We build a React component that implements our Figma design with the intent of creating screenshots in a headless browser later. We use Emotion, a CSS-in-JS library, Flexbox, and React Textfit to build a scalable OpenGraph image. We cover using complex CSS selectors with multiple pseudo-selectors as well as common Flexbox layout techniques such as space-between.
Chris Biscardi: [00:00] The open-graph image layout that we created in Figma is on the left here. On the right, we have a CodeSandbox. We'll create a new React CodeSandbox although you can do this with any technology that you want. This is not limited to React.
[00:12] It's a little hard to see on the left so we'll use a feature of Figma called Present. Note that it comes up with the first item in our Figma template so we need to click once to go to the right to get to our open-graph image. We also need to go up to the top right to the percentage and do scale down to fit.
[00:29] Now that we have a view of our template, we can scale it down as much as we want. I'll let the CodeSandbox take up the rest of the screen while we work.
[00:35] The first thing we're going to do is import JSX from a motion core. Motion core hasn't been added as a dependency yet so we need to click the button and do that. We can also go into the package JSON and add if we want to.
[00:47] We'll also use a pragma at the top of the file to specify that we want the JSX that we're writing to convert into this JSX function instead of react.createElement. I'm using a CSS from JS library because later, we're going to take this code and put it inside of a browser inside of a serverless function.
[01:20] We can enter Zen mode to get rid of the top bar and click on the menu on the left on the currently active Explorer pane to get rid of the sidebar. We'll keep the console around just in case. We can also get rid of React because we're not using createElement.
[01:33] The CSS prop allows us to use all the features of CSS right in our JSX. We'll set the width and the height to 1200 and 630, which is the same size as our Open Graph image on the left. The first thing we want to do is get this gradient in as a background color for our div. If we switch back to the gradient magic page, we can copy CSS out that represents our background.
[01:54] We weren't able to do this in Figma because Figma doesn't have CSS import. Now that we're working in CodeSandbox, we can just use the CSS linear gradient instead. Now, we've set a background image with our gradient.
[02:06] For our purposes, this is fairly large, so we're going to temporarily chop both values in half so that we can see more of it on the screen at the same time. Before we ship, we'll want to change these values back. There are a couple global styles we want to add.
[02:18] You might be used to seeing these from a CSS reset or a normalized CSS file. We're setting box sizing to border box. We're setting the margin in the padding to zero, which is fairly opinionated.
[02:29] We've set the font family to system UI. Remember that we used Roboto in our Figma template. We've also set a default font size here. This is using the global component from Emotion, which will inject the styles and apply them globally.
[02:41] If we ever didn't render this global component, the styles would get removed. Since we only render this once and it's not an application, this global component will never get removed. The next thing we want to do is wrap this image text in a div of its own.
[02:57] We're creating this dark background area inside of our opengraph image. We can start with any arbitrary color to figure out that our size is not actually what we wanted to be, as you see this black bar at the top.
[03:09] If we add a margin of one RAM, we could see that we get the margin that we want from the sides, but the top actually pushes the entire thing down. We can fix this by setting an absolute position for our bounding div. Now, we have our black box but it's quite thin and it doesn't fill the entire area we wanted to.
[03:26] If we use the display flex on the containing div, we get the height that we want, as we can see if we scroll down, but we no longer have the width that we want. If we set display flex on the child or woman as well, it doesn't seem to do anything.
[03:38] If we set flex one, which implies that the content should grow to fit the area, we can see that our box now fills the entire area that we wanted to with an appropriate margin offset.
[03:49] Note that if we remove display flex on the child, flex one will still work for us but we'll use flexbox to layout our text so we're going to keep it around. We'll put an H1 tag again to represent the title of our blog post. I'll set the color to white even though it's not the color that we end up wanting.
[04:05] Remember that right now, we're just laying everything out. I'm going to fine-tune it later. If we look at the way the text is laid out, we're going to need the H1 tag for the heading. We're going to need a list for these tags down here, and we're going to need some kind of span for the for the username.
[04:19] We can separate the text on the top, which is our title, from a box bounding the bottom, which will allow us to use flexbox to separate these two to the top and the bottom of the box. Again, we'll use another flexbox to separate the list from the span that contains the username inside of the bottom div.
[04:37] If we specify a text color for the list and for the span, we see that we have actually our title on the left and the list and the span on the right. That's not quite what we want, so we use flex direction column on the containing div that contains the H1, and the lower div to put them into a column.
[04:54] This puts one on top of the other. If we use just five content space between, you can see the any of the free space that's left over from that containing div is put between the two elements. This is what we want because we can use margins to offset them from the edges, but the space between them will always be where the extra space goes.
[05:15] If we put the display flex on the containing div that contains the list and the span, we get a flexDirection row by default. This is what we want, but we also want space-between again. Now space-between has put the list on the left and our username on the right.
[05:32] We want to get rid of the bullets before the list because this isn't going to be a bulleted list. We'll do, list-style-type: none, on the UL element. Remember, that we scaled down the bounding box. The font sizes for image, the username, and our title don't match up with the bounding box yet.
[05:50] If we add a couple more list elements, we'll notice that our unordered list is in a column format as well. We need to change this to be a row which will use flexbox four again. The gap property hasn't quite shipped yet. Whereas, if we were using CSS Grid, we would be able to use a grid-gap to separate these items apart. When using flexbox, we still have to refer to margins.
