From 822df5f40d335084c8bca3b91124b267d42b4754 Mon Sep 17 00:00:00 2001
From: stijn-uva <42036349+stijn-uva@users.noreply.github.com>
Date: Thu, 1 Oct 2026 13:46:01 +0000
Subject: [PATCH] chore: sync map_item for twitter-import from 4CAT
 7cba15f91a5321aa5049bbe8814c3c85009f6d12

---
 modules/twitter.js | 374 ++++++++++++++++++++++-----------------------
 1 file changed, 185 insertions(+), 189 deletions(-)

diff --git a/modules/twitter.js b/modules/twitter.js
index 5920ca7..2840506 100644
--- a/modules/twitter.js
+++ b/modules/twitter.js
@@ -189,10 +189,10 @@ export function capture(response, source_platform_url, source_url) {
 // === auto-generated by 4cat map_item sync — BLOCK REPLACED AUTOMATICALLY ===
 // (regenerated from datasources/twitter-import/search_twitter.py)
 function map_item_modern(tweet) {
-    // Resolve author information
-    const userResult = (tweet.core && tweet.core.user_results && tweet.core.user_results.result) || {};
+    // Resolve author information – user object may be missing
+    const userResult = (tweet.core?.user_results?.result) || {};
     const author = map_user(userResult);
-    const authorScreenName = author.screen_name || _screen_name_from_media(py_get(tweet, "legacy", {}));
+    const authorScreenName = author.screen_name || _screen_name_from_media(py_get(tweet, 'legacy') || {});
     const authorFullname = author.fullname;
     const authorAvatarUrl = author.avatar_url;
     const authorBannerUrl = author.banner_url;
@@ -203,211 +203,207 @@ function map_item_modern(tweet) {
     const authorLocation = author.location;
 
     const tweetLink = authorScreenName
-        ? `https://x.com/${authorScreenName}/status/${py_get(tweet, "id")}`
-        : `https://x.com/i/web/status/${py_get(tweet, "rest_id")}`;
+        ? `https://x.com/${authorScreenName}/status/${tweet.id}`
+        : `https://x.com/i/web/status/${tweet.rest_id}`;
 
-    const createdAtStr = py_get(py_get(tweet, "legacy", {}), "created_at");
-    const createdDate = new Date(createdAtStr);
-    const unixTimestamp = Math.floor(createdDate.getTime() / 1000);
-    const timestamp = formatUtcTimestamp(unixTimestamp);
-    let withheld = false;
+    // Parse timestamp – X uses a format compatible with Date parsing
+    const dateObj = new Date(py_get(tweet, 'legacy').created_at);
+    const unixSeconds = Math.floor(dateObj.getTime() / 1000);
+    const timestamp = formatUtcTimestamp(unixSeconds);
 
+    let withheld = false;
     let body = get_full_text(tweet);
 
-    const retweetObj = py_get(py_get(tweet, "legacy", {}), "retweeted_status_result");
-    let retweetedUser = "";
-    if (retweetObj) {
-        let rtResult = retweetObj.result;
-        if (rtResult && rtResult.tweet) {
-            rtResult = rtResult.tweet;
+    // Retweet handling
+    const retweet = py_get(tweet, 'legacy').retweeted_status_result;
+    let retweetedUser = '';
+    if (retweet) {
+        if (retweet.result && retweet.result.tweet) {
+            retweet.result = retweet.result.tweet;
         }
-        const rtUserResult = (rtResult && rtResult.core && rtResult.core.user_results && rtResult.core.user_results.result) || {};
-        retweetedUser = map_user(rtUserResult).screen_name || _screen_name_from_media(py_get(rtResult, "legacy", {}));
-        if (py_get(rtResult, "legacy", {}) && py_get(py_get(rtResult, "legacy", {}), "withheld_scope")) {
+        const rtResult = retweet.result;
+        const rtUserResult = (rtResult.core?.user_results?.result) || {};
+        retweetedUser = map_user(rtUserResult).screen_name || _screen_name_from_media(rtResult.legacy || {});
+        if (rtResult.legacy && rtResult.legacy.withheld_scope) {
             withheld = true;
             body = get_full_text(rtResult);
         } else {
-            body = "RT @" + retweetedUser + ": " + get_full_text(rtResult);
+            body = `RT @${retweetedUser}: ${get_full_text(rtResult)}`;
         }
     }
 
