What’s new in EFX 2

Expressions can access privacy-related properties of fields withheld from publication. These properties allow rules and templates to reason about the privacy status of fields without having to reference the underlying privacy fields (e.g. BT-197, BT-198) by name:

field:rawValue provides access to the raw string value of a field, bypassing type-specific parsing:

min, max, average — aggregate functions on numeric sequences.
absolute — absolute value.
round, round-down, round-up — rounding functions.

substring-before, substring-after — split strings around a delimiter.
normalize-space — collapse and trim whitespace.
trim, trim-left, trim-right — whitespace trimming.
pad-left, pad-right — string padding.
repeat — repeat a string N times.
replace, replace-regex — string replacement (literal and regex).
split — split a string into a sequence by delimiter.
url-encode — URL encoding.
capitalize-first — capitalize the first character.
index-of-substring — find the position of a substring within a string.
empty — test whether a string is empty.

year, month, day — extract components from a date.
hours, minutes, seconds — extract components from a time.
years, months, days — extract components from a duration.
current-date, current-time — return the current date or time.

sort, reverse — sequence ordering.
subsequence — extract a sub-range from a sequence.
index-of — find the position of an element in a sequence.
get-duplicates — extract duplicate values from a sequence.
count-duplicates — count the number of duplicate values.

text(boolean), text(date), text(time), text(duration) — convert values to string.
number(boolean) — convert boolean to number.
indicator(number) — convert number to boolean.

Template lines now use WITH context DISPLAY template; instead of the EFX-1 {context} template syntax. Template lines are now semicolon-terminated, and the context declaration is optional — allowing template lines that display content without iterating over a context.

The EFX-1 syntax is still supported for backward compatibility but will likely be removed by beta.1.

Template lines can now conditionally render different content using WHEN / OTHERWISE:

View templates can now be structured with ---- NAVIGATION ---- and ---- SUMMARY ---- section headers, separating the main content from navigation aids and summary views.

Templates can be defined once and reused across the template file:

Reusable typed functions can be declared with scalar or sequence return types:

Functions are invoked using the ?name(args) syntax.

Dictionaries pre-compute key-value lookups, improving performance by avoiding repeated traversals:

Template content can be linked to URLs using the @{url} syntax:

Template text now supports HTML character references (&, ©, ©) and escape sequences.

Locale-aware date/time formatting is available through format-short, format-medium, and format-long (template-only functions):

Rules are organised into stages (---- STAGE stage-id ----), each representing a validation phase (e.g. "1a", "3b", "4"). Each stage contains optional stage-level variable declarations and one or more rule-sets.

Multiple rules can share the same context via a single WITH clause, avoiding repetition of context declarations.

Two rule types with distinct semantics:

Rules can be conditionally applied using WHEN / OTHERWISE:

Each rule specifies a severity (ERROR, WARNING, INFO) and an identifier following the R-XXX-XXX pattern, used for error message translation lookup:

FOR field|node specifies which field or node the rule validates. This is used to organise validators by target for efficient lookup.

IN specifies which notice types the rule applies to. Supports individual types, ranges, and wildcards:

SCOPE specifies rule applicability and categorisation:

Omitting @PRE/@POST means the rule applies in both contexts.