If you insert invalid UTF-16 text into a string and save it in MongoDB, the database will happily store it, but if you try to run mapReduce() on it, you'll get one of the following two errors:

map reduce failed:{
"errmsg" : "exception: map invoke failed: JS Error: InternalError: buffer too small (anon):1",
"code" : 9014,
"ok" : 0
}

map reduce failed:{
"errmsg" : "exception: map invoke failed: JS Error: TypeError: bad surrogate character 0x61 (anon):1",
"code" : 9014,
"ok" : 0
}

Suggested fixes:

• Change MongoDB to not accept broken UTF-16 surrogate pairs
• Print a more informative error message when encountering this

Reproducable test case attached