-    let quoteTweet = py_get(tweet, "quoted_status_result");
+    // Quote tweet handling
+    let quoteTweet = tweet.quoted_status_result;
     if (quoteTweet && quoteTweet.result && quoteTweet.result.tweet) {
         quoteTweet.result = quoteTweet.result.tweet;
     }
     const quoteWithheld = !!(quoteTweet && quoteTweet.result && quoteTweet.result.tombstone);
-
-    let quoteAuthor = "";
-    let quoteBody = "";
+    let quoteAuthor = '';
+    let quoteBody = '';
     let quoteImages = [];
     let quoteVideos = [];
     if (quoteTweet && !quoteWithheld) {
         const quoteResult = quoteTweet.result;
-        const quoteUserResult = (quoteResult && quoteResult.core && quoteResult.core.user_results && quoteResult.core.user_results.result) || {};
-        quoteAuthor = map_user(quoteUserResult).screen_name || _screen_name_from_media(py_get(quoteResult, "legacy", {}));
+        const quoteUserResult = (quoteResult.core?.user_results?.result) || {};
+        quoteAuthor = map_user(quoteUserResult).screen_name || _screen_name_from_media(quoteResult.legacy || {});
         quoteBody = get_full_text(quoteResult);
-        const media = get_media(quoteResult);
-        quoteImages = media[0];
-        quoteVideos = media[1];
+        const mediaRes = get_media(quoteResult);
+        quoteImages = mediaRes[0];
+        quoteVideos = mediaRes[1];
     }
 
-    let quoteTweetId = py_get(py_get(tweet, "legacy", {}), "quoted_status_id_str", "");
+    // Quote tweet ID – may be present even when the quoted tweet object is missing
+    let quoteTweetId = py_get(tweet, 'legacy').quoted_status_id_str || '';
     if (!quoteTweetId && quoteTweet && quoteTweet.result) {
-        quoteTweetId = py_get(quoteTweet.result, "rest_id", "");
+        quoteTweetId = py_get(quoteTweet.result, 'rest_id') || '';
     }
-    const isQuoteTweet = Boolean(quoteTweetId || quoteTweet || py_get(py_get(tweet, "legacy", {}), "is_quote_status"));
+    const isQuoteTweet = Boolean(quoteTweetId || quoteTweet || py_get(tweet, 'legacy').is_quote_status);
     if (!quoteAuthor) {
-        const permalink = py_get(py_get(py_get(tweet, "legacy", {}), "quoted_status_permalink", {}), "expanded", "");
-        quoteAuthor = _screen_name_from_url(permalink);
+        quoteAuthor = _screen_name_from_url(py_get(py_get(tweet, 'legacy').quoted_status_permalink, 'expanded', ''));
     }
 
-    const mediaResult = get_media(tweet);
-    const images = mediaResult[0];
-    const videos = mediaResult[1];
+    // Media and entities for the main tweet
+    const mediaRes = get_media(tweet);
+    const images = mediaRes[0];
+    const videos = mediaRes[1];
     const entities = get_entities(tweet);
 
