PDF Techniques
for Web Content Accessibility Guidelines

W3C Internal Working Draft 8 November 2000

This version:

http://www.w3.org/WAI/GL/2000/12/pdf.html

Latest version:

http://www.w3.org/WAI/GL/2000/12/pdf.html

Editors:

Loretta Guarino Reid, Adobe Systems

Katie Haritos-Shea, Paradigm Solutions Corp. & NTIS - US Dept. of Commerce

Wendy Chisholm, W3C

Status

This document is a Draft.

This document has been produced as part of the W3C Web Accessibility Initiative. The goal of the Web Content Guidelines Working Group is discussed in the Working Group charter. Last up-dated 2001-03-15

Please send comments on this document to w3c-wai-gl@w3.org.

Copyright 2000 W3C (MIT, INRIA, Keio ), All Rights Reserved. W3C liability, trademark, document useand software licensingrules apply. Your interactions with this site are in accordance with our public and Member privacy statements.

Abstract

This document describes techniques for creating accessible Adobe Portable Document Format (PDF) content (refer to PDF Reference Second Edition, Version 1.3). This document is intended to help authors of Web content who wish to claim conformance to "Web Content Accessibility Guidelines 1.0" ([WCAG10]). Because PDF is a Page Description Language, not intended to be edited directly, these techniques are intended particularly for the developers of authoring tools that generate PDF as an output format.

For each technique, we identify the version of PDF in which the language support is first available. Where no version is specified, the technique can be applied in all versions of PDF. (Some of these items refer to language features in PDF 1.4, which has not yet been released.)

This document is part of a series of documents about techniques for authoring accessible Web content. For information about the other documents in the series, please refer to "Techniques for Web Content Accessibility Guidelines 1.0" WCAG 1.0

PDF Techniques

Guideline 1. Presentation:
Design content that allows presentation according to the user's needs and preferences

Provide text equivalents for images and graphics

Checkpoint:

Technique:

WCAG 1.0 Checkpoints in section 2:

1.1 Provide a text equivalent for every non-text element. [Priority 1]

WCAG 2.0 Checkpoints in section 2:

NA yet

Identify the natural language of all text in the document

Use the language tagging facilities (Lang) to specify the natural language of all text in the document. (PDF 1.4)

Checkpoint:

Technique:

WCAG 1.0 Checkpoints in section 4:

4.1 Clearly identify changes in the natural language of a document's text and any text equivalents (e.g., captions). [Priority 1]

4.3 Identify the primary natural language of a document. [Priority 3]

WCAG 2.0 Checkpoints in section 4:

NA yet

Foreground and Background colors

Checkpoint:

Technique:

Acrobat lets the user control foreground and background color of a document. Text in the original document will use the foreground color. Background elements will use the background color. A background element is defined to be any rectangle that is aligned with the edges of the page (not skewed or rotated) and that covers 50% or more of the page.

Avoid drawing rectangles behind text that are not background elements. Even if the rectangle color matches the background color, when the user changes the background color, the rectangle will not change and may cause contrast problems with the new foreground color.

Example: A document with a white background may draw a title by placing characters on a pink rectangle. Since the pink rectangle will not be recognized as part of the background, when the user sets the foreground color to white and the background color to black, the white characters will be hard to see on the pink background.

Images are not affected by the background and foreground color settings, so text should not be placed on top of images.

Example: Black text will be visible on a pastel-colored background image, but yellow text may not be, and changing the background color to black will not change the colors in the image.

WCAG 1.0 Checkpoints in section 7:

2.2 Ensure that foreground and background color combinations provide sufficient contrast when viewed by someone having color deficits or when viewed on a black and white screen. [Priority 3]

WCAG 2.0 Checkpoints in section 7:

NA yet

Guideline 2, Interaction:
Design content that allows interaction according to the user's needs and preferences

Generate text that is accessible

Checkpoint:

Technique:

Generate the text of your document so that it can be extracted reliably in logical reading order.

Render characters and words in reading order.

Render words, and characters within words, in reading order within the page-content stream. The ReverseChars marked content may be used when rendering right-to-left text that will be typeset left-to-right (PDF 1.4).

Serves the same function as the non-text element.

May contain structured content or metadata.

Deprecated Technique Examples:

• Some PDF authoring applications save space by rendering all the characters in one font at a time. Hence, the PDF file may render all the bold characters on a page, then all the normal weight characters. Generally, this will cause characters not to be rendered in reading order.
• A PDF page may render all characters left-to-right, top-to-bottom. For a multi-column document, this does not render the words in reading order, since the first line of the second column will be rendered before the second line of the first column.

