Skip to main content

PDF testing rules

  • Updated

PDF accessibility testing rules provide a framework for ensuring that PDF content is usable by everyone. 

Note: To test a PDF, you must upload it to the platform. 

On this page: 

About accessibility testing rules 

Testing determines if your uploaded PDF complies with the supported accessibility standards. For example, to meet core accessibility requirements, your uploaded PDF document must include proper tagging, Alt text for images, and a logical structure that assistive technologies can navigate.  

Note: Non-compliance with these rules can lead to legal and operational consequences, making accessibility testing a critical process for your organization.    

Current PDF testing rules

Learn more about our testing categories and their rules. 

Categories of testing rules 

The categories group rules pertaining to similar test items (for example, rules that pertain to document metadata, headings, tables of contents, figures, annotations, fonts and so on). 

The platform tests the following accessibility issues, grouped in their respective categories:

Rule category Description and example rules
Document metadata & identification

Rules that ensure the file declares its standard, title, and required metadata streams.

  • PDF/UA identification schema & parts: p5 t1, p5 t2, p5 t3, p5 t4, p5 t5.

  • XMP metadata stream present: p7.1 t8.

  • Dublin Core title required: p7.1 t9.

  • Language for document metadata determined: p7.2 t33.
Language (natural language) requirements

Rules ensuring that all user‑facing text and attributes declare language.

  • Language on outline entries: p7.2 t2.

  • Language on page text content: p7.2 t34.

  • Language on Span attributes (Alt/ActualText/E): p7.2 t30, t31, t32.

  • Language in annotations contents: p7.2 t24.

  • Language in form field tooltips (TU): p7.2 t25.

  • Valid RFC‑3066 language tags anywhere a Lang entry appears: p7.2 t29.
Structure tree and semantics (global)

Core rules about the structure tree presence, integrity, and role mapping.

  • StructTreeRoot required: p7.1 t11.

  • No circular references in PDStructElem: p7.1 t6. 

  • Parent (P) entry required on structure elements: p7.1 t12.

  • All page content tagged or artifacted: p7.1 t3.

  • No artifacts inside tags & no tags inside artifacts: p7.1 t1, p7.1 t2.

  • Use standard tags or role‑map non‑standard to standard: p7.1 t5, p7.1 t7.
Headings (H / H# rules)

Consistency and sequencing for headings.

  • Use either strong H or weak H# — not both: p7.4.4 t2/t3 (duplicates expressed). 

  • One child H per node: p7.4.4 t1.

  • Heading sequence rules for “not strongly structured” docs: p7.4.2 t1 (H1 first; don’t skip levels when descending; levels can repeat; sequence may continue without restarting at H1).
Tables

Containment, structure, and header association requirements.

  • Containment (table parts):

    • Table contains only TR/THead/TBody/TFoot/Caption: p7.2 t3.  

    • THead/TBody/TFoot in Table: p7.2 t5, t6, t7.

    • TR in Table/THead/TBody/TFoot: p7.2 t4.

    • TH/TD in TR: p7.2 t8, t9; TR only contains TH/TD: p7.2 t10.

    • THead/TBody/TFoot contain only TR: p7.2 t36, t37, t38.

  • Optional children counts:

    • Table: zero/one THead; zero/one TFoot: p7.2 t11, t12.

    • If THead present ⇒ one or more TBodies: p7.2 t14; if TFoot present ⇒ one or more TBodies: p7.2 t13.

    • Caption rules (position & single instance): p7.2 t16, t39.

  • Structural consistency:

    • No intersecting cells: p7.2 t15.

    • Columns have same number of rows; rows have same number of columns (accounting for spans): p7.2 t41, t42/t43.

  • Header associations:

    • If headers/IDs insufficient ⇒ TH must have Scope: p7.5 t1/t2 (worded twice).

  • Containment:

    • L may contain only L, LI, Caption; Caption must be first if present: p7.2 t19, t40.

    • LI must contain only Lbl and LBody; LBody inside LI: p7.2 t20, t18.
Table of contents (TOC/TOCI)

Defines table of contents rules. 

Containment:

  • TOC may contain TOC/TOCI/Caption; Caption must be first if present: p7.2 t27, t28.

  • TOCI must be inside TOC: p7.2 t26.
Figures, media, math, and alternative text

Requirements ensuring that non‑text content is accessible.

  • Figures need Alt or ActualText: p7.3 t1.

  • Media clips require Alt and CT key: p7.18.6.2 t1/t2.

  • Formula elements require Alt or ActualText: p7.7 t1.
Annotations & links (including forms & widgets)

  • General annotation accessibility:

    • Non‑Widget/PrinterMark/Link annotations must be inside Annot tag: p7.18.1 t1.

    • Visible, non‑Widget annotations inside crop box need description (Contents or Alt on parent element): p7.18.1 t2.  

    • TrapNet annotations disallowed if outside crop box or printed: p7.18.2 t1.

    • PrinterMark handling (hidden/outside crop box only): p7.18.8 t1.

  • Link annotations:

    • Links must be tagged as Link elements and need alternate description (Contents); don’t mark as artifact or nest in other tags: p7.18.5 t1/t2.

  • Page with annotations:

    • Page dictionary Tabs=S (structural order): p7.18.3 t1. 

  • Notes:

    • Note element must have non‑empty unique ID: p7.9 t1/t2. 

  • Forms & widgets:

    • PDWidgetAnnot must be inside a Form tag (or artifact): p7.18.4 t1. 

