instagram-to-ghost.js

const axios         = require('axios').default;
const mysql         = require('mysql');
const util          = require('util');
const slugify       = require('slugify');
const GhostAdminAPI = require('@tryghost/admin-api');
const Downloader    = require('nodejs-file-downloader');

//Helper function to setup a database wrapper that works with promises
//https://codeburst.io/node-js-mysql-and-async-await-6fb25b01b628
function makeDatabase() {

    console.log(getDate() + "Running with !! PRODUCTION !! database");

    const host = '0.0.0.0';

    const connection = mysql.createConnection({
        host       : host,
        user       : 'username',
        password   : 'password',
        database   : 'database_name',
        charset    : 'utf8mb4'
    });

    return {
        query( sql, args ) {
            return util.promisify( connection.query )
                .call( connection, sql, args );
        },
        close() {
            return util.promisify( connection.end ).call( connection );
        },
        beginTransaction() {
          return util.promisify( connection.beginTransaction )
            .call( connection );
        },
        commit() {
          return util.promisify( connection.commit )
            .call( connection );
        },
        rollback() {
          return util.promisify( connection.rollback )
            .call( connection );
        }
    };
}


//Helper function that just returns a formatted date. Used for console output.
function getDate() {

    return new Date().toISOString() + ": ";
}

//Promisify our database connection
const db = makeDatabase();

// Ghost API Connection
const api = new GhostAdminAPI({
    url:     'https://thomasclowes.com',
    version: "v3",
    key:     'ghost-api-key'
});

const instagramToken = "user-instagram-token";


async function pullInstagramPosts() {

    const url = "https://graph.instagram.com/me/media?fields=id,caption,media_type,media_url,permalink,thumbnail_url,timestamp,children{id,media_type,media_url,permalink,thumbnail_url,timestamp}&access_token=" + instagramToken;

    var response = await axios.get(url);

    const responseData = response.data;

    const posts = responseData.data;
    const paging = responseData.paging;

    var postData = [];

    for (post of posts) {

        console.log(post);

        postData.push([
            post.id,
            post.id,
            post.caption,
            post.media_type,
            post.media_url,
            post.permalink,
            post.timestamp
        ]);

        if ('children' in post) {

            console.log("Parent " + post.id);

            for (child of post.children.data) {

                const childId = child.id;

                var caption = null;

                postData.push([
                    post.id,
                    child.id,
                    caption,
                    child.media_type,
                    child.media_url,
                    child.permalink,
                    child.timestamp
                ]);

                console.log("should get " + childId);
            }
        }
    }


    const insertSql = "INSERT IGNORE INTO `ig_posts` (`parent_instagram_id`, `instagram_id`, `caption`, `media_type`, `media_url`, `permalink`, `datetime`) VALUES ?";

    //Insert the Strava activity row
    await db.query(insertSql, [postData]);
}




//Helper function that returns an appropriate mobiledoc card for the post type
function getAppropriateCard(post, uploadedMediaUrl) {

    if (post.media_type == "VIDEO") {

        return ["html",{"html":"<p><video width=\"100%\" controls><source src=\"" + uploadedMediaUrl + "\" type=\"video/mp4\">Your browser does not support the video tag.</video></p>"}];

    } else {

        return ["image",{"src": uploadedMediaUrl,"alt":"","title":""}];
    }
}


async function doWork() {

    await pullInstagramPosts();

    //Get all the Instagram photos/videos that we have get to post to the blog
    const posts = await db.query("SELECT * FROM ig_posts WHERE postId IS NULL ORDER BY parent_instagram_id DESC");

    const processedPosts = {};

    var mediaIndex   = 0;
    var lastParentId = null;

    //Loop through each one
    for (post of posts) {

        //Download the media to our server
        const parts    = post.media_url.split('/');
        const fileName = parts[parts.length - 1].split('?')[0];

        const downloader = new Downloader({     
          url: post.media_url,//If the file name already exists, a new file with the name 200MB1.zip is created.     
          directory: "./ig-images",//This folder will be created, if it doesn't exist.    
          fileName: fileName           
        })
        
        await downloader.download();//Downloader.download() returns a promise.
 
        //Upload it to the Ghost blog file directory
        const upload = await api.images.upload({
            file: "ig-images/" + fileName,
        })

        //This is the return uploaded media url
        const uploadedMediaUrl = upload.url;

        //If the parent Instagram ID differs to the last one then we are working in relation to a different parent post
        if (post.parent_instagram_id != lastParentId) {

            //Reset our media index
            mediaIndex = 0;
            lastParentId = post.parent_instagram_id;
        }


        if (post.parent_instagram_id in processedPosts) {

            processedPosts[post.parent_instagram_id]["media_urls"].push(post.media_url);
            processedPosts[post.parent_instagram_id]["rowIds"].push(post.ID);