Separate words explicitly with spacing characters.

Separate words explicitly with spacing characters. Do not rely on the location of the characters or the division of characters into showstring operations to indicate word breaks. Note that this implies that lines of text for western languages usually end with a trailing space character.

Consider rendering the two line example:

Now is the winter
of our discontent.

Correct Technique Example:

• Position to the beginning of line 1
• Show String ("Now is the winter ")
• Position to the beginning of line 2
• Show String ("of ")
• Position to the beginning of "our"
• Show String ("our ")
• Position to the beginning of "discontent"
• Show String ("discontent. ")

Deprecated Technique Example:

• Position to the beginning of line 1
• Show String ("Now is the winter")
• Position to the beginning of line 2
• Show String ("of")
• Position to the beginning of "our"
• Show String ("our")
• Position to the beginning of "discontent"
• Show String ("discontent.")

Note: In the deprecated example above, there are no spacing characters at the end of either line, and there are no spacing characters between the words of the second line.

Use soft hyphens and hard hyphens appropriately.

Use a soft hyphen, identified by a character that maps to the Unicode value U+00AD or 173 decimal, when a line-break hyphen is introduced into the middle of a word.

Example: If the word "father-in-law" is hyphenated after the first syllable ("fa-ther-in-law"), the first hyphen should be a soft hyphen, and the second and third hyphens should be hard hyphens.

Use the ActualText attribute.

If characters are not rendered using the showstring operation, they must be marked in the page as a Span element with an ActualText value reflecting the desired Unicode value. (PDF 1.4)

Example: Suppose the word "Arthur" is rendered using an illuminated A.
The structure subtree for this word might contain
<Figure> graphics or image for illuminated A
<Span> "rthur"
The <Figure> structural element should have the ActualText attribute, with value "A", so the word "Arthur" could be extracted.

Ensure that all characters codes map reliably to Unicode.

Within a PDF page, show string operations operate on a sequence, each with a sequence of Character Codes with associated fonts. Every such sequence of character codes must map unambiguously into a sequence of Unicode code points. Mapping is done as follows:

1.5.1 If the Font contains a ToUnicode entry, convert the Character Code to Unicode via the ToUnicode CMap.

1.5.2 If the Font uses one of the PDF predefined encodings MacRomanEncoding, MacExpertEncoding, or WinAnsi Encoding (perhaps as modified by a DIFFERENCES array in the fonts encoding resource), use the DIFFERENCES array or Appendix D of the PDF Reference Manual to convert the Character Code to an Adobe glyph name. Then use the Adobe glyph name and look up the corresponding Unicode value.

1.5.3 If the Font uses one of the predefined CMaps listed in Table 5.14 on page 320 of the PDF Reference Manual except Identity-H and Identity-V, convert the Character Code to a Unicode value via the following steps.

1.5.3.1 Obtain the Registry and Ordering of the predefined CMap from the CIDSystemInfo of the appropriate CMap.

1.5.3.2 Concatenate the Registry and the Ordering according to the format "--UCS2" to obtain a second CMap name, e.g. "Adobe-Japan1-UCS2". Obtain that CMap.

1.5.3.3 Index into the predefined Cmap, using the Character Code, and obtain an Intermediate Value

1.5.3.4 Index into the CMap obtained in step 1.5.3.2, using the Intermediate Value, and obtain a Unicode Value.

If any of these four steps fail, e.g. there is no CMap of that name or the indexing value is missing or undefined in the CMap, then there is no mapping of the character code to Unicode.

1.5.4 If the font is a Type 0 font whose descendant CIDFont uses the Adobe-Japan, Adobe-Korea, Adobe-CNS1, or Adobe-GB1 character collection, as specified in the CIDSystemInfo dictionary, follow the same steps as in 1.5.4 to obtain the character code mapping.

1.5.5 If the Font is a Type 1 font whose character names are taken from the Adobe standard Latin character set and the set of named characters in the Symbol font, documented in Appendix C, use the corresponding Unicode value found by looking up the glyph name.

WCAG 1.0 Checkpoints in section 1:

??? Ideas anyone [Priority  3]

WCAG 2.0 Checkpoints in section 1:

NA yet

Provide structural grouping

WCAG 1.0 Checkpoints in section 3:

5.1 For data tables, identify row and column headers. [Priority 1]

5.2 For data tables that have two or more logical levels of row or column headers, use markup to associate data cells and header cells. [Priority 1]

5.4 If a table is used for layout, do not use any structural markup for the purpose of visual formatting.. [Priority 1]

12.3 Divide large blocks of information into more manageable groups where natural and appropriate. [Priority 2]

13.6 Group related links, identify the group (for user agents), [Priority 1]

WCAG 2.0 Checkpoints in section 3:

NA yet

Document navigation

Checkpoint:

Technique:

Use bookmarks to provide navigation aids into a document.

