text position is advanced from left to right after each glyph is rendered. The 'direction' property is set to 'ltr'.

rl-tb | rl

Sets the inline-progression to right-to-left, and the block-progression to top-to-bottom, as is common in Arabic or Hebrew scripts. The 'direction' property is set to 'rtl'.

tb-rl | tb

Sets the inline-progression to top-to-bottom, and the block-progression to right-to-left, as is common in some East Asian scripts. The baseline alignment may be different in this context. Typically, the dominant baseline runs through the center of the upright glyphs. The 'direction' property is set to 'ltr'.

tb-lr

Sets the inline-progression to top-to-bottom, and the block-progression to left-to-right, as is common in the Mongolian script. This type of vertical layout also occurs in Latin-based documents, particularly in table column or row labels. The baseline alignment may be different in this context. Typically, the dominant baseline runs through the center of the upright glyphs. The 'direction' property is set to 'rtl'.

bt-rl

Sets the inline-progression to bottom-to-top, and the block-progression to right-to-left. This value exists only to cover the case of the 'direction' property value 'rtl' applied to an element where the current writing-mode property value is 'tb-rl' or 'tb'. The 'direction' property is set to 'rtl'.

bt-lr

Sets the inline-progression to bottom-to-top, and the block-progression to left-to-right. This value exists only to cover the case of the 'direction' property value 'rtl' applied to an element where the current writing-mode property value is 'tb-lr'. The 'direction' property is set to 'ltr.

The combination of inline-progression and block-progression set by the 'writing-mode' property is also referred as a flow orientation. In such contexts, the values 'lr-tb', 'lr', 'rl-tb', 'rl' correspond to horizontal flow orientations, and the others 'tb-rl', 'tb', 'tb-lr', 'bt-rl', 'bt-lr' correspond to vertical flow orientations.

For horizontal flow orientations, the top and bottom margins can be collapsed. For vertical flow orientations, the left and right margin can be collapsed. See "Collapsing margins" in the CSS3 Box module [forthcoming] for the details of collapsing margins.

This property also specifies the direction of table column layout, the direction of the overflow when determined by the inline-progression (such as the 'start' and 'end' value of the 'text-align' property), the initial alignment of text and the position of an incomplete last line in a block in case of 'text-align: justify'.

For the 'writing-mode' property to have any effect on inline-level elements, one or both of the following conditions must be met:

• the new inline-progression is perpendicular to the parent's inline-progression or
• the 'unicode-bidi' property's value is 'embed' or 'bidi-override' and one of the following conditions is met:
  • the flow is vertical and the 'glyph orientation' is 'auto', '90deg (or equivalent), or -90eg (or equivalent)
  • the flow is horizontal and the 'glyph orientation' is '0deg (or equivalent), or 180deg (or equivalent).

An inline-level element that has a different block-progression than its parent becomes an 'inline-block' element.

.date {writing-mode: lr-tb;}
<span class="date">1996</span>

Values for this property have the following meanings:

ltr

Left-to-right direction.

rtl

Right-to-left direction.

This property specifies the inline-progression and the direction of embeddings and overrides (see 'unicode-bidi') for the Unicode bidirectional algorithm. The block-progression is not affected by this property. The values 'ltr' and 'rtl' have to be interpreted relative to the line direction. In addition, it specifies the direction of table column layout, the direction of the overflow when determined by the inline-progression (such as the 'start' and 'end' values of the 'text-align' property), the initial alignment of text and the position of an incomplete last line in a block in case of 'text-align: justify'.

For the 'writing-mode' property to have any effect on inline-level elements, the 'unicode-bidi' property's value must be 'embed' or 'bidi-override' and, in addition, one of the following conditions must be met:

• the flow is vertical and the 'glyph orientation' is 'auto', '90deg (or equivalent), or -90eg (or equivalent)
• the flow is horizontal and the 'glyph orientation' is '0deg (or equivalent), or 180deg (or equivalent).

Note: The 'writing-mode' and 'direction' properties, when specified for table column elements, are not inherited by cells in the column since column elements are never the ancestors of their constituent cell elements. Thus, CSS cannot easily capture the "dir" attribute inheritance rules described in [[HTML4.01], section 11.3.2.

Note: The 'writing-mode' and 'direction' properties interact with each other. As such, 'writing-mode' resets the 'direction' value. Similarly, if modifying 'direction' after 'writing-mode' introduces a change of inline-progression, the 'writing-mode' value will be updated to reflect the new inline-progression. For example, 'direction: rtl' applied to an element with 'writing-mode: lr-tb' effectively makes 'writing-mode: rl-tb'. This is one of the main reason why the mixed usage of these two properties is discouraged or at least they should be used with great caution. To illustrate the case, in the following example, the computed value of 'writing-mode' is 'rl-tb':

  e { writing-mode: lr-tb; direction: rtl;}

3.3. Glyph orientation within a text run: the 'glyph-orientation-vertical' and 'glyph-orientation-horizontal' properties

In some cases, it is required to alter the orientation of a sequence of characters relative to the inline-progression. The requirement is particularly applicable to vertical layouts of East Asian documents, where sometimes half-width Latin text is to be displayed horizontally and other times vertically.

Two properties control the glyph orientation relative to the inline-progression. 'glyph-orientation-vertical' controls glyph orientation when the inline-progression is vertical. 'glyph-orientation-horizontal' controls glyph orientation when the inline-progression is horizontal. It is necessary to distinguish between vertical and horizontal for the following reasons:

• When the inline-progression is vertical the typical glyph orientation depends on the related character. See the description of the 'auto' value. This value is not necessary in horizontal layout.
• The initial value is different for the two properties.

<angle>

The user agent may round the value of the angle to the values of glyph rotation supported by the user agent. This does not affect the computed value. Conforming user agents must at least support either the value 90deg or all of the following values: 0deg, 90deg, 180deg and 270deg. User agents may support other values.

A value of "0deg" indicates that all glyphs are oriented with the bottom of the glyphs toward the inline-progression, resulting in glyphs which are stacked vertically on top of each other. A value of "90deg" indicates a rotation of 90 degrees clockwise from the "0deg" orientation. For characters which have this property set to 90 or 270 degree, reordering is first applied according to the Unicode Bidi algorithm and then the resulting glyphs are rotated according to the <angle> value. The rotation specified by this value is applied to the glyph representations of all assigned Unicode character codes.

auto

The glyph orientation relative to the inline-progression is determined automatically based on the Unicode character code of the rendered character.

Full-width ideographic and full-width Latin glyphs (excluding some ideographic punctuation and bracket symbol) are oriented as if an <angle> of "0deg" had been specified for top to bottom inline-progression and "180deg" for bottom to top inline-progression (resulting in glyphs which are stacked vertically on top of each other).

Ideographic punctuation and other ideographic characters having alternate horizontal and vertical glyph shapes shall use the vertical shape of the glyph.

Text which is not full-width will be set as if an <angle> of "90deg" had been specified; thus, half-width Latin text will be rotated 90 degree clockwise versus full-width ideographic and full-width Latin text.

Hebrew and Arabic text are also rotated 90 degree clockwise. The visual order of this text is determined by the bidirectional algorithm applied prior to the rotation.

Note: A value of auto will generally produce the expected results in common uses of mixing Japanese with European characters; however, the exact algorithms are based on complex interactions between many factors, including font design, and thus different algorithms might be employed in different processing environments. For precise control, specify explicit <angle> values.

This property specifies the orientation of glyphs relative to the inline-progression and block-progression determined by the 'writing-mode' property. This property is applied only to text written in a vertical writing-mode. Conforming user agents may do the following in increasing levels of supports:

1. support only the 90deg value,
2. support only the 'auto' and 90deg value,
3. support the 'auto', 0deg, 90deg, 180deg and 270deg values,
4. support all values above and any number of additional values.

The glyph orientation affects the amount that the current text position advances as each glyph is rendered. It also affects how the glyph is aligned relative to the baseline. When the inline-progression is vertical and the 'glyph-orientation-vertical' value results in a glyph orientation angle which is a multiple of 180deg, then the current text position is incremented according to the vertical metrics of the glyph, and the glyph is aligned using the vertical alignment-point as determined by 'alignment-adjust' [link to the CSS3 line module] and 'alignment-baseline' [link to the CSS3 line module]. Otherwise, the text position is incremented according to the horizontal metrics of the glyph and the glyph is aligned using the horizontal alignment-point as determined by 'alignment-adjust' and 'alignment-baseline'.

Note: The concepts of alignment-point and alignment-baseline are described in the CSS3 line module.

The diagrams below illustrate different uses of 'glyph-orientation-vertical'. The diagram on the left shows the result of the mixing of full-width ideographic characters with normal-width Latin characters when 'glyph-orientation-vertical' for the span containing the Latin characters is either auto or "90deg". The diagram on the right show the result of mixing full-width ideographic characters with normal-width Latin characters when the span containing the Latin characters is specified to have a 'glyph-orientation-vertical' of "0deg".

Note: The effect on the right can be also be achieved by using full-width Latin characters and using 'glyph-orientation-vertical: auto' for the span containing the ideographic characters and the full-width Latin characters.

The bidi algorithm and the 'glyph-orientation-vertical' property have the following interaction:

1. The bidi algorithm is applied separately to each contiguous text range having the same glyph-orientation-vertical value. In other words a change in the property value resets the bidi algorithm.
2. When the glyph-orientation-vertical value is 270 degree, all mirroring symbols after all due bidi processing are mirrored before being rotated 270 degree clockwise. This is done to achieve the desired rendering result, which is to have the mirroring characters pointing 'inward' the text they are enclosing.

<angle>

The user agent may round the value of the angle to the values of glyph rotation supported by the user agent. This does not affect the computed value. Conforming user agents must at least support either the 0 deg value or all of the following values: 0deg, 90deg, 180deg and 270deg. User agents may support other values.

A value of "0deg" indicates that all glyphs are oriented with the top edge of the glyphs toward the before edge of the line box, resulting in glyphs which are positioned side by side. A value of "90deg" indicates an orientation of 90 degrees clockwise from the "0" orientation. For characters which have this property set to 0 or 180 degree, reordering is first applied according to the Unicode Bidi algorithm and then the resulting glyphs are rotated clockwise according to the <angle> value.

This property specifies the orientation of glyphs relative to the inline-progression determined by the 'writing-mode' property. This property is applied only to text written in a horizontal writing-mode.

The glyph orientation affects the amount that the current text position advances as each glyph is rendered. It also affects how the glyph is aligned relative to the baseline. When the inline-progression is horizontal and the 'glyph-orientation-horizontal' value results in a glyph orientation angle which is a multiple of 180deg, then the current text position is incremented according to the horizontal metrics of the glyph, and the glyph is aligned using the horizontal alignment-point as determined by 'alignment-adjust' [link to the CSS3 line module] and 'alignment-baseline' [link to the CSS3 line module]. Otherwise, the text position is incremented according to the vertical metrics of the glyph and the glyph is aligned using the vertical alignment-point as determined by 'alignment-adjust' and 'alignment-baseline'.

