gn format: add call wrapping

gn format: add call wrapping

For function call arguments:
- if all arguments plus ')' fit at current point, insert
- if all arguments fit at current, one per line
- otherwise at minimum of current and margin+4, one per line, recurse into children. Similar for binary expressions so that "a + b" can wrap.

Also, add basic output of AccessorNode + test.

(not as fancy as clang-format, but there's not that many long
function calls in .gn files.)

R=brettw@chromium.org
BUG=348474
Committed: https://crrev.com/8a4459122520cd7d59be4f3b03a2a7eb74344a71
Cr-Commit-Position: refs/heads/master@{#297765}

[scottmg, 2014-10-01 21:43:01]

Patchset #7 (id:120001) has been deleted

[scottmg, 2014-10-01 21:46:17]

It doesn't diff very well, but it's not as big as it looks initially.

OTOH, I think I'll refactor |Sequence| next and pull function calls out of that
because they're different enough that it's just making for way too many booleans
and execution paths in that function.

[brettw, 2014-10-01 23:52:38]

lgtm

https://codereview.chromium.org/607243002/diff/140001/tools/gn/format_test_da...
File tools/gn/format_test_data/022.golden (right):

https://codereview.chromium.org/607243002/diff/140001/tools/gn/format_test_da...
tools/gn/format_test_data/022.golden:3: x = a.b
This is one case where I think I would like to change in the future. In some
cases it's nice to have blank lines separating assignments for clarity, and in
other cases not. Like for chromium_code config, I would put the subtraction and
addition one-after-another. Or public_deps and deps. And some targets are two
variables only and the extra whitespace looks silly.

I wonder if it can preserve this?

[scottmg, 2014-10-02 00:12:31]

On 2014/10/01 23:52:38, brettw wrote:
> lgtm
> 
>
https://codereview.chromium.org/607243002/diff/140001/tools/gn/format_test_da...
> File tools/gn/format_test_data/022.golden (right):
> 
>
https://codereview.chromium.org/607243002/diff/140001/tools/gn/format_test_da...
> tools/gn/format_test_data/022.golden:3: x = a.b
> This is one case where I think I would like to change in the future. In some
> cases it's nice to have blank lines separating assignments for clarity, and in
> other cases not. Like for chromium_code config, I would put the subtraction
and
> addition one-after-another. Or public_deps and deps. And some targets are two
> variables only and the extra whitespace looks silly.
> 
> I wonder if it can preserve this?

Yeah, I've been thinking the same thing, but I'm not sure what the rule is
exactly. I'm happy to add some special case(s) to make things prettier if we can
figure out what they are though.

Size of RHSs? Number of statements in the block?

[scottmg, 2014-10-02 00:12:36]

The CQ bit was checked by scottmg@chromium.org

[commit-bot: I haz the power, 2014-10-02 00:13:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/607243002/140001

[commit-bot: I haz the power, 2014-10-02 01:00:52]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-02 01:00:54]

Try jobs failed on following builders:
  win_chromium_x64_rel_swarming on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[scottmg, 2014-10-02 01:26:26]

The CQ bit was checked by scottmg@chromium.org

[commit-bot: I haz the power, 2014-10-02 01:27:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/607243002/160001

[commit-bot: I haz the power, 2014-10-02 02:06:16]

Committed patchset #8 (id:160001) as c38649ed037c5c88680e9294286d9f48b5146b69

[commit-bot: I haz the power, 2014-10-02 02:06:53]

Patchset 8 (id:??) landed as
https://crrev.com/8a4459122520cd7d59be4f3b03a2a7eb74344a71
Cr-Commit-Position: refs/heads/master@{#297765}