AO3: [Wrangling] Mark Illegal Characters in Canonicals

Warns about any canonical tag that includes characters which should, per guidelines, be avoided. Checks on new tag, edit tag, search results, wrangle bins, and tag landing pages
As of 2024-08-02. See the latest version.
Install this script ?
Ask a question, post a review, or report the script.
Wrap lines
// ==UserScript==
// @name         AO3: [Wrangling] Mark Illegal Characters in Canonicals
// @namespace    https://greasyfork.org/en/users/906106-escctrl
// @version      2.0
// @description  Warns about any canonical tag that includes characters which should, per guidelines, be avoided. Checks on new tag, edit tag, search results, wrangle bins, and tag landing pages
// @author       escctrl
// @match        *://*.archiveofourown.org/tags/*
// @license      MIT
// @grant        none
// ==/UserScript==
 
(function() {
    'use strict';
 
    // we wanna check on a bunch of different pages, and everywhere the check is slightly different
 
    var page_url = window.location.pathname;
    // just in case the URL ended with a / we get rid of that
    // that usually doesn't happen from AO3 links on the site, but may be how browsers store bookmarks or history
    if (page_url.endsWith("/")) { page_url = page_url.slice(0, page_url.length-1); }
 
    if (page_url == "/tags/new") checkAsYouType(); // New Tag page
    else if (page_url == "/tags/search") checkSearchResults(); // Tag Search page
    else if (page_url.match(/^\/tags\/.+\/edit$/gi)) checkEditTag(); // Edit page
    else if (page_url.match(/^\/tags\/.+\/wrangle$/gi)) checkBinTags(); // Wrangle page
    else if (page_url.match(/^\/tags\/[^\/]+$/gi)) checkTag(); // Tag Landing page
      // that excludes anything including another slash, which would only incorrectly match on tags/new and tags/search
      // but those would have already jumped into the other functions and would never get here
})();
 
// *************** GENERAL FUNCTIONS ***************
 
// a holistic function to check
// not allowed: non-latin (including accented) characters and special chars (with a few exceptions)
//              two apostrophes '' (used instead of a quote ")
//              a slash with spaces before or after
//              an apersand without spaces before and after
//              space at the beginning or end of the string
//              multiple spaces after each other
// this returns the matched characters in an array
function hasIllegalChars(string) {
    return string.match(/[^\p{Script=Latin}0-9 \-().&/'"|:!]|'{2,}| \/|\/ |[^ ]&|&[^ ]| {2,}|^ | $/gui);
}
 
// similar to above, but in fandoms we allow letters, numbers and tone/accent marks of ANY script, not just Latin
// also more special characters are allowed
function hasFandomIllegalChars(string) {
    return string.match(/[^\p{L}\p{M}\p{N} \-().&/'"|:!#?_]|'{2,}| {2,}|^ | $/gui);
}
 
// print a box to explain the problem
function insertHeadsUp(illegalChars, refNode, befNode = null, inline = false) {
    // describe non-printable chars and other hard to identify issues
    illegalChars.forEach((val, ix) => {
        if (val == "''") illegalChars[ix] = "2 single quotes";
        else if (val == "/ " || val == " /") illegalChars[ix] = "space around the /";
        else if (val.slice(0,1) == "&" || val.slice(-1) == "&") illegalChars[ix] = "no space around the &";
        else if (val.trim() == "")
            illegalChars[ix] = (val == "\t") ? "tab" :
                               (val === " " && ix == 0 && refNode.childNodes[0].value.slice(0, 1) === " ") ? "space in front" :
                               (val === " " && refNode.childNodes[0].value.slice(-1) === " ") ? "space at end" :
                               "multiple spaces";
    });
    // setting up the div to contain the heads-up to the user
    const warningNode = document.createElement("div");
    warningNode.id = "illegalChars";
    warningNode.classList.add("notice");
 
    warningNode.innerHTML = "<p>Questionable: " + illegalChars.join(", ") + "</p>";
 
    if (inline) {
        warningNode.style.display = "inline-block";
        warningNode.style.padding = "0";
        warningNode.style.margin = "0.1em 0.1em 0.1em 0.5em";
        warningNode.children[0].style.padding = "0.1em 0.3em";
        warningNode.children[0].style.fontWeight = "normal";
    }
 
    // if that already exists, we're gonna replace it rather than add more divs
    if (refNode.querySelector("#illegalChars")) refNode.replaceChild(warningNode, refNode.querySelector("#illegalChars"));
    else refNode.insertBefore(warningNode, befNode);
}
 
// remove the explain box again
function removeHeadsUp(refNode) {
    if (refNode.querySelector("#illegalChars")) refNode.removeChild(refNode.querySelector("#illegalChars"));
}
 
