Upload changes limits

mergin/client_push.py

@@ -24,7 +24,14 @@
 
 from .local_changes import LocalChange, LocalChanges
 
-from .common import UPLOAD_CHUNK_ATTEMPT_WAIT, UPLOAD_CHUNK_ATTEMPTS, UPLOAD_CHUNK_SIZE, ClientError, ErrorCode
+from .common import (
+    MAX_UPLOAD_VERSIONED_SIZE,
+    UPLOAD_CHUNK_ATTEMPT_WAIT,
+    UPLOAD_CHUNK_ATTEMPTS,
+    UPLOAD_CHUNK_SIZE,
+    MAX_UPLOAD_MEDIA_SIZE,
+    ClientError,
+)
 from .merginproject import MerginProject
 from .editor import filter_changes
 from .utils import get_data_checksum
@@ -296,28 +303,23 @@ def push_project_async(mc, directory) -> Optional[UploadJob]:
         mp.log.info(f"--- push {project_path} - nothing to do")
         return
 
-    mp.log.debug("push changes:\n" + pprint.pformat(changes))
+    mp.log.debug("push changes:\n" + pprint.pformat(asdict(changes)))
     tmp_dir = tempfile.TemporaryDirectory(prefix="python-api-client-")
 
     # If there are any versioned files (aka .gpkg) that are not updated through a diff,
     # we need to make a temporary copy somewhere to be sure that we are uploading full content.
     # That's because if there are pending transactions, checkpointing or switching from WAL mode
     # won't work, and we would end up with some changes left in -wal file which do not get
     # uploaded. The temporary copy using geodiff uses sqlite backup API and should copy everything.
-    for f in changes["updated"]:
-        if mp.is_versioned_file(f["path"]) and "diff" not in f:
+    for f in changes.updated:
+        if mp.is_versioned_file(f.path) and not f.diff:
             mp.copy_versioned_file_for_upload(f, tmp_dir.name)
 
-    for f in changes["added"]:
-        if mp.is_versioned_file(f["path"]):
+    for f in changes.added:
+        if mp.is_versioned_file(f.path):
             mp.copy_versioned_file_for_upload(f, tmp_dir.name)
 
-    local_changes = LocalChanges(
-        added=[LocalChange(**change) for change in changes["added"]],
-        updated=[LocalChange(**change) for change in changes["updated"]],
-        removed=[LocalChange(**change) for change in changes["removed"]],
-    )
-    job = create_upload_job(mc, mp, local_changes, tmp_dir)
+    job = create_upload_job(mc, mp, changes, tmp_dir)
     return job
 
 
@@ -471,12 +473,27 @@ def remove_diff_files(job: UploadJob) -> None:
                 os.remove(diff_file)
 
 
-def get_push_changes_batch(mc, mp: MerginProject) -> Tuple[dict, int]:
+def get_push_changes_batch(mc, mp: MerginProject) -> Tuple[LocalChanges, int]:
     """
     Get changes that need to be pushed to the server.
     """
     changes = mp.get_push_changes()
     project_role = mp.project_role()
     changes = filter_changes(mc, project_role, changes)
 
-    return changes, sum(len(v) for v in changes.values())
+    local_changes = LocalChanges(
+        added=[LocalChange(**change) for change in changes["added"]],
+        updated=[LocalChange(**change) for change in changes["updated"]],
+        removed=[LocalChange(**change) for change in changes["removed"]],
+    )
+    if local_changes.get_media_upload_size() > MAX_UPLOAD_MEDIA_SIZE:
+        raise ClientError(
+            f"Total size of media files to upload exceeds the maximum allowed size of {MAX_UPLOAD_MEDIA_SIZE / (1024**3)} GiB."
+        )
+
+    if local_changes.get_gpgk_upload_size() > MAX_UPLOAD_VERSIONED_SIZE:
+        raise ClientError(
+            f"Total size of GPKG files to upload exceeds the maximum allowed size of {MAX_UPLOAD_VERSIONED_SIZE / (1024**3)} GiB."
+        )
+
+    return local_changes, sum(len(v) for v in changes.values())

mergin/common.py

@@ -24,6 +24,12 @@
 # seconds to wait between sync callback calls
 SYNC_CALLBACK_WAIT = 0.01
 
