[PATCH 7 of 8] ignore: add support for including subdir .hgignores

[Durham Goode]

On 5/13/15, 8:03 PM, "Siddharth Agarwal" <sid at less-broken.com> wrote:

>On 05/13/2015 08:13 AM, Durham Goode wrote:
>> # HG changeset patch
>> # User Durham Goode <durham at fb.com>
>> # Date 1431485142 25200
>> #      Tue May 12 19:45:42 2015 -0700
>> # Node ID 42bafc92b436181ba5d6568967577a214198648d
>> # Parent  f058e14998c947d86ccbdd17edaf2392b4190cce
>> ignore: add support for including subdir .hgignores
>
>This should have a (BC) note.
>
>> This adds support for including sub-hgignore files in a .hgignore. The
>>syntax is
>> '%include path/to/.hgignore' and is relative to the current .hgignore.
>
>I actually suspect a better syntax might '#include path/to/.hgignore',
>so that older Mercurial doesn't try and interpret the ignore rules as a
>pattern. I know this makes it different from config rules, though.

Talked a bit in IRC about this.  Moving to a
'subinclude:path/to/.hgignore' style syntax might be nice, since then
users could use a similar pattern of 'include:path/to/.hgignore' with -I
and -X and stuff.  I'll see if I can make that happen, then resend.

>
>>  The rules
>> inside the sub-ignore file only apply to files in that subdirectory.
>>
>> At the moment we only support globs in sub-ignore files. regexs will
>>cause an
>> exception. This is because we can't reliabily
>(sp)
>>  modify a regex to have a prefix
>> (ex: adding a prefix to '^foo|^bar' would require parsing the regex).
>>In a
>> subsequent patch we will add a config option to allow users to opt-in
>>to regexs
>> in sub-ignores if they're certain their regexs are sane.
>>
>> diff --git a/mercurial/ignore.py b/mercurial/ignore.py
>> --- a/mercurial/ignore.py
>> +++ b/mercurial/ignore.py
>> @@ -19,6 +19,13 @@ def ignorepats(ui, root, filepath):
>>      syntax = 'relre:'
>>      patterns = []
>>      warnings = []
>> +    subpaths = set()
>> +
>> +    common = os.path.commonprefix([filepath, root])
>> +    subdir = None
>> +    if root and common == root:
>> +        # Find the prefix that should be applied to these patterns
>> +        subdir = os.path.dirname(filepath[len(root) + 1:])
>>  
>>      fp = open(filepath)
>>  
>> @@ -35,6 +42,18 @@ def ignorepats(ui, root, filepath):
>>          if not line:
>>              continue
>>  
>> +        if line.startswith('%include '):
>> +            subpath = line[9:]
>> +            fullpath = os.path.join(os.path.dirname(filepath), subpath)
>> +            fullpath = os.path.normpath(fullpath)
>> +            subpats, subsubpaths = readignorefile(ui, root, fullpath)
>
>'subpats' is close enough to 'subpath' that my eyes crossed over while
>reading this. Could we change one of the names?
>
>> +
>> +            subpaths.add(fullpath)
>> +            subpaths.update(subsubpaths)
>> +
>> +            patterns.extend(subpats)
>> +            continue
>> +
>>          if line.startswith('syntax:'):
>>              s = line[7:].strip()
>>              try:
>> @@ -53,22 +72,33 @@ def ignorepats(ui, root, filepath):
>>                  linesyntax = rels
>>                  line = line[len(s) + 1:]
>>                  break
>> +
>> +        if subdir:
>> +            # The ignore file is inside the repo, so make its patterns
>> +            # relative to its location.
>> +            if linesyntax == 'relre:':
>> +                message = _('cannot use sub-ignore files that use
>>regular '
>> +                            'expressions')
>> +                raise util.Abort(message)
>> +            elif linesyntax == 'relglob:':
>> +                line = '%s/*%s' % (subdir, line)
>
>What happens if linesyntax is something else?

I don't think ignore files ever worked with anything besides glob and
regex. But if it's something else, we'll just not prefix it.