☰

Image Webinfo

Extract JSON-/string-style info from Pixiv/Twitter/Booru image pages.

질문, 리뷰하거나, 이 스크립트를 신고하세요. 
  1. // ==UserScript==
  2. // @name          Image Webinfo
  3. // @name:zh-CN    图片网页信息提取
  4. // @version       1.3.2
  5. // @description   Extract JSON-/string-style info from Pixiv/Twitter/Booru image pages.
  6. // @description:zh-CN 将 Pixiv/Twitter/Booru 图片页面中的信息以 JSON 或字符串形式导出。
  7. // @author        wklchris
  8. // @match         https://danbooru.donmai.us/posts/*
  9. // @match         https://www.pixiv.net/artworks/*
  10. // @match         https://x.com/*/status/*
  11. // @grant         none
  12. // @license       MIT
  13. // @namespace https://greasyfork.org/users/672201
  14. // ==/UserScript==
  15.  
  16. (function() {
  17.   'use strict';
  18.   
  19.   // --- Customization ---
  20.   // Specify returned keys when copy string
  21.   let pixiv_str_keys = [
  22.     // 'artist',
  23.     // 'illust_id',
  24.     'artist_id',
  25.     'post_date',
  26.     'post_title'
  27.   ];
  28.   let booru_str_keys = [
  29.     // 'artists',
  30.     // 'copyrights',
  31.     // 'characters',
  32.     // 'general',
  33.     'booru_id',
  34.     'rating_level'
  35.   ];
  36.   
  37.   // --- Main ---
  38.   let artwork_data = {};
  39.   let export_json_btn = document.createElement("button");
  40.   export_json_btn.innerHTML = "Export WebInfo as JSON";
  41.   let export_str_btn = document.createElement("button");
  42.   export_str_btn.innerHTML = "Export WebInfo as String";
  43.  
  44.   // General functions
  45.   function logMessage(msg) {
  46.     console.log("[ImageWebInfo] " + msg);
  47.   }
  48.   
  49.   let popMessage = 'Image webinfo exported to clipboard.';
  50.   function copyToClipboard(s) {
  51.     try {
  52.       navigator.clipboard.writeText(s);
  53.     } catch (err) {
  54.       popMessage = 'Image webinfo export failed.';
  55.       logMessage(popMessage);
  56.     }
  57.   }
  58.  
  59.   function attemptedQuerySelector(qs, callback) {
  60.     // querySelector may not find the element when the website loads slowly.
  61.     // Keep attempting until found or exceed the given maximum attempts.
  62.     var attempts = 0, max_attempts = 20;
  63.     var attemptCall = function() {
  64.         var elem = document.querySelector(qs);
  65.         if (elem) {
  66.           console.log(`Found querySelector '${qs}' in ${attempts} attempts`);
  67.           return callback(elem);
  68.         } else {
  69.             attempts++;
  70.             if (attempts >= max_attempts) {
  71.                 console.warn(`Could not find: ${qs} within ${max_attempts} attempts`);
  72.             } else {
  73.                 setTimeout(attemptCall, 4000*(attempts + 1)/max_attempts + 1000)
  74.             }
  75.         }
  76.     };
  77.     attemptCall();
  78.   }
  79.   
  80.   // --- Pixiv posts ---
  81.  
  82.   if (location.hostname.includes("pixiv")) {
  83.     function getPixivInfo(as_JSON=true) {
  84.       let data = {};
  85.       // Get user
  86.       let user_tag = document.querySelector('aside div[aria-haspopup="true"] > div > a');
  87.       data['artist'] = user_tag.childNodes[0].innerText 
  88.       let split_user_url = user_tag.href.split('/');
  89.       data['artist_id'] = split_user_url[split_user_url.length - 1];
  90.       // Get post id
  91.       let split_url = location.href.split('/');
  92.       data['illust_id'] = split_url[split_url.length - 1];
  93.       // Get upload date
  94.       let timestr = document.getElementsByTagName('time')[0].innerText;
  95.       let time_seps = {'year': '年', 'month': '月', 'day': '日'};
  96.       let i_start = 0, i_end = -1, tmp_data = {};
  97.       for (const [key, sep] of Object.entries(time_seps)) {
  98.         i_start = i_end+1;
  99.         i_end = timestr.search(sep);
  100.         tmp_data[key] = timestr.substring(i_start, i_end);
  101.       }
  102.       data['post_date'] = `${tmp_data['year']}-${tmp_data['month'].padStart(2, 0)}-${tmp_data['day'].padStart(2, 0)}`;
  103.       // Get post title (if having a title)
  104.       let title_dom = document.querySelector('figcaption h1');
  105.       data['post_title'] = title_dom ? title_dom.innerText : '' ;
  106.  
