hgweb: make templater mostly compatible with log templates Prepares for gradually switching templatekw.showsuccsandmarkers() to new API. This was a PoC showing how to reuse templatekw functions in hgweb. We could remove rev, node, author, etc. from the commonentry() table, but we'll have to carefully remove all corresponding symbols from webcommands.*(). Otherwise, we would face the issue5612. Still templatekw.keywords aren't exported. Otherwise some tests would fail because the atom template expects {files} to be empty in filelog, but templatekw.showfiles() provides the {files} implementation.

log: do not invoke templatekw.showobsfate() as a function Prepares for switching to the (context, mapping) API. I tried, but it appeared not an one-off change to extract a non-template function from showobsfate(), which deeply depends on the templater internals.

templatekw: inline getfiles() It's just three lines. We don't need a separate function for that.

templatekw: factor out function to build a list of files per status Removes copy-paste code before switching to the (context, mapping) API.

templatekw: switch non-showlist template keywords to new API

templatekw: extract non-templatekw function as getgraphnode() Prepares for switching to the (context, mapping) API. We still need (repo, ctx) function for the fast path.

convert: avoid closing ui.fout in subversion code (issue5807) Don't close 'fp' (= 'ui.fout') stream to prevent 'ValueError: I/O operation on closed file' (Bug #5807). Regression of changeset 30261:6bed17ba00a1 (https://www.mercurial-scm.org/repo/hg/rev/6bed17ba00a1)

cmdutil: expand filename format string by templater (BC) This is BC because '{}' could be a valid filename before, but I believe good programmers wouldn't use such catastrophic output filenames. On the other hand, '\' has to be escaped since it is a directory separator on Windows. Thanks to Matt Harbison for spotting this weird issue. This patch also adds cmdutil.rendertemplate(ctx, tmpl, props) as a simpler way of expanding template against single changeset. .. bc:: '{' in output filename passed to archive/cat/export is taken as a start of a template expression.

annotate: do not poorly split lines at CR (issue5798) mdiff and lines(text) take only LF as a line separator, but str.splitlines() breaks our assumption. Use mdiff.splitnewlines() consistently. It's hard to read \r in tests, so \r is replaced with [CR]. I had to wrap sed by a shell function to silence check-code warning.

templater: add option to parse template string just like raw string literal This seems a bit odd because the template syntax has no raw string literal containing template fragments, but is necessary to port filename format string to templater. See the next patch.

cmdutil: reorder optional arguments passed to makefileobj() **props will be passed directly to templater.

cmdutil: strip "%m" pattern (first line of commit message) from both ends This matches the behavior of the template keyword {desc}.

test-acl: mock up util.getuser() to trust $LOGNAME on Windows The test relies on POSIX-like getuser() behavior, so we can't use windows.getuser().

fileset: drop bad "elif:" trying to check invalid size expression Since str.isdigit is a function, the last "elif" was always true. An invalid expression is rejected by util.sizetoint(), so we don't need "elif".

py3: fix test-command-template.t to write files in binary mode

py3: use bytestr() to coerce position carried by ParseError to string The position value is either int or byte string.

py3: use bytes.endswith('\n') to strip off '\n' from debug color output

py3: fix type of attribute names forwarded by templatekw._hybrid # skip-blame because just r'' prefixes

py3: move between bytes and unicode when re-raising IOError IOError is a Python exception, which argument must be a system string.

py3: use '%d' to format diffstat sum

py3: make regexp literal bytes in templatefilters.py # skip-blame because just b'' prefixes

templatefilters: use encoding.unifromlocal/unitolocal() for py3 compatibility

py3: replace str() with it's bytes equivalent in hgext/shelve.py Internally we are dealing with bytes everywhere, so anything returning a unicode results in an error or some change in behaviour. Differential Revision: https://phab.mercurial-scm.org/D2520

py3: make sure we write bytes in a file open in bytes mode # skip-blame just b'' prefix Differential Revision: https://phab.mercurial-scm.org/D2519

py3: add b'' prefixes in tests/test-obsolete.t # skip-blame because it's just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D2518

py3: add a b'' prefix in tests/test-fncache.t # skip-blame because it's just b'' prefix Differential Revision: https://phab.mercurial-scm.org/D2517

py3: use pycompat.bytestr() to convert error instances to bytes Differential Revision: https://phab.mercurial-scm.org/D2516

py3: listify the return value of filter() filter() on Python 3 returns a filter object. Differential Revision: https://phab.mercurial-scm.org/D2515

py3: use '%d' instead of '%s' for ints Differential Revision: https://phab.mercurial-scm.org/D2514

py3: add 14 new passing tests to whitelist Differential Revision: https://phab.mercurial-scm.org/D2511