Content transforms gracefully when mechanisms provided by the author are not supported or turned off but the content is still usable and readable by the user.

Use links within a document.

Markup languages, multimedia formats, software interface standards, etc., vary in their support of accessibility. When choosing which technologies to use, consider how easy it is apply these guidelines. Where feasible, favor technologies that:

If the value of the link does not describe the target clearly and accurately, provide Alt attributes.

Markup languages, multimedia formats, software interface standards, etc., vary in their support of accessibility. When choosing which technologies to use, consider how easy it is apply these guidelines. Where feasible, favor technologies that:

Provide a user name (/TU key) for all form fields

In Acrobat, this field is called the Short Description.

WCAG 1.0 Checkpoints in section 6:

0 Create a style of presentation that is consistent across pages. [Priority 3]

Create a logical tab order through links, form controls, and objects.

WCAG 2.0 Checkpoints in section 6:

NA yet

Guideline 3, Comprehension:
Make it as easy as possible to use and understand

Provide expansions for acronyms and abbreviations (PDF 1.4)

Checkpoint:

Technique:

Note: this guideline applies only where the content provides its own user interface

Checkpoint:

Technique:

(for example as a form or programmatic object).

Need example

WCAG 1.0 Checkpoints in section 5:

4.2 Specify the expansion of each abbreviation or acronym in a document where it first occurs. [Priority 3]

WCAG 2.0 Checkpoints in section 5:

NA yet

Guideline 4, Technology considerations:
Design for compatibility and interoperability

Set document protections to permit access

Checkpoint:

Technique:

Set the data access restrictions on the document to permit the contents to be accessed.

Checkpoint:

Technique:

In PDF 1.3 and earlier, permit the text and graphics in the document to be copied.

Checkpoint:

Technique:

In PDF 1.4, set accessibility permission for the document.

Checkpoint:

Technique:

WCAG 1.0 Checkpoints in section 8:

??? Ideas anyone [Priority 3]

WCAG 2.0 Checkpoints in section 8:

NA yet

PDF Glossary

Accessibility permission NEW 01-01-08

A PDF file can be encrypted (PDF 1.1) to protect its contents from unauthorized access. PDF's standard security handler defines a set of access privileges for a document, including privileges such as modifying the document's contents, copying text and graphics from the document, and printing the document. In PDF 1.4, this set includes accessibility permission, which controls whether the contents of the document are available via standard accessibility APIs to screen readers and other assistive technology.

ActualText value NEW 01-01-08

Sometimes characters are rendered by graphics commands other than showstring. For instance, an illuminated character may be rendered by an image or a series of graphics commands. In this situations, the Actual Text property is used to identify the character being rendered. This character may be concantentated with adjoining text to form a word.

Adobe glyph name NEW 01-01-08

The name of a character in the Adobe standard character encodings, in Appendix D of the PDF 1.3 Reference Manual. The encodings list characters, character names, and character codes used in platform standard encodings.

Artifacts NEW 00-12-14

A page element that is a side effect of rendering, rather than an intrinsic part of the document or story. For example, artifacts of the printing process might include crop-box markings or the document file name printed outside the crop box. Artifacts of the pagination of a document are elements that would be absent (or present in a different form) if the document was always one very big page. So pagination artifacts include running headers and page. A horizontal rule above a foornate would be an artifact of the layout process and typographic style.

Characters NEW 00-12-14

A character is a printable symbol having phonetic or pictographic meaning and usually forming part of a word of text, depicting a numeral, or expressing grammatical punctuation. A character is generally one of a limited number of symbols, including the letters of a particular language's alphabet, the numerals in the decimal number system, and certain special symbols such as the ampersand (&) and "atsign" (@). Several standards of computer encoding have been developed for characters. The most commonly used in personal computers is ASCII. IBM mainframe systems use extended binary-coded decimal interchange code. A new standard, Unicode, is supported by the Windows NT system. A distinction is sometimes made between a character and a glyph. In this distinction, a character can be distinguished from other characters in terms of meaning and sound and a glyph is the graphic image used to portray the character. In different implementations, a character can have more than one possible glyph, and a glyph can represent more than one possible character.