  107.       let s = "";
  108.       if (as_JSON == true) {
  109.         s = JSON.stringify(data)
  110.       } else {
  111.         s = pixiv_str_keys.map((x, i) => data[x]).join("\t");
  112.       }
  113.       copyToClipboard(s);
  114.       return data;
  115.     }
  116.  
  117.     // Add buttons & their click functions
  118.     let search_booru_btn = document.createElement("button");
  119.     search_booru_btn.innerHTML = "Search artist on Booru";
  120.     let search_booru_illust_btn = document.createElement("button");
  121.     search_booru_illust_btn.innerHTML = "Search illust on Booru";
  122.  
  123.     attemptedQuerySelector('aside section button', (exist_dom) => {
  124.       exist_dom.parentNode.insertBefore(export_json_btn, exist_dom.nextSibling);
  125.       exist_dom.parentNode.insertBefore(export_str_btn, export_json_btn.nextSibling);
  126.       exist_dom.parentNode.insertBefore(search_booru_btn, export_str_btn.nextSibling);
  127.       exist_dom.parentNode.insertBefore(search_booru_illust_btn, search_booru_btn.nextSibling);
  128.     });
  129.  
  130.     export_json_btn.onclick = function() {
  131.       artwork_data = getPixivInfo(true);
  132.       export_json_btn.style.background = 'lightgreen';
  133.       export_str_btn.style.removeProperty("background");
  134.     };
  135.     export_str_btn.onclick = function() {
  136.       artwork_data = getPixivInfo(false);
  137.       export_str_btn.style.background = 'lightgreen';
  138.       export_json_btn.style.removeProperty("background");
  139.     };
  140.     search_booru_btn.onclick = function() {
  141.       if (!('artist_id' in artwork_data)) {
  142.         artwork_data = getPixivInfo(false);
  143.       }
  144.       let _prefix = "https://danbooru.donmai.us/artists?commit=Search&search%5Border%5D=created_at&search%5Burl_matches%5D=https%3A%2F%2Fwww.pixiv.net%2Fusers%2F";
  145.       let _url = _prefix + artwork_data["artist_id"];
  146.       window.open(_url, "_blank");
  147.     };
  148.     search_booru_illust_btn.onclick = function() {
  149.       if (!('illust_id' in artwork_data)) {
  150.         artwork_data = getPixivInfo(false);
  151.       }
  152.       let illust_search_prefix = "https://danbooru.donmai.us/posts?tags=pixiv%3A";
  153.       let illust_search_url = illust_search_prefix + artwork_data["illust_id"];
  154.       window.open(illust_search_url, "_blank");
  155.     };
  156.   }
  157.  
  158.   // --- Twitter/X Post ---
  159.   if (["twitter.com", "x.com"].includes(location.hostname)) {
  160.     let twitter_url_id = '';
  161.     function getTwitterInfo(as_JSON=true) {
  162.       let data = {};
  163.       // Get user
  164.       let user_tag = document.querySelector('article > div div[data-testid="User-Name"]');
  165.       data['artist'] = user_tag.childNodes[0].innerText;
  166.       // -- Twitter user only has username, doesn't have a user ID number.
  167.       //    The matched id is only used for creating search buttons.
  168.       data['artist_id'] = '';
  169.       let split_user_url = user_tag.childNodes[1].querySelector('a').href.split('/');
  170.       twitter_url_id = split_user_url[split_user_url.length - 1];
  171.       // Get post id
  172.       data['illust_id'] = "T" + location.href.split('status/')[1].split('/')[0];
  173.       // Get upload date
  174.       data['post_date'] = document.getElementsByTagName('time')[0].dateTime.split('T')[0];
  175.       // Get post title (if having a title)
  176.       let tweet_nodes = document.querySelector('article div[data-testid="tweetText"]').querySelectorAll('span, img');
  177.       data['post_title'] = Array.from(tweet_nodes).map(
  178.         elem => elem.alt || elem.innerText.trim()).filter(d => d.length && d[0] != '#'
  179.       ).join('').split('\n')[0];
  180.  