// *************** PAGE HANDLING FUNCTIONS ***************
 
// New tag page
function checkAsYouType() {
    // a little JS magic to quickly add the same event listener to all elements
    [ document.getElementById("tag_name"),
      document.getElementById('tag_type_fandom'),
      document.getElementById('tag_type_character'),
      document.getElementById('tag_type_relationship'),
      document.getElementById('tag_type_freeform')
    ].forEach((el) => {
        el.addEventListener("input", () => {
            var checkNode = document.getElementById("tag_name");
 
            // which tag type are you trying to create? fandom or anything else?
            const isFandom = document.getElementById('tag_type_fandom').checked;
            var issues = (isFandom) ? hasFandomIllegalChars(checkNode.value) : hasIllegalChars(checkNode.value);
            if (issues !== null) insertHeadsUp(issues, checkNode.parentNode);
            else removeHeadsUp(checkNode.parentNode);
 
            // length counter
            let label = document.querySelector('dt label[for="tag_name"]');
            label.innerText = "Name (" + checkNode.value.length +")";
 
            // extra special handling: tag length>100 error
            const refNode = checkNode.parentNode;
            if (checkNode.value.length > 100) {
                const errorNode = document.createElement("div");
                errorNode.id = "tooLong";
                errorNode.classList.add("error");
                errorNode.innerHTML = "<p>Sorry, you'll need to trim this down. You're at "+ checkNode.value.length +" characters!</p>";
 
                // if that already exists, we're gonna replace it rather than add more divs
                if (refNode.querySelector("#tooLong")) refNode.replaceChild(errorNode, refNode.querySelector("#tooLong"));
                else refNode.insertBefore(errorNode, null);
            }
            else if (refNode.querySelector("#tooLong")) refNode.removeChild(refNode.querySelector("#tooLong"));
        });
    });
    // on page load, trigger event once. browser remembers previous form selections/input upon page refresh and box would otherwise not appear until another change is made
    document.getElementById("tag_name").dispatchEvent(new Event("input"));
}
 
// Landing page
function checkTag() {
    // only if the viewed tags is canonical
    var tagDescr = document.querySelector(".tag>p").innerText;
    if (tagDescr.indexOf("It's a common tag") < 0) return true;
 
    // first the viewed tag itself
    var checkNode = document.querySelector(".tag .header h2.heading");
    var tagType = tagDescr.match(/This tag belongs to the (.+) Category/i);
    tagType = tagType[1];
    var issues = (tagType == "Fandom") ? hasFandomIllegalChars(checkNode.innerText) : hasIllegalChars(checkNode.innerText);
    if (issues !== null) insertHeadsUp(issues, checkNode.parentNode.parentNode, checkNode.parentNode.parentNode.children[1]);
 
    // then the meta and subtags (if any)
    checkNode = document.querySelectorAll("div.meta.listbox a.tag, div.sub.listbox a.tag");
    checkNode.forEach((n) => {
        var issues = (tagType == "Fandom") ? hasFandomIllegalChars(n.innerText) : hasIllegalChars(n.innerText);
        if (issues !== null) insertHeadsUp(issues, n.parentNode, n.parentNode.children[1], true);
    });
    // it would be really cool if we could check Parent Tags as well, but we can't tell which of those are fandoms vs. anything else
}
 
// Wrangle Bin Page
// sadly we can't tell here at all if we're ever looking at fandoms
function checkBinTags() {
    // this needs a different approach to the logic:
    // don't check show=mergers at all, too repetitive
    var searchParams = new URLSearchParams(window.location.search);
    if (searchParams.get('show') == "mergers") return true;
 
    // create a key -> value pair Map of the table columns, so we know which column to check
    var tableIndexes = new Map();
    document.querySelectorAll("#wrangulator table thead th").forEach((th, ix) => {
        tableIndexes.set(th.innerText, ix);
    });
 
    // now we can loop through the list of tags
    var issues, checkNode;
    var checkRows = document.querySelectorAll("#wrangulator table tbody tr");
    checkRows.forEach((r) => {
        // if there's a column "Canonical" and the cell says "Yes" then we check the tag itself
        if (tableIndexes.has("Canonical") && r.cells[tableIndexes.get("Canonical")].innerText == "Yes") {
            checkNode = r.cells[0].querySelector("label");
            issues = searchParams.get('show') == "fandoms" ? hasFandomIllegalChars(checkNode.innerText) : hasIllegalChars(checkNode.innerText);
            if (issues !== null) insertHeadsUp(issues, checkNode.parentNode);
        }
 
        // if there's a column "Synonym", we check the content of that cell (there'll only be one tag)
        if (tableIndexes.has("Synonym") && r.cells[tableIndexes.get("Synonym")].innerText.trim() !== "") {
            checkNode = r.cells[tableIndexes.get("Synonym")].querySelector("a");
            issues = searchParams.get('show') == "fandoms" ? hasFandomIllegalChars(checkNode.innerText) : hasIllegalChars(checkNode.innerText);
            if (issues !== null) insertHeadsUp(issues, checkNode.parentNode);
        }
 