     return {
-        "collected_from_url": normalize_url_encoding(py_get(py_get(tweet, "__import_meta", {}), "source_platform_url", "")),
-        "id": py_get(tweet, "rest_id"),
-        "thread_id": py_get(py_get(tweet, "legacy", {}), "conversation_id_str"),
-        "timestamp": timestamp,
-        "unix_timestamp": unixTimestamp,
-        "link": tweetLink,
-        "body": body,
-        "author": authorScreenName,
-        "author_fullname": authorFullname,
-        "author_id": py_get(py_get(tweet, "legacy", {}), "user_id_str"),
-        "author_avatar_url": authorAvatarUrl,
-        "author_banner_url": authorBannerUrl,
-        "author_followers": authorFollowers,
-        "author_following": authorFollowing,
-        "author_bio": authorBio,
-        "author_location": authorLocation,
-        "verified": authorVerified,
-        "source": strip_tags(py_get(tweet, "source", "")),
-        "language_guess": py_get(py_get(tweet, "legacy", {}), "lang"),
-        "possibly_sensitive": (py_get(tweet, "possibly_sensitive", false) || py_get(py_get(tweet, "legacy", {}), "possibly_sensitive", false)) ? "yes" : "no",
-        "retweet_count": py_get(py_get(tweet, "legacy", {}), "retweet_count"),
-        "reply_count": py_get(py_get(tweet, "legacy", {}), "reply_count"),
-        "like_count": py_get(py_get(tweet, "legacy", {}), "favorite_count"),
-        "quote_count": py_get(py_get(tweet, "legacy", {}), "quote_count"),
-        "impression_count": py_get(py_get(tweet, "views", {}), "count", ""),
-        "is_retweet": retweetObj ? "yes" : "no",
-        "retweeted_user": retweetedUser,
-        "is_quote_tweet": isQuoteTweet ? "yes" : "no",
-        "quote_tweet_id": quoteTweetId,
-        "quote_author": quoteAuthor,
-        "quote_body": quoteBody,
-        "quote_images": quoteImages.join(","),
-        "quote_videos": quoteVideos.join(","),
-        "is_quote_withheld": quoteWithheld ? "yes" : "no",
-        "is_reply": String(py_get(py_get(tweet, "legacy", {}), "conversation_id_str")) !== String(py_get(tweet, "rest_id")) ? "yes" : "no",
-        "replied_author": py_get(py_get(tweet, "legacy", {}), "in_reply_to_screen_name", ""),
-        "is_withheld": withheld ? "yes" : "no",
-        "hashtags": (entities.hashtags || []).map(h => h.text).join(","),
-        "urls": (entities.urls || []).map(u => u.expanded_url || u.display_url).join(","),
-        "images": images.join(","),
-        "videos": videos.join(","),
-        "mentions": (entities.user_mentions || []).map(m => m.screen_name).join(","),
-        "long_lat": py_get(py_get(tweet, "legacy", {}), "place") ? get_centroid(py_get(py_get(py_get(tweet, "legacy", {}), "place", {}), "bounding_box", {}).coordinates) : "",
-        "place_name": py_get(py_get(py_get(tweet, "legacy", {}), "place", {}), "full_name", "")
+        collected_from_url: normalize_url_encoding(py_get(item, '__import_meta', {}).source_platform_url || ''),
+        id: tweet.rest_id,
+        thread_id: py_get(tweet, 'legacy').conversation_id_str,
+        timestamp: timestamp,
+        unix_timestamp: unixSeconds,
+        link: tweetLink,
+        body: body,
+        author: authorScreenName,
+        author_fullname: authorFullname,
+        author_id: py_get(tweet, 'legacy').user_id_str,
+        author_avatar_url: authorAvatarUrl,
+        author_banner_url: authorBannerUrl,
+        author_followers: authorFollowers,
+        author_following: authorFollowing,
+        author_bio: authorBio,
+        author_location: authorLocation,
+        verified: authorVerified,
+        source: strip_tags(py_get(tweet, 'source', '')),
+        language_guess: py_get(tweet, 'legacy').lang,
+        possibly_sensitive: (tweet.possibly_sensitive || py_get(tweet, 'legacy').possibly_sensitive) ? 'yes' : 'no',
+        retweet_count: py_get(tweet, 'legacy').retweet_count,
+        reply_count: py_get(tweet, 'legacy').reply_count,
+        like_count: py_get(tweet, 'legacy').favorite_count,
+        quote_count: py_get(tweet, 'legacy').quote_count,
+        impression_count: py_get(py_get(tweet, 'views') || {}, 'count', ''),
+        is_retweet: retweet ? 'yes' : 'no',
+        retweeted_user: retweetedUser,
+        is_quote_tweet: isQuoteTweet ? 'yes' : 'no',
+        quote_tweet_id: quoteTweetId,
+        quote_author: quoteAuthor,
+        quote_body: quoteBody,
+        quote_images: quoteImages.join(','),
+        quote_videos: quoteVideos.join(','),
+        is_quote_withheld: quoteWithheld ? 'yes' : 'no',
+        is_reply: String(py_get(tweet, 'legacy').conversation_id_str) !== String(tweet.rest_id) ? 'yes' : 'no',
+        replied_author: py_get(tweet, 'legacy').in_reply_to_screen_name || '',
+        is_withheld: withheld ? 'yes' : 'no',
+        hashtags: (entities.hashtags || []).map(h => h.text).join(','),
+        urls: (entities.urls || []).map(u => u.expanded_url || u.display_url).join(','),
+        images: images.join(','),
+        videos: videos.join(','),
+        mentions: (entities.user_mentions || []).map(m => m.screen_name).join(','),
+        long_lat: py_get(tweet, 'legacy').place ? get_centroid(py_get(tweet, 'legacy').place.bounding_box.coordinates) : '',
+        place_name: py_get(tweet, 'legacy').place ? py_get(py_get(tweet, 'legacy').place, 'full_name', '') : ''
     };
 }
 