  181.       let s = "";
  182.       if (as_JSON == true) {
  183.         s = JSON.stringify(data)
  184.       } else {
  185.         s = pixiv_str_keys.map((x, i) => data[x]).join("\t");
  186.       }
  187.       copyToClipboard(s);
  188.       return data;
  189.     }
  190.  
  191.     // Add buttons & their click functions
  192.     let search_booru_btn = document.createElement("button");
  193.     search_booru_btn.innerHTML = "Search artist on Booru";
  194.     let search_booru_illust_btn = document.createElement("button");
  195.     search_booru_illust_btn.innerHTML = "Search illust on Booru";
  196.  
  197.     attemptedQuerySelector('div[data-testid="User-Name"]', (exist_dom) => {
  198.       exist_dom.parentNode.insertBefore(export_json_btn, exist_dom.nextSibling);
  199.       exist_dom.parentNode.insertBefore(export_str_btn, export_json_btn.nextSibling);
  200.       exist_dom.parentNode.insertBefore(search_booru_btn, export_str_btn.nextSibling);
  201.       exist_dom.parentNode.insertBefore(search_booru_illust_btn, search_booru_btn.nextSibling);
  202.     });
  203.  
  204.     export_json_btn.onclick = function() {
  205.       artwork_data = getTwitterInfo(true);
  206.       export_json_btn.style.background = 'lightgreen';
  207.       export_str_btn.style.removeProperty("background");
  208.     };
  209.     export_str_btn.onclick = function() {
  210.       artwork_data = getTwitterInfo(false);
  211.       export_str_btn.style.background = 'lightgreen';
  212.       export_json_btn.style.removeProperty("background");
  213.     };
  214.     search_booru_btn.onclick = function() {
  215.       if (!('artist_id' in artwork_data)) {
  216.         artwork_data = getTwitterInfo(false);
  217.       }
  218.       let _prefix = "https://danbooru.donmai.us/artists?commit=Search&search%5Border%5D=created_at&search%5Burl_matches%5D=https%3A%2F%2Fwww.twitter.com%2F";
  219.       let _url = _prefix + twitter_url_id;
  220.       window.open(_url, "_blank");
  221.     };
  222.     search_booru_illust_btn.onclick = function() {
  223.       if (!('illust_id' in artwork_data)) {
  224.         artwork_data = getTwitterInfo(false);
  225.       }
  226.       let illust_search_prefix = "https://danbooru.donmai.us/posts?tags=source%3A";
  227.       let illust_search_url = illust_search_prefix + encodeURIComponent(location.href.replace('x.com/','twitter.com/'));
  228.       window.open(illust_search_url, "_blank");
  229.     };
  230.   }
  231.  
  232.   // --- Booru posts ---
  233.   if (location.hostname.includes("booru")) {
  234.     // Add renamed link
  235.     let renamed_download_link = document.createElement("a");
  236.     renamed_download_link.href = document.getElementById('post-option-download').querySelector('a').href;
  237.     renamed_download_link.style.display = 'inline-block';
  238.     renamed_download_link.style.paddingBottom = '3px';
  239.     renamed_download_link.text = "RENAMED FILENAME";
  240.     // Add listener so that the renamed text is copied when right-click
  241.     renamed_download_link.addEventListener('contextmenu', function(event) {
  242.       let _renamed_filename = renamed_download_link.download;
  243.       navigator.clipboard.writeText(_renamed_filename)
  244.     });
  245.  
  246.     function getBooruInfo(as_JSON=true) {
  247.       let data = {};
  248.       let tag_list = document.getElementById("tag-list").querySelector("div");
  249.  
  250.       function query_tags(ul_class, sep=", ") {
  251.         // Get tag list by ul_class name. Join by sep if multiple.
  252.         let query_str = `ul.${ul_class} a.search-tag`;
  253.         let _tags = tag_list.querySelectorAll(query_str);
  254.         return Array.from(_tags).map((x, i) => x.innerText).join(sep)
  255.       }
  256.       data["artists"] = query_tags("artist-tag-list");
  257.       data["copyrights"] = query_tags("copyright-tag-list");
  258.       data["characters"] = query_tags("character-tag-list");
  259.       data["general"] = query_tags("general-tag-list");
  260.  
  261.       // Get information (rating-levels, etc.)
