view tests/test-convert-hg-source.t @ 36820:f33a87cf60cc

xdiff: add a preprocessing step that trims files

xdiff has a `xdl_trim_ends` step that removes common lines, unmatchable
lines. That is in theory good, but happens too late - after splitting,
hashing, and adjusting the hash values so they are unique. Those splitting,
hashing and adjusting hash values steps could have noticeable overhead.

Diffing two large files with minor (one-line-ish) changes are not uncommon.
In that case, the raw performance of those preparation steps seriously
matter. Even allocating an O(N) array and storing line offsets to it is
expensive.  Therefore my previous attempts [1] [2] cannot be good enough
since they do not remove the O(N) array assignment.

This patch adds a preprocessing step - `xdl_trim_files` that runs before
other preprocessing steps. It counts common prefix and suffix and lines in
them (needed for displaying line number), without doing anything else.

Testing with a crafted large (169MB) file, with minor change:

```
  open('a','w').write(''.join('%s\n' % (i % 100000) for i in xrange(30000000) if i != 6000000))
  open('b','w').write(''.join('%s\n' % (i % 100000) for i in xrange(30000000) if i != 6003000))
```

Running xdiff by a simple binary [3], this patch improves the xdiff perf by
more than 10x for the above case:

```
  # xdiff before this patch
  2.41s user 1.13s system 98% cpu 3.592 total
  # xdiff after this patch
  0.14s user 0.16s system 98% cpu 0.309 total
  # gnu diffutils
  0.12s user 0.15s system 98% cpu 0.272 total
  # (best of 20 runs)
```

It's still slightly slower than GNU diffutils. But it's pretty close now.

Testing with real repo data:

For the whole repo, this patch makes xdiff 25% faster:

```
  # hg perfbdiff --count 100 --alldata -c d334afc585e2 --blocks [--xdiff]
  # xdiff, after
  ! wall 0.058861 comb 0.050000 user 0.050000 sys 0.000000 (best of 100)
  # xdiff, before
  ! wall 0.077816 comb 0.080000 user 0.080000 sys 0.000000 (best of 91)
  # bdiff
  ! wall 0.117473 comb 0.120000 user 0.120000 sys 0.000000 (best of 67)
```

For files that are long (ex. commands.py), the speedup is more than 3x, very
significant:

```
  # hg perfbdiff --count 3000 --blocks commands.py.i 1 [--xdiff]
  # xdiff, after
  ! wall 0.690583 comb 0.690000 user 0.690000 sys 0.000000 (best of 12)
  # xdiff, before
  ! wall 2.240361 comb 2.210000 user 2.210000 sys 0.000000 (best of 4)
  # bdiff
  ! wall 2.469852 comb 2.440000 user 2.440000 sys 0.000000 (best of 4)
```

[1]: https://phab.mercurial-scm.org/D2631
[2]: https://phab.mercurial-scm.org/D2634
[3]:

```
// Code to run xdiff from command line. No proper error handling.
#include <stdlib.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include "mercurial/thirdparty/xdiff/xdiff.h"
#define ensure(x) if (!(x)) exit(255);
mmfile_t readfile(const char *path) {
  struct stat st; int fd = open(path, O_RDONLY);
  fstat(fd, &st); mmfile_t file = { malloc(st.st_size), st.st_size };
  ensure(read(fd, file.ptr, st.st_size) == st.st_size); close(fd);
  return file;
}
int main(int argc, char const *argv[]) {
  mmfile_t a = readfile(argv[1]), b = readfile(argv[2]);
  xpparam_t xpp = {0}; xdemitconf_t xecfg = {0}; xdemitcb_t ecb = {0};
  xdl_diff(&a, &b, &xpp, &xecfg, &ecb);
  return 0;
}
```

Differential Revision: https://phab.mercurial-scm.org/D2686

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > convert=
  > [convert]
  > hg.saverev=False
  > EOF
  $ hg init orig
  $ cd orig
  $ echo foo > foo
  $ echo bar > bar
  $ hg ci -qAm 'add foo bar' -d '0 0'
  $ echo >> foo
  $ hg ci -m 'change foo' -d '1 0'
  $ hg up -qC 0
  $ hg copy --after --force foo bar
  $ hg copy foo baz
  $ hg ci -m 'make bar and baz copies of foo' -d '2 0'
  created new head

Test that template can print all file copies (issue4362)
  $ hg log -r . --template "{file_copies % ' File: {file_copy}\n'}"
   File: bar (foo)
   File: baz (foo)