+# maximum size of media files able to upload in one push (in bytes)
+MAX_UPLOAD_MEDIA_SIZE = 10 * (1024**3)
+
+# maximum size of GPKG files able to upload in one push (in bytes)
+MAX_UPLOAD_VERSIONED_SIZE = 5 * (1024**3)
+
 # default URL for submitting logs
 MERGIN_DEFAULT_LOGS_URL = "https://g4pfq226j0.execute-api.eu-west-1.amazonaws.com/mergin_client_log_submit"
 
@@ -39,7 +45,9 @@ class ErrorCode(Enum):
 
 
 class ClientError(Exception):
-    def __init__(self, detail: str, url=None, server_code=None, server_response=None, http_error=None, http_method=None):
+    def __init__(
+        self, detail: str, url=None, server_code=None, server_response=None, http_error=None, http_method=None
+    ):
         self.detail = detail
         self.url = url
         self.http_error = http_error

mergin/editor.py

@@ -1,7 +1,7 @@
 from itertools import filterfalse
 from typing import Callable, Dict, List
 
-from .utils import is_mergin_config, is_qgis_file, is_versioned_file
+from .utils import is_qgis_file
 
 EDITOR_ROLE_NAME = "editor"
 

mergin/local_changes.py

@@ -1,6 +1,10 @@
 from dataclasses import dataclass, field
 from datetime import datetime
-from typing import Dict, Optional, List, Tuple
+from typing import Optional, List, Tuple
+
+from .utils import is_versioned_file
+
+MAX_UPLOAD_CHANGES = 100
 
 
 @dataclass
@@ -55,6 +59,18 @@ class LocalChanges:
     updated: List[LocalChange] = field(default_factory=list)
     removed: List[LocalChange] = field(default_factory=list)
 
+    def __post_init__(self):
+        """
+        Enforce a limit of changes combined from `added` and `updated`.
+        """
+        total_changes = len(self.get_upload_changes())
+        if total_changes > MAX_UPLOAD_CHANGES:
+            # Calculate how many changes to keep from `added` and `updated`
+            added_limit = min(len(self.added), MAX_UPLOAD_CHANGES)
+            updated_limit = MAX_UPLOAD_CHANGES - added_limit
+            self.added = self.added[:added_limit]
+            self.updated = self.updated[:updated_limit]
+
     def to_server_payload(self) -> dict:
         return {
             "added": [change.to_server_data() for change in self.added],
@@ -96,3 +112,24 @@ def update_chunks(self, server_chunks: List[Tuple[str, str]]) -> None:
 
         for change in self.updated:
             change.chunks = self._map_unique_chunks(change.chunks, server_chunks)
+
+    def get_media_upload_size(self) -> int:
+        """
+        Calculate the total size of media files in added and updated changes.
+        """
+        total_size = 0
+        for change in self.get_upload_changes():
+            if not is_versioned_file(change.path):
+                total_size += change.size
+        return total_size
+
+    def get_gpgk_upload_size(self) -> int:
+        """
+        Calculate the total size of gpgk files in added and updated changes.
+        Do not calculate diffs (only new or overwriten files).
+        """
+        total_size = 0
+        for change in self.get_upload_changes():
+            if is_versioned_file(change.path) and not change.diff:
+                total_size += change.size
+        return total_size

mergin/merginproject.py

@@ -21,7 +21,7 @@
     conflicted_copy_file_name,
     edit_conflict_file_name,
 )
-
+from .local_changes import LocalChange
 
 this_dir = os.path.dirname(os.path.realpath(__file__))
 
@@ -470,20 +470,20 @@ def get_push_changes(self):
         changes["updated"] = [f for f in changes["updated"] if f not in not_updated]
         return changes
 