    • If form field (SEForm) lacks Role, it must have exactly one child referencing its widget: p7.18.4 t2.

    • Form field must have TU key or all Widgets must have Alt via enclosing structure: p7.18.1 t3.
Fonts, encodings, and CMaps

Complete font embedding, Unicode mapping, widths, and CID/CMap integrity.

  • Embedding & Unicode mapping:

    • Every font should be embedded: p7.21.4.1 t1.

    • Every font must map to Unicode (ToUnicode etc.): p7.21.7 t1. 

    • No .notdef glyph used: p7.21.8 t1.

    • Rendered glyphs must exist in embedded font; TT mapping uses ISO 32000‑1 §9.6.6.4: p7.21.4.1 t2.

    • Glyph widths must match font program: p7.21.5 t1.

  • CMaps / CID integrity:

    • All non‑standard CMaps must be embedded; allowed externals per ISO table: p7.21.3.3 t1/t3; WMode consistency: t2.

    • Type‑0 font CIDSystemInfo must match CMap (except Identity‑H/V): p7.21.3.1 t1.

  • TrueType specifics:

    • Non‑symbolic TT fonts: use MacRoman/WinAnsi; Differences only AGL names; include Unicode (3,1) cmap: p7.21.6 t2. 

    • Must embed at least one non‑symbolic cmap (or two if 3.0 present): p7.21.6 t1.

    • Symbolic TT fonts: no Encoding entry; cmap contains exactly one or includes Microsoft Symbol (3,0): p7.21.6 t3/t4.

  • CID/Type‑1 font extras:

    • Type‑1 with CharSet must list all glyph names: p7.21.4.2 t1.

    • CIDSet stream (if present) lists all CIDs in program: p7.21.4.2 t2.

  • ToUnicode values valid: no 0, U+FEFF, U+FFFE: p7.21.7 t2.
Core PDF syntax / file‑level conformance

Low‑level PDF requirements that PDF/UA builds on.

  • Header & version marker: file starts “%PDF‑1.n” with single EOL, nothing before: p6.1 t1.

  • Catalog contains MarkInfo/Marked=true: p6.2 t1.

  • ViewerPreferences with DisplayDocTitle=true: p7.1 t10. 

  • Document not marked as Suspects: p7.1 t4.

  • No reference XObjects: p7.20 t1.

  • Form XObjects containing MCIDs used only once and properly associated: p7.20 t2.

  • Optional content config dictionaries: have Name (p7.10 t1); AS key must not appear (p7.10 t2).

  • Embedded file spec must have F and UF: p7.11 t1.

  • Encryption permissions: encryption dictionary P has 10th bit set: p7.16 t1.
Pages & navigation

Small but important page‑level settings.

  • Tabs=S when annotations present: p7.18.3 t1 (also listed above under annotations).

 

List of testing rules

These rules  ensure that the file declares its standard, title, and required metadata streams:

Rule Short description Tag

PDF/ua1 p7.5 t1

If a table's header relationships can't be deduced from the Headers or ID attributes, every header cell (TH) must specify its Scope (e.g., "row", "col")

structure, table

PDF/ua1 p7.2 t10

TR (row cell) element may contain only TH (header cell) and TD (data cell) elements

structure, table

PDF/ua1 p7.4.4 t1

Each node in the tag tree may contain at most one child H (heading) tag

structure, heading

PDF/ua1 p5 t3

Property "part" of the PDF/UA Identification Schema shall have namespace prefix "pdfuaid"

metadata

PDF/ua1 p7.18.6.2 t2

PDMediaClip (media clip) must include an Alt key (alternative text) in its data dictionary

alt-text

PDF/ua1 p7.2 t18

LBody (list body) element should be contained in LI (list item) element

structure, list

PDF/ua1 p7.2 t24

Natural language in the Contents entry for annotations shall be determined

lang, annotation, structure

PDF/ua1 p7.1 t10

Catalog dictionary must contain a ViewerPreferences dictionary, and that dictionary must include the key DisplayDocTitle set to true

syntax

PDF/ua1 p7.2 t5

THead (table header) element should be contained in Table element

structure, table

PDF/ua1 p7.18.4 t1

PDWidgetAnnot (widget annotations) must be nested inside a Form tag (or treated as an artifact)

annotation

PDF/ua1 p7.2 t25

Natural language in the tooltip text of a document form field (TU) shall be determined

structure, lang

PDF/ua1 p7.21.4.1 t2

Any glyph that is actually rendered in a document must be defined in the embedded font. Fonts that appear only in text rendering mode-3 are exempt, because their glyphs are never drawn. For TrueType fonts that will be rendered, the mapping from character codes to glyphs must use the standard ISO 32000-1:2008, 9.6.6.4 mapping (no custom mapping)

font

PDF/ua1 p7.1 t8

Catalog dictionary must contain a metadata stream that has a /Type /Metadata entry and a /Subtype /XML entry, in accordance with ISO-32000-1:2008, 14.3.2

metadata

PDF/ua1 p7.2 t8

TH (header cell) element should be contained in TR (row cell) element

structure, table

PDF/ua1 p7.2 t23

Natural language for text in E attribute shall be determined

structure, lang

PDF/ua1 p7.21.6 t3

For a TrueType font that is marked as symbolic (the isSymbolic flag is true), the font dictionary must not contain an Encoding entry

