gip/

Command absolutise rewrites every relative URL in an HTML document against a base URL, streaming, and reports what it changed.

Command abtest picks a variant for each experiment, keeps the markup for the chosen one, removes the rest, and marks the document with what it chose.

Command alt reports images with no alt attribute, and images whose alt text says nothing a screen reader could not already say.

Command article finds a page's article body by scoring elements as it streams past them, and then emits that element's subtree and nothing else.

Command autocomplete adds autocomplete tokens to form fields whose purpose the markup makes plain.

Command autoplay stops media that would start playing on its own.

Command backpressure measures what a rewrite costs a slow destination.

Command bailout shows what a memory limit does to a response, with and without graceful bail-out.

Command beacon injects an analytics beacon and then proves it changed nothing else.

Command bindings turns framework attribute syntax into plain HTML attributes where it can, and says why it cannot everywhere else.

Command bom rewrites a document that may begin with a byte-order mark, and does what a browser would do with it - which is not what this library does on its own.

Command breadcrumb emits schema.org BreadcrumbList JSON-LD derived from a page's own breadcrumb nav.

Command buildinfo stamps a page with the commit that produced it, and reports where it managed to put the stamp.

Command bust puts a build's content hash on every asset URL it knows.

Command bytewise measures what a write size costs.

Command cachetags collects cache tags from a document and turns them into a header value, which is a thing a streaming rewriter cannot quite do.

Command canonical enforces exactly one canonical link element.

Command captions gives a video with no captions track a placeholder to fill in.

Command charset makes sure a document declares the encoding it is actually in.

Command chunkinvariance checks what does and does not depend on how the input was written.

Command classmap renames classes through a mapping, the way a CSS-modules build does.

Command clicktoload turns every third-party iframe into a placeholder that loads on click, so a page does not hand an embed to the reader before they ask for it.

Command clientgone rewrites a page to a destination that stops accepting bytes partway through, which is what a browser closing a connection looks like from inside a handler.

Command collapse collapses runs of insignificant whitespace to a single space and leaves the elements where whitespace is significant alone.

Command comments renders an untrusted comment: it removes what a comment has no business containing, and turns bare URLs into links.

Command consentgate stops third-party scripts from running until consent is given, by rewriting their type attribute rather than removing them.

Command controls gives media elements a control bar and stops them downloading until someone asks for them.

Command cookiebanner injects a consent banner before the closing body content, without a script and without a template engine.

Command corpus reports which of the rewriter's documented hazards a document actually contains.

Command cspnonce prepares a document for a nonce-based Content Security Policy: it stamps every script and style with the nonce, removes any nonce the document already carried, hashes inline script bodies so a hash-based policy can be emitted as well, and reports the constructs that no nonce can rescue.

Command csrf inserts a hidden token field into every form that posts.

Command darkmode adds the two things a page needs to respect a reader's colour preference: a theme-color meta per scheme, and a stylesheet link that only loads in dark mode.

Command decomment strips comments from a document without stripping the things that only look like comments.

Command deferscripts adds defer to the scripts that can take it.

Command deployid echoes a deploy identifier from the environment into a meta tag, and says when it could not put it somewhere a browser will read.

Command deprecated reports the obsolete elements and attributes a page still uses.

Command depth enforces a maximum element nesting depth and reports the deepest path it found.

Command descript removes script elements and reports what that saved.

Command detached shows what every rewritable unit answers after its handler has returned.

Command dimensions reports every image, iframe, video, embed and object that does not declare its size, which is what makes a page shift under the reader as it loads.

Command dir adds dir="rtl" to the elements whose text reads right to left.

Command doctypepick chooses what a rewrite does from the document's doctype, and only from a doctype the document's own parser will honour.

Command dupsection duplicates a section of a document, renaming the ids in the copy, without ever holding the whole document.

Command email prepares an HTML page for an email client: it inlines the stylesheet, makes every URL absolute, and removes what a mail client would refuse to run.

Command emailstrip removes everything a mail client would reject and says what it removed and why.

Command emoji expands :shortcodes: in a document's text.

Command encodingmatrix runs a document through every encoding the rewriter accepts and compares what comes out.

Command envbadge marks a page as belonging to a non-production environment: a visible badge in the corner and a prefix on the browser tab.

Command esi expands Edge Side Include markers two ways - with lolhtml.WithESITags and without it - and reports where the two disagree.

Command etag computes an entity tag for a rewritten page, and does it without waiting for the page to be rewritten.

Command firstlink links each glossary term once, the first time it is mentioned, and leaves alone any term the page already links.

Command flags gates blocks of markup on feature flags, removing what is off.

Command fontpreload injects preload hints for the fonts a stylesheet uses.

Command formschema reads every form on a page and prints what it would take to submit it.

Command glossary reads the definition lists in a document and links every mention of their terms in the body.

Command greet injects a personalised greeting taken from a request header into four kinds of place, and treats the header as what it is: a string an attacker chooses.

Command gunzip rewrites a document that arrives gzipped, decompressing as it goes, and refuses to be surprised by the two things that go wrong.