Note: The concepts of alignment-point and alignment-baseline are described in the CSS3 line module.

3.4. Embedding and override: the 'unicode-bidi' property

This property allows further control of the Unicode bidirectional algorithm by allowing new embedding levels or direction overrides. Values for this property have the following meanings:

normal

The element does not open an additional level of embedding with respect to the bidirectional algorithm. For inline-level elements, implicit reordering works across element boundaries.

embed

If the element is inline-level, this value opens an additional level of embedding with respect to the bidirectional algorithm. The direction of this embedding level is given by the 'direction' property. Inside the element, reordering is done implicitly. This corresponds to adding a LRE (U+202A; for 'direction: ltr') or RLE (U+202B; for 'direction: rtl') at the start of the element and a PDF (U+202C) at the end of the element.

bidi-override

If the element is inline-level or a block-level element that contains only continuous stretches of inline elements, this creates an override. This means that inside the element, reordering is strictly in sequence according to the 'direction' property; the implicit part of the bidirectional algorithm is ignored. This corresponds to adding a LRO (U+202D; for 'direction: ltr') or RLO (U+202E; for 'direction: rtl') at the start of the element and a PDF (U+202C) at the end of the element.

The final order of characters in each block-level element is the same as if the bidi control codes had been added as described above, mark-up had been stripped, non-textual entities such as images treated as neutral characters, and the resulting character sequence had been passed to an implementation of the Unicode bidirectional algorithm for plain text that produced the same line-breaks as the styled text.

Note: For non-textual entities such as images, when their 'unicode-bidi' property has a value other than 'normal', and the 'direction' has the value 'rtl', the resulting directionality is equivalent to character type R (according to the types defined by the Unicode Bidirectional algorithm).

Note: In order to be able to flow inline boxes in a uniform direction (either entirely left-to-right or entirely right-to-left), more inline boxes (including anonymous inline boxes) may have to be created, and some inline boxes may have to be split up and reordered before flowing.

Because the Unicode algorithm has a limit of 61 levels of embedding, care should be taken not to use 'unicode-bidi' with a value other than 'normal' unless appropriate. In particular, a value of 'inherit' should be used with extreme caution. However, for elements that are, in general, intended to be displayed as blocks, a setting of 'unicode-bidi: embed' is preferred to keep the element together in case display is changed to inline (see example below).

The following example shows an XML document with bidirectional text. It illustrates an important design principle: DTD designers should take bidi into account both in the language proper (elements and attributes) and in any accompanying style sheets. The style sheets should be designed so that bidi rules are separate from other style rules. The bidi rules should not be overridden by other style sheets so that the document language's or DTD's bidi behavior is preserved.

<div xml:lang="he">
  <par>HEBREW1 HEBREW2 english3 HEBREW4 HEBREW5</par>
  <par>HEBREW6 <emph>HEBREW7</emph> HEBREW8</par>
</div>
<div xml:lang="en">
  <par>english9 english10 english11 HEBREW12 HEBREW13</par>
  <par>english14 english15 english16</par>
  <par>english17 <quo xml:lang=he">HEBREW18 english19 HEBREW20</quo></par>
</div>

/* Rules for bidi */
div:lang(he)   {direction: rtl}
quo:lang(he)   {direction: rtl; unicode-bidi:  embed}
par:lang(en)   {direction: ltr}

/* Rules for presentation */
div, par  {display: block}
emph      {font-weight: bold}

               5WERBEH 4WERBEH english3 2WERBEH 1WERBEH

                                8WERBEH 7WERBEH 6WERBEH

english9 english10 english11 13WERBEH 12WERBEH

english14 english15 english16

english17 20WERBEH english19 18WERBEH

       2WERBEH 1WERBEH
  -EH 4WERBEH english3
                 5WERB

   -EH 7WERBEH 6WERBEH
                 8WERB

english9 english10 en-
glish11 12WERBEH
13WERBEH

english14 english15
english16

english17 18WERBEH
20WERBEH english19

3.5. Script character classification: the 'text-script' property

In text layout, many of the behaviors are related to a character classification based on scripts.

For some operations, such as baseline alignment, there is a requirement for a dominant script which is used to determine an alignment strategy for the whole element. A dominant script is established by setting the 'text-script' property to an explicit script identifier in conformance with the Unicode Technical Report [UTR-24]: "Script names", or by using the heuristic determination computed by the user agent when the 'text-script' value is set to 'auto'.

In many other cases, such as white-space handling or text justification, the script property is used on a character by character basis. In those cases, the 'text-script' property can be used to set an homogeneous value for all characters of the element through the usage of an explicit script identifier. But, if the 'text-script' is set to 'auto', the user agent will establish a script property value for each character of the element.

Values have the following meanings:

auto

Use the first character descendant which has an unambiguous script identifier [UTR-24] to determine the dominant script of the element's content. An unambiguous script identifier is any script value other than 'COMMON' or 'INHERITED'. In the absence of any textual components with an unambiguous script identifier (or no textual content at all), the dominant script is 'LATIN'. For operations that do not require a dominant scrip, but instead use a script determination on a character by character basis within the element, the user agent will determine a script value for each character based on its inherent script property as established by [UTR-24]. In such cases, the values 'COMMON' or 'INHERITED' are valid.

<script>

A script identifier in conformance with [UTR-24]. If the script identifier is either 'COMMON' or 'INHERITED' the dominant script value is 'LATIN', otherwise the dominant script value is the specified value. For operations not using the dominant script value, all characters of the element will behave as if their script property was the set value, not their inherent value.

Note: The Unicode technical report [UTR-24]: 'Script Names' specifies an inherent script value for each character of the Unicode Standard character repertoire [UNICODE]. There is also an ISO draft standard [ISO15924] addressing script identification.

Note: Setting an explicit script property value on an element reclassifies all its textual content to the given script. For example setting the script to a script belonging to the CJK group (Chinese, Japanese, Korean) makes the content behave as a CJK content for line-breaking rules. And setting an Arabic text to Latin would prevent the context to be affected by the Kashida justification effect. Typically, this property should be set to an explicit script value only when the textual content is script ambiguous and a specific behavior is sought.

4. Text alignment and justification

4.1. Text alignment: the 'text-align' property

This property describes how inline content of a block is aligned. Values have the following meanings:

start

The text is aligned on the start of the inline-progression.

end

The text is aligned on the end of the inline-progression.

left, right

In horizontal inline-progression, the text is aligned on the left or right respectively. In vertical inline-progression, the alignment is UA dependent. Because these two values are not related to the current inline-progression, the values 'start' and 'end' are typically preferred.

center

The text is center aligned.

justify

The text is justified. The justification algorithm can be further refined by using the 'text-justify' property. Although conforming CSS2 user agents could interpret the value 'justify' as 'start',  conforming CSS3 user agents may not, unless a profile specifies otherwise.

<string>

Specifies a string on which cells in a table column will align (see the section on horizontal alignment in a column for details and an example). This value applies only to table cells. If set on other elements, the computed value is 'start'.

A block of text is a stack of line boxes. In the case of 'start', 'end', 'left', 'right' and 'center', this property specifies how the inline boxes within each line box align with respect to the line box's start and end sides; alignment is not with respect to the viewport. In the case of 'justify', the UA may stretch the inline boxes in addition to adjusting their positions. (See also 'letter-spacing' and 'word-spacing'.)

Example(s):

In this example, note that since 'text-align' is inherited, all block-level elements inside the div element with 'class=important' will have their inline content centered.

div.important { text-align: center }

Note: The property initial value has changed between CSS2 and CSS3 from being UA dependent in CSS2 to be related to the current text advance direction in CSS3 (through the usage of the 'start' value).

4.2. Justification: the 'text-justify' property

This property selects the type of justify alignment for last lines when 'text-align-last' is set to 'justify' and other lines when 'text-align' is set to 'justify'. Most of the text-justify values affects writing systems in very specific ways. These writing systems (or group of) are:

• CJK and Hangul and by extension all 'wide' characters,
• Devanagari and all South Asian scripts using baseline connector such as Bengali and Gurmukhi,
• South Eastern Asian scripts that do not use space between words (such as Thai, Lao, Khmer, Myanmar),
• Cursive scripts like Arabic,
• Scripts using space between word without connector (such as Latin, Greek, Cyrillic, Hebrew, etc...) and symbol characters.

Depending on script classification value (controlled by the 'text-script' property value) and the 'text-justify' property value, spacing may be altered between words or letters or both.

The possible values for the text-justify property are:

auto

The UA determines the justification algorithm to follow, based on a balance between performance and adequate presentation quality. Inter-word expansion is typically used for all scripts that use space as word delimiter. The concept of a word is script dependent, although the user agent determines the exact algorithm. If the 'text-kashida-space' property has a non zero percent value it is recommended to use kashida elongation for Arabic text.

inter-word

Selects the simplest and fastest full justification behavior, which spreads the text evenly across the line by increasing the width of the space between words only. The concept of a word is script dependent, although the user agent determines the exact algorithm. At minimum, justification is expected to occur at each white space boundary. No expansion or compression occurs within the words, i.e. no additional letter spacing is created. No kashida effect takes place.

Note: White space does not include zero-width-space, therefore justification is not expected for these characters. However justification is expected for white space with explicit width set by the 'word-spacing' property.

The diagram below illustrates this mode, by showing how the glyphs are laid out in the last two lines of an element:

Mixed glyph layout in the last two lines in an inter-word justified element

For example a viewer could render an 'inter-word' justified paragraph in the following way:

Inter-word justification applied to mixed text

newspaper

Selects the justification behavior in which both inter-word and inter-letter spacing can be expanded or reduced to spread the text across the whole line. Also, text distribution on any given line may depend on the layout or the contents of the previous or the following several lines. This is the significantly slower and more sophisticated type of the full justify behavior preferred in newspaper and magazines, as it is especially useful for narrow columns. For example, typically, compression is tried first. If unsuccessful, expansion occurs: inter-word spaces are expanded up to a threshold, and finally inter-letter expansion is performed. Inter-letter spacing is not applied to Devanagari and other South Asian writing systems using baseline connectors. The threshold value may be related to the ratio of column width to font size. The exact layout algorithm is determined by the user agent. Further explanation about multi-column layout can be found in the CSS3 Multi-column layout module.