font

PDF/ua1 p7.21.7 t2

Every Unicode value assigned to a glyph in a document's ToUnicode CMap should be valid: each value must be greater than zero and cannot be the BOM (U+FEFF) or the non-character (U+FFFE)

font

PDF/ua1 p7.18.1 t2

Any visible annotation that is not a Widget, not hidden, and whose rectangle lies within the crop-box must provide descriptive text. This is enforced by requiring the annotation to contain either a non-empty Contents key or an Alt entry in its parent structure element

annotation, alt-text

PDF/ua1 p7.2 t11

Table element should contain zero or one THead (table header) child

structure, table

PDF/ua1 p7.15 t1

Dynamic XFA forms shall not be used (dynamicRender is set to required). Only static XFA forms are permitted

syntax

PDF/ua1 p7.9 t2

Each Note tag shall have a unique ID key

structure, note

PDF/ua1 p7.21.6 t4

The 'cmap' table in the embedded font program for a symbolic TrueType font shall contain either exactly one encoding or it shall contain, at least, the Microsoft Symbol (3,0 - Platform ID=3, Encoding ID=0) encoding

font

PDF/ua1 p7.20 t1

A document shall not contain any reference XObjects (objects that use a Ref key in the XObject dictionary)

syntax

PDF/ua1 p7.21.7 t1

Every font used in the document must provide a mapping from its character codes to Unicode, usually via a ToUnicode map or other mechanisms as defined in ISO14289‑1, 7.21.7

font

PDF/ua1 p7.2 t34

Natural language for text in page content shall be determined

lang, text

PDF/ua1 p7.2 t14

If Table element contains THead (table header) child, Table element should contain one or more TBody (table body) children

structure, table

PDF/ua1 p7.2 t36

THead (table header) element may contain only TR (row cell) elements

structure, table

PDF/ua1 p7.2 t31

Natural language for text in Alt attribute in Span Marked Content shall be determined

lang, alt-text

PDF/ua1 p7.2 t41

Table columns shall have the same number of rows (taking into account row spans)

structure, table

PDF/ua1 p7.21.5 t1

When a font is embedded and actually used to render content, the width of every glyph must match the width stored in the font's dictionary

font

PDF/ua1 p7.1 t5

Any structure element in the document that uses a non-standard type must be mapped, via the role-map dictionary of the structure-tree root, to the closest functionally equivalent standard type defined in ISO-32000-1:2008, 14.8.4. The mapping may be indirect (a non-standard type can point to another non-standard type), but it must eventually resolve to a standard type

structure

PDF/ua1 p7.1 t9

The document must contain a Dublin Core title in its XMP metadata stream. This requires that the document's metadata include a dc:title entry that explicitly identifies the title of the document

metadata

PDF/ua1 p6.1 t1

The document file must begin with a header that reads exactly "%PDF-1.n", where n is a single digit from 0 to 7 (e.g., %PDF-1.4), followed by a single end-of-line marker. No extra characters, spaces or line breaks are allowed before this header

syntax

PDF/ua1 p7.10 t2

The AS key shall not appear in any optional content configuration dictionary

syntax

PDF/ua1 p7.18.3 t1

If a page contains any annotations, the page's dictionary must include the key Tabs set to the value S (indicating that annotations are to be sorted in structural order)

page

PDF/ua1 p7.2 t3

Table element may contain only TR (row cell), THead (table header), TBody (table body), TFoot (table footer) and Caption elements

structure, table

PDF/ua1 p7.21.3.3 t1

All CMaps that appear inside a PDF/UA document must be embedded in the file, except for the standard CMaps explicitly listed in ISO 32000-1:2008, 9.7.5.2, Table 118

font

PDF/ua1 p7.2 t43

Table rows shall have the same number of columns (taking into account column spans)

structure, table

PDF/ua1 p7.2 t40

L (list) element may contain a Caption element only as its first child. No other element type may precede the Caption

structure, list

PDF/ua1 p7.11 t1

The file specification dictionary for an embedded file shall contain the non-empty F (file name) and UF (Unicode file name) keys

syntax

PDF/ua1 p7.18.6.2 t1

In the media clip data dictionary (PDMediaClip), the optional CT key (ISO 32000-1:2008, 13.2.4.2, Table 274) is required

syntax

PDF/ua1 p7.2 t2

Natural language in the Outline entries shall be determined

lang

PDF/ua1 p7.9 t1

Every Note element shall contain a non-empty ID attribute

structure, note

PDF/ua1 p5 t4

Property "amd" of the PDF/UA Identification Schema shall have namespace prefix "pdfuaid"

metadata

PDF/ua1 p7.4.2 t1

For documents that are not strongly structured, as described in ISO 32000-1:2008, 14.8.4.3.5, heading tags shall be used as follows: (*) If any heading tags are used, H1 shall be the first. (*) A document may use more than one instance of any specific tag level. For example, a tag level may be repeated if document content requires it. (*) If document semantics require a descending sequence of headers, such a sequence shall proceed in strict numerical order and shall not skip an intervening heading level. (*) A document may increment its heading sequence without restarting at H1 if document semantics require it

structure, heading

PDF/ua1 p7.21.3.1 t1

For any composite (Type-0) font in a compliant document, the CIDSystemInfo section of the CIDFont dictionary must match the CIDSystemInfo of its CMap, unless the font's Encoding key is Identity-H or Identity-V. In those special cases, the Registry, Ordering, and Supplement values may be arbitrary. Otherwise, the Registry and Ordering strings must be identical, and the Supplement value of the CIDFont must be less than or equal to that of the CMap

