Allow downloading a single file by project name without a local checkout

Author: varmar05

.gitignore

@@ -14,3 +14,4 @@ deps
 venv
 debug.py
 .vscode/
+.python-version

mergin/cli.py

@@ -335,17 +335,32 @@ def share(ctx, project):
 @click.argument("filepath")
 @click.argument("output")
 @click.option("--version", help="Project version tag, for example 'v3'")
+@click.option(
+    "--project",
+    help="Full project name ('<workspace>/<project>') to download the file directly from the server. "
+    "If not given, the current directory is used and must be an existing checked out project.",
+)
 @click.pass_context
-def download_file(ctx, filepath, output, version):
+def download_file(ctx, filepath, output, version, project):
     """
-    Download project file at specified version. `project` needs to be a combination of namespace/project.
-    If no version is given, the latest will be fetched.
+    Download project file at specified version. If no version is given, the latest will be fetched.
     """
     mc = ctx.obj["client"]
     if mc is None:
         return
+    if project is None:
+        # no --project given, so we default to the current directory - make sure that's actually a checked out project
+        try:
+            MerginProject(os.getcwd()).project_full_name()
+        except InvalidProject:
+            click.secho(
+                "Current directory is not a Mergin Maps project. Run this command from within a "
+                "checked out project directory, or pass --project <workspace>/<project>.",
+                fg="red",
+            )
+            return
     try:
-        job = download_file_async(mc, os.getcwd(), filepath, output, version)
+        job = download_file_async(mc, project or os.getcwd(), filepath, output, version)
         with click.progressbar(length=job.total_size) as bar:
             last_transferred_size = 0
             while download_project_is_running(job):

mergin/client.py

@@ -1199,7 +1199,7 @@ def download_file(self, project_dir, file_path, output_filename, version=None):
         """
         Download project file at specified version. Get the latest if no version specified.
 
-        :param project_dir: project local directory
+        :param project_dir: project local directory or a full project name ("<workspace>/<project>")
         :type project_dir: String
         :param file_path: relative path of file to download in the project directory
         :type file_path: String
@@ -1401,11 +1401,11 @@ def download_files(
         """
         Download project files at specified version. Get the latest if no version specified.
 
-        :param project_dir: project local directory
+        :param project_dir: project local directory or a full project name ("<workspace>/<project>")
         :type project_dir: String
         :param file_path: List of relative paths of files to download in the project directory
         :type file_path: List[String]
-        :param output_paths: List of paths for files to download to. Should be same length of as file_path. Default is `None` which means that files are downloaded into MerginProject at project_dir.
+        :param output_paths: List of paths for files to download to. Should be same length of as file_path. Default is `None` which means that files are downloaded into MerginProject at project_dir (only valid when project_dir is an existing local checkout).
         :type output_paths: List[String]
         :param version: optional version tag for downloaded file
         :type version: String

mergin/client_pull.py

@@ -22,11 +22,11 @@
 import concurrent.futures
 
 
-from .common import CHUNK_SIZE, ClientError, DeltaChangeType, PullActionType
+from .common import CHUNK_SIZE, ClientError, DeltaChangeType, InvalidProject, PullActionType
 from .models import ProjectDelta, ProjectDeltaChange, PullAction
 from .merginproject import MerginProject
-from .utils import cleanup_tmp_dir, save_to_file
-from typing import List, Optional
+from .utils import cleanup_tmp_dir, is_versioned_file, save_to_file
+from typing import List, Optional, Union
 
 # status = download_project_async(...)
 #
@@ -54,7 +54,7 @@ def __init__(
         update_tasks,
         download_queue_items,
         tmp_dir: tempfile.TemporaryDirectory,
-        mp,
+        mp: Union[MerginProject, "DownloadScratchContext"],
         project_info,
     ):
         self.project_path = project_path
@@ -64,7 +64,7 @@ def __init__(
         self.update_tasks = update_tasks
         self.download_queue_items = download_queue_items
         self.tmp_dir = tmp_dir
-        self.mp = mp  # MerginProject instance
+        self.mp = mp
         self.is_cancelled = False
         self.project_info = project_info  # parsed JSON with project info returned from the server
         self.failure_log_file = None  # log file, copied from the project directory if download fails
@@ -80,6 +80,25 @@ def dump(self):
         print("--- END ---")
 
 
+class DownloadScratchContext:
+    """
+    Minimal stand-in for MerginProject, used by download_files_async() when downloading files
+    directly by project name ("<workspace>/<project>") without an existing local project checkout.
+
+    Provides only what the shared download job code actually needs from MerginProject.
+    """
+
+    def __init__(self, mc, cache_dir: str):
+        self.log = mc.log
+        self.cache_dir = cache_dir
+        # only used by _cleanup_failed_download() to look for a log file
+        self.dir = cache_dir
+
+    def remove_logging_handler(self):
+        # no-op: self.log is mc's shared logger, not owned by this throwaway context
+        pass
+
+
 class DownloadQueueItem:
     """
     a piece of data from a project that should be downloaded - it can be either a chunk or it can be a diff.
@@ -398,7 +417,7 @@ def __init__(
         self.download_queue_items = download_queue_items
         self.latest_version = latest_version
 
-    def apply(self, directory, mp):
+    def apply(self, directory, mp: Union[MerginProject, "DownloadScratchContext"]):
         """assemble downloaded chunks into a single file"""
 
         if self.destination_file is None:
@@ -411,14 +430,14 @@ def apply(self, directory, mp):
         os.makedirs(file_dir, exist_ok=True)
 