            //Mobiledoc data
            processedPosts[post.parent_instagram_id]["cards"].push(getAppropriateCard(post, uploadedMediaUrl));
            processedPosts[post.parent_instagram_id]["sections"].push([10, mediaIndex]);

            mediaIndex++;
        //This is the first media for this parent ID
        } else {

            const caption = post.caption != null ? post.caption.replace(/(?:\r\n|\r|\n)/g, '<br>') : "";
            const captionMobileDoc = ["html",{"html": caption}];

            var captionLines = post.caption.split("\n")
            captionLines = captionLines.filter(function(e){return e}); 

            console.log(captionLines);

            var mobileDocCards = [];
            var mobileDocSections = [];

            for (line of captionLines) {

                mobileDocSections.push([1,"p",[[0,[],0,line]]]);
            }

            if (post.media_type != "CAROUSEL_ALBUM") {

                mobileDocCards.push(getAppropriateCard(post, uploadedMediaUrl));
                mobileDocSections.push([10, mediaIndex]);

                mediaIndex++;
            }

            var indexOfFullStop = caption.indexOf('.');
            var indexOfNewLine  = caption.indexOf('\n');
            indexOfFullStop     = indexOfFullStop !== -1 ? indexOfFullStop : 1000;
            indexOfNewLine      = indexOfNewLine !== -1 ? indexOfNewLine : 1000;

            const indexOf = Math.min(indexOfFullStop, indexOfNewLine);

            const postTitle = post.caption.substr(0, indexOf !== 1000 ? indexOf : caption.length);
            const postSlug  = slugify("Instagram " + postTitle);

            var postTags    = [];
            var matchedTags = caption.match(/#[A-Za-z0-9\-]+/gi);
            matchedTags     = matchedTags != null && matchedTags.length > 0 ? matchedTags.map(tag => tag.replace("#", "")) : [];
            postTags.push("instagram");
            postTags = postTags.concat(matchedTags);

            console.log(post.datetime.toISOString());

            var publishedAt = post.datetime.toISOString();

            processedPosts[post.parent_instagram_id]                 = {};
            processedPosts[post.parent_instagram_id]["featureImage"] = uploadedMediaUrl;
            processedPosts[post.parent_instagram_id]["publishedAt"]  = publishedAt;
            processedPosts[post.parent_instagram_id]["postTitle"]    = postTitle;
            processedPosts[post.parent_instagram_id]["postCaption"]  = post.caption;
            processedPosts[post.parent_instagram_id]["postSlug"]     = postSlug;
            processedPosts[post.parent_instagram_id]["postTags"]     = postTags;
            processedPosts[post.parent_instagram_id]["permalink"]    = post.permalink;
            processedPosts[post.parent_instagram_id]["caption"]      = post.caption;
            processedPosts[post.parent_instagram_id]["media_urls"]   = [post.media_url];
            processedPosts[post.parent_instagram_id]["rowIds"]       = [post.ID];

            //Mobiledoc data
            processedPosts[post.parent_instagram_id]["cards"]      = mobileDocCards;
            processedPosts[post.parent_instagram_id]["sections"]   = mobileDocSections;
        }
    }

    //We will now submit the posts to the blog..

    //Loop through our processed posts
    for (instagramParentId of Object.keys(processedPosts)) {

        const processedPost = processedPosts[instagramParentId];

        let featureImage = processedPost["featureImage"];
        let publishedAt  = processedPost["publishedAt"];
        let postTitle    = processedPost["postTitle"];
        let postCaption  = processedPost["postCaption"];
        let postSlug     = processedPost["postSlug"];
        let postTags     = processedPost["postTags"];
        let permalink    = processedPost["permalink"];

        let rowIds    = processedPost["rowIds"];
        let cards     = processedPost["cards"];
        let sections  = processedPost["sections"];

        //Add the attribution card
        cards.push(["markdown",{"markdown":"*This post was originally posted on [my Instagram](" + permalink + ").*"}]);
        sections.push([10, cards.length - 1]);

        console.log(cards);
        console.log(sections);

        const mobiledoc = {
            version:  "0.3.1",
            atoms:    [],
            markups:  [],
            cards:    cards,
            sections: sections,
        }

        const res = await api.posts
            .add(
                {
                    title:            postTitle, 
                    slug:             postSlug, 
                    tags:             postTags, 
                    meta_description: postCaption,
                    custom_excerpt:   postCaption.substring(0, 300), //Apparently there is a 300 character limit on this
                    meta_title:       postTitle,
                    feature_image:    featureImage, 
                    status:           "published", 
                    published_at:     publishedAt, 
                    created_at:       publishedAt, 
                    updated_at:       publishedAt, 
                    mobiledoc:        JSON.stringify(mobiledoc)
                },
            )


        const postId = res.id;

        console.log("POSTED: " + postId);

        //Mark the media rows submitted as part of this post as complete
        const updateStravaAccessTokenSql = "UPDATE ig_posts SET postId = ? WHERE ID IN (?)";
        const updateResponse             = await db.query(updateStravaAccessTokenSql, [postId, rowIds]);

        console.log("MARKED AS POSTED");
    }
}

doWork();