font

PDF/ua1 p7.1 t3

All page content must be either explicitly tagged as real content or must be nested inside a tag marked as Artifact

artifact

PDF/ua1 p7.18.4 t2

When a form field (SEForm) is missing the Role attribute, it must contain exactly one child element that references the widget annotation it controls (ISO 32000-1:2008, 14.8.4.5, Table 340). It ensures that every form field is properly linked to its visual widget and can be interpreted correctly by assistive technologies

structure

PDF/ua1 p7.2 t13

If Table element contains TFoot (table footer) child, Table element should contain one or more TBody (table body) children

structure, table

PDF/ua1 p7.1 t4

A document shall not be marked as suspect. The Suspects entry in the document's syntax shall be set to false (ISO 32000-1:2008, Table 321)

syntax

PDF/ua1 p7.2 t30

Natural language for text in ActualText attribute in Span Marked Content shall be determined

lang, alt-text

PDF/ua1 p7.2 t38

TFoot (table footer) element may contain only TR (row cell) elements

structure, table

PDF/ua1 p7.21.3.3 t2

For those CMaps that are embedded, the integer value of the WMode entry in the CMap dictionary shall be identical to the WMode value in the embedded CMap stream

font

PDF/ua1 p7.18.5 t2

Every link annotation (PDLinkAnnot) must contain an alternate description in its Contents entry as described in ISO 32000-1:2008, 14.9.3

structure, annotation, alt-text

PDF/ua1 p7.1 t2

No tagged (semantic) content may appear inside a region marked as Artifact. Artifacts are non-content elements such as decorative graphics, background images, or hidden text that should be ignored by assistive technology

artifact

PDF/ua1 p7.1 t11

A document must contain a StructTreeRoot entry in its catalog as required by ISO-32000-1:2008, 14.7

structure

PDF/ua1 p7.2 t27

A Table of Contents (TOC) element may only contain child elements of the types TOC, Table of Contents Items (TOCI), or Caption

structure, toc

PDF/ua1 p7.2 t9

TD (data cell) element should be contained in TR (row cell) element

structure, table

PDF/ua1 p7.4.4 t3

A document must use either strong H headings or weak H# headings, but not both

structure, heading

PDF/ua1 p7.2 t15

A table cell shall not have an intersection with other cells

structure, table

PDF/ua1 p7.2 t17

LI (list item) element should be contained in L (list) element

structure, list

PDF/ua1 p7.2 t26

Table of Contents Items (TOCI) must be placed inside a Table of Contents (TOC) element

structure, toc

PDF/ua1 p5 t2

The value of "pdfuaid:part" shall be the part number of the International Standard to which the file conforms

metadata

PDF/ua1 p7.2 t6

TBody (table body) element should be contained in Table element

structure, table

PDF/ua1 p7.18.2 t1

Annotations of subtype TrapNet shall not be present. Any annotation of this subtype shall be considered non-conformant if it lies outside the document's crop box or if it carries the Print flag

annotation

PDF/ua1 p6.2 t1

Catalog dictionary must contain a MarkInfo dictionary, and the Marked flag inside that dictionary has to be set to true

syntax

PDF/ua1 p7.2 t19

L (list) element may contain only L, LI (list item) and Caption elements

structure, list

PDF/ua1 p7.20 t2

Form XObjects containing MCIDs shall be used only once, and their content shall be incorporated into structure elements in accordance with ISO 32000-1:2008, 14.7.2

syntax

PDF/ua1 p7.2 t33

Natural language for document metadata shall be determined

lang, metadata

PDF/ua1 p5 t5

Property "corr" of the PDF/UA Identification Schema shall have namespace prefix "pdfuaid"

metadata

PDF/ua1 p7.18.8 t1

PrinterMark annotations are considered incidental artifacts, so they should be treated like hidden page elements. They may only be present if they lie outside the crop box or are explicitly marked hidden as defined in ISO 32000-1:2008, 14.8.2.2.3

annotation

PDF/ua1 p7.2 t7

TFoot (table footer) element should be contained in Table element

structure, table

PDF/ua1 p7.4.4 t2

A document must use either strong H headings or weak H# headings, but not both

structure, heading

PDF/ua1 p7.7 t1

Every mathematical expression in the document must be wrapped in a Formula element (as defined in ISO 32000‑1:2008, 14.8.4.5), and that element must provide an alternate description using either an Alt attribute or an ActualText attribute

structure, alt-text

PDF/ua1 p7.21.6 t2

A non‑symbolic embedded TrueType font shall use a standard Windows or Mac encoding. The font dictionary shall specify an Encoding value of MacRomanEncoding or WinAnsiEncoding. If a Differences array is present, it shall contain only glyph names defined in the Adobe Glyph List, and the embedded font file shall include a Microsoft Unicode (3,1 – Platform ID=3, Encoding ID=1) entry in its 'cmap' table

font

PDF/ua1 p7.2 t37

TBody (table body) element may contain only TR (row cell) elements

structure, table

PDF/ua1 p7.16 t1

In any document that is encrypted, the encryption dictionary has to contain a P entry, and that value must have its 10th bit (the 512-bit flag) set to 1

