New one... ;o

0     *===========*       
1       =*     *=         
2     *=         =       *
3       *            *    
4       *     *           
5        *              * 
6 *             *         
7  *                   *  
8  *                     *
9 *                     * 
0       *             *   
1 *                     * 
2    *                   *
3       *              *  

It's never explained what are the rules about the supports:

• where do they start under the brigde?
• what size should they have?
• what happens when the bridge is at the last or second to last row of the matrix? (not tested, apparently)
• it's not said if the support should override blocks characters or not (for now, they do...)

for instance, the current rules call for "non sensical" brigdes. Imo, you should ask for supports that actually support something, meaning, they should be extended until they either reach the blocks, or the beginning/end of the "matrix" (considere this as a suggestion only). And they should stop before overwritting a block character (this part is rather an issue... Well, let's say a "strong" suggestion...)

another point: your tests allow the user the mutate the input. Not a problem in the way we usual write this, because it doesn't affect the tests, but that's bad practice in general, for this kind of input. So I'd suggest that you go for one of those two options:

• either you clearly state the user can mutate the input
• or or you state he cannot, and then you need to add specific tests for that.

edit:

and you need real sample tests, as I said below, testing for meangful cases (like, all the edge cases I talked about above!)

Duplicate.