From 2719e2ede9696554e9a72d779c7a287b31eb8e23 Mon Sep 17 00:00:00 2001
From: Christian Monch <christian.moench@web.de>
Date: Sat, 28 Oct 2023 21:13:23 +0200
Subject: [PATCH] use iter_gitworktree to determine git-files

This commit uses the "real" iter_gitworktree
iterator instead a run-context that executes
`git ls-files`
---
 .../iter_collections/annexworktree.py         | 53 +------------------
 1 file changed, 2 insertions(+), 51 deletions(-)

diff --git a/datalad_next/iter_collections/annexworktree.py b/datalad_next/iter_collections/annexworktree.py
index e5cf284c8..05eb5cb92 100644
--- a/datalad_next/iter_collections/annexworktree.py
+++ b/datalad_next/iter_collections/annexworktree.py
@@ -17,64 +17,18 @@
 from typing import Generator
 
 from datalad.support.annexrepo import GeneratorAnnexJsonNoStderrProtocol
-from datalad_next.runners import StdOutCaptureGeneratorProtocol
 from datalad_next.runners.batch import annexjson_batchcommand
 from datalad_next.runners.run import run
 
 from .gitworktree import (
     GitWorktreeItem,
     GitWorktreeFileSystemItem,
-    _lsfiles_line2props,
-    _mode_type_map,
-    lsfiles_untracked_args,
+    iter_gitworktree,
 )
 
 lgr = logging.getLogger('datalad.ext.next.iter_collections.annexworktree')
 
 
-# The following protocol is part of the POC implementation. It allows a simple
-# `with run()`-commandline that returns `GitWorkTreeItem`-objects or
-# `GitWorkTreeFileSystemItem`-objects. In the real implementation we would
-# probably use `iter_gitworktree()`.
-class GitLsFilesProtocol(StdOutCaptureGeneratorProtocol):
-    def __init__(self, done_future=None, encoding=None) -> None:
-        from datalad_next.runners import LineSplitter
-
-        StdOutCaptureGeneratorProtocol.__init__(self, done_future, encoding)
-        self.encoding = encoding or 'utf-8'
-        self.line_splitter = LineSplitter(separator='\x00', keep_ends=False)
-
-    def send_line_result(self, line: str) -> None:
-        ipath, lsfiles_props = _lsfiles_line2props(line)
-        if not lsfiles_props:
-            self.send_result(
-                GitWorktreeItem(
-                    name=ipath,
-                    gittype=None,
-                    gitsha=None,
-                )
-            )
-        else:
-            self.send_result(
-                GitWorktreeItem(
-                    name=ipath,
-                    gittype=_mode_type_map[lsfiles_props['mode']],
-                    gitsha=lsfiles_props['gitsha'],
-                )
-            )
-
-    def pipe_data_received(self, fd: int, data: bytes) -> None:
-        assert fd == 1
-        for line in self.line_splitter.process(data.decode()):
-            self.send_line_result(line)
-
-    def pipe_connection_lost(self, fd: int, exc: BaseException | None) -> None:
-        if fd == 1:
-            remaining_string = self.line_splitter.finish_processing()
-            if remaining_string:
-                self.send_line_result(remaining_string)
-
-
 # TODO Could be `StrEnum`, came with PY3.11
 class AnnexTreeItemType(Enum):
     """Enumeration of item types of Git trees
@@ -149,9 +103,6 @@ def iter_annexworktree(
     # we cannot use git-annex-contentlocation, because it only reports on
     # present annex objects, and here we also need to report on would-be
     # locations
-    git_ls_files_cmd = ['git', 'ls-files', '-z', '--stage', '--cached']
-    if untracked:
-        git_ls_files_cmd.extend(lsfiles_untracked_args[untracked])
     git_annex_find_cmd = [
         'git', 'annex', 'find', '--include=*',
         '--json', '--json-error-messages', '.'
@@ -161,9 +112,9 @@ def iter_annexworktree(
     glf_store = dict()
     with \
             run(git_annex_find_cmd, protocol_class=GeneratorAnnexJsonNoStderrProtocol, **common_args) as git_annex_find, \
-            run(git_ls_files_cmd, protocol_class=GitLsFilesProtocol, **common_args) as git_ls_files, \
             annexjson_batchcommand(['git', 'annex', 'examinekey', '--json', '--batch'], **common_args) as examine_key:
 
+        git_ls_files = iter_gitworktree(path=path, untracked=untracked)
         for gaf_item, glf_item in zip_longest(git_annex_find, git_ls_files):
             if gaf_item:
                 gaf_store[PurePath(PurePosixPath(gaf_item['file']))] = gaf_item