Skip to content

v0.3.0

Choose a tag to compare

View all tags
@onetogregorio onetogregorio released this 04 Jun 17:56
· 10 commits to main since this release
v0.3.0
d541eee
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Minor Changes

  • #52 01b5335 Thanks @onetogregorio! - Client bundle: real <a href> fallback (SEO), touch support, scheme guard, aria-live.

    • SEO (issue #49): the client-only fallback now renders real <a href>
      anchors       instead of <span> — so when a consuming site forgets to register
      the SSR middleware, Googlebot still sees real links to midvash.com and link
      equity flows. URLs match the SSR shape exactly (/{lang}/{version}/{slug}/ {ch}/{verse}), with the slug pre-localized server-side. No nofollow, no
      target="_blank" — by design.
    • Touch / mobile (issue #36): on coarse-pointer devices a single tap on a
      reference opens the tooltip and blocks navigation; a second tap on the same
      reference (tooltip still open) lets the click through so the user reaches
      midvash.com. Tap-outside closes the tooltip. Desktop (pointer: fine)
      behavior is unchanged — hover/focus still drives the tooltip, click
      navigates normally.
    • Security defense-in-depth (issue #42): payload.readMoreUrl is now
      validated before being placed in an href. Only http: / https: schemes
      are accepted; javascript: / data: / etc. cause the "Ler mais" link to
      be omitted (the verse text still renders).
    • a11y (issue #38): the tooltip body is marked aria-live="polite" /
      aria-atomic="true", so screen readers announce the verse when it
      replaces the "Carregando…" placeholder (follow-up to #13).

    buildClientPattern is now capturing (m[1..4] = book, chapter, verse,
    verseEnd) so the client can extract the parts needed to build a URL; the
    non-capturing shape is no longer exported. New buildNameToSlug helper
    exports a small name→localized-slug map alongside the pattern.

  • #53 cf8c6e2 Thanks @onetogregorio! - Tooltip distinguishes "verse not found" from "couldn't load" (issue #41).

    fetchVerse now returns a tagged result ({ ok: true, data } | { ok: false, kind: "not-found" | "fetch-error" }) so the plugin can tell an upstream 404
    (the reference parses but doesn't exist — typically a typo like
    "John 99:99") apart from a transient network/timeout/5xx failure. The
    /lookup route propagates { error, reference, version } on failure; the
    client renders the matching localized string from the new notFound entry
    in ClientStrings (added in pt-br / en / es).

    404 results are NOT cached, so fixing a typo and re-hovering hits the
    upstream again. Other failures still aren't cached either (no behavior
    change there).

    Breaking (internal): fetchVerse's return type changed from
    VerseResponse | null to VerseResult. No public re-export touches this,
    so external consumers should be unaffected.

  • #50 49767f6 Thanks @onetogregorio! - SSR linkifier: scope references to article content, robustness + perf fixes.

    • Scope (issue #37): the SSR linkifier no longer wraps references inside
      page chrome (<nav>, <header>, <footer>, <aside>, <head>) or non-
      content widgets (<title>, <option>, <select>, <optgroup>,
      <button>, <svg>, <math>, <noscript>, <iframe>). This matches the
      client scanner's default selectors (article, .prose, .post-content,
      main) and avoids sitewide-repeated links that Google reads as
      over-optimization.
    • Robustness (issue #40): the tag scanner is now attribute-aware — a >
      inside a quoted attribute value (e.g. <img alt="2 > 1">) is correctly
      ignored when seeking the tag end, so the body that follows is no longer
      mis-parsed.
    • Perf (issue #39): a cheap probe regex runs first; pages with no
      reference candidate are returned unchanged (same string instance), skipping
      the full streaming parse entirely.
    • SEO: linkified anchors now include a title attribute matching the
      reference (e.g. title="João 3:16") for crawler context. The plugin still
      does NOT add rel="nofollow" or target="_blank" — link equity flows
      in-document to midvash.com, by design.

    This is a minor bump because the default scope changed: references that
    used to be linkified inside <nav> / <header> / <footer> (etc.) no
    longer are. Sites that relied on that behavior should move the references
    into article content.

Patch Changes

  • #54 c78dbd7 Thanks @onetogregorio! - Docs: clarify the SEO model + document 0.3.0 changes.

    • Reframe the SSR middleware section as "recommended" (was "optional") — SEO
      is the plugin's stated goal.
    • Document the two-layer model: SSR middleware + client-side fallback both
      produce real <a href> anchors (since 0.3.0 in the client fallback too).
    • List the SEO contract explicitly: no nofollow, no target="_blank",
      title attribute, scope limited to article content (no nav/footer/title
      pollution).
    • Document the new mobile / touch behavior (tap-to-toggle, second-tap to
      navigate).
    • Updates to README.md, README.pt-BR.md, README.es.md — kept in sync.

    Patch-level bump (docs only).

Assets 2
Loading