Character Codes NEW 00-12-14

(a la Loretta)A show string is the encoded representation of a sequence of non-negative integers. Each of those integers is a Character Code. The interpretation of a show string depends on the associated font: some fonts imply a one-byte representation whie others imply a more complicated representation.
A mapping from a set of integers to a set of characters. This mapping is generally 1:1 (i.e., bijective), for example, the code position 65 in ASCII maps only to "A", and it's the only position that maps to "A". There are several standard coded character sets, the most widely used is ASCII, generally in its Latin-1 dialect (the ASCII coded character set, encoded directly as single-byte values), or UTF-8 (the Unicode coded character set, encoded with an 8-bit transformation method), with Unicode becoming slowly more common; while EBCDIC and Baudot are extinct except in legacy systems. A coded character set may include letters, digits, punctuation, control codes, various mathematical and typographic symbols, and other characters. Each character in the set is represented by a unique character code (or "code position").

Column headers NEW 00-12-14

@@

CMap NEW 01-01-08

A CMap specifies the mapping from character codes to character selectors (CIDs, character names, or character codes) in one or more associated fonts or CIDFonts. It serves a function analogous to the Encoding dictionary for a simple font. A Cmap also specifies the writing mode - horizontal or vertical - for any CIDFont with which the CMap is combined.
Also a CMap (character map) file specifies the correspondence between character codes and the CID (character identifier) numbers used to identify characters. For composite (Type 0) fonts, it is the equivalent to the concept of an encoding in a simple font. A CMap can describe a mapping from multiple-byte codes to thousands of characters in a large CID-keyed font.

Concatenate NEW 00-12-14

To combine character strings, to join together two or more files or lists to form one big one. Example: The Unix cat command can be used to concatenate files.

Crop box NEW 01-01-08

The crop box defines the region to which the contents of the page are to be clipped (cropped) when displayed or printed.

Data tables NEW 00-12-14

@@

Expansion NEW 00-12-14

@@

Form fields NEW 01-01-08

@@

Glyph NEW 00-12-14

An image used in the visual representation of characters; roughly speaking, how a character looks. A font is a set of glyphs. In the simple case, for a given font (typeface and size), each character corresponds to a single glyph but this is not always the case, especially in a language with a large alphabet where one character may correspond to several glyphs or several characters to one glyph (a character encoding). A glyph can be an alphabetic or numeric font or some other symbol that pictures an encoded character. The following quote is from a document written as background for the Unicode character set standard. An ideal characterization of characters and glyphs and their relationship may be stated as follows: A character conveys distinctions in meaning or sounds. A character has no intrinsic appearance. A glyph conveys distinctions in form. A glyph has no intrinsic meaning. One or more characters may be depicted by one or more glyph representations (instances of an abstract glyph) in a possibly context dependent fashion. Glyph is from a Greek word for "carving."

Indexing value NEW 00-12-14

@@

Line-break hyphen NEW 00-12-14

Hyphens that you add explicitly by entering the dash character are called line-break or hard hyphens. A hyphen that is always set; for example, the hyphen in "cost-effective." A soft hyphen, by contrast, will only be set when a word that is not normally hyphenated falls at the end of a line, and must be broken for proper type spacing. Word processors use two basic techniques to perform hyphenation. The first employs an internal dictionary of words that indicates where hyphens may be inserted. The second uses a set of logical formulas to make hyphenation decisions. The dictionary method is more accurate but is usually slower. The most sophisticated programs use a combination of both methods. Most word processors allow you to override their own hyphenation rules and define yourself where a word should be divided.

Link text NEW 00-12-14

@@

MacRomanEncoding, MacExpertEncoding, or WinAnsi Encoding NEW 01-01-08

The regular font encodings used for Latin-text fonts on mac OS and Windows systems are named MacRomanEncoding and WinAnsiEncoding, respectively. Additionally, an encoding named MacExpertEncoding is used with "expert" fonts that contain addiitonal characters useful for sophisticated typography. Complete details of these encodings and the characters present in typical fonts are found in Appendix D of the PDF Version 1.3 Reference Manual.

Map, mapped NEW 00-12-14

@@.

Markup NEW 00-12-14

@@

The rendered text content of a link.

Objects NEW 00-12-14

An object is an identifiable, encapsulated entity that provides one or more services requested by a client. Objects can refer to the objects in OOP (object-oriented programming) or the objects in OLE (Object Linking and Embedding). In object-oriented programming, objects are the things you think about first in designing a program and they are also the units of code that are eventually derived from the process. In between, each object is made into a generic class of object and even more generic classes are defined so that objects can share models and reuse the class definitions in their code. Each object is an instance of a particular class or subclass with the class's own method or procedures and data variable. An object is what actually runs in the computer. An object can be a spell checker or a piece of a graphics program used to draw squares or circles. Do you remember the crazy story people used to try to tell about a word processer where you could pick all of your favorite pieces (favorite spell checker, grammar checker, text editor, font manager, etc.) and piece them together to form the ultimate customizable word processer? Well, those pieces are objects. In OLE, an object is a piece of a document, a graphic, or some multimedia. In general multimedia terms, an object is a stored data element, such as a video clip, an audio file, or a graphic representation of an object.

