This fixes#53. Note that this means the eye creature isn't observable from the UI, as moving the slider to "bonkers" shows the eye with parts randomly replaced. The "pristine" eye creature can still be shown for debugging purposes by adding `&eye=on` to the end of the URL.
This fixes#39 by adding a "Randomly invert symbols" checkbox.
If the inversion is undesirable, disabling it will keep the existing creature but remove the inversion.
Note that this checkbox is only displayed when the creature complexity isn't bonkers.
This refactors `creature-symbol.tsx` so that it doesn't have to rely on awkwardly introspecting `JSX.Element` instances to do its job. Now all of that mumbo-jumbo, which is only really useful for when we want to manually construct symbols like the eye creature, is encapsulated in `creature-symbol-factory.tsx`.
This adds a bit of debugging information on mouseover.
For example, a tooltip with the text `bird@tail.arm[0]` can be interpreted as "a bird symbol attached to the tail symbol's first arm attachment point."
The implementation is a bit funky: we basically annotate the SVG DOM with various `data` attributes, and on mouseover we traverse the DOM from the element the mouse is over all the way up to the SVG root element, picking out relevant `data` attributes and building a tooltip out of it. This ended up being easier than e.g. passing a bunch of props down the whole tree in React.
This builds upon #30 and #33 by adding support for randomly nested symbols in complexity levels 0-4 of the creature page.
Currently, we don't _recursively_ nest symbols, so e.g. we won't make an eye containing an eye containing a cube. And for now we'll only add a nested symbol if the parent symbol has the `always_nest` metadata property defined.
This fixes#20 by adding support for TOML-based metadata for symbols.
There's now a template TOML file in `svg/_template.toml` that documents all the different metadata properties. The file can be copied to match the name of an SVG file (but with a `.toml` extension) and it will be processed by the vocabulary builder when the site is generated.
Currently there are only two properties documented, `always_nest` and `always_be_nested` (as per the needs outlined in https://github.com/toolness/mystic-symbolic/issues/17#issuecomment-786696966), but they don't actually do anything yet (actual support for _using_ the metadata will come in another file).
Right now the TOML files are validated quite stringently: if a file contains the name of a property it doesn't understand, or the type of the property is wrong, it will raise an error. We can revisit this if it becomes burdensome.
This adds basic support for nesting (#17). It does not swap fill/stroke colors based on the position of the nesting box, though. It also only nests a symbol in the eye creature and "bonkers" complexity setting for now--support for nesting in complexity levels 1-4 is forthcoming.
The nested symbol is essentially uniformly scaled as much as possible without extending outside the boundaries of its parent's nesting box.
This fixes#21 by changing the existing "tail" attachment point (`#ff0000`) to be called the "anchor" attachment point, since that's primarily what we're using it for. It also adds a brand-new "tail" attachment point with color `#be0027`, which is expected to be oriented in the direction the tail should point (not the _opposite_ direction of the tail, which is how the old tail attachment point was oriented).
Note that this change means, for now, that no symbols have tail attachment points. An error will be logged in the console when viewing the "Creature!" page for now because of this (since the default composition attaches a symbol to the tail), but as soon as Nina updates the SVGs, this should go away.
This attempts to manually scale the strokes on creatures so they aren't affected by scale.
The weird thing here is that applying `vector-effect="non-scaling-stroke"` does not appear to do the trick--rather, it makes all the strokes look _super heavy_. So in this PR we scale the stroke by the inverse of whatever we're cumulatively scaling by when we're drawing something.
This fixes a bunch of bugs in our attachment-related code, and in so doing refactors things to make more sense, adds more tests, and also adds documentation.
Here we attempt to use arrowhead symbols in SVGs to allow artists to specify the orientation of attached symbols, rather than attempting to auto-detect it via projecting the circles onto the shape and using the normal as we did in #2.
It seems like this logic is both much simpler and less error-prone, especially around corners. It also gives artists much more control over the placement of attachments.
The one fragile aspect of the implementation is that we assume the first point of the arrow shape is its tip, and that the third point is its indented bottom. This appears to be consistent across all of Nina's SVGs, so far at least...
This auto-detects the direction of normals on which to attach attachments.
Right now it's getting very confused by corners, but otherwise it seems to mostly work.
Atul Varma