 function map_item_legacy(tweet) {
-    const createdAtStr = py_get(py_get(tweet, "legacy", {}), "created_at");
-    const createdDate = new Date(createdAtStr);
-    const unixTimestamp = Math.floor(createdDate.getTime() / 1000);
-    const timestamp = formatUtcTimestamp(unixTimestamp);
-    const tweetId = py_get(py_get(tweet, "legacy", {}), "id_str");
+    const dateObj = new Date(py_get(tweet, 'legacy').created_at);
+    const unixSeconds = Math.floor(dateObj.getTime() / 1000);
+    const timestamp = formatUtcTimestamp(unixSeconds);
+    const tweetId = py_get(tweet, 'legacy').id_str;
     let withheld = false;
 
-    const retweetObj = py_get(py_get(tweet, "legacy", {}), "retweeted_status_result");
-    if (retweetObj) {
-        const rtLegacy = py_get(retweetObj, "result", {});
-        if (py_get(py_get(rtLegacy, "legacy", {}), "withheld_status")) {
+    const retweet = py_get(tweet, 'legacy').retweeted_status_result;
+    if (retweet) {
+        if (py_get(retweet, 'result').legacy && py_get(retweet, 'result').legacy.withheld_status) {
             withheld = true;
-            py_get(tweet, "legacy").full_text = py_get(rtLegacy, "legacy", {}).full_text;
+            py_get(tweet, 'legacy').full_text = py_get(retweet, 'result').legacy.full_text;
         } else {
-            const rtUserScreen = py_get(py_get(py_get(rtLegacy, "core", {}), "user_results", {}), "result", {}).legacy.screen_name;
-            const tText = "RT @" + rtUserScreen + " " + py_get(rtLegacy, "legacy", {}).full_text;
-            py_get(tweet, "legacy").full_text = tText;
+            const rtUserScreen = py_get(retweet, 'result').core.user_results.result.legacy.screen_name;
+            const tText = `RT @${rtUserScreen} ${py_get(retweet, 'result').legacy.full_text}`;
+            py_get(tweet, 'legacy').full_text = tText;
         }
     }
 
-    let quoteTweet = py_get(tweet, "quoted_status_result");
+    let quoteTweet = tweet.quoted_status_result;
     if (quoteTweet && quoteTweet.result && quoteTweet.result.tweet) {
         quoteTweet.result = quoteTweet.result.tweet;
     }
 