         # ignore check if we download not-latest version of gpkg file (possibly reconstructed on server on demand)
-        check_size = self.latest_version or not mp.is_versioned_file(self.file_path)
+        check_size = self.latest_version or not is_versioned_file(self.file_path)
         # merge chunks together (and delete them afterwards)
         file_to_merge = DownloadFile(dest_file_path, self.download_queue_items, check_size)
         file_to_merge.from_chunks()
 
         # Make a copy of the file to meta dir only if there is no user-specified path for the file.
-        # destination_file is None for full project download and takes a meaningful value for a single file download.
-        if mp.is_versioned_file(self.file_path) and self.destination_file is None:
+        # destination_file is None for full project download and takes a meaningful value for a single file download
+        if self.destination_file is None and is_versioned_file(self.file_path):
             mp.geodiff.make_copy_sqlite(mp.fpath(self.file_path), mp.fpath_meta(self.file_path))
 
 
@@ -902,9 +921,30 @@ def download_files_async(
     """
     Starts background download project files at specified version.
     Returns handle to the pending download.
+
+    `project_dir` can either be an existing local project directory (previously fetched with
+    download_project()), or a full project name ("<workspace>/<project>") to download files
+    directly from the server without needing a local checkout. In the latter case, `output_paths`
+    must be provided explicitly, as there is no project directory to place files into by default.
     """
-    mp = MerginProject(project_dir)
-    project_path = mp.project_full_name()
+    # temporary directory to stage downloaded chunks in
+    tmp_dir = tempfile.TemporaryDirectory(prefix="python-api-client-")
+
+    mp: Union[MerginProject, "DownloadScratchContext"]
+    try:
+        mp = MerginProject(project_dir)
+        project_path = mp.project_full_name()
+    except InvalidProject:
+        # project_dir is not an existing local checkout - treat it as a full project name
+        # ("<workspace>/<project>") and download straight from the server instead
+        if output_paths is None:
+            cleanup_tmp_dir(mc, tmp_dir)
+            raise ClientError(
+                "output_paths must be provided when downloading files without an existing local project checkout"
+            )
+        project_path = project_dir
+        mp = DownloadScratchContext(mc, tmp_dir.name)
+
     ver_info = f"at version {version}" if version is not None else "at latest version"
     mp.log.info(f"Getting [{', '.join(file_paths)}] {ver_info}")
     latest_proj_info = mc.project_info(project_path)
@@ -914,9 +954,6 @@ def download_files_async(
         project_info = latest_proj_info
     mp.log.info(f"Got project info. version {project_info['version']}")
 
-    # set temporary directory for download
-    tmp_dir = tempfile.TemporaryDirectory(prefix="python-api-client-")
-
     if output_paths is None:
         output_paths = []
         for file in file_paths:
@@ -991,6 +1028,4 @@ def download_files_finalize(job: DownloadJob):
     for task in job.update_tasks:
         task.apply(job.tmp_dir, job.mp)
 
-    # Remove temporary download directory
-    if job.tmp_dir is not None and os.path.exists(job.tmp_dir.name):
-        cleanup_tmp_dir(job.mp, job.tmp_dir)
+    cleanup_tmp_dir(job.mp, job.tmp_dir)

mergin/test/test_client.py

@@ -1354,6 +1354,42 @@ def test_download_file(mc):
         mc.download_file(project_dir, f_updated, f_downloaded, version="v5")
 
 
+def test_download_file_without_checkout(mc):
+    """Test downloading a single file directly by project name, without an existing local checkout."""
+    test_project = "test_download_file_without_checkout"
+    project = create_project_path(test_project, mc)
+    project_dir = os.path.join(TMP_DIR, test_project)
+    f_updated = "base.gpkg"
+
+    create_versioned_project(mc, test_project, project_dir, f_updated)
+
+    # download straight from the server by "workspace/project" name into a fresh directory
+    # that has never been used as a project checkout
+    download_dir = os.path.join(TMP_DIR, test_project + "_no_checkout")
+    remove_folders([download_dir])
+    os.makedirs(download_dir, exist_ok=True)
+    f_downloaded = os.path.join(download_dir, f_updated)
+
+    expected_content = "inserted_1_A.gpkg"
+    mc.download_file(project, f_updated, f_downloaded, version="v2")
+    expected = os.path.join(TEST_DATA_DIR, expected_content)
+    assert check_gpkg_same_content(MerginProject(project_dir), f_downloaded, expected)
+    assert not os.path.exists(os.path.join(download_dir, ".mergin"))
+
+    # output_paths must be provided explicitly when there is no local checkout
+    with pytest.raises(ClientError, match="output_paths must be provided"):
+        mc.download_files(project, [f_updated])
+
+    # non-existent file in an existing project - same error as with a local checkout
+    with pytest.raises(ClientError, match=r"No \[does_not_exist\.gpkg\] exists at version v2"):
+        mc.download_file(project, "does_not_exist.gpkg", f_downloaded, version="v2")
+
+    # non-existent / inaccessible project should fail clearly too
+    nonexistent_project = create_project_path("this_project_does_not_exist", mc)
+    with pytest.raises(ClientError):
+        mc.download_file(nonexistent_project, f_updated, f_downloaded)
+
+
 def test_download_diffs(mc):
     """Test download diffs for a project file between specified project versions."""
     test_project = "test_download_diffs"