syntax

PDF/ua1 p7.3 t1

Figure elements (SEFigure) must provide an accessible description. According to ISO-32000-1:2008, 14.7.2 (Table-323), each figure tag must include either an Alt attribute (non-empty) or an ActualText replacement

alt-text, structure, figure

PDF/ua1 p5 t1

The PDF/UA version and conformance level of a file shall be specified using the PDF/UA Identification extension schema

metadata

PDF/ua1 p7.1 t6

The document's structure tree (PDStructElem) shall not contain any circular references among its elements

structure

PDF/ua1 p7.2 t16

Table element may contain a Caption element as its first or last child

structure, table

PDF/ua1 p7.21.4.2 t2

When a document contains an embedded CID font whose FontDescriptor dictionary includes a CIDSet stream, that stream must list every CID that is actually present in the font program, regardless of whether the document ever uses those glyphs

font

PDF/ua1 p7.21.3.3 t3

A CMap shall not reference any other CMap except those listed in ISO 32000-1:2008, 9.7.5.2, Table 118. The allowed CMap names are Identity-H, GB-EUC-H, UniJIS-UCS2-H, etc.

font

PDF/ua1 p7.1 t12

A structure element dictionary shall contain the P (parent) entry according to ISO 32000-1:2008, 14.7.2, Table 355. This ensures the hierarchical integrity of the document's structural tree

structure

PDF/ua1 p7.10 t1

Every optional content configuration dictionary in a document should contain a non-empty Name entry. Any dictionary that forms the value of the D key or any element in the Configs array of the OCProperties dictionary must have a Name key whose value is a non-blank string

syntax

PDF/ua1 p7.2 t20

LI (list item) element may contain only Lbl (label) and LBody (body) elements

structure, list

PDF/ua1 p7.2 t42

Table rows shall have the same number of columns (taking into account column spans)

structure, table

PDF/ua1 p7.18.5 t1

Links in the document must be tagged as standard Link elements (ISO-32000-1:2008, 14.8.4.4.2). A link annotation should not be marked as an artifact or nested inside any other tag type

It may also be flagged as an external link or lie outside the crop box.

PDF/ua1 p7.2 t22

Natural language for text in Alt attribute shall be determined

structure, lang

PDF/ua1 p7.21.3.2 t1

When a document embeds a Type-2 CIDFont dictionary, the font must include a CIDToGIDMap entry. This entry can either be a stream that maps each character ID (CID) to a glyph index, or the special name Identity (meaning a direct one-to-one mapping)

font

PDF/ua1 p7.21.6 t1

Non-symbolic TrueType fonts that are used for rendering must embed at least one non-symbolic-cmap table (or, if a 3.0 cmap is present, at least two) so that all glyph lookups can be performed

font

PDF/ua1 p7.2 t29

If the document's Catalog dictionary, any structure element dictionary, or any property list contains a Lang entry, its value must be a language identifier that follows the format of a Language-Tag defined in RFC-3066 (e.g., en, en-US, es-ES)

lang

PDF/ua1 p7.18.1 t3

A form field shall have a TU key present or all its Widget annotations shall have alternative descriptions (in the form of an Alt entry in the enclosing structure elements)

annotation, alt-text

PDF/ua1 p7.2 t21

Natural language for text in ActualText attribute shall be determined

structure, lang

PDF/ua1 p7.1 t7

Every structural element (PDStructElem) in a document should be one of the standard PDF tags defined by ISO-32000-1:2008, 14.8.4. These tags such as Paragraph, Heading, Table, and so on, must retain their original names

They may not be remapped to a different tag value.

PDF/ua1 p7.1 t1

Any content marked as an Artifact shall not be placed inside a tagged structure element

artifact

PDF/ua1 p7.21.8 t1

A PDF/UA compliant document shall not contain a reference to the .notdef glyph from any of the text showing operators, regardless of text rendering mode, in any content stream

font

PDF/ua1 p7.2 t39

Table element may contain only one Caption element

structure, table

PDF/ua1 p7.2 t32

Natural language for text in E attribute in Span Marked Content shall be determined

lang, alt-text

PDF/ua1 p7.18.1 t1

An annotation, excluding annotations of subtype Widget, PrinterMark or Link, shall be nested within an Annot tag

annotation

PDF/ua1 p7.2 t4

TR (row cell) element should be contained in Table, THead (table header), TBody (table body) or TFoot (table footer) element

structure, table

PDF/ua1 p7.5 t2

When a table's header relationships can't be determined automatically (because Headers and IDs are missing or ambiguous), every header cell (TH) must explicitly declare its Scope attribute

structure, table

PDF/ua1 p7.21.4.1 t1

Every font used in a document should be fully embedded, which is required by ISO 32000-1:2008, 9.9

font

PDF/ua1 p7.2 t28

A Table of Contents (TOC) element may only contain a Caption as its first child

structure, toc

PDF/ua1 p7.21.4.2 t1

If a document embeds a Type‑1 font and the font’s FontDescriptor contains a CharSet string, that string must list every glyph name that appears in the font program. Regardless of whether the document actually uses those glyphs, all glyph names must be present

font

PDF/ua1 p7.2 t12

Table element should contain zero or one TFoot (table footer) child

structure, table

Legacy PDF testing rules

The table below describes the legacy testing rules for the PDF accessibility checker. 

Title Short description Longer description
Untagged content All content must be tagged or marked as artifacts.

To ensure a PDF is fully accessible, every content element in the document must be either associated with a Tag element (to be presented by the screen reader) or an Artifact (to be ignored by the screen reader).