-    let quoteTweetId = py_get(py_get(tweet, "legacy", {}), "quoted_status_id_str", "");
-    if (!quoteTweetId && quoteTweet && quoteTweet.result) {
-        quoteTweetId = py_get(quoteTweet.result, "rest_id", "");
+    const quoteTweetId = py_get(tweet, 'legacy').quoted_status_id_str || '';
+    let finalQuoteTweetId = quoteTweetId;
+    if (!finalQuoteTweetId && quoteTweet && quoteTweet.result) {
+        finalQuoteTweetId = py_get(quoteTweet.result, 'rest_id') || '';
     }
-    const isQuoteTweet = Boolean(quoteTweetId || quoteTweet || py_get(py_get(tweet, "legacy", {}), "is_quote_status"));
-    const retweetUser = retweetObj
-        ? py_get(py_get(py_get(py_get(retweetObj, "result", {}), "core", {}), "user_results", {}), "result", {})
-        : {};
-    let quoteAuthor = "";
+    const isQuoteTweet = Boolean(finalQuoteTweetId || quoteTweet || py_get(tweet, 'legacy').is_quote_status);
+    let quoteAuthor = '';
     if (quoteTweet) {
-        const quoteUser = py_get(py_get(py_get(py_get(quoteTweet, "result", {}), "core", {}), "user_results", {}), "result", {});
-        quoteAuthor = py_get(py_get(quoteUser, "legacy", {}), "screen_name", "");
+        quoteAuthor = py_get(quoteTweet.result.core.user_results.result.legacy, 'screen_name', '');
     }
     if (!quoteAuthor) {
-        const permalink = py_get(py_get(py_get(tweet, "legacy", {}), "quoted_status_permalink", {}), "expanded", "");
-        quoteAuthor = _screen_name_from_url(permalink);
+        quoteAuthor = _screen_name_from_url(py_get(py_get(tweet, 'legacy').quoted_status_permalink, 'expanded', ''));
     }
 