The diagram below illustrates this mode:

Mixed glyph layout in the last two lines of a newspaper justified element

Note: In CSS3 a value of 'letter-spacing: 0' no longer strictly inhibits spacing-out of words for justification. The letter-spacing value is just an entry to the letter-spacing process that occurs prior to the possible justification process. Justification may alter the initial spacing between letters, especially with the 'text-justify: newspaper' value.

inter-ideograph

In this mode, letter-spacing modification only occurs for the CJK group. Others only use inter-word expansion. No kashida effect takes place. This is the preferred justification in the context of the Japanese writing system, but not Latin nor Korean.

The diagram below illustrates this mode:

Mixed glyph layout in the last two lines in an inter-ideograph justified element

Below is an example of how this mode would work:

Inter-ideograph justification applied to mixed text

distribute

Like 'newspaper' it allows letter spacing modification for most script groups (except the Devanagari group), but unlike newspaper, it does not prioritize between word spacing and letter spacing, i.e. the space character gets the same letter spacing modification as others. And by consequence there are no variations between narrow and wide columns. This value is best used in East Asian context.

The diagram below illustrates this mode:

Mixed glyph layout in the last two lines of a distribute justified element

For example a viewer could render a 'distribute' justified paragraph in the following way:

Distribute justification applied to mixed text

inter-cluster

This is the Southeast Asian counterpart to 'inter-ideograph'. That is letter spacing only occurs between script-defined grapheme clusters occurring in Southeast Asian scripts.

Note: A grapheme cluster is what a language user consider to be a character or a basic unit of the language. The term is described in detail in the Unicode technical report [UTR-29]: Text Boundaries.

kashida

This is the Arabic counterpart to 'inter-ideograph'. Letter spacing may be increased between Arabic letters, the extra space being filled by kashida. The amount of kashida elongation is controlled by the 'text-kashida-space' property. If 'text-kashida-space' is set to '0%' (which is the initial value), there will be no kashida effect. No letter spacing occurs for other scripts.

The following table describes the expansion/compression strategy for the combination of each script groups and the text-justify property value for each relevant text-justify property value:

*The values shown for the auto column are only a recommendation. The UAs might implement a different strategy.

**The Latin entry represents as well other scripts and writing systems used in Europe and America that use the same typographic convention for justification such as Greek, Cyrillic, etc.

***The Devanagari entry represents as well other scripts and writing systems used in India that use baseline connectors such as  Bengali and Gurmukhi.

4.3. Last line alignment: the 'text-align-last' property

This property describes how the last line of the inline content of a block is aligned. This also applies to the only line of a block if it contains a single line, the line preceding a br element in a XHTML context, or a hard line break in other languages, and to last lines of anonymous blocks. Depending on the value of the property, the last line alignment may also be influenced by the value of  the  'text-align' and 'text-justify' properties. Possible values:

relative

If the 'text-align' property is set to the value 'justify', the last line will be aligned to the start of the inline-progression. If the 'text-align' property is set to any other values, the last line alignment will be determined by the value of the 'text-align' property.

start, end and center

Start, end and center text respectively.

justify

The last line will be justified according to the 'text-justify' property value. However, if there is no expansion opportunity in the last line, the line might not appear justified.

size

The line content is scaled to fit on the line. All the fonts on the line must be scaled by the same factor. Typically this value is used for single line element. Finally, this value, unlike the others, may change (i.e. decrease) the number of lines in a block element.

p.distributealllines
 { text-align: justify;
   text-justify: distribute;
   text-align-last: justify }

4.4. Minimum and maximum font size: the 'min-font-size' and 'max-font-size' property

The two following properties are only used in conjunction with the 'text-align-last' property set to 'size'. They control the font-size adjustments allowed to to fit the line content within the line.

Possible values:

<'font-size'>

The font sizes of the last line of an element are not allowed to become smaller than the smaller of the computed 'font-size' value and the <'font-size'> value set to 'min-font-size'.

auto

The user agent determine the minimum readable font-size for the media. For example, a value of '8px' (relative to the viewing device) is recommended for Latin scripts.

Possible values:

<font-size>

The font sizes of the last line of an element are not allowed to become larger than the larger of the computed 'font-size' value and the value <'font-size'> set to 'max-font-size'.

auto

There is no upper limit to the font sizes of the last line of an element.

4.5. Additional compression: The 'text-justify-trim' property

This sets the individual font blank space compression permissions for the text justification algorithm, when 'text-justify' is anything other than 'inter-word'. This special type of space compression occurs on the font level, i.e. the blank space within the glyphs themselves may be reduced without affecting the appearance of the filled parts of glyphs. This applies to wide-cell glyphs only. Possible values:

none

No wide-cell font space compression is allowed.

Glyph layout with no compression

punctuation

Space can be taken away only from wide-cell punctuation glyphs.

Glyph layout with punctuation compression

punct-and-kana

Space compression is allowed on wide-cell punctuation and wide-cell Kana glyphs.

Glyph layout with punctuation and Kana compression

4.6. Kashida effect: the 'text-kashida-space' property'

Kashida is a typographic effect used in Arabic writing systems that allows glyph elongation at some carefully chosen points. Each elongation can be accomplished using a number of kashida glyphs, a single graphic or character elongation on each side of the kashida point. (The user agent may use either mechanism based on font or system capability). The 'text-kashida-space' property expresses the ratio of the kashida expansion size to the white space expansion size. The value '0%' means no kashida expansion. The value '100%' means kashida expansion only. This property has a visible effect with any justification style where kashida expansion is allowed (currently if the 'text-justify' property is set to: auto, kashida, distribute or newspaper).

In the diagram below showing two identical paragraphs of Arabic text, the blue line in the second line (not justified) shows the length that is used for kashida and divided among the elongation opportunities in the first line (justified), as indicated by the red underlines:

In that example no expansion occurs between the word themselves, indicating that the text-kashida-space property is set to 100%.

5. Text indentation

5.1. Summary

Since CSS1 it has been possible to indent the first line of a block element using the 'text-indent' property. However until now there was no easy way to indent the other lines of the block element. Changing margin or padding values was possible, but while indentation is always on the starting edge of the text, margin and padding are always given in absolute terms (independent from the layout direction of the block). Furthermore, unlike text indentation, the margin and padding effects are not inherited. Because of all of this, creating a hanging indent was not easy as it implied a mix of properties controlling text indent and padding or margin with different inheritance rules.

CSS3 solves the problem by making 'text-indent' a shorthand property of two new properties: 'text-first-indent' which controls the text indent of the first line within a block element and 'text-block-indent' which controls the text-indent of the other lines. To maintain compatibility with prior version of CSS and to facilitate authoring,  'text-block-indent' does not take length values but instead strategy hints. This allows 'text-indent' to be use in a compatible way or in a new way allowing hanging indent without resorting to margin or padding.

5.2. First Line indentation: the 'text-first-indent' property

This property specifies the indentation of the first line of text in a block. More precisely, it specifies the indentation of the first box that flows into the block's first line box. The box is indented with respect to the starting edge of the line box. User agents should render this indentation as blank space. When the 'text-align' property is not set to align the text at the starting edge, this property only specifies a minimum indentation. When the 'text-align' property is set to 'center', the content of the first line is centered within the whole line box with the additional constraint created by the indentation.

Values have the following meanings:

<length>

The indentation is a fixed length.

<percentage>

The indentation is a percentage of the containing block width.

The value of 'text-first-indent' may be negative, but there may be implementation-specific limits.

Note: Since the 'text-first-indent' property inherits, when specified on a block element, it will affect descendent inline-block elements. For this reason, it is often wise to specify 'text-first-indent: 0' (or 'text-indent: 0', see below) on elements that are specified 'display: inline-block'.

5.3. Block Line indentation: the 'text-block-indent' property

This property specifies the indentation of all lines of text in a block except the first line to allow the creation of an 'hanging indent' effect. User agents should render this indentation as blank space. When the 'text-align' property is not set to align the text at the starting edge, this property only specifies a minimum indentation. When the 'text-align' property is set to 'center', the content of the lines is centered within the whole line box with the additional constraint created by the indentation.

Values have the following meanings:

auto

All lines are indented the minimum length necessary to keep the first line inside the starting text edge. If the computed value of

'text-first-indent' is negative, the computed value of 'text-block-indent' is the corresponding positive value. If the computed value of 'text-first-indent' is positive, the computed value of 'text-block-indent' is zero.

For example, if 'text-first-indent' were set to '-2em', the computed value of the 'text-block-indent' would be 2em, but if 'text-first-indent' were set to '2em', the computed value would be zero.

none

There is no indentation for the remaining lines of the block.  The computed value is zero.

5.4. All lines indentation: the 'text-indent' shorthand property

This property specifies the indentation of the lines of text in a block. It sets the value of 'text-first-indent' to the input value and 'text-block-indent' to 'inherit' (note that this is not its initial value).

Note: Since the 'text-indent' property inherits, when specified on a block element, it will affect descendent inline-block elements. For this reason, it is often wise to specify 'text-indent: 0' (or 'text-first-indent: 0', see above) on elements that are specified 'display: inline-block'.

     :root {text-block-indent: auto} 
     p     { text-indent: -3em }

The 'float-displace' property [link TBD from CSS3 box] determines the line wrapping policy in the presence of floats. As such it is influenced by text indentation. When that property is set to the 'line' value, a float may appear within a hanging indent created by text indentation. When the same property is set to the 'content' value, and the 'indent-edge-reset' property [link from CSS3 box] which determines the reference indent edge is set to 'content-edge', the actual reference edge is the starting content edge of the least indented line or lines of the block element.

6. Line breaking

6.1. Types of line breaking

In documents written in Latin-based languages, where runs of characters make up words and words are separated by spaces or hyphens, line breaking is relatively simple. In the most general case, (assuming no hyphenation dictionary is available to the UA), a line break can occur only at white-space characters or hyphens, including U+00AD SOFT HYPHEN.

In ideographic typography, however, where what appears as a single glyph can represent an entire word and no spaces nor any other word separating characters are needed, a line breaking opportunity is not as obvious as a space. It can occur after or before many other characters. Certain line breaking restrictions still apply, but they are not as strict as they are in Latin typography.

Thai is another interesting example with its own special line breaking rules. Since Thai words are made up of runs of characters, it resembles Latin in that respect. But the lack of spaces as word delimiters, or in fact any consistent word delimiters, makes it similar to CJK. Thai, like Latin in the absence of a hyphenating dictionary, never breaks inside of words. In fact, a knowledge of the vocabulary is necessary to be able to correctly break a line of Thai text. To specify an explicit line breaking opportunity, the character U+200B ZERO WIDTH SPACE can be inserted in documents of Thai and similar scripts .

A number of levels of line-breaking strictness can be used in Japanese typography. These levels add or remove line breaking restrictions. The model presented in this specification distinguishes between two most commonly used line breaking levels for Japanese text, using the 'line-break' property.

In ideographic typography, it is also possible, though not always preferred, to allow line breaks to occur inside of quoted Latin and Hangul (Korean) words without following the line breaking rules of those particular scripts. The model proposed in this document gives the author control over that behavior through the 'word-break-cjk' property.