        // if there's a column "Characters", we check the content of that cell (there might be multiple tags)
        if (tableIndexes.has("Characters") && r.cells[tableIndexes.get("Characters")].innerText.trim() !== "") {
            checkNode = r.cells[tableIndexes.get("Characters")].querySelectorAll("a");
            checkNode.forEach((n) => {
                issues = hasIllegalChars(n.innerText);
                if (issues !== null) insertHeadsUp(issues, n.parentNode);
            });
        }
 
        // if there's a column "Metatag", we check the content of that cell (there might be multiple tags)
        if (tableIndexes.has("Metatag") && r.cells[tableIndexes.get("Metatag")].innerText.trim() !== "") {
            checkNode = r.cells[tableIndexes.get("Metatag")].querySelectorAll("a");
            checkNode.forEach((n) => {
                issues = searchParams.get('show') == "fandoms" ? hasFandomIllegalChars(checkNode.innerText) : hasIllegalChars(n.innerText);
                if (issues !== null) insertHeadsUp(issues, n.parentNode);
            });
        }
    });
}
 
// Tag Search
function checkSearchResults() {
    // with search results table userscript enabled
    var checkNodes = document.querySelectorAll("table#resulttable .resulttag.canonical a, table#resulttable .resultName.canonical a");
    checkNodes.forEach((n) => {
        var issues = (n.parentNode.parentNode.querySelector('td.resulttype').title == "Fandom") ? hasFandomIllegalChars(n.innerText) : hasIllegalChars(n.innerText);
        if (issues !== null) insertHeadsUp(issues, n.parentNode, null, true);
    });
 
    // with plain search results page
    checkNodes = document.querySelectorAll("ol.tag li span.canonical a.tag");
    checkNodes.forEach((n) => {
        var issues = (n.parentNode.firstChild.textContent.trim() == "Fandom:") ? hasFandomIllegalChars(n.innerText) : hasIllegalChars(n.innerText);
        if (issues !== null) insertHeadsUp(issues, n.parentNode.parentNode, null, true);
    });
}
 
// Edit Tag Page
function checkEditTag() {
    const tagCanonical = document.getElementById('tag_canonical');
    const tagType = document.querySelector('#edit_tag fieldset:first-of-type dd strong').innerText;
    var issues;
 
    // initial check only if the tag is already canonical
    if (tagCanonical.checked) {
        var checkNode = document.getElementById("tag_name");
        issues = (tagType == "Fandom") ? hasFandomIllegalChars(checkNode.value) : hasIllegalChars(checkNode.value);
        if (issues !== null) insertHeadsUp(issues, checkNode.parentNode);
    }
 
    // if the tag's canonical status is changed
    tagCanonical.addEventListener("input", (event) => {
        var checkNode = document.getElementById("tag_name");
        if (event.target.checked) {
            var issues = (tagType == "Fandom") ? hasFandomIllegalChars(checkNode.value) : hasIllegalChars(checkNode.value);
            if (issues !== null) insertHeadsUp(issues, checkNode.parentNode);
            else removeHeadsUp(checkNode.parentNode);
        }
        else removeHeadsUp(checkNode.parentNode);
    });
 
    // if this is a synonym, check the canonical tag it's synned to
    const synonym = document.querySelector('#edit_tag fieldset:first-of-type dd ul.autocomplete .added.tag');
    if (synonym !== null) {
        issues = (tagType == "Fandom") ? hasFandomIllegalChars(synonym.firstChild.textContent.trim()) : hasIllegalChars(synonym.firstChild.textContent.trim());
        if (issues !== null) insertHeadsUp(issues, synonym.parentNode.parentNode, synonym.parentNode.parentNode.children[1]);
    }
 
    // if this is canonical, check its sub- and metatags
    const metasubs = document.querySelectorAll('#parent_MetaTag_associations_to_remove_checkboxes ul li a, #child_SubTag_associations_to_remove_checkboxes ul li a');
    if (metasubs !== null) {
        metasubs.forEach((n) => {
            issues = (tagType == "Fandom") ? hasFandomIllegalChars(n.innerText) : hasIllegalChars(n.innerText);
            if (issues !== null) insertHeadsUp(issues, n.parentNode);
        });
    }
 
    // if this is any other type of tag that's in a fandom, check the fandom tag
    const fandoms = document.querySelectorAll('#parent_Fandom_associations_to_remove_checkboxes ul li a');
    if (fandoms !== null) {
        fandoms.forEach((n) => {
            issues = hasFandomIllegalChars(n.innerText);
            if (issues !== null) insertHeadsUp(issues, n.parentNode);
        });
    }
 
    // if this is a relationship, check the tagged characters
    const chars = document.querySelectorAll('#parent_Character_associations_to_remove_checkboxes ul li a');
    if (chars !== null) {
        chars.forEach((n) => {
            issues = hasIllegalChars(n.innerText);
            if (issues !== null) insertHeadsUp(issues, n.parentNode);
        });
    }
}