It is possible to do cluster analysis on text files. (in fact, anytime you can caclulate some metric between elements in a set, you can cluster that set)

use a spatial index — the DBSCAN article notes that this results in a O(n log n) runtime
- more generally, any technique that speeds up the nearest neighbor search
MinHash
use an algorithm that approximates the edit distance
use an algorithm that clusters suboptimally
locality-sensitive hashing

modules that do both metric+cluster

metric modules, when doing metric+cluster

clustering modules, when doing metric+cluster

similar fields