In addition, hyphenation is controlled by 'word-break-inside'.

The 'word-break' shorthand property sets 'word-break-cjk' and 'word-break-inside'.

Finally, there is an additional property 'wrap-option' which may influence line-breaking, especially the property value 'wrap-option: emergency' which provides for emergency word-breaking for long words.

Note: Line breaking is covered by the Unicode Standard Annex [UAX-14], available from the Unicode Web site. It contains a detailed recommendation and corresponding data for each Unicode character. The line breaking data for a character is formally independent from its inherent script value, although both are tightly correlated. Consequently, the 'text-script' property has no influence on line breaking and word breaking processing.  The following properties descriptions use commonly script classification because the classification conveniently describes the specific cases of line breaking and word breaking.

6.2. Line breaking: the 'line-break' property

This property selects the set of line breaking rules to be used for text. The values described below are especially useful to CJK authors, but the property itself is open to other, not yet specified settings for non-CJK authors as well. (This is an area for future expansion.)

normal

Selects the normal line breaking mode for CJK. While the UA is free to define its own line breaking restrictions for the 'normal' mode, it is recommended that breaks between a standard katakana or hiragana character and a small katakana or hiragana (respectively) character be allowed. That is the preference in modern Japanese typography, and is especially desirable for narrow columns. Japanese kana words may be long, and it is preferable to allow line breaks to occur among such characters than to have excessive expansion due to justification.

strict

Selects a more restrictive line breaking mode for CJK text. While the UA is free to define its own line breaking restrictions for the 'strict' mode, it is recommended that the restrictions specified by the Unicode Standard Annex [UAX-14] be followed. That implies that in this mode, small katakana and hiragana characters are not allowed to start a line if they follow a standard katakana or hiragana character.

Note: In Japanese, a set of line breaking restrictions is referred to as "Kinsoku". JIS X-4051 [JIS-X-4051] is a popular source of reference for this behavior using the strict set of rules. The rules described by JIS X-4051 have been superseded by the Unicode Technical Report #14.

Note: Both values: 'normal' and 'strict' imply that a set of line-breaking restrictions is in use.

6.3. Word breaking: the 'word-break-cjk', 'word-break-inside' properties and the shorthand 'word-break' property

This property controls line-breaking behavior inside of words from a CJK point of view. Possible values:

normal

Keeps non-CJK scripts together (according to their own rules), while Hangul and CJK ideographs (including the Korean Hanja characters) break according to the rules of the 'line-break' mode. However, the behavior of non-CJK scripts can be superseded by using the value 'emergency' in the 'wrap-option' property, or the value 'hyphenate' in the 'word-break-inside' property.

break-all

Same as 'normal' for CJK ideographs and Hangul, but non-CJK scripts can break anywhere. This option is used mostly in a context where the text is predominantly using CJK characters with few non-CJK excerpts and it is desired that the text be better distributed on each line. The UAs may however limit the break everywhere behavior for script using clusters such as Thai.

keep-all

Same as 'normal' for all non-CJK scripts. CJK ideographs and Hangul are kept together. This option should only be used in the context of CJK ideographs used in small clusters like in the Korean writing system.

p.anywordbreaks { word-break: break-all }

This property controls the hyphenation behavior inside of words. Possible values:

normal

A word should always stay in a single line. However, this can be superseded by using the value 'break-all' in the 'word-break-cjk' property, or the value 'emergency' in the 'wrap-option' property. Moreover, explicit hyphenation characters (hyphen, soft hyphen, etc...) still create line breaking opportunities.

hyphenate

Words can be broken at an appropriate hyphenation point. It requires that the user agent have an hyphenation dictionary for the language of the text being broken. Setting this value activates the hyphenation engine in the user agent.

Note: Intra-word breaks may or may not be indicated by a visible hyphen, depending on the language. The hyphenation glyph may appear at the end of the line or at the start of the next line, and its actual shape may depend on the text language.

The 'word-break' property is a shorthand property for setting 'word-break-cjk', and 'word-break-inside', at the same place in the style sheet.

The properties 'word-break-cjk' and 'word-break-inside' are first reset to their initial values (all 'normal'). Then, those properties that are given explicit values in the 'word-break' shorthand are set to those values.

7. Text Wrapping, White-space Control and Text Overflow

The following section describes text wrapping, white-space handling and text overflow. Text wrapping and white-space handling are interrelated through the CSS2 'white-space' property combining these two effects together. Text wrapping and text overflow both deal with situation where the text reaches the flow after-edge of its containing box.

CSS3 clearly separates these three effects in different sets of property while keeping the 'white-space' property for compatibility reason.

The following section frequently use the term line feed character to specify the normalized newline indicator. In XML and HTML context, the line feed character is the LINE FEED (U+000A). In other contexts, it may be represented differently, for example by a CARRIAGE RETURN (U+000A). The term 'line feed character' represents the normalized newline character native to a given framework.

7.1. Text wrapping: the 'wrap-option' property

This property controls whether or not text wraps when it reaches the flow edge of its containing block box. Several value descriptions use the term preserved line feed characters. A preserved line-feed character (either from the source content or from occurrence of "\A" in generated content) is maintained for presentation purpose and may therefore influence text wrapping. The preserved status of line-feed characters is determined by the 'linefeed-treatment' property. The 'wrap-option' possible values are:

wrap

The text is wrapped at the best line-breaking opportunity (if required) within the available block inline-progression dimension (block width in horizontal text flow). The best line-breaking opportunity is determined in priority by the existence of preserved line-feed characters, or by the line-breaking algorithm controlled by the 'line-break' and word-break' properties.

no-wrap

The text is only wrapped where explicitly specified by preserved line feed characters. In the case when lines are longer than the available block width, the overflow will be treated in accordance with the 'overflow' property specified in the element.

soft-wrap

The text is wrapped after the last character which can fit before the ending-edge of the line and where explicitly specified by preserved line feed characters. No line-breaking algorithm is invoked. The intended usage is the rendering of a character terminal emulation.

emergency

The text is wrapped like for the 'wrap' case, except that the line-breaking algorithm will allow as a last resort option a text wrap after the last character which can fit before the ending edge of the line box, independently of 'line-break', 'word-break-cjk' and 'word-break-inside' properties. For example, this addresses the situation of very long words constrained in a fixed-width container with no scrolling allowed.

7.2. White-space control: the 'linefeed-treatment', 'white-space-treatment', 'all-space-treatment' properties and the 'white-space' shorthand property

White-space processing in the context of CSS is the mechanism by which all white-space characters are interpreted for rendering purpose. The white-space set is determined by the XML [XML1.0] specification as being a combination of one or more space characters (Unicode value U+0020), carriage returns (U+000D), line feeds (U+000A), or tabs (U+0009).

Note: [HTML401] also defines the form feed character (U+000C) as a white space character, but that character is not part of any XHTML versions as they are all based on XML.

The amount of white space processing that can be achieved by a user agent that supports CSS is directly related to the CSS processing model, especially the document parsing and validation. After parsing and possible validation, the document tree may contain text nodes that contain unprocessed white space characters, or the document tree may already have been processed in a way that white space characters have been collapsed and partially removed (white space normalization).

In that respect, the CSS properties related to white space processing can only be effective if the CSS processor has access to the white space characters that were originally encoded in the document. However, end-of-line characters are typically handled (like by XML processors) in such a way that any arbitrary combination of end-of-line characters is replaced by a single line feed character.

Note: The first version of XML [XML1.0] only normalizes two characters sequences of (U+000D U+000A) or any U+000D not followed by U+000A to a single U+000A. The forthcoming version of XML [XML1.1] adds U+0085 (NEL) and U+2028 (LINE SEPARATOR) to the linefeed normalization process. However the set of white-space characters is unchanged. Notably, the character U+2029 (PARAGRAPH SEPARATOR) is not part of that set. If the characters U+2028 and U+2029 appears in text, they are treated as zero-width characters without semantic meaning.

Note: XML Schema, through its 'whiteSpace' facet can constrain exactly the type of white space characters still available to a rendering process like CSS for elements containing string datatype. In addition, some XML languages like [XHTML1.0] may have their own white-space processing rules when parsing and validating documents with white-space characters. Therefore, some of the behaviors described below may be affected by these limitations and may be user agent dependent in these contexts.

In addition, line feed characters can be inserted in generated content by using the '\A' string. The behavior of these inserted line feed characters is identical to original line feed characters part of the source document and is controlled by the same set of properties.

White-space processing

Any text that is directly contained inside a block (not inside an inline) should be treated as an anonymous inline element.

For each inline (including anonymous inlines), the following steps are performed, ignoring bidirectional formatting characters as if they were not there:

1. Each non-linefeed white-space character is treated as per the 'white-space-treatment' property.
2. If 'all-space-treatment' is set to 'preserve', any sequence of spaces (U+0020) unbroken by an element boundary is treated as a sequence of non-breaking spaces. However, a line breaking opportunity exists at the end of the sequence.
3. Each linefeed character is treated as per the 'linefeed-treatment' property.
4. If 'all-space-treatment' is set to 'collapse',
   1. every tab (U+0009) is converted to a space (U+0020)
   2. any space (U+0020) following another space (U+0020)--even a space before the inline, if that space also has 'all-space-treatment' set to collapse--is removed.

Then, the entire block is rendered. Inlines are laid out, taking bidirectional reordering into account, and wrapping as specified by the 'wrap-option', 'line-break' and 'word-break' properties.

As each line is laid out,

1. If a space (U+0020) at the beginning of a line has all-space-treatment' set to 'collapse', it is removed.
2. All tabs (U+0009) are rendered as a horizontal shift that lines up the start edge of the next glyph with the next tab stop. Tab stops occur at points that are multiples of 8 times the width of a space (U+0020) rendered in the block's font from the block's starting content edge.
3. If a space (U+0020) at the end of a line has 'all-space-treatment' set to 'collapse', it is also removed.

Note: Tab stops line up in the block regardless of font change.

These rendering rules make no assumption about the storage model of these white-space character sequences. It is outside the scope of CSS to determine the character code values accessible through programming interface such as DOM. These rules do not apply to elements that have an explicit white-space rendering behavior (like the pre element in XHTML).

When white-space characters are collapsed for rendering purpose, the text decoration style applied to the collapsed set is the one that would be applied to the first white-space character of the original sequence.

The 'white-space' property is a shorthand property for 'linefeed-treatment', 'white-space-treatment', 'all-space-treatment' and 'wrap-option'.