[06:09] Note, that we can still use CSS selectors in their full capacity in the exact same way that you're used to in a CSS file. In this case, we've used AND, which is a popular term for the parent element, and then we've set any child of LI. Since we won't have nested LI elements, we don't have to worry about specifying the direct child or anything like that. This will be fine for us.
[06:30] We can use the after pseudo selector with the UTF-8 dot in a content rule on the selector, along with a margin-left of 05rem. This gives us 05rem of margin on the right and on the left of the dot, making our spacing nice and consistent.
[06:44] Note, that we do have a problem, in that the last element in our list also has a dot, which we don't want to keep. If we include the not() pseudo selector with the last child pseudo selector, in addition to our after pseudo selector, we can see that when the element is not the last child, we can specify the after content to be the dot, which gets rid of the dot after the Figma element.
[07:04] We're starting to look pretty close to what we want. We can start inserting the correct sizing for everything and the correct colors for everything. If we click into our Figma presentation, we can go back to the original page, which has all of our colors on it. We can just pull the colors out and start inserting them.
[07:21] If we go back in the Presentation mode and check out our page, we can see that all of the colors now match. A bit of the spacing is off, but since we scaled it down right now, we're not too worried about that. We can set the border radius to 15 on the div, that is the background color, and when we scale this up, it'll look a little bit smaller than it does now.
[07:38] We could specify these values in rems or something else that's relative to the font size. Since we know how big this is going to be exactly, we can be a little bit more loose with it. If we make the size of the containing div as big as it's going to be, we can see that our font size, which looked really big before, really doesn't look big at all.
[07:57] We'll start with the size of 48. If you click this little Open in New Window button in CodeSandbox, we can pop out the frame and look more at the actual content of our Open Graph image. We can always refer back to Figma. We've come far enough at this point that we can work at the spacing out ourselves.
[08:14] Note, that I've zoomed out on the left so we can get a better view while we work on the actual spacing. We could set margin on the h1 tag and margins on the list and the username as well. Instead, we're going to add a padding to the containing div.
[08:29] In this case, we've added a padding of 2rem, which looks pretty decent, and guarantees that we'll have the right spacing on all different sides. Remember, that because we set the margin and the padding to zero earlier for all elements, we don't have to worry as much about margin collapsing or some padding sticking around.
[08:47] We can bump the font size of the list elements and the username. Notice, that if we bump the font size, which we've set on the global level, all of our spacing increases as well. This is because we specified our padding in rems, which is relative to the root font size, in this case 24, which produces a decent effect.
[09:05] However, we did have pretty specific spacing requirements for how much of the outside border we show. Since that was 40 pixels, I'll change it to 40 pixels. Because we already know how big this square is going to be when it shows up as an Open Graph image, pixels allows us a little bit of more control when it comes to the final image.
[09:22] This is one of the longer titles for my blog post. I want to make sure that it fills out enough of the space. This is why I've bumped the font size. In other scenarios, I might have shorter blog posts, in which case I still want it to fill up a decent amount of space.
[09:37] Note, that if we make the font size too big and we don't know how long our blog post titles are going to be, it will overflow the bounds. If we want to completely avoid having to choose a font size, we can use a library called react-textfit. Note, that we will have to add it as a dependency, which our preview window still lets us do, even though we popped it out earlier.
[09:56] Note, that even though we have a font size of 100 pixels here, the div below it has a font size of 24 because we set it globally. This was probably a mistake, as global styles typically are. You can see that now we've removed it, textfit has blown up the font size to make it very big and fit the entire card. We will need to add the font size back to our sub-elements on the bottom of the card.
[10:16] Note that we can specify the textfit mode to be single, but that will make sure that the text actually fills out the entire width, which we aren't currently constraining. Even with this much text, we probably won't be able to fit it all on a single line anyway.
[10:30] We can use textfit based on dynamic content that comes in to determine whether we should lay out the title on a single line or multiple lines, if this matters to you. Remember that the default is multiline.
[10:42] You may also notice that if we go down to fewer characters, the font size doesn't change that much. This is because textfit takes a min and a max. The default maximum is 100 pixels, while the default minimum is 1. We can use these values to control how much of the card our font will actually take up. We could, for example, have single-line cards be very big.
[11:06] Of course, if our titles get very, very long, we hit the minimum font size. If you set a minimum font size, you will have to take care to note that your titles are not too long, or you'll end up running over the card. This is OK though because most headlines should not be 256 characters or more. If they are, that's more than most people tweet.
[11:27] Note that if we don't like it filling out the entire card for some reason, we can also limit how much of the width it takes up. Note that if we limit the width, that does give us less room to operate in with the number of characters in our titles, but it can also give us a different visual representation. Note also that doing things like setting the line height will affect the sizing.
[11:46] There's one final thing I want to do with these cards and that's add a slight drop shadow to the back of this card. If we go to this site as brumm.af/shadows, we can generate a new box shadow. You can see the box shadow on the back of this card. We'll take the default box shadow and put it on the back of our background card.
[12:03] Note that those overflow a little on the entire container, but it gives us a nicer little gradient that pops up the card a little bit more off the background. To finish off a little bit on the end and make sure that none of the overflow box shadow from card spreads over, we can use overflow hidden. This gives us a nice, clean line around our 1200 by 630 pixel Open Graph card.
instead of using the /** @jsx jsx / pragma, use /* @jsxImportSource @emotion/react */