Command gzipout rewrites a document into a gzip writer and checks the round trip, because two Closers in a chain is a thing people get wrong in a way that costs bytes.

Command handlerstats reports how many times each registered handler fired.

Command headings reports headings that skip a level, and the other things a heading outline can get wrong.

Command headonly rewrites the head of a document and passes the body through without parsing it.

Command highlight marks search terms in a document's text without ever injecting markup.

Command histogram counts the elements in a document by tag name and prints them as a bar chart.

Command hoiststyle moves inline style attributes into a stylesheet.

Command honeypot adds a decoy field to every form, and reports where each one went.

Command hreflang injects alternate-language links from a locale table.

Command idempotent runs a rewrite twice and reports whether the second pass changed anything.

Command idmerge concatenates several documents into one and keeps every id unique, rewriting the references as well as the ids.

Command ids reports duplicate id attributes, and the references they make ambiguous.

Command imgcdn points a page's images at an image CDN, with a width and a format.

Command importmap injects an import map before the first module script.

Command include expands Edge Side Include markers into fetched content, streaming, without buffering either the page or what it pulls in.

Command inlinesvg replaces small SVG images with the file's own markup, so that CSS can style them and the page can stop asking for them.

Command inputtype upgrades generic text inputs to email, tel and url where the field's name says what it holds - and, more often, declines to.

Command inventory lists the custom elements a page uses and says which of them nothing defines.

Command islands annotates the interactive regions of a page for partial hydration: which ones there are, which are inside which, and what each one needs to hydrate.

Command jsonld extracts every JSON-LD block from a document and reports what is wrong with each one.

Command keywords counts word frequencies in a document's own content, leaving out the parts every page on the site shares.

Command labels reports form controls with no label, and labels that point at nothing.

Command landmarks adds ARIA landmark roles to a document that has none.

Command lang adds a lang attribute to the elements whose own text is in a script other than the document's.

Command lazyload defers off-screen images and iframes.

Command linkify turns bare URLs in a document's text into links, without breaking the links it already has.

Command linkreport collects every anchor in a document with its target and text, and reports the ones a reader or a screen reader would struggle with.

Command linktext finds links whose text says nothing about where they go, and either flags or fixes them.

Command localise formats dates, numbers, currencies and percentages that a document has marked, in the locale a request asked for.

Command locate reports the source location of every match and proves each report by slicing the caller's own copy of the input.

Command mapembed replaces interactive map embeds with a static image and a link, so a page that shows a map does not load a map SDK to do it.

Command markdown converts the subset of HTML it understands to Markdown, and says what it dropped.

Command mentions turns @names and #tags in a document's text into links.

Command microdata extracts HTML microdata into a flat key-value report.

Command middleware wraps an http.Handler so that its HTML output is rewritten on the way to the client, without giving up streaming.

Command minifydiff minifies a document conservatively and then proves it did no harm, by parsing the input and the output and diffing what a parser sees.

Command mixed finds mixed content on an https page and, told to, refuses the page.

Command modernise replaces obsolete elements with the markup that means the same thing, moving their presentation into classes.

Command modulesplit turns a classic script into a module-and-fallback pair.

Command mojibake finds text that has been decoded with the wrong encoding, and says which one it probably was.

Command multipart rewrites the HTML parts of a multipart body and passes everything else through byte for byte.

Command needsrewrite decides whether a document is worth rewriting before rewriting it, which is the question a proxy asks thousands of times a second and mostly answers "no".

Command noevents removes the ways markup can execute script: inline event handler attributes, and javascript: URLs.

Command noindex adds a robots meta to pages whose path matches a pattern, so a staging host, a print view or a search-results page is not indexed.

Command noopener hardens links that open a new browsing context.

Command numbering prefixes each heading with its position in the document's outline: 1, 1.1, 1.2, 2, and so on.

Command observe reads a document, reports what is in it, and proves it changed nothing.

Command ogcompute fills in missing Open Graph tags from the page's own first heading and first image.

Command origins reports every origin a page would contact, and what asked for it.

Command pagenav adds rel=next and rel=prev link elements from a page's own pagination markup.

Command panics shows where a handler panic comes out, and what is left afterwards.

Command passthrough checks that a rewrite which changes nothing changes nothing.

Command pipeline runs a document through two rewriters, the output of the first being the input of the second, and shows what that buys over doing both in one pass.

Command placeholders resolves handlebars-style {{ name }} placeholders, choosing the escape by where the placeholder sits, and refuses the positions where no escape is enough.

Command plaintext converts a document to text, keeping the block structure as blank lines and newlines.

Command poisoned walks the ways a rewrite can fail and prints what each call returns afterwards.

Command preconnect adds resource hints for the third-party origins a page actually uses.

Command preserve runs a set of rewrites over a document and says which of them left it alone.

Command printstyles makes a page printable: a print stylesheet in the head, and a hint before every h2 so a section does not start at the bottom of a page.

Command proxy rewrites HTML response bodies in a reverse proxy, and skips the ones it must not touch.

Command queue runs a rewriter per goroutine over a queue of documents, checks that no worker sees another's work, and says how much of the time went on building rewriters rather than on rewriting.

