Annotations (TextformatterAnnotations)

A ProcessWire Textformatter that automatically appends a configurable mark to configurable words during output formatting. The mark can be anything (a symbol like ©, ®, ™, ℠, a footnote marker, or any short string) and can optionally be wrapped in a <sup> tag per mapping.

Each string is configured in a small table (operation, mark/part, tag, and the match options). Examples of what it can do:

• frameless → frameless®
• Term → Term<sup>1</sup> (footnote, first mention only)
• H2O → H<sub>2</sub>O
• frameless → <strong>frameless</strong>

Why not Find/Replace?

ProcessWire's TextformatterFindReplace is a full regex engine, so to be clear: you can do all of this with it. With enough PCRE (lookarounds for idempotency, patterns to avoid tags, attributes and e-mail addresses, alternations for every entity spelling, and so on) every behaviour below is expressible.

The point of Annotations is not capability, it is who writes and maintains it:

• No regex to write or maintain. The tricky parts (only touch text not markup, do not double an existing ®/®/<sup>®</sup>, skip e-mails, longest-match across overlapping phrases, first-occurrence-only) are built in as defaults, not patterns you have to craft and keep correct per term.
• Editable by clients. Configuration is a list of strings plus a small per-string table, so non-technical editors manage it themselves (also under Setup, see below). A page of hand-written regex is not something you hand to a client.

Both are output textformatters that run on every render, and capability-wise Find/Replace is a superset. The choice is only this: use Find/Replace if you are comfortable writing and maintaining the regex, use Annotations if you want this annotation job as ready-made, no-regex defaults that a client can edit.

Installation

1. Copy the TextformatterAnnotations folder into /site/modules/.
2. In the ProcessWire admin go to Modules → Refresh, then install Annotations. This also installs the companion ProcessAnnotations.
3. Edit a text/textarea field, open the Details tab and add Annotations to the field's Text formatters.

Client-editable settings page

Installing the module adds a page under Setup > Annotations (the companion ProcessAnnotations module) that shows the same settings as the module config. It is guarded by the annotations-edit permission, so you can let editors manage the strings without giving them access to the Modules section: assign annotations-edit to the relevant role. Superusers can still configure it the usual way under Modules > Configure > Annotations.

Configuration

Open the module configuration (Modules → Configure → Annotations). The same form is also available under Setup → Annotations (/setup/annotations/), which is reachable with the annotations-edit permission, no Modules access required.

1. In Strings, enter one search string per line: nothing else. A string may contain spaces (e.g. frameless Media).
2. Save. A settings row is generated per string under Per-string settings.
3. Configure each row and save again.

Per-string settings (one row per string)

Column	Meaning
Operation	append after: add a mark after the word. wrap inside: wrap part of the word in a tag. both: do both (e.g. bold a word and append ®).
Mark (append)	The mark to add: a symbol, footnote, any text (symbol shortcuts below). Shown for append and both.
Part (wrap)	The part of the word to wrap: leave empty to wrap the whole word. Shown for wrap and both.
Tag	The tag to wrap in. append: (none) = inline, or any tag to wrap the mark. wrap: the tag for the part (defaults to sub). both: styles the wrap; the appended mark stays inline.
Options	Whole word (complete words only: cat won't match in category; unicode-aware), Case (case-sensitive), First only (annotate only the first occurrence).

New rows default to append, whole word on, case on, first off. With both, e.g. frameless → <strong>frameless</strong>® (wrap whole word in strong, append ®).

Once configured, each row collapses to a one-line summary of its operation and options:

Allowed wrap tags: sub, sup, b, strong, i, em, u, s, mark, small, ins, del, code, kbd, samp, var, abbr, cite, dfn, q, time.

Symbol shortcuts for the Mark field of an append row:

Shortcut(s)	Symbol
(c), copyright	©
(r), reg, registered	®
(tm), tm, trademark	™
(sm), sm, servicemark	℠

A single global option remains: Skip inside these tags: text inside the listed HTML elements (and descendants) is left untouched. Default: code pre script style. Add a if you do not want link text annotated.

How rows combine

Append rows are applied first, then wrap rows layer on top, so a wrapped string also styles inside an appended phrase. With frameless Media (append (r)) and frameless (wrap whole word in strong), the text frameless Media becomes <strong>frameless</strong> Media®. Within each phase the longest matching string wins.

The tag is authoritative: an existing mark is normalised to it, keeping its spelling. An append row with a tag wraps a bare mark (and rewraps a different tag); an append row with (none) unwraps an existing wrapper. A different mark next to the word is never touched.

First only annotates the string exactly once per field value. For append, the first occurrence keeps/normalises its mark and every later one has its mark removed (including marks already in the source: ©, ©, <sup>…</sup>). For wrap, only the first occurrence is wrapped. Useful for footnotes. Protected regions (attributes, e-mails, skip-tags) are ignored when finding occurrences.

HTML-aware

Replacements are applied to text content only. HTML tags, attributes (href, alt, class, title, …) and comments are never modified, so a word inside a URL, an alt text or a class name is left alone:

<a href="/frameless">frameless</a>   →  <a href="/frameless">frameless®</a>
<img alt="frameless logo">           →  <img alt="frameless logo">   (unchanged)
<code>frameless</code>               →  <code>frameless</code>        (unchanged)

Notes

• E-mail addresses are protected. A configured word that is part of an address is left untouched, e.g. with a frameless → ® mapping the text info@frameless.at stays as-is (no info@frameless®.at).
• The formatter never adds a mark twice. If a word is already followed by its mark (tolerating surrounding whitespace and an existing <sup> wrapper), it is normalised rather than duplicated.
• Symbol entity forms are recognised. For the symbol shortcuts, the named entity in lower and upper case (®/®, ©/©, ™/™) and numeric references (®, ®, with leading zeros or either hex case) count as the symbol, so frameless® is never turned into frameless®®.
• When matching is case-insensitive, the original casing of the matched word is preserved.
• Anything between < and > is treated as markup. In plain-text fields a literal a < b may therefore be skipped; on rich-text/HTML fields (the intended use) this is exactly the desired behaviour.

License

Released under the MIT License. See LICENSE.