     return {
-        "collected_from_url": normalize_url_encoding(py_get(py_get(tweet, "__import_meta", {}), "source_platform_url", "")),
-        "id": tweetId,
-        "thread_id": py_get(py_get(tweet, "legacy", {}), "conversation_id_str"),
-        "timestamp": timestamp,
-        "unix_timestamp": unixTimestamp,
-        "link": `https://x.com/${py_get(tweet, "user", {}).screen_name}/status/${tweetId}`,
-        "body": py_get(py_get(tweet, "legacy", {}), "full_text"),
-        "author": py_get(tweet, "user", {}).screen_name,
-        "author_fullname": py_get(tweet, "user", {}).name,
-        "author_id": py_get(tweet, "user", {}).id_str,
-        "author_avatar_url": "",
-        "author_banner_url": "",
-        "author_followers": py_get(py_get(tweet, "user", {}), "followers_count", ""),
-        "author_following": py_get(py_get(tweet, "user", {}), "friends_count", ""),
-        "author_bio": py_get(py_get(tweet, "user", {}), "description", ""),
-        "author_location": py_get(py_get(tweet, "user", {}), "location", ""),
-        "verified": "",
-        "source": strip_tags(py_get(py_get(tweet, "legacy", {}), "source", "")),
-        "language_guess": py_get(py_get(tweet, "legacy", {}), "lang"),
-        "possibly_sensitive": py_get(py_get(tweet, "legacy", {}), "possibly_sensitive") ? "yes" : "no",
-        "retweet_count": py_get(py_get(tweet, "legacy", {}), "retweet_count"),
-        "reply_count": py_get(py_get(tweet, "legacy", {}), "reply_count"),
-        "like_count": py_get(py_get(tweet, "legacy", {}), "favorite_count"),
-        "quote_count": py_get(py_get(tweet, "legacy", {}), "quote_count"),
-        "impression_count": py_get(py_get(tweet, "ext_views", {}), "count", ""),
-        "is_retweet": retweetObj ? "yes" : "no",
-        "retweeted_user": retweetObj ? py_get(py_get(retweetUser, "legacy", {}), "screen_name", "") : "",
-        "is_quote_tweet": isQuoteTweet ? "yes" : "no",
-        "quote_tweet_id": quoteTweetId,
-        "quote_author": quoteAuthor,
-        "quote_body": "",
-        "quote_images": "",
-        "quote_videos": "",
-        "is_quote_withheld": "",
-        "is_reply": String(py_get(py_get(tweet, "legacy", {}), "conversation_id_str")) !== tweetId ? "yes" : "no",
-        "replied_author": py_get(py_get(tweet, "legacy", {}), "in_reply_to_screen_name", ""),
-        "is_withheld": withheld ? "yes" : "no",
-        "hashtags": (py_get(py_get(py_get(tweet, "legacy", {}), "entities", {}), "hashtags", [])).map(h => h.text).join(","),
-        "urls": (py_get(py_get(py_get(tweet, "legacy", {}), "entities", {}), "urls", [])).map(u => u.expanded_url || u.display_url).join(","),
-        "images": (py_get(py_get(py_get(tweet, "legacy", {}), "extended_entities", {}), "media", []).filter(m => m.type === "photo").map(m => m.media_url_https)).join(","),
-        "videos": (py_get(py_get(py_get(tweet, "legacy", {}), "extended_entities", {}), "media", []).filter(m => m.type === "video").map(m => m.video_info.variants[0].url)).join(","),
-        "mentions": (py_get(py_get(py_get(tweet, "legacy", {}), "entities", {}), "user_mentions", [])).map(m => m.screen_name).join(","),
-        "long_lat": py_get(py_get(tweet, "legacy", {}), "place") ? get_centroid(py_get(py_get(py_get(tweet, "legacy", {}), "place", {}), "bounding_box", {}).coordinates) : "",
-        "place_name": py_get(py_get(py_get(tweet, "legacy", {}), "place", {}), "full_name", "")
+        collected_from_url: normalize_url_encoding(py_get(item, '__import_meta', {}).source_platform_url || ''),
+        id: tweetId,
+        thread_id: py_get(tweet, 'legacy').conversation_id_str,
+        timestamp: timestamp,
+        unix_timestamp: unixSeconds,
+        link: `https://x.com/${tweet.user.screen_name}/status/${tweetId}`,
+        body: py_get(tweet, 'legacy').full_text,
+        author: tweet.user.screen_name,
+        author_fullname: tweet.user.name,
+        author_id: tweet.user.id_str,
+        author_avatar_url: '', // todo: add
+        author_banner_url: '', // todo: add
+        author_followers: tweet.user.followers_count || '',
+        author_following: tweet.user.friends_count || '',
+        author_bio: tweet.user.description || '',
+        author_location: tweet.user.location || '',
+        verified: '', // todo: add
+        source: strip_tags(py_get(tweet, 'legacy').source),
+        language_guess: py_get(tweet, 'legacy').lang,
+        possibly_sensitive: py_get(tweet, 'legacy').possibly_sensitive ? 'yes' : 'no',
+        retweet_count: py_get(tweet, 'legacy').retweet_count,
+        reply_count: py_get(tweet, 'legacy').reply_count,
+        like_count: py_get(tweet, 'legacy').favorite_count,
+        quote_count: py_get(tweet, 'legacy').quote_count,
+        impression_count: py_get(py_get(tweet, 'ext_views') || {}, 'count', ''),
+        is_retweet: retweet ? 'yes' : 'no',
+        retweeted_user: retweet ? py_get(retweet, 'result.core.user_results.result.legacy').screen_name || '' : '',
+        is_quote_tweet: isQuoteTweet ? 'yes' : 'no',
+        quote_tweet_id: finalQuoteTweetId,
+        quote_author: quoteAuthor,
+        quote_body: '', // todo: add
+        quote_images: '', // todo: add
+        quote_videos: '', // todo: add
+        is_quote_withheld: '', // todo: add
+        is_reply: String(py_get(tweet, 'legacy').conversation_id_str) !== tweetId ? 'yes' : 'no',
+        replied_author: py_get(tweet, 'legacy').in_reply_to_screen_name || '',
+        is_withheld: withheld ? 'yes' : 'no',
+        hashtags: (py_get(tweet, 'legacy').entities.hashtags || []).map(h => h.text).join(','),
+        urls: (py_get(tweet, 'legacy').entities.urls || []).map(u => u.expanded_url || u.display_url).join(','),
+        images: (py_get(tweet, 'legacy').extended_entities.media || []).filter(m => m.type === 'photo').map(m => m.media_url_https).join(','),
+        videos: (py_get(tweet, 'legacy').extended_entities.media || []).filter(m => m.type === 'video').map(m => m.video_info.variants[0].url).join(','),
+        mentions: (py_get(tweet, 'legacy').entities.user_mentions || []).map(m => m.screen_name).join(','),
+        long_lat: py_get(tweet, 'legacy').place ? get_centroid(py_get(tweet, 'legacy').place.bounding_box.coordinates) : '',
+        place_name: py_get(tweet, 'legacy').place ? py_get(py_get(tweet, 'legacy').place, 'full_name', '') : ''
     };
 }
 