This property specifies the treatment of linefeed characters for rendering purpose. Values have the following meanings:

auto

The user agent either transforms each line feed character to a space character (U+0020), transforms each line feed character to a zero width space character (U+200B), or removes the line feed characters, following the line feed conversion algorthim. The choice of the resulting character is conditioned by the script value of the characters preceding and following the line feed character which are part of the same inline text flow in the same block element. The script value of each character is determined by the 'text-script' property.

ignore

Linefeed characters are ignored. They are removed and are not rendered.

preserve

Linefeed characters indicate an end of line of boundary.

treat-as-space

Linefeed characters are transformed for rendering purpose into a space character (U+0020). The result of the transformation can be treated by subsequent CSS processing (including white space collapsing).

treat-as-zero-width-space

Linefeed characters are transformed for rendering purpose into a zero width space character (U+200B). The result of the transformation can be treated by subsequent CSS processing (including white space collapsing).

ignore-if-after-linefeed

Specifies that any linefeed characters that immediately follows a linefeed character, shall be discarded. This collapses multiple consecutive linefeed characters into a single linefeed.

Note: The Unicode Standard [UNICODE] specifies that the zero width space is considered a valid line-break point and that if two characters with a zero width space in between are placed on the same line they are placed with no space between them; and that if they are placed on two lines no additional glyph area, such as for a hyphen, is created at the line-break.

Line feed conversion algorithm

This algorithm is used when 'linefeed-treatment' is set to 'auto'. In determining how to convert a line feed character, a user agent should consider the following cases, whereby the scripts value of characters preceding and following the line feed determine the choice of the replacement. The script value of each character is determined by the 'text-script' property. (Note that if 'text-script' is set to 'auto', the determination is done character by character within the element, otherwise all characters share the same script value within the element.) Characters of COMMON script (such as punctuation) are treated as the same as the script on the other side:

1. If the characters preceding and following the line feed character have a script value in which the space character (U+0020) is used as a word separator, the line feed character should be converted into a space character. Examples of such scripts are Latin, Greek, and Cyrillic.

2. If the characters preceding and following the line feed character have either a ideographic-based script value or a script value which make them part of an ideographic-based writing system in which there is no word separator, the line feed should be converted into no character. Examples of such scripts or writing systems are Chinese, Japanese.

3. If the characters preceding and following the line feed character have a non ideographic-based script vale in which there is no word separator, the line feed should be converted into a zero width space character (U+200B) or no character. Examples of such scripts are Thai, Khmer.

4. If none of the conditions in (1) through (3) are true, the line feed character should be converted into a space character (U+0020).

This property specifies the treatment for rendering purpose of the space character (U+0020) and other white-space characters (except for linefeed characters, since their treatment is determined by the 'linefeed-treatment' property). White-space characters, when rendered as an advance width, use the width of the empty glyph normally used for the space character (U+0020). Values have the following meanings:

ignore

White-space characters, except for linefeed characters, are ignored. They are removed and are not rendered.

preserve

All white-space characters other than line feed are rendered as they are (with advance width).

ignore-if-before-linefeed

Specifies that any white-space characters, except for linefeeds, that immediately precedes a linefeed character, shall be discarded. This action shall take place regardless of the setting of the linefeed-treatment property.

ignore-if-after-linefeed

Specifies that any white space characters, except for linefeeds, that immediately follows a linefeed character, shall be discarded. This action shall take place regardless of the setting of the linefeed-treatment property.

ignore-if-surrounding-linefeed

Specifies that any white space characters, except for linefeeds, that immediately precedes or follows a linefeed character, shall be discarded. This action shall take place regardless of the setting of the linefeed-treatment property.

The 'all-space-treatment' property specifies the treatment of all consecutive white-space characters (with no exception for linefeed characters, unlike the 'white-space-treatment' property). Values have the following meanings:

preserve

All white-space characters are rendered as they are. The rendering of tab characters (U+0009) is described in the white-space processing section.

collapse

Specifies that a character is not rendered if:

• it is a white-space character (according to XML), and
• it is not a preserved linefeed (due to 'linefeed-treatment: preserve'), and
• the immediately preceding (non-ignored) character is a white-space character or the immediately following (non-ignored) character is a preserved linefeed.

This property declares how 'white-space' inside the element is handled. Setting a value on the 'white-space' property sets the respective values on 'wrap-option', 'linefeed-treatment', 'white-space-treatment' and 'all-space-treatment'. Although, strictly speaking, the property has no initial value, it is equivalent to 'normal'. The definition of the property values are established by the following table, which shows the settings of the constituent properties.

pre        { white-space: pre }
p          { white-space: normal }
td[nowrap] { white-space: nowrap }
:before, :after {white-space: pre-line }

pre[wrap]  {white-space: pre-wrap }

7.3. Text overflow: the 'text-overflow-mode', 'text-overflow-ellipsis' properties and the shorthand 'text-overflow' property

Text overflow deals with the situation where some textual content is clipped when it overflows the element's box in its inline-progression direction as determined by the 'writing-mode' property value. This situation only occurs when the 'overflow' property has the values: hidden, scroll or auto (in the latter case only when the user agent introduces a scrolling mechanism).

Text overflow allows the author to introduce a visual hint at the two ending boundaries of the text flow within the element box (after and end). The hint is typically a horizontal ellipsis character (U+2026), although the hint may be some other string or even an image. Setting a non empty string (or an uri for an image) for either text flow boundary enables the presentation of the hint. If both hints are enabled, only the 'after' hint is rendered. Initially, only the end of line hint is shown (correspond to the right of any over flown lines for left to right inline-progression).

Control over text-overflow is divided among properties: 'text-overflow-mode' controls the presence and position of the hint, 'text-overflow-ellipsis' controls what constitutes the hint. The shorthand property 'text-overflow' sets the other text flow properties.

clip

Clip text as appropriate for the text content. Glyphs representation of the text may be only partially rendered.

ellipsis

A visual hint is inserted at each box boundary where text overflow occurs. The 'text-overflow-ellipsis' property determines the content of the hint. The insertions take place after the the last letter that entirely fits on the line.

ellipsis-word

A visual hint is inserted at each box boundary where text overflow occurs. The 'text-overflow-ellipsis' property determines the content of the hint. The insertions take place after the last word that entirely fits on the line.

The overflow hints are active only for textual content. That is, the user agent must not render an overflow hint when only replaced content overflows.

Although the property is not inherited, overflowing children blocks that are either statically or relatively positioned and do not have a specified width or height will be hinted as specified by their parent text-overflow-mode property value.

<blockquote>
<p class="sentence"><span class="nowrap">I didn't like the play,</span> but then I saw
it under adverse conditions - the curtain was up.
<div class="attributed-to">_Groucho_Marx_</div>
</p>
</blockquote>

blockquote        { width:100px; border: thin solid red; overflow: hidden;
                    text-overflow-mode:ellipsis;font-size:14px }
span.nowrap       { white-space : nowrap; }
div.attributed-to { position: relative;left:8px }

p.sentence { width :100px; margin-top : 50px; margin-left : 50px; }

The <ellipsis> value is defined as: [<string> | <uri>]. The <string> value determines a textual value for the overflow visual hint. The font-size used for the <string> is the element computed font-size. An empty string disables the hint. The <uri> value determines an image to be used for the overflow visual hint.

When only one <ellipsis> value is set by 'text-overflow-ellipsis', it determines the overflow visual hint at the end of the element box. If two <ellipsis> values area provided, they determine the overflow visual hint at the end and the overflow visual hint after the element box respectively. The visual hint after the element box only appear if there is content which is clipped because of the block-progression dimension of the block, not because the last line cannot fit. If the visual hint after the element box is enabled and would appear at the same location as the visual hint at the end of the element box, only the visual hint after the element box is rendered.

Note: Because the initial value (U+2026) of the overflow visual hint after the element box may not be easily rendered in some situations, the user agent may replace it by a sequence of 3 FULL STOP characters (U+002E).

This property is the shorthand for 'text-overflow-mode' and 'text-overflow-ellipsis'.

8. Text spacing

8.1. Letter spacing: the 'letter-spacing' property

This property specifies spacing behavior between grapheme clusters.

Note: A grapheme cluster is what a language user consider to be a character or a basic unit of the language. The term is described in detail in the Unicode technical report [UTR-29]: Text Boundaries.

Values have the following meanings:

normal

The spacing is the normal spacing for the current font. It is typically zero-length. However, this value allows the user agent to alter the space between grapheme clusters in order to justify text.

<length>

This value indicates spacing added between grapheme clusters in addition to the default spacing between grapheme clusters. For justification purposes, user agents should minimize effect on letter-spacing as much as possible (priority to word-spacing expansion/compression as opposed to character-spacing expansion/compression).

Values may be negative, but there may be implementation-specific limits. For justification purposes, user agents should minimize alteration of spacing within words. The priority should be to alter spacing between words.

Because of the visually disruptive effect of modifying this spacing in writing systems, such as Arabic, which use joined glyphs, the usage of this property is discouraged in those cases.

There are cases, like in Japanese or Chinese writing systems, where justification will change all spacing between grapheme clusters, as there is no other opportunity in the line to expand or compress the textual content in order to fit the line.

The user agent determines the exact algorithm for spacing between grapheme clusters. Furthermore this property should not be set to a <length> for scripts and/or fonts that ligate glyphs with connecting strokes; such scripts and fonts include cursive Latin fonts, Arabic, Devanagari. Spacing between grapheme clusters may also be influenced by justification (see the 'text-align' property).

blockquote { letter-spacing: 0.1em }

blockquote { letter-spacing: 0cm }   /* Same as '0' */

When the resultant spacing is not the default, user agents should consider avoiding the use of ligatures. (There are cases such as Arabic where ligatures may still be used)

8.2. Word spacing: the 'word-spacing' property

This property specifies spacing behavior between words. Values have the following meanings:

normal

The normal inter-word space, as defined by the current font and/or the user agent. If the inter-word boundary is delimited by one or several white-space characters, they should be visible. If there are no word-separating characters, the user agent doesn't have to create an additional character advance width between words.

<length>

This value indicates inter-word space in addition to the default space between words. Values may be negative, but there may be implementation-specific limits.

Word spacing algorithms are user agent-dependent. Determining word boundary is typically done by detecting white space characters. There are however many scripts and writing systems that do not separate their words by any character (like Japanese, Chinese, Thai, etc...), detecting word boundaries in these cases require dictionary based algorithms that may not be supported by all user agents. Word spacing is also influenced by justification (see the 'text-align' property).

h1 { word-spacing: 1em }

8.3. Punctuation trimming: the 'punctuation-trim' property

This property determines whether or not a full-width punctuation mark character should be trimmed if it appears at the beginning of a line, so that its "ink" lines up with the first glyph in the line above and below. In some scenarios, it may be preferable for the author not to allow leading punctuation marks to be trimmed, for example when it is more important that the glyphs tend to line up vertically. In other scenarios such an effect is desirable, for example when it is more important for the author that as much text as possible fits on a single line.