  $ hg bookmark premerge1
  $ hg merge -r 1
  merging baz and foo to baz
  1 files updated, 1 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)
  $ hg ci -m 'merge local copy' -d '3 0'
  $ hg up -C 1
  1 files updated, 0 files merged, 1 files removed, 0 files unresolved
  (leaving bookmark premerge1)
  $ hg bookmark premerge2
  $ hg merge 2
  merging foo and baz to baz
  1 files updated, 1 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)
  $ hg ci -m 'merge remote copy' -d '4 0'
  created new head

Make and delete some tags

  $ hg tag that
  $ hg tag --remove that
  $ hg tag this

#if execbit
  $ chmod +x baz
#else
  $ echo some other change to make sure we get a rev 5 > baz
#endif
  $ hg ci -m 'mark baz executable' -d '5 0'
  $ cd ..
  $ hg convert --datesort orig new 2>&1 | grep -v 'subversion python bindings could not be loaded'
  initializing destination new repository
  scanning source...
  sorting...
  converting...
  8 add foo bar
  7 change foo
  6 make bar and baz copies of foo
  5 merge local copy
  4 merge remote copy
  3 Added tag that for changeset 88586c4e9f02
  2 Removed tag that
  1 Added tag this for changeset c56a7f387039
  0 mark baz executable
  updating bookmarks
  $ cd new
  $ hg out ../orig
  comparing with ../orig
  searching for changes
  no changes found
  [1]
#if execbit
  $ hg bookmarks
     premerge1                 3:973ef48a98a4
     premerge2                 8:91d107c423ba
#else
Different hash because no x bit
  $ hg bookmarks
     premerge1                 3:973ef48a98a4
     premerge2                 8:3537b15eaaca
#endif

Test that redoing a convert results in an identical graph
  $ cd ../
  $ rm new/.hg/shamap
  $ hg convert --datesort orig new 2>&1 | grep -v 'subversion python bindings could not be loaded'
  scanning source...
  sorting...
  converting...
  8 add foo bar
  7 change foo
  6 make bar and baz copies of foo
  5 merge local copy
  4 merge remote copy
  3 Added tag that for changeset 88586c4e9f02
  2 Removed tag that
  1 Added tag this for changeset c56a7f387039
  0 mark baz executable
  updating bookmarks
  $ hg -R new log -G -T '{rev} {desc}'
  o  8 mark baz executable
  |
  o  7 Added tag this for changeset c56a7f387039
  |
  o  6 Removed tag that
  |
  o  5 Added tag that for changeset 88586c4e9f02
  |
  o    4 merge remote copy
  |\
  +---o  3 merge local copy
  | |/
  | o  2 make bar and baz copies of foo
  | |
  o |  1 change foo
  |/
  o  0 add foo bar
  

check shamap LF and CRLF handling

  $ cat > rewrite.py <<EOF
  > import sys
  > # Interlace LF and CRLF
  > lines = [(l.rstrip() + ((i % 2) and b'\n' or b'\r\n'))
  >          for i, l in enumerate(open(sys.argv[1], 'rb'))]
  > open(sys.argv[1], 'wb').write(b''.join(lines))
  > EOF
  $ $PYTHON rewrite.py new/.hg/shamap
  $ cd orig
  $ hg up -qC 1
  $ echo foo >> foo
  $ hg ci -qm 'change foo again'
  $ hg up -qC 2
  $ echo foo >> foo
  $ hg ci -qm 'change foo again again'
  $ cd ..
  $ hg convert --datesort orig new 2>&1 | grep -v 'subversion python bindings could not be loaded'
  scanning source...
  sorting...
  converting...
  1 change foo again again
  0 change foo again
  updating bookmarks

init broken repository

  $ hg init broken
  $ cd broken
  $ echo a >> a
  $ echo b >> b
  $ hg ci -qAm init
  $ echo a >> a
  $ echo b >> b
  $ hg copy b c
  $ hg ci -qAm changeall
  $ hg up -qC 0
  $ echo bc >> b
  $ hg ci -m changebagain
  created new head
  $ HGMERGE=internal:local hg -q merge
  $ hg ci -m merge
  $ hg mv b d
  $ hg ci -m moveb

break it

  $ rm .hg/store/data/b.*
  $ cd ..
  $ hg --config convert.hg.ignoreerrors=True convert broken fixed
  initializing destination fixed repository
  scanning source...
  sorting...
  converting...
  4 init
  ignoring: data/b.i@1e88685f5dde: no match found
  3 changeall
  2 changebagain
  1 merge
  0 moveb
  $ hg -R fixed verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  3 files, 5 changesets, 5 total revisions

manifest -r 0

  $ hg -R fixed manifest -r 0
  a

manifest -r tip

  $ hg -R fixed manifest -r tip
  a
  c
  d