bzrlib.groupcompress.PyrexGroupCompressor : API documentation

Produce a serialised group of compressed texts.

It contains code very similar to SequenceMatcher because of having a similar task. However some key differences apply:

there is no junk, we want a minimal edit not a human readable diff.
we don't filter very common lines (because we don't know where a good range will start, and after the first text we want to be emitting minmal edits only.
we chain the left side, not the right side
we incrementally update the adjacency matrix as new lines are provided.
we look for matches in all of the left side, so the routine which does the analagous task of find_longest_match does not need to filter on the left side.

Method	__init__	Create a GroupCompressor.
Method	_compress	see _CommonGroupCompressor._compress
Method	_output_chunks	Output some chunks.

Method	compress	Compress lines with label key.
Method	extract	Extract a key previously added to the compressor.
Method	flush	Finish this group, creating a formatted stream.
Method	pop_last	Call this if you want to 'revoke' the last compression.
Method	ratio	Return the overall compression ratio.

def _compress(self, key, bytes, max_delta_size, soft=False):

see _CommonGroupCompressor._compress

b.g.PyrexGroupCompressor(_CommonGroupCompressor) : class documentation