Possible values:

none

Leading punctuation is not trimmed.

Glyph layout with no leading punctuation compression

start

Leading punctuation is trimmed.

Glyph layout with leading punctuation compression

Note: This property may in the future be expanded to cover other punctuation behaviors for other types of punctuation as well, not just wide-cell.)

8.4. Adding space: the 'text-autospace' property

When a run of non-ideographic or numeric characters appears inside of ideographic text, a certain amount of space is often preferred on both sides of the non-ideographic text to separate it from the surrounding ideographic glyphs. This property controls the creation of that space when rendering the text. That added width does not correspond to the insertion of additional space characters, but instead to the width increment of existing glyphs.

(A commonly used algorithm for determining this behavior is specified in JIS X-4051 [JIS-X-4051].)

This property is additive with the 'word-spacing' and 'letter-spacing' [CSS2] properties. That is, the amount of spacing contributed by the 'letter-spacing' setting (if any) is added to the spacing created by 'text-autospace'. The same applies to 'word-spacing'.

Possible values:

none

No extra space is created.

ideograph-numeric

Creates extra spacing between runs of ideographic text and numeric glyphs.

ideograph-alpha

Creates extra spacing between runs of ideographic text and non-ideographic text, such as Latin-based, Cyrillic, Greek, Arabic or Hebrew.

ideograph-space

Extends the width of the space character while surrounded by ideographs.

ideograph-parenthesis

Creates extra spacing between normal (non wide) parenthesis and ideographs.

span.autospace { text-autospace:none; }

<span class="autospace">[ideographs]1997[ideographs]</span>

span.autospace { text-autospace:ideograph-numeric; }

8.5. Text kerning: the 'kerning-mode' and 'kerning-pair-threshold' property

This property controls all kerning effects. Some kerning effects are based on information located explicitly on fonts (pair-kerning). Others are based on context and typical ink placement within characters and don't rely on font information. Pair kerning is used mainly for Latin, Greek and Cyrillic scripts, while contextual kerning is more common in East Asian context. A typical example of pair kerning is the pair 'Wa'. A good example of contextual kerning is the pair '[[' (when using the wide width variant). Possible values:

none

no kerning is enabled

pair

enables pair kerning

contextual

enables contextual kerning

This property controls the font size threshold for the pair kerning effect. At the threshold value and above, pair kerning is active if enabled by the 'kerning-mode' property.  Because pair kerning is not very effective at small font size, this property can be used to adjust the threshold.

Possible values:

auto

The user agent determines the font size threshold from which the kerning takes place.

<'font-size'>

<'font-size'> threshold.

9. Text decoration

9.1. Introduction

In CSS1 and CSS2, the 'text-decoration' property controls text decorations: underline, overline, line-through, and blinking. CSS2 adds text shadows, which the 'text-shadow' property controls. However, the 'text-decoration' property has limitations stemming from its syntax, precluding fine control over each of those formatting effects. Specifically, it offers no way to control the color or line style of the underline, overline or line-through.

CSS3 extends the model by introducing new properties allowing additional controls over those formatting effects. CSS3 also makes turning these formatting effects on or off possible without affecting any other 'text-decoration' settings.

Furthermore, to reflect the usage of underline in East Asian vertical writing, a new control is offered for the underline positioning; this allows the underline to appear before (on the right in vertical text flow) or after (on the left in vertical text flow) the formatted text. The property is called 'text-underline-position'.

Note: In East Asian typography it is typical to 'emphasize' text using glyph elements such as an accent mark, a dot, a hollow circle, etc. This feature is tightly connected to the rendered font and is therefore described in the font module. See 'font-emphasis' [Link TBD to the CS3 fonts module] for more details.

The 'text-decoration' property itself is now a shorthand property for all these new properties. However, the 'text-decoration' values are not composites built from the values of constituent properties. Rather, 'text-decoration' values come from a small set specific to the 'text-decoration' shorthand property.

These properties describe decorations that are added to the text of an element. If they are specified for a block-level element, they affect the root inline box (the anonymous inline box which wraps all the inline children of an element).

If they are specified for (or affects) an inline-level element, they affect all boxes generated by the element. If an element is empty or is a replaced element (e.g., img element in XHTML), user agents must refrain from rendering text decorations on the element. In deciding whether an element is empty, user agents must ignore white-space characters that are collapsed during the white-space processing.

Text decoration properties are not inherited, but descendant boxes must be formatted with the same decoration (e.g., they must all be underlined). The color of decorations must remain the same even if descendant elements have different 'color' values.

When a text decoration could result in text being unintentionally overdrawn by the text decoration style, the user agents may chose to skip the text decoration over the intersecting area by using 'text-underline-mode', 'text-line-through-mode' or 'text-overline-mode' with the appropriate values.

In determining the position of and thickness of text decoration lines, user agents may consider the font sizes of and dominant baselines of children. Of course, user agents may ignore children in these determinations. Such an averaging is done on a line per line basis.

Note: User agents typically consider superscript segments but ignore subscript segments. The baseline-shift is not considered for superscript segments.

9.2. Text decoration style: the 'text-underline-style', 'text-line-through-style' and 'text-overline-style' properties

These properties specifies the line style for underline, line-through and overline text decoration. Possible values:

none

Produces no line.

solid

Produces a solid line.

double

Produces a double line.

dotted

Produces a dotted line.

dashed

Produces a dashed line style.

dot-dash

Produces a line whose repeating pattern is a dot followed by a dash.

dot-dot-dash

Produces a line whose repeating pattern is two dots followed by a dash. 

wave

Produces a wavy line.

9.3. Text decoration width: the 'text-underline-width', 'text-line-through-width' and 'text-overline-width' properties

These property specifies the line width for the underline, line-through and overline text decorations. Possible values:

auto

The user agent may use any algorithm to determine the text decoration width. Implementers are encouraged to consider the font-sizes and baselines of all affected text. The text decoration style should affect the resulting value (the style 'double', for instance, would produce a width greater than the style 'solid' produces). 

normal

The text decoration width is the normal text decoration width for the nominal font. If no font characteristic exists for the width of the text decoration in question, the user agent should proceed as though 'auto' were specified.

<number>

The text decoration width is the product of the <number> and the computed 'font-size'. If inherited, the inherited value is the <number> rather than the computed <length>.

<length>

The text decoration width is the length

thin

This value is equivalent to a <number> value kept by the user agent. The <number> must be constant through a given view of a document and should yield a thin line.

medium

This value is equivalent to a <number> value kept by the user agent. The <number> must be constant through a given view of a document, must be greater than or equal to the 'thin' number and should yield a medium line.

thick

This value is equivalent to a <number> value kept by the user agent. The <number> must be constant through a given view of a document, must be greater than or equal to the 'medium' number and should yield a thick line.

9.4. Text decoration color: the 'text-underline-color', 'text-line-through-color' and 'text-overline-color' properties

These property specifies the line colors for the underline, line-through and overline text decorations. Possible values:

auto

The color of the text decoration is determined by the 'color' property.  This is the computed value.

<color>

Specifies a color value.

9.5. Text decoration mode: the 'text-underline-mode', 'text-line-through-mode' and 'text-overline-mode' properties

These properties set the mode for the underline, line-through and overline text decorations, determining whether the text decoration affects the space characters or not. 'Space characters' are all characters classified by the Unicode Standard [UNICODE] as category 'Zs', in addition to the white-space characters. Possible values:

continuous

This value means that the line is continuous.

skip-space

This means that space characters will not be lined.

skip-glyph

This means that portion of the character glyphs that would collide with the line will not be lined.

skip-glyph-and-space

This means that space characters and and portion of the character glyphs that would collide with the line will not be lined.

9.6. Other text decoration simple properties: 'text-underline-position' and 'text-blink'

This property sets the position of the underline when set through the 'text-underline-style' property. It can appear either 'before' (above in an horizontal flow) or after (below in an horizontal flow) the run of text in relation to its baseline orientation. This property is typically used in vertical writing context where it may be desired to have the underline appear 'before' the run of text. This results in having the underline appearing on the right side of the vertical writing column. Possible values:

auto

The user agent may use any algorithm to determine the underline position. The following algorithm is recommended:

In horizontal inline-progression, the underline, if set ,should be aligned with the alphabetic baseline. In vertical inline-progression, if the language is set to Japanese or Korean, the underline, if set, should be aligned with the 'before-edge' of the line box.

before-edge

the underline, if set, is aligned with the 'before-edge' of the line box.

alphabetic

the underline, if set, will appear after the alphabetic baseline. In this case the underline may cross some descenders.

after-edge

the underline, if set, is aligned with the 'after-edge' baseline of the line box. In this case the underline does not cross the descenders. This is sometimes called 'accounting' underline.

This property specifies the blink mode. Possible values:

none

Produces no blinking.

blink

Blinks.

Conforming CSS3 user agents may simply not blink the text.

9.7. Text decoration shorthand properties: 'text-underline', 'text-line-through', 'text-overline' and 'text-decoration'

The 'text-underline' property is the shorthand for 'text-underline-style', 'text-underline-width', 'text-underline-color', 'text-underline-mode' and 'text-underline-position'. Any constituent properties not explicitly assigned values in a 'text-underline' declaration take their respective initial values.

The 'line-through' property is the shorthand for 'text-line-through-style', 'text-line-through-width', 'text-line-through-color' and 'text-line-through-mode'. Any constituent properties not explicitly assigned values in a 'text-line-through' declaration take their respective initial values.

The 'text-overline' property is the shorthand for 'text-overline-style', 'text-overline-width', 'text-overline-color' and 'text-overline-mode'. Any constituent properties not explicitly assigned values in a 'text-overline' declaration take their respective initial values.

The 'text-decoration' property is a shorthand that takes its own set of values.

The 'text-decoration' shorthand sets the properties 'text-underline-style', 'text-underline-width', 'text-underline-color', 'text-underline-mode', 'text-underline-position', 'text-line-through-style', 'text-line-through-width', 'text-line-through-color', 'text-line-through-mode', 'text-overline-style', 'text-overline-width', 'text-overline-color', 'text-overline-mode' and 'text-blink' to their initial values, and then the values of the shorthand change these values as follows:

none

All the  constituent properties maintain their initial values.

underline

The 'text-underline-style' value is 'solid' rather than 'none'.

overline

The 'text-overline-style' value is 'solid' rather than 'none'.

line-through

The 'text-line-through-style' value is 'solid' rather than 'none'.

blink

The 'text-blink' value is 'blink' rather than 'none'.

Conforming user agents may simply not blink the text.

Although the property has no explicit initial value, it is equivalent to the value 'none.

a[href] { text-decoration: underline blink }

9.8. Text shadows: the 'text-shadow' property

<shadow> itself is defined as "[<color>  <length> <length> <length>? | <length> <length> <length>? <color>?]".