  262.       function query_by_id(dom_id, remove_str) {
  263.         let _text = document.getElementById(dom_id).innerText;
  264.         return _text.replace(remove_str, "").trim();
  265.       }
  266.       data["booru_id"] = query_by_id("post-info-id", "ID:");
  267.       data["rating_level"] = query_by_id("post-info-rating", "Rating:");
  268.  
  269.       let s = "";
  270.       if (as_JSON == true) {
  271.         s = JSON.stringify(data)
  272.       } else {
  273.         s = booru_str_keys.map((x, i) => data[x]).join("\t");
  274.       }
  275.       copyToClipboard(s);
  276.       return data;
  277.     }
  278.  
  279.     function updateRenamedDownloadLink() {
  280.       let renamed_filename = '';
  281.       const file_link = document.getElementById('post-info-size').querySelector('a').href;
  282.       const post_sources = document.getElementById('post-info-source').querySelectorAll('a');
  283.       const post_url = post_sources[0].href;
  284.       const post_imgurl = post_sources[post_sources.length - 1].href;
  285.  
  286.       function getSitePostId() {
  287.         const _site_url = post_sources[0].href;
  288.         const _site_id = _site_url.substring(_site_url.lastIndexOf('/') + 1);
  289.         return _site_id;
  290.       }
  291.  
  292.       function getSiteFilenamePrefix(site_url) {
  293.         const prefix_dict = {
  294.           'pixiv.net': '',
  295.           'fanbox.cc': 'PF',
  296.           'patreon.com': 'PT',
  297.           'twitter.com': 'T',
  298.           'x.com': 'T'
  299.         };
  300.         for (const key in prefix_dict) {
  301.           if (site_url.includes(key)) {
  302.               return prefix_dict[key];
  303.           }
  304.         }
  305.         return '#';
  306.       }
  307.  
  308.       if (post_url.includes('pixiv.net') ) {
  309.         // If pixiv, simply return original filename (from last <a> in sources)
  310.         renamed_filename = post_imgurl.substring(post_imgurl.lastIndexOf('/') + 1);
  311.       } else {
  312.         // Otherwise, find the parent/children relations of the current post
  313.         const children_previews = document.querySelector('.posts-container');
  314.         let id_child = 0;  // If not found parent
  315.         if (children_previews) {
  316.           const articles = children_previews.querySelectorAll('article');
  317.           for (let i = 0; i < articles.length; i++) {
  318.               if (articles[i].classList.contains('current-post')) {
  319.                 id_child = i;
  320.                 break;
  321.               };
  322.           }
  323.         }
  324.         // Concat the filename
  325.         const file_prefix = getSiteFilenamePrefix(post_url);
  326.         const site_id = getSitePostId();
  327.         const file_ext = file_link.substring(file_link.lastIndexOf('.'));
  328.         renamed_filename = file_prefix + site_id + "_p" + id_child + file_ext;
  329.       }
  330.       // Update the renamed link element
  331.       renamed_download_link.download = renamed_filename;
  332.       const meta_tags = document.querySelectorAll('ul.meta-tag-list a');
  333.       // If it is bad id (the post has revised version), mark it out in the text
  334.       let msg_bad_id = '';
  335.       for (const _meta_tag of meta_tags) {
  336.         if (_meta_tag.innerText == 'bad id') {
  337.           msg_bad_id = '(bad) ';
  338.           break;
  339.         }
  340.       }
  341.       renamed_download_link.text = msg_bad_id + renamed_filename;
  342.     }
  343.     // Update link for (renamed) image file download
  344.     updateRenamedDownloadLink();
  345.  
  346.     // Add renamed link and buttons
  347.     attemptedQuerySelector('section[id="search-box"]', (exist_dom) => {
  348.       exist_dom.parentNode.insertBefore(renamed_download_link, exist_dom.nextSibling);
  349.       exist_dom.parentNode.insertBefore(export_json_btn, renamed_download_link.nextSibling);
  350.       exist_dom.parentNode.insertBefore(export_str_btn, export_json_btn.nextSibling);
  351.     });
  352.  
  353.     export_json_btn.onclick = function() {
  354.       artwork_data = getBooruInfo(true);
  355.       export_json_btn.style.background = 'lightgreen';
  356.       export_str_btn.style.removeProperty("background");
  357.     };
  358.     export_str_btn.onclick = function() {
  359.       artwork_data = getBooruInfo(false);
  360.       export_str_btn.style.background = 'lightgreen';
  361.       export_json_btn.style.removeProperty("background");
  362.     };
  363.   }
  364.  
  365. })();