Mercurial > hg > hg-fastimport

comparison hgext3rd/fastimport/__init__.py @ 95:3b398a887b95

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Use a sqlite3 database to store the blob data if available This is much more performant than using a filesystem when we are dealing with a large number of blobs. If sqlite3 is not available, then fallback to writing to the filesystem. In both cases, the blob data is compressed before writing to save space. A new option has also been added to specify a path for persistent blob data. This is only really important for large continuous interations where the source data has no concept of export marks and thus only gets bigger. What we gain here is a reduction in the write load on the disk.
author Roy Marples <roy@marples.name>
date Thu, 21 Jan 2021 23:59:21 +0000
parents dc1d11466aa6
children 7eb15a5c4cad
comparison
equal deleted inserted replaced
94:b18c5670f6c0 95:3b398a887b95
30 command = cmdutil.command(cmdtable) 30 command = cmdutil.command(cmdtable)
31 31
32 @command(b'fastimport', 32 @command(b'fastimport',
33 [(b'', b'branchsort', None, _(b'try to sort changesets by branches')), 33 [(b'', b'branchsort', None, _(b'try to sort changesets by branches')),
34 (b'', b'datesort', None, _(b'try to sort changesets by date')), 34 (b'', b'datesort', None, _(b'try to sort changesets by date')),
35 (b'', b'sourcesort', None, _(b'preserve source changesets order'))], 35 (b'', b'sourcesort', None, _(b'preserve source changesets order')),
36 (b'', b'blobpath', b'', _(b'path for persistent blob data'))],
36 _(b'hg fastimport SOURCE ...'), 37 _(b'hg fastimport SOURCE ...'),
37 norepo=False) 38 norepo=False)
38 39
39 def fastimport(ui, repo, *sources, **opts): 40 def fastimport(ui, repo, *sources, **opts):
40 '''Convert a git fastimport dump into Mercurial changesets. 41 '''Convert a git fastimport dump into Mercurial changesets.
53 # Boo, hiss. 54 # Boo, hiss.
54 55
55 if not sources: 56 if not sources:
56 sources = (b'-') 57 sources = (b'-')
57 58
59 opts = pycompat.byteskwargs(opts)
60
58 # assume fastimport metadata (usernames, commit messages) are 61 # assume fastimport metadata (usernames, commit messages) are
59 # encoded UTF-8 62 # encoded UTF-8
60 convcmd.orig_encoding = encoding.encoding 63 convcmd.orig_encoding = encoding.encoding
61 encoding.encoding = b'UTF-8' 64 encoding.encoding = b'UTF-8'
62 65
63 # sink is the current repo, src is the list of fastimport streams 66 # sink is the current repo, src is the list of fastimport streams
64 destc = hg.mercurial_sink(ui, b'hg', repo.root) 67 destc = hg.mercurial_sink(ui, b'hg', repo.root)
65 srcc = fastimport_source(ui, b'fastimport', repo, sources) 68 srcc = fastimport_source(ui, b'fastimport', repo, sources, opts[b'blobpath'])
66 69
67 opts = pycompat.byteskwargs(opts)
68 defaultsort = b'branchsort' # for efficiency and consistency 70 defaultsort = b'branchsort' # for efficiency and consistency
69 sortmodes = (b'branchsort', b'datesort', b'sourcesort') 71 sortmodes = (b'branchsort', b'datesort', b'sourcesort')
70 sortmode = [m for m in sortmodes if opts.get(m)] 72 sortmode = [m for m in sortmodes if opts.get(m)]
71 if len(sortmode) > 1: 73 if len(sortmode) > 1:
72 raise error.Abort(_(b'more than one sort mode specified')) 74 raise error.Abort(_(b'more than one sort mode specified'))