Page-content stream NEW 01-01-08

A page's content stream contains operands and operators used to place "paint" on a page in selected areas. By executing the actions described in the page content stream, an application builds up the image of the page described by the stream.

Pagination NEW 00-12-14

@@

ReverseChars NEW 00-12-14

Font characteristics may suggest that right-to-left text be typeset left-to-right. The ReverseChars marked content indicates that the show strings within the marked content are individually reversed in reading order.

Running headers NEW 00-12-14

@@.

Showstring NEW 00-12-14

(a la Loretta) The strings that are the arguments to the PDF and Postscript text-showing operators that show text on a page. The show string is interpreted as a sequence of character codes identifying the glyphs to be painted.

Soft hyphen NEW 00-12-14

(a la Loretta) A character that is used to mark conditional hyphenation points. Unicode and ISO_Latin-1 code-point 0xAD.
A hyphen that will only be set if the word falls at the end of a line which is too long, and has to be broken. Hyphens inserted automatically by a hyphenation utility are called discretionary or soft hyphens. Word processors use two basic techniques to perform hyphenation. The first employs an internal dictionary of words that indicates where hyphens may be inserted. The second uses a set of logical formulas to make hyphenation decisions. The dictionary method is more accurate but is usually slower. The most sophisticated programs use a combination of both methods. Most word processors allow you to override their own hyphenation rules and define yourself where a word should be divided.

Trailing space character NEW 01-01-08

A white space character inserted into the text for a page after the last word on a line. A trailing space character is not needed to produce the correct page image, but is important for determining word breaks in the text of the page.

Type 0 font, Type 1 font NEW 01-01-08

Type 0 font: a composite font, that is, a font composed of other fonts, organized hierarchically.
Type 1 font: a font represented using the Adobe Type 1 Font Format. A Type 1 font program is a stylized PostScript program that describes glyph shapes.

Typographic styleNEW 00-12-14

@@.

Unicode NEW 00-12-14

A character coding scheme that uses 16 bits for each character, designed to extend the capabilities of ASCII, which uses seven bits. Nearly all letters and symbols in all languages can be represented in a standard way with Unicode. The first 128 characters of Unicode are identical to those in standard ASCII. Unicode is an entirely new idea in setting up binary codes for text or script characters. Officially called the Unicode Worldwide Character Standard, it is a system for "the interchange, processing, and display of the written texts of the diverse languages of the modern world." It also supports many classical and historical texts in a number of languages. Currently, the Unicode standard contains 57709 distinct coded characters derived from 24 supported language scripts. These characters cover the principal written languages of the world. Originally Unicode was designed to be universal, unique, and uniform, i.e., the code was to cover all major modern written languages (universal), each character was to have exactly one encoding (unique), and each character was to be represented by a fixed width in bits (uniform). Parallel to the development of Unicode an ISO/IEC standard was being worked on that put a large emphasis on being compatible with existing character codes such as ASCII or ISO Latin 1. To avoid having two competing 16-bit standards, in 1992 the two teams compromised to define a common character code standard, known both as Unicode and BMP. Since the merger the character codes are the same but the two standards are not identical. The ISO/IEC standard covers only coding while Unicode includes additional specifications that help implementation. Unicode is not a glyph encoding. The same character can be displayed as a variety of glyphs, depending not only on the font and style, but also on the adjacent characters. A sequence of characters can be displayed as a single glyph or a character can be displayed as a sequence of glyphs. Which will be the case, is often font dependent.

Unicode value NEW 00-12-14

(a la Loretta)Unicode value or code point: The Unicode Consortium defined a set of sixteen-bit code points, 57709 of which are currently assigned and named Unicode Characters. The lowest 65536 code-points in ISO 10646-1 1993 are idential to the Unicode Standard and are sometimes called the Basic Multilingual Plane. See http://www.unicode.org

User name (/TU key) NEW 01-01-08

Any interactive form field may contain the optional /TU entry in its dictionary. This entry, known as the user name or short description, is used to identify this field when generating an error message or naming the field to a screen reader.

Word breaks NEW 01-01-08

Applications divide the text of a page into words; word breaks are the points in the text stream that separate adjoining words. Different applications may use different rules for defining words; for example, one application may consider everything between white space characters to be a word. Another application may not include leading or trailing punctuation as part of a word.