When a failure in this checkpoint occurs it indicates that there is content somewhere in the document that has not yet been assigned to a Tag or an Artifact. These elements are considered untagged content.
Incomplete mappings Custom defined tags must provide a mapping to one of the standard known tag types so that assistive technologies understand how to present the custom tag.

The role mapping functionality described above is not limited to a single mapping. Custom tags can be mapped to other custom tags. This can even occur multiple times, creating a chain of mappings. Eventually, the chain must end at one of the standard PDF tags so the screen reader knows how to process all of the custom tags in the chain.

For example:

<MyBoldText> maps to <StyleSpan>

<StyleSpan> maps to <Span>
Circular mappings Custom defined tags are not allowed to map to each other to create a circular mapping. This would prevent assistive technologies from knowing how to present the custom tags.

A failure in this checkpoint means that the role mapping of custom tags described above has an issue where one custom tag <Tag1> is mapped to another custom tag <Tag2> , but <Tag2> is mapped to <Tag1> . This is called a circular mapping. Circular mappings are not limited to just two tags, so some circular mappings of tags could be even larger.

For accessibility, we need all custom tags to map to one of the PDF standard tags so the screen reader knows how to process and present all tagged content. Therefore, any circular mappings need to be eliminated.
Standard type remapped Custom defined tags cannot use the same name as one of the standard tag types.

A failure in this checkpoint means that the role mapping of tags described above has an issue where one of the standard PDF tags such as <H1> , <P> , <Sect> or <Table> has been remapped.

Screen readers and other technologies understand how to correctly process each of the standard tags. Therefore it is important that Role Mapping entries in a PDF file do not try to confuse the screen readers.

For example, an invalid Role Map entry that could map <H1> to <P>.
Document title is missing All documents must have a defined title in the Document Properties that provides a short, descriptive summary of the document. A failure in this checkpoint means that the document does not yet have a title. Screen readers will read this title to the user when the document is opened.
Table structure issues Table tags should follow the correct structure to ensure they can be presented correctly to the user using assistive technologies.

A failure in this checkpoint means that the document contains or more <Table> tags, but there is something wrong with the structure of the table.

Tables are complex tags that have other internal tags inside them to represent each row, and each cell. Some cells are header cells (usually the first row and/or first column) and other are data cells. Rows are represented by <TR> tags and will contain their own internal tags for header cells <TH> or data cells <TD>.

Another common error is called table irregularity. This means the table is not perfectly square. For example, the table may have four columns which means each row must have four heading or data cells. If this table was found to be irregular, there may be one or more incorrect rows with three or five cells. Some cells are the size of two or more cells. For example, a table header cell that represents two columns is said to have column span of two. Complex tables can contain cells that have column span and/or row span of multiple cells.
List structure issues List tags should follow the correct structure to ensure they can be presented correctly to the user using assistive technologies.

A failure in this checkpoint means that the document contains or more <L> list tags, but there is something wrong with the structure of the list.


 

Lists are complex tags that have other internal tags inside them to represent each item in the list. These are called list items and represented by the <LI> tag. Each list item can contain a list label

<Lbl> tag to represent a bullet or number prefix followed by a list body <LBody> tag. The list body contains the text or other content associated with the list item.

Some lists have multiple levels, often represented using indentation and/or alternating label styles. These are represented by a new list <L> tag inside their parent list item <LI> tag.
Table of Contents issues Table of Contents (TOC) tags should follow the correct structure to ensure they can be presented correctly to the user using assistive technologies.

A failure in this checkpoint means that the document contains or more <TOC> table of contents tags, but there is something wrong with the structure of those tags.

Table Of Contents are complex tags that have other internal tags inside them to represent each item in the TOC. These are called table of contents items and represented by the <TOCI> tag.

Some TOC structures have multiple levels, which are often used to break down sub-sections within the document. These are represented by a new list <TOC> tag inside their parent table of contents item <TOCI> tag.
Ruby structure issues Ruby tags (used for Japanese typography) should follow the correct structure to ensure they can be presented correctly to the user using assistive technologies.

Ruby tags are used to represent specific structures in Japanese typography and will only be applicable to use in such documents.

A failure in this checkpoint means that the document contains or more <Ruby> table of contents tags, but there is something wrong with the structure of those tags.
Warichu structure issues Warichu tags (used for Japanese typography) should follow the correct structure to ensure they can be presented correctly to the user using assistive technologies. Warichu tags are used to represent specific structures in Japanese typography and will only be applicable to use in such documents.A failure in this checkpoint means that the document contains or more <Warichu> table of contents tags, but there is something wrong with the structure of those tags.
 
Characters have Unicode Unicode values must be provided for all text to ensure it can be understood correctly by screen readers and other assistive technologies.

The Unicode specification defines unique ID values for all characters in all known languages in the world.

A failure in this checkpoint means that there is some text data in the document that cannot be properly read by a screen reader or other assistive technology because it is missing its required Unicode values.

Typically, the fonts associated with the text in the document will provide all of the required Unicode information required in a PDF file. In some cases, the fonts may have missing or incorrect Unicode information for some or all of the characters defined in the font.
Document Language The primary language of the document must be set in the Document Properties. All tags will inherit this language but can specify a different language when required in a multi-language document.
Figures have alternative text All Figure tags must define alternate text that summarizes and describes the image for screen reader users.

A failure in this checkpoint means that at least one Figure or image in the document does not have an alternate text description provided.