Command readingtime counts the words in a document and estimates how long it takes to read, ignoring anything that is not prose.

Command rebase rewrites a <base> element away by resolving the URLs it affected.

Command redact removes email addresses and phone numbers from a document's text and from its attributes.

Command reencode converts a document from a single-byte legacy encoding to UTF-8 and proves the text survived.

Command references decodes the character references a document did not need.

Command regions applies a different set of handlers to each region of one document, split at offsets the caller gives, and refuses a split that would change what the document means.

Command rollinghash digests the rewritten output as it streams, without holding it.

Command sandbox hardens third-party iframes: it adds a sandbox attribute and a referrer policy, and reports the sandboxes that do not sandbox anything.

Command scrollwrap puts a scroll container around the elements that overflow a narrow screen.

Command selectorcheck reports every selector a rewrite cannot use, before the rewrite starts.

Command selectorcoverage reports which of a stylesheet's selectors never match a document.

Command servertiming times a rewrite and writes what it measured into the document it rewrote, as a Server-Timing comment at the end.

Command shadow gives every custom element a declarative shadow root, and gives it exactly once, so the same page can go through twice without gaining two.

Command shard spreads a page's asset URLs across a set of hostnames, always putting the same asset on the same one.

Command shrink reduces a failing document to its essence: the smallest input it can find that still fails the same way.

Command sizebudget streams HTML through a rewrite under a size budget, and stops as soon as the budget is spent.

Command slots fills named slots in a template with supplied fragments.

Command slugs assigns stable id attributes to a document's headings.

Command socialmeta reports the Open Graph and Twitter card metadata a page carries, and what is missing for a link to it to render well.

Command split cuts a document into parts at a chosen heading level, and makes each part stand on its own.

Command sprite injects an SVG sprite once and points the page's icons at it.

Command srcset builds a responsive srcset for every image from its src, a width list and an image CDN template.

Command sri adds subresource integrity attributes to a document's scripts and stylesheets, from a manifest of hashes, and reports every subresource the manifest does not cover.

Command stopafter rewrites a document until it meets a marker and copies everything after it.

Command stopwhen rewrites a stream that never ends and stops when it has what it came for.

Command streamvsmemory runs the same rewrite twice - once in memory, once streamed - and reports what differs.

Command strictmode shows what each of the two parsing modes makes of a document.

Command summary extracts a page's summary and stops reading as soon as it has one.

Command tableaudit reports tables whose header cells cannot be associated with their data.

Command tablecsv extracts every table in a document as CSV, expanding colspan and rowspan so every row has the same number of fields.

Command tablejson converts each table in a document to JSON, one object per row, keyed by the header cells.

Command tablelayout converts a div-based page into the table markup an email client will render, and refuses the conversions whose result would depend on the document's doctype.

Command tailcomment emits a summary of what a rewrite changed as a trailing HTML comment, which is the shape a build stamp or a debug trace usually takes: invisible in the page, there in the source.

Command tailreport appends a generated report to the end of every document it rewrites, and it is here because the obvious way to do that holds the whole report in memory.

Command tee streams a document to two destinations at once - one rewritten, one exactly as it arrived - from a single read of the input, and reports how far apart they ran.

Command textmap reports the source location of every text chunk and reconstructs the document from what it was told.

Command texttruth reconstructs a document's text - the characters, not the source bytes - from the text chunks a rewrite reports, and it exists because the two are not the same thing and the difference is four rules with three different element lists.

Command toc builds a table of contents from a document's headings and inserts it at a marker.

Command transitions gives view-transition names to the elements that appear on both of two pages, so a browser can animate between them.

Command translate adds translate="no" to the elements whose text is not prose, so a machine translator leaves them alone.

Command tweetquote rewrites embedded tweets into plain blockquotes with attribution, so a page that quotes someone does not load a third-party script to show the quote.

Command twoways runs two rewriters over the same document at the same time: one that transforms it and one that only reports on it.

Command typography applies typographic quotes and dashes to a document's prose, and leaves code alone.

Command unbounded rewrites a document larger than any buffer worth holding, and says which handler patterns keep the memory flat.

Command units converts imperial quantities in prose to metric, wrapping each conversion in a span whose title keeps what the page said.

Command untrack removes tracking from an HTML document as it streams past: tracking parameters from every URL, and tracking pixels from the markup.

Command upgrade rewrites http:// subresources to https:// as a document streams past, and reports what it changed and what it could not.

Command viewport fixes a missing or harmful viewport meta tag.

Command weight reports the total byte weight of the scripts, styles and images a page references.

Command widgets turns legacy widget markup into web component markup: a container becomes a custom element, the state it kept in classes and data attributes becomes properties, and the parts it kept in nested divs become slots.

Command widows joins the last two words of every heading with a non-breaking space, so a heading cannot wrap with one word alone on the last line.

Command worstshape finds the document shape a handler set is slowest on, by running the same handlers over documents of the same size in different shapes and ranking them.

Command ytembed replaces YouTube embeds with a thumbnail that loads the player on click, so a page does not fetch several hundred kilobytes of player from a third party before the reader asks for it.