This property accepts a comma-separated list of shadow effects to be applied to the text of the element. The shadow effects are applied in the order specified and may thus overlay each other, but they will never overlay the text itself. Shadow effects do not alter the size of a box, but may extend beyond its boundaries. The stack level of the shadow effects is the same as for the element itself.

Each shadow effect must specify a shadow offset and may optionally specify a blur radius and a shadow color.

A shadow offset is specified with two <length> terms that indicate the distance from the text. The first length term specifies the horizontal distance to the right of the text. A negative horizontal length term places the shadow to the left of the text. The second length term specifies the vertical distance below the text. A negative vertical length term places the shadow above the text.

A blur radius may be specified after the shadow offset. The blur radius is a length value that indicates the boundaries of the blur effect. The exact algorithm for computing the blur effect is not specified. If no blur radius is specified, the treatment is as if a blur radius of zero were specified and the shadow has the same size and shape as the glyphs that cast it. User agents may only implement only part of this property by ignoring blur effects. Such user agents should consider declarations that specify the blur radius to be parser errors, as described in the Syntax module [link TBD].

A color term may be specified before or after the length terms of the shadow effect. The color term will be used as the basis for the shadow effect. If no color is specified, the value of the 'color' property will be used instead.

Text shadows may be used with the ::first-letter and ::first-line pseudo-elements.

h1 { text-shadow: 0.2em 0.2em }

h2 { text-shadow: 3px 3px 5px red }

h2 { text-shadow: 3px 3px red, yellow -3px 3px 2px, 3px -3px }

span.glow {
    background: white;
    color: white;
    text-shadow: black 0px 0px 5px;
}

Note: This property is not defined in CSS1. Some shadow effects (such as the one in the last example) may render text invisible in user agents that only support CSS1. The usage of this property is discouraged as much better filter effects are available in SVG [SVG1.0].

10. Document grid

10.1. What is document grid?

It is very common for the glyphs in documents written in East Asian languages, such as Chinese or Japanese, to be laid out on the page according to a specified one- or two-dimensional grid. The concept of grid can also be used in other, non-ideographic contexts such as Braille or monospaced layout.

The diagram below represents a fragment of horizontal text on a page with mixed wide-cell and narrow-cell glyphs that a Japanese user intended to be laid out on a grid which resulted in 9 glyphs per line (gray grid lines shown for clarity):

The grid behavior can be set on the inline-progression, on the block-progression or both. The grid on the block-progression dimension is determined by the following properties:

• 'line-stacking-strategy' [link to CSS3 line module] set to the 'grid-height' value,
• 'line-height' [link to CSS3 line module] of the block element set to the desired grid block-progression value,
• 'text-height' [link to CSS3 line module] which defines the block-progression value for inline boxes. Depending on the content size, one or a multiple of 'line-height' will be necessary to accommodate a given inline box.

The block-progression grid is not described in this section as it can be achieved simply by using the properties mentioned above and described in the CSS3 Line module.

The grid on the inline-progression dimension is obtained by altering the glyphs advance width (or inline-progression value) of inline elements.  There are several modes:

• Adding extra space between each grapheme clusters by using the 'letter-spacing' property. This is sometimes called "loose" grid.
• Giving a fixed advance width to ideograph grapheme clusters only. Other grapheme clusters are spaced normally. This is called "genko" in Japanese typography.
• Giving a fixed advance width to all grapheme clusters, excepted connected glyphs (like Arabic).

Two properties control this advance width modification: 'line-grid-mode' enables it and 'line-grid-progression' determines its value. The shorthand 'line-grid' allows setting both together.

10.2. Line grid mode: the 'line-grid-mode' property

Specifies the line grid behavior. Each line grid mode value entails a different set of rules for rendering inline contents (the term 'horizontal' is used in the following description to indicate the inline-progression dimension). Possible values:

none

No line grid. Standard text alignments apply to the block element.

ideograph

Content is divided into units that we will call strips. Each strip is horizontally centered within the smallest number of grid spaces that contain the strip. The width of the grid space is determined by the 'line-grid-progression' setting.

Each grapheme cluster with a wide base character is a strip. Each grapheme cluster with a narrow kana character as its base is a strip. Each non-breakable object (e.g. an image) is a strip. Other grapheme clusters are treated as a single strip bounded by the strips described prior. That single strip may be decomposed in several strips if line breaking occurs within it.

The strip are arranged in the grid as follows:

• Each strip corresponding to wide base character (as well as narrow kana character) that can fit within a single grid space is rendered in the horizontal center of the grid space.
• Each strip corresponding to other characters are placed in the center of the smallest number of grid spaces necessary for it to fit. If a line break occurs within such a strip, the strip is treated as two or more separate strips whose individual placement follows the same rules as those for a single strip.
• Strip corresponding to non-breakable objects and strip corresponding to wide base character (as well as narrow kana character) that are wider than a single grid space, are each centered within the smallest number of grid cells necessary for them to fit.

Mixed glyph layout in strict grid

Object layout in strict grid

The 'ideograph' mode disables all special text justification and glyph width adjustment normally applied to the contents of the block element.

If a line break opportunity cannot be found in a text run going over the line boundary, then that text run will be pushed down to the next line and the last part of the previous line will be left blank.

Here is an example of mixed text in 'ideograph' grid mode:

Strict grid applied to mixed text

all

This type of grid can be used to achieve mono-spaced layout. As with 'ideograph', content is divided into strips and each strip is horizontally centered within the smallest number of grid spaces that can contain the grid. The rules for determining strips differs.

Each grapheme cluster with a non-joining base character is a strip. Each non-breakable object (e.g. an image) is a strip. Each run of grapheme clusters with joining base characters that join to each other is a strip.

Mixed glyph layout in fixed grid

For example:

Fixed grid applied to mixed text

The 'letter-spacing' property does not apply to characters in a grid but does apply for all characters not in the grid (i.e. all characters for 'line-grid-mode: none', non-ideographs for 'line-grid-mode: ideograph' and all non-connected glyphs for 'line-grid-mode: all').

10.3. Line grid progression: the 'line-grid-progression' property

This property affects the inline-progression dimension of characters which are subject to the fixed advance width as determined by the 'line-grid-mode' property.

Possible values:

text-height

The computed value of the block element 'text-height' [link to CSS3 line module] is used.

line-height

The computed value of the block element 'line-height' [link to CSS3 line module] is used.

<length>

inline-progression dimension of the line grid's unit space.

div.section1 { line-grid-progression: .5in }

10.4. Line grid: the 'line-grid' shorthand property

The 'line-grid' property is a shorthand property for setting 'line-grid-mode' and 'line-grid-progression'.

div.grid { line-height:20pt;
           text-height: max-size;
           line-stacking-strategy: grid-height;
           line-grid: ideograph line-height; }

11. Miscellaneous text formatting

11.1. Capitalization: the 'text-transform' property

This property controls capitalization effects of an element's text. Values have the following meanings:

capitalize

Puts the first letter of each word in uppercase.

uppercase

Puts all characters of each word in uppercase.

lowercase

Puts all characters of each word in lowercase.

none

No capitalization effects.

Although limited, the case mapping process has some language dependencies. Some well known examples are Turkish and Greek. See HTML [HTML401] for ways to find the language of an HTML element. XML, and consequently [XHTML1.0], uses an attribute called xml:lang. There may be other ways, specific to certain document languages, to determine the human language.

The case mapping rules for the character repertoire specified by the Unicode Standard 3.2 can be found on the Unicode Consortium web site at [UNICODE-casing].

Conforming user agents must support case mapping rules according to the Unicode Standard 3.2 for all characters specified by that standard.

Note: The conformance rule is more stringent than the ones specified in lower levels of CSS.

h1 { text-transform: uppercase }

11.2. Hanging punctuation: the 'hanging-punctuation' property

This property determines whether a punctuation mark, if one is present, may be placed outside the content area at the start or at the end of a full line of text. Allowing a punctuation to hang at the end of a line is is a common practice in East Asian typography. It is the responsibility of the designer to create meaningful padding and margin areas to allow effective rendering of the punctuation. If the padding and margin are of zero width or not wide enough to show the punctuation, the punctuation may not be rendered.

Note: Compression, including compression of the punctuation characters, may also occur in the line independently of this effect.

Possible values:

none

No punctuation marks are allowed to be placed outside the content area.

start

A leading punctuation mark, if present, may overhang at the start of the content area.

end

A trailing punctuation mark, if present, may overhang at the end of the content area.

both

A leading punctuation mark, if present, may overhang at the start of the content area. A trailing punctuation mark, if present, may overhang at the end of the content area.

Note: User agents should follow the script-specific and language-specific conventions about which lines allow which hanging punctuation. For example, in Latin text, a leading punctuation mark may overhang only on the first line and a trailing punctuation may overhang only on the last line. It would look improper in other lines. The following figure shows an example of correct usage.

11.3. Combining text: the 'text-combine' property'

This property controls the creation of grapheme clusters ("kumimoji") or lines ("warichu").

Possible values:

none

Glyphs arrange normally.

letters

Combines glyphs to fit within the space of a single wide-cell glyph, by reducing them in size and stacking them next to and/or on top of each other. This effect is commonly used in East Asian typography.

No more than five grapheme clusters can combine into a unit. If an element with 'text-combine' set to 'letters' has more than five, only the first five are combined. The rest is rendered as regular text. If the text to be combined has only a single grapheme cluster, no special combination effect occurs. The following texts shows the arrangement for two, three, four and five grapheme clusters:

The valid "Kumimoji" arrangements

The following styling rule:

span.kumimoji { text-combine: letters }

applied to the following fragment:

<span>&#x304f;&#x307f;&#x3082;&#x3058;
<span class="kumimoji">&#x304f;&#x307f;&#x3082;&#x3058;</span></span>

would make the 4 glyphs of the second span element appear as one (shown in blue for clarity):

"Kumimoji" applied to four glyphs

lines

Combines the glyphs so they fit into two lines of equal length and height, whose combined height is equal to or slightly greater than the height of the line in which they appear. The combined lines appear inline with the surrounding text. The Japanese Standard [JIS-X-4051] describes recommended guidelines for this feature.  A summary of these guidelines is provided here:

1. The combination should be preceded and followed by bracket or parenthesis characters
2. This combination is restricted to two lines and is rendered as a single inline box or multiple inline boxes if it does not fit in a single line box .
3. If the combination does not fit in a single line box, it can be split across several line boxes, using multiple inline boxes. In that case, each inline box represents its own character flow subset of the combination. That is, each of these inline boxes is rendered as a two-lines inline block with characters flowing from the ending edge of the first line to the starting edge of the second line of the inline box.

The following figure shows a typical usage for this feature.