Screen readers will read this description to user when a Figure tag is encountered in the document.
Headings start at H1 Documents that contain numbered headings must use Heading Level 1 (H1 tag) as the first heading in the document.

A failure in this checkpoint means that the document contains some headings, but does not include an H1 (heading level 1). This prevents the screen readers and other assistive technologies from providing the required.

To resolve this error, review the current headings in the document and edit the tags to ensure there is at least one H1 tag.
Heading levels skipped Numbered heading levels in descending sequence are skipped (Example: H3 follows directly after H1).

A failure in this checkpoint means that the document contains more than one numbered heading tags such as H1 or H2, but they do not follow the correct sequence.

For example, you cannot jump from an H1 to an H3 tag. If you only had these two tags, the typical solution would be to change the H3 to an H2.

Some examples:

Valid:

  • <H1>
    • <H2> (+1 level OK)
    • <H3> (+1 level OK)
    • <H4> (+1 level OK)
       

Invalid:

  • <H4> 
    • <H2> (-2 levels invalid)
    • <H1> (-3 levels invalid)

To resolve this error, review the current headings in the document and edit the tags to ensure that they follow the correct sequence.
Invalid use of H tag In documents that use unnumbered headings (H tag), only one H tag can be added as a direct child of any other tag. Use nested tags such as Sect and Div to hold additional H tags.

A failure in this checkpoint means that the document contains generic unnumbered heading tags ( <H> ) and they do not follow the required hierarchy in the tag tree.

The <H> tags can be used alternatively to numbered headings such as <H1> and <H2> , but they are not as common. Screen reader users can still navigate them easily, but may lose understanding of the document structure.

Specifically, this rule in PDF/UA-1 states that any node in the tag tree such as <Document> , <Div> , <Sect> cannot have more than one <H> tag. To address this you can use grouping tags such as <Div> or <Sect> to create sub structures that hold the H tags.

Some examples:

Valid:

<Document>

<H>

<Sect>

<H>

<Div>

<H>

There are no tags that have more than one direct child that is an H tag.

Invalid:

All H tags are direct children of the same tag (Document)

TIP

Using numbered headings instead of <H> tags can avoid this checkpoint and provide a better experience for screen reader users.

<Document>

<H>

<P>

<H>

<P>

<H>
Mixed use of heading types Documents should only use either numbered headings (H1, H2, H3, H4, H5, H6 tags) or unnumbered headings (H tag). Using a combination may confuse users and assistive technologies like screen readers.
Table header relationships All table data cells (TD tag) must be associated with at least one table heading cell (TH tag). These relationships can be modified by opening the Table Editor when editing a table tag and switching to the Headers view.

A failure in this checkpoint means that there is not enough information in one or more <Table> structures to associate some table data cells ( <TD> ) with their corresponding table header cell(s).

Screen readers provide options for users to navigate table cells to understand the content they are interested in. They also provide the ability to read the table header cells for each TD cell as the user is navigating the table.

In the tagged PDF format, there are two options for Tables to define the relationships of TH and TD cells:

  1. Setting the Scope property on the TH tag to be either Row or Column. Row scope means the TH cell is a header cell for all TD cells on the same row (positioned to the right of the TH). Column scope means the TH cell is a header cell for all TD cells in the same column (positioned below the TH).
  2. Defining ID values for each TH, and then adding those TH IDs to any TD that is associated with the TH.

Using the scope option is suitable for simple tables with a single set of TH tags where either only the first row contains TH tags, or only the first column contains TH tags.

However, it is not uncommon for tables to be more complex and have multiple rows and/or columns that contain TH tags. In these instances, ID relationships are required.
Formulas have alternate text Formula tags must define alternate text that summarizes and describes the math expression contained in them.

A failure in this checkpoint means that the document contains one or more <Formula> tags but those tags do not have the required alternate text provided.

The <Formula> tag is used to encapsulate mathematical expressions. These could be simple expressions or highly complex expressions that use any combination of text, images and vector graphics to display expression on the page. Screen readers cannot automatically describe complex expressions; therefore, we must provide alternate text descriptions to explain the expression.
 
Incorrect OC Entries Documents that contain Optional Content elements that can be hidden or unhidden by the user during viewing must be structured correctly.

Some PDF files can consist of Optional Content that can be viewed or hidden by the authors or users who are viewing the PDF files. Examples can be CAD drawings, maps with multiple layers, or multi language documents where users can select which language to view the document in.

 

These PDF/UA checkpoints provide technical requirements on ensuring any optional content present in the document is configured correctly.

 
Filenames provided All embedded file attachments must provide a filename attribute. A failure in this checkpoint means that the PDF file contains embedded file attachments, but the required filename property was not provided for some or all of those files. Screen readers can present this filename to describe the file attachments to the user.
Dynamic Forms found Documents that use forms in the deprecated XFA (XML Forms Architecture) format are not allowed to use features that dynamically alter the documents structure during viewing.

A failure in this checkpoint means that the document contains interactive form elements in the XFA format that can modify the document layout dynamically. For example, the form may contain a button that adds additional rows of fillable text fields to an invoice when a user clicks on it. In some cases, this could add additional pages to the document as the number of rows increases.

To reduce complexity for accessibility users and tools such screen readers, these types of dynamic forms are not permitted by PDF/UA.