 function map_user(userResult) {
-    if (typeof userResult !== "object" || userResult === null) {
+    if (typeof userResult !== 'object' || userResult === null) {
         userResult = {};
     }
     const core = userResult.core || {};
@@ -420,19 +416,19 @@ function map_user(userResult) {
 
     function first(...values) {
         for (const v of values) {
-            if (v !== undefined && v !== null) {
+            if (v !== null && v !== undefined) {
                 return v;
             }
         }
-        return "";
+        return '';
     }
 
     return {
-        screen_name: core.screen_name || legacy.screen_name || "",
-        fullname: core.name || legacy.name || "",
-        avatar_url: avatar.image_url || legacy.profile_image_url_https || "",
-        banner_url: banner.image_url || legacy.profile_banner_url || "",
-        verified: py_get(userResult, "is_blue_verified", ""),
+        screen_name: core.screen_name || legacy.screen_name || '',
+        fullname: core.name || legacy.name || '',
+        avatar_url: avatar.image_url || legacy.profile_image_url_https || '',
+        banner_url: banner.image_url || legacy.profile_banner_url || '',
+        verified: userResult.is_blue_verified || '',
         followers: first(counts.followers, legacy.followers_count),
         following: first(counts.following, legacy.friends_count),
         bio: first(profileBio.description, legacy.description),
@@ -441,26 +437,24 @@ function map_user(userResult) {
 }
 
 function get_note_result(tweet) {
-    const note = py_get(tweet, "note_tweet", {});
-    return py_get(py_get(note, "note_tweet_results", {}), "result", {});
+    const note = py_get(tweet, 'note_tweet') || {};
+    return py_get(note, 'note_tweet_results', {}).result || {};
 }
 
 function get_full_text(tweet) {
-    const text = py_get(py_get(tweet, "legacy", {}), "full_text", "");
-    const noteText = py_get(get_note_result(tweet), "text", "");
+    const text = py_get(tweet, 'legacy').full_text || '';
+    const noteText = py_get(get_note_result(tweet), 'text', '');
     return noteText.length > text.length ? noteText : text;
 }
 