Glyph layout in Warichu within a single line box

Glyph layout in Warichu split within two line boxes

The following styling rule:

span.warichu { text-combine: lines }

applied to the following fragment:

<span>&#x5272;&#x6ce8;(
<span class="warichu">&#x3053;&#x308c;&#x306f;
&#x308f;&#x308a;&#x3061;&#x3085;&#x3067;&#x3059;&#x3002;</span>
)&#x3067;&#x3059;&#x3002;</span>

would make the enclosed text look like the following (shown in blue for clarity):

"Warichu" applied to 10 characters

12. Properties index

In addition to the specified values, all properties take the initial and inherit values.

The following properties are defined in other specifications:

• background, see "CSS3 module: Backgrounds" [CSS3-background]
• color, see "CSS3 module: color" [CSS3-color]
• font-family, see "CSS3 module: fonts" [CSS3-fonts]
• font-size, see "CSS3 module: fonts" [CSS3-fonts]

13. Profiles

This document defines three modules: The CSS1 text module, the CSS2 text module and the CSS3 text module.

The CSS1 text module has the following properties and values:

Concerning the 'text-align' property, a conforming CSS1 user agent my interpret the value 'justify' as 'left' or 'right', depending on whether the element's current text direction is left-to-right or riht-to-left respectively.

The following table describes the CSS2 text module. Because all properties have added the 'inherit' value and have a media type, all CSS1 properties have been specified below as well. The properties added are 'direction', 'text-shadow' and 'unicode-bidi'. In addition, the 'text-align' has a new value: <string>.

The CSS3 module adds the following properties:

• Text layout flow: 'glyph-orientation-horizontal', 'glyph-orientation-vertical', 'writing-mode', 'text-script'.
• Text alignment and justification: 'min-font-size', 'max-font-size', 'text-align-last', 'text-justify', 'text-justify-trim', 'text-kashida-space'.
• Text indent: 'text-first-indent', 'text-block-indent'.
• Line breaking: 'line-break', 'word-break', 'word-break-cjk', 'word-break-inside'.
• Text wrapping, white-space control and text overflow: 'all-space-treatment', 'linefeed-treatment', 'text-overflow', 'text-overflow-ellipsis', 'text-overflow-mode', 'white-space-treatment', 'wrap-option'.
• Text spacing: 'kerning-mode', 'kerning-pair-threshold', 'punctuation-trim', 'text-autospace'.
• Text decoration: 'text-line-through', 'text-line-through-color', 'text-line-through-mode', 'text-line-through-style', 'text-line-through-width', 'text-overline', 'text-overline-color', 'text-overline-mode', 'text-overline-style', 'text-overline-width', 'text-underline', 'text-underline-color', 'text-underline-mode', 'text-underline-position', 'text-underline-style', 'text-underline-width', text-blink.
• Document grid: 'line-grid', 'line-grid-mode', 'line-grid-progression'.
• Miscellaneous: 'hanging-punctuation', 'text-combine'.

It also modifies the following properties as described:

• 'text-align' has two new values: 'start' and 'end'.
• 'text-decoration', in addition to its CSS2 behavior, becomes a shorthand property for the following properties: 'text-line-through', 'text-line-through-color', 'text-line-through-mode', 'text-line-through-style', 'text-line-through-width', 'text-overline', 'text-overline-color', 'text-overline-mode', 'text-overline-style', 'text-overline-width', 'text-underline', 'text-underline-color', 'text-underline-mode', 'text-underline-position', 'text-underline-style', 'text-underline-width', text-blink.
• 'white-space' has two new values: 'pre-wrap' and 'pre-line' and becomes a shorthand property for the following properties: 'wrap-option', 'all-space-treatment', 'linefeed-treatment' and 'white-space-treatment'.
• all properties take the value 'initial'.

14. Glossary

Hangul

Subset of the Korean writing system in which a few letters combine to form syllables. Visually, each written syllable occupies the same amount of space.

Hanja

Subset of the Korean writing system that utilizes ideographic characters borrowed or adapted from the Chinese writing system. Also see Kanji.

Hiragana

Japanese syllabic script, or character of that script. Rounded and cursive in appearance. Subset of the Japanese writing system, used together with kanji and katakana. In recent times, mostly used to write Japanese words when kanji are not available or appropriate, and word endings and particles. Also see Katakana.

Ideograph

A character that is used to represent an idea, word, or word component, in contrast to a character from an alphabetic or syllabic script. The most well-known ideographic script is used (with some variation) in East Asia (China, Japan, Korea,...).

Kana

Collective term for hiragana and katakana.

Kanji

Japanese term for ideographs; ideographs used in Japanese. Subset of the Japanese writing system, used together with hiragana and katakana. Also see Hanja.

Kashida

Any of a number of elongation glyphs that sit on the alphabetic baseline; used in Arabic and Syriac. Also a character (U+0640, Arabic tatweel) used to indicate the glyph.

Katakana

Subset of the Japanese writing system consisting of phonetic characters used to represent mostly Latin words. Also see Hiragana.

Kinsoku

Japanese term for a set (or sets) of line breaking restrictions.

Kumimoji

Composition of two to five grapheme clusters that are reduced in size and arranged to fit within the space of a single wide-cell glyph. Some of these compositions are encoded in the Unicode Standard.

Logograph, Logogram

Character or glyph that represents an entire word. Common in the Han script used in East Asia.

Ruby

A run of text that appears in the vicinity of another run of text and serves as an annotation or a pronunciation guide for that text.

Tate chu yoko

Run of horizontal text inside of a column of vertical text; frequently used in East Asian documents for displaying certain numbers, such as years.

Warichu

A run of text of reduced font size that appears inside of a line of text as two lines of equal height and length

Appendix A: Vertical Layout Effect on CSS Properties

In general, the existing [CSS2] properties that imply directionality or position are absolute, i.e. "left" means "left" and "top" means "top" regardless of the writing mode of the page. The purpose of this appendix however is to list the exceptions to that rule and clarify ambiguities. If a property does not appear in this list, it is intended to be interpreted as absolute, i.e. it does not rotate when the layout mode changes.

Acknowledgments

This specification would not have been possible without the help from:

Ayman Aldahleh, Stephen Deach, Martin Dürst, Laurie Anna Edlund, Ben Errez, Yaniv Feinberg, Arye Gittelman, Richard Ishida, Koji Ishii, Masayasu Ishikawa, Michael Jochimsen, Eric LeVine, Chris Pratley, Rahul Sonnad, Frank Tang, Chris Thrasher, Etan Wexler, Masafumi Yabe.

References

Normative references

[UTR-24]

Mark Davis. Script Names. 1 April 2002. Unicode Technical Report #24. URL: http://www.unicode.org/unicode/reports/tr24/tr24-4.html

[UNICODE-casing]

Unicode Consortium. Case folding. March 2002. Unicode Character data base Version 3.2, case folding information. URLs: http://www.unicode.org/Public/3.2-Update/UnicodeData-3.2.0.txt and http://www.unicode.org/Public/3.2-Update/CaseFolding-3.2.0.txt

[XML1.0]

Tim Bray; Jean Paoli; C. M. Sperberg-McQueen; Eve Maler. Extensible Markup Language (XML) 1.0 (Second Edition). October 2000. W3C Recommendation. URL: http://www.w3.org/TR/2000/REC-xml-20001006

Other references

[CSS2]

Bert Bos; Håkon Wium Lie; Chris Lilley; Ian Jacobs. Cascading Style Sheets, level 2. 1998. W3C Recommendation. URL: http://www.w3.org/TR/REC-CSS2

[CSS3-background]

Tim Boland; Bert Bos. CSS3 module: backgrounds. 19 February 2002. W3C working draft. (Work in progress.) URL: http://www.w3.org/TR/2002/WD-css3-background-20020219

[CSS3-color]

Tantek Çelik; Chris Lilley. CSS3 module: color. 18 April 2002. W3C working draft. (Work in progress.) URL: http://www.w3.org/TR/2002/WD-css3-color-20020418/

[CSS3-fonts]

Michel Suignard; Chris Lilley. CSS3 module: fonts. 2 August 2002. W3C working draft. (Work in progress.) URL: http://www.w3.org/TR/2002/WD-css3-fonts-20020802/

[HTML401]

Dave Raggett; Arnaud Le Hors; Ian Jacobs. HTML 4.01 Specification. December 1999. W3C Recommendation. URL: http://www.w3.org/TR/1999/REC-html401-19991224

[ISO15924]

Code for the representation of names of scripts. International Organization for Standardization.. 1998. ISO 15924:1998. Draft International Standard

[JIS-X-4051]

Line composition rules for Japanese documents. Japanese Standards Association. 1995. JIS X 4051-1995. In Japanese

[RFC2119]

Scott Bradner. Key words for use in RFCs to Indicate Requirement Levels. Internet RFC 2119. URL: http://www.ietf.org/rfc/rfc2119.txt

[SVG1.0]

Various. Scalable Vector Graphics (SVG) 1.0 Specification. September 2001. W3C Recommendation. URL: http://www.w3.org/TR/2001/REC-SVG-20010904

[UAX-11]

Asmus Freytag. East Asian Width. 15 March 2002. Unicode Standard Annex #11. URL: http://www.unicode.org/unicode/reports/tr11/tr11-10

[UAX-14]

Asmus Freytag. Line Breaking Properties. 15 March 2002. Unicode Standard Annex #14. URL: http://www.unicode.org/unicode/reports/tr14/tr14-12

[UNICODE]

The Unicode Consortium. Book version: The Unicode Standard: Version 3.0. Addison Wesley Longman. 2000. ISBN 0-201-61633-5. Online version: The Unicode Standard: Version 3.2.0, URL:
http://www.unicode.org/unicode/standard/versions/enumeratedversions.html#Unicode_3_2_0
For more information, consult the Unicode Consortium's home page at http://www.unicode.org/

[UTR-29]

Mark Davis. Text Boundaries. 11th October 2002. Unicode Technical Report #29. URL: http://www.unicode.org/unicode/reports/tr29/tr29-3.html

[XHTML1.0]

Steven Pemberton; et al. XHTML 1.0: The Extensible HyperText Markup Language. Jan 2000. W3C Recommendation. URL: http://www.w3.org/TR/2000/REC-xhtml1-20000126

[XHTMLMOD]

Robert Adams; Murray Altheim; et al. Modularization of XHTML. April 2001. W3C Recommendation. URL: http://www.w3.org/TR/2001/REC-xhtml-modularization-20010410

[XML1.1]

John Cowan. Extensible Markup Language (XML) 1.1. October 2002. W3C Candidate Recommendation. URL: http://www.w3.org/TR/2002/CR-xml11-20021015

[XSL1.0]

Adler, Sharon; Berglund, Anders; et al. Extensible Stylesheet Language (XSL) Version 1.0. October 2001. W3C Recommendation. URL: http://www.w3.org/TR/2001/REC-xsl-20011015