At the technical level, there are two types of interactive forms allowed in PDF: AcroForms and XFA. XFA was intended to provide more robust capabilities of PDF forms, but its requirements were overly complex and it did not receive wide adoption by PDF software tools beyond Adobe’s tools.

Additionally, in the PDF 2.0 specification (ISO 32000-2) the use of any XFA forms has been deprecated and going forward AcroForms will remain the only standard for PDF forms. Support for XFA will eventually be removed in all PDF creation software.
Untagged annotations Annotation objects in the document are required to use the Annot tag if they are not one of the common annotation types: Form, Link or Print Marks. A failure in this checkpoint means that there is an untagged PDF annotation object that is not one of the common types (interactive link, form field or printer mark). These types of annotations are required to be tagged using the <Annot> tag.
Annotations have alternate text Annotation objects that are tagged using the Annot tag must specify alternate text that describes the annotation. A failure in this checkpoint means that there is a PDF annotation object that is correctly tagged using the <Annot> tag type but the tag does not provide alternate text required for screen readers to describe the annotation.
Forms have alternate text Form tags are used to tag interactive form objects but do not contain alternative text to describe the form field. For example, "Last name" for a text field or "Country of residence" for a drop-down list of country codes. A failure in this checkpoint means that there is a form field annotation object that is correctly tagged using the <Form> tag type but the tag does not provide alternate text required for screen readers to describe the form field.
Undefined annotations All annotation objects used in the PDF must specify what type of annotation they are.

A failure in this checkpoint means the PDF file contains an annotation object, but the type of annotation is not specified.

This is a very uncommon event as most PDF creation software will not omit this information. However, the PDF/UA specification includes this requirement to ensure all PDF files can be properly supported by assistive technologies.
TrapNet annotations exist Accessible documents are not permitted to contain TrapNet type annotation objects.

A failure in this checkpoint means the PDF file contains a TrapNet type annotation object. These uncommon types of annotation objects are specialized for use in PDF print focused environments to provide quality control of printed documents.

Experience this failure is a very uncommon event as most PDF creation software will not include these types of artifacts unless they are intended for print production environments. However, the PDF/UA specification includes this requirement to ensure all content in a PDF file can be properly supported by assistive technologies.
Form tag required All interactive form fields in the document must be tagged using the Form tag. A failure in this checkpoint means that there is an interactive form field annotation object that is using a tag other than <Form>.
Link tag required All interactive links in the document must be tagged using the Link tag. A failure in this checkpoint means that there is an interactive link annotation object that is using a tag other than <Link>.
Links have descriptions Documents that contain interactive links using the Link tag must also provide a "contents" description that allows screen readers to describe the link. A failure in this checkpoint means that there is an interactive link annotation object that does not provide a description of the link for screen reader users.
Media have MIME types Documents that contain media clip annotations (such as an audio or video clip) embedded into the PDF file must define a MIME type that defines what type of media file it is.
Multimedia have alternate text Documents that contain media clip annotation objects (such as an audio or video clip) embedded into the PDF file must provide alternate text to describe the content of the media clip. A failure in this checkpoint means that there is a media clip annotation object (such as a video or audio clip) embedded into the PDF file but that object does not provide any alternate text to describe the content of the media clip.
PrinterMarks are artifacts Documents that contain Printer Mark objects intended for machine-based processing (e.g., mail sorting) of printed documents. These objects should be marked as artifacts.
Type0 font issues When Type0 font files are used in the document they must define visual glyphs and Unicode values for all characters in the font. This ensures the document will be displayed and read correctly.

A failure in one of these checkpoints indicates that the PDF file contains a Type 0 font resource that does not properly define the visual glyphs and Unicode values for all characters defined in the font.

This could lead to incorrect viewing and screen reader presentation of the document.
Type2 font issues When Type2 font files are used in the document they must define visual glyphs and Unicode values for all characters in the font. This ensures the document will be displayed and read correctly.
Unicode character mappings All font resources embedded into the PDF must properly define Unicode values for all of the characters in the font to ensure the content can be read correctly by screen readers.
Fonts are embedded All fonts used in the document must include the font file as an embedded resource to ensure the file will view consistently in all environments. If a PDF only includes a font name, the application used to display the PDF must attempt to locate a matching font on the user's system which may provide inconsistent results.

A failure in one of these checkpoints indicates that the PDF file uses one or more fonts but only provides the name of the font; it does not provide the actual font file as an embedded resource.

When this occurs, the PDF viewing application must choose the appropriate substitute font from the set of fonts that are installed on the user’s system. This means the PDF file may not view consistently in all operating systems or environments.
Missing glyphs All embedded font files used in the PDF must contain the correct set of visual glyphs to represent the text characters used by those fonts in the document. When glyphs are missing, the PDF may not display correctly on some or all platforms.

A failure in one of these checkpoints indicates that one or more font files embedded in the PDF do not contain the correct set of visual glyphs to represent the text characters used by those fonts in the document.

When this occurs, the PDF may not display correctly on some or all platforms.
Font widths inconsistent All fonts embedded in the PDF must define characters widths that are consistent with how the PDF file positions text on the page. This requirement ensures that PDF files view consistently on all platforms by all software applications.

A failure in this checkpoint indicates that one or more fonts embedded in the PDF define characters widths that are inconsistent with how the PDF file positions text on the page.

This requirement in PDF/UA is intended to ensure that PDF files view consistently on all platforms by all software applications.

Related to

Related articles

Comments

0 comments

Article is closed for comments.