 function get_entities(tweet) {
-    const entities = py_get(py_get(tweet, "legacy", {}), "entities", {});
+    const entities = py_get(tweet, 'legacy').entities || {};
     const note = get_note_result(tweet);
     const noteEntities = note.entity_set;
     if (!noteEntities) {
         return entities;
     }
-    const legacyFullText = py_get(py_get(tweet, "legacy", {}), "full_text", "");
-    const noteText = note.text || "";
-    if (noteText.length <= legacyFullText.length) {
+    if (note.text && note.text.length <= (py_get(tweet, 'legacy').full_text || '').length) {
         return entities;
     }
     const combined = Object.assign({}, noteEntities);
@@ -473,22 +467,20 @@ function get_entities(tweet) {
 function get_media(tweet) {
     const images = [];
     const videos = [];
-    const legacy = py_get(tweet, "legacy", {});
-    const extended = py_get(legacy, "extended_entities", {});
-    const entities = py_get(legacy, "entities", {});
+    const legacy = py_get(tweet, 'legacy') || {};
     const mediaItems = [];
-    if (Array.isArray(extended.media)) {
-        mediaItems.push(...extended.media);
+    if (legacy.extended_entities && legacy.extended_entities.media) {
+        mediaItems.push(...legacy.extended_entities.media);
     }
-    if (Array.isArray(entities.media)) {
-        mediaItems.push(...entities.media);
+    if (legacy.entities && legacy.entities.media) {
+        mediaItems.push(...legacy.entities.media);
     }
     for (const media of mediaItems) {
         if (!media.media_url_https) continue;
         images.push(media.media_url_https);
-        if (media.type !== "video" && media.type !== "animated_gif") continue;
-        const variants = (media.video_info && media.video_info.variants) || [];
-        const videoVariants = variants.filter(v => (v.content_type || "").startsWith("video/"));
+        if (media.type !== 'video' && media.type !== 'animated_gif') continue;
+        const variants = (media.video_info || {}).variants || [];
+        const videoVariants = variants.filter(v => (v.content_type || '').startsWith('video/'));
         if (videoVariants.length) {
             videoVariants.sort((a, b) => (b.bitrate || 0) - (a.bitrate || 0));
             videos.push(videoVariants[0].url);
@@ -501,45 +493,49 @@ function get_media(tweet) {
 }
 
 function _screen_name_from_url(url) {
-    if (typeof url !== "string") return "";
+    if (typeof url !== 'string') return '';
     const m = url.match(/^https?:\/\/(?:x|twitter)\.com\/([^/]+)\/status\//);
-    return m ? m[1] : "";
+    return m ? m[1] : '';
 }
 
 function _screen_name_from_media(legacyObj) {
-    if (typeof legacyObj !== "object" || legacyObj === null) return "";
-    for (const container of ["extended_entities", "entities"]) {
-        const mediaArray = py_get(legacyObj, container, {}).media || [];
+    if (typeof legacyObj !== 'object' || legacyObj === null) return '';
+    const containers = ['extended_entities', 'entities'];
+    for (const container of containers) {
+        const mediaArray = (legacyObj[container] && legacyObj[container].media) || [];
         for (const m of mediaArray) {
-            const url = typeof m === "object" ? py_get(m, "expanded_url", "") : "";
+            const url = typeof m === 'object' ? (m.expanded_url || '') : '';
             const screen = _screen_name_from_url(url);
             if (screen) return screen;
         }
     }
-    return "";
+    return '';
 }
 
 function get_centroid(box) {
     try {
         const ring = box[0];
         if (!Array.isArray(ring) || ring.length < 2 || !ring[0] || !ring[1]) {
-            return "";
+            return '';
         }
         const lon = String(Number(((ring[0][0] + ring[1][0]) / 2).toFixed(6)));
         const lat = String(Number(((ring[0][1] + ring[1][1]) / 2).toFixed(6)));
         return `${lon},${lat}`;
     } catch (e) {
-        return "";
+        return '';
     }
 }
 
 export function map_item(item) {
-    if (py_get(item, "rest_id")) {
+    // Determine which mapping function to use based on the shape of the imported tweet
+    const hasRestId = py_get(item, 'rest_id') !== null && py_get(item, 'rest_id') !== undefined;
+    const isLegacy = py_get(item, 'type') === 'adaptive';
+    if (hasRestId) {
         return new MappedItem(map_item_modern(item));
-    } else if (py_get(item, "type") === "adaptive") {
+    } else if (isLegacy) {
         return new MappedItem(map_item_legacy(item));
     } else {
-        throw new MapItemException("Unsupported item shape");
+        throw new MapItemException('Unrecognised tweet shape');
     }
 }
 // === end auto-generated ===