-    def copy_versioned_file_for_upload(self, f, tmp_dir):
+    def copy_versioned_file_for_upload(self, f: LocalChange, tmp_dir: str) -> str:
         """
         Make a temporary copy of the versioned file using geodiff, to make sure that we have full
         content in a single file (nothing left in WAL journal)
         """
-        path = f["path"]
+        path = f.path
         self.log.info("Making a temporary copy (full upload): " + path)
         tmp_file = os.path.join(tmp_dir, path)
         os.makedirs(os.path.dirname(tmp_file), exist_ok=True)
         self.geodiff.make_copy_sqlite(self.fpath(path), tmp_file)
-        f["size"] = os.path.getsize(tmp_file)
-        f["checksum"] = generate_checksum(tmp_file)
-        f["chunks"] = [str(uuid.uuid4()) for i in range(math.ceil(f["size"] / UPLOAD_CHUNK_SIZE))]
-        f["upload_file"] = tmp_file
+        f.size = os.path.getsize(tmp_file)
+        f.checksum = generate_checksum(tmp_file)
+        f.chunks = [str(uuid.uuid4()) for i in range(math.ceil(f.size / UPLOAD_CHUNK_SIZE))]
+        f.upload_file = tmp_file
         return tmp_file
 
     def get_list_of_push_changes(self, push_changes):

mergin/test/test_common.py

@@ -1,5 +1,6 @@
 from ..common import ClientError, ErrorCode
 
+
 def test_client_error_is_blocked_sync():
     """Test the is_blocked_sync method of ClientError."""
     error = ClientError(detail="", server_code=None)
@@ -12,6 +13,7 @@ def test_client_error_is_blocked_sync():
     error.server_code = ErrorCode.ProjectVersionExists.value
     assert error.is_blocking_sync() is True
 
+
 def test_client_error_is_rate_limit():
     """Test the is_rate_limit method of ClientError."""
     error = ClientError(detail="", http_error=None)
@@ -21,6 +23,7 @@ def test_client_error_is_rate_limit():
     error.http_error = 429
     assert error.is_rate_limit() is True
 
+
 def test_client_error_is_retryable_sync():
     """Test the is_retryable_sync method of ClientError."""
     error = ClientError(detail="", server_code=None, http_error=None)
@@ -43,4 +46,4 @@ def test_client_error_is_retryable_sync():
     error.http_error = 500
     assert error.is_retryable_sync() is False
     error.http_error = 429
-    assert error.is_retryable_sync() is True
+    assert error.is_retryable_sync() is True

mergin/test/test_local_changes.py

@@ -118,3 +118,62 @@ def test_local_changes_get_upload_changes():
     assert len(upload_changes) == 2  # Only added and updated should be included
     assert upload_changes[0].path == "file1.txt"  # First change is from added
     assert upload_changes[1].path == "file2.txt"  # Second change is from updated
+
+
+def test_local_changes_get_media_upload_size():
+    """Test the get_media_upload_size method of LocalChanges."""
+    # Create sample LocalChange instances
+    added = [
+        LocalChange(path="file1.txt", checksum="abc123", size=1024, mtime=datetime.now()),
+        LocalChange(path="file2.jpg", checksum="xyz789", size=2048, mtime=datetime.now()),
+    ]
+    updated = [
+        LocalChange(path="file3.mp4", checksum="lmn456", size=5120, mtime=datetime.now()),
+        LocalChange(path="file4.gpkg", checksum="opq123", size=1024, mtime=datetime.now()),
+    ]
+
+    # Initialize LocalChanges
+    local_changes = LocalChanges(added=added, updated=updated)
+
+    # Call get_media_upload_size
+    media_size = local_changes.get_media_upload_size()
+
+    # Assertions
+    assert media_size == 8192  # Only non-versioned files (txt, jpg, mp4) are included
+
+
+def test_local_changes_get_gpgk_upload_size():
+    """Test the get_gpgk_upload_size method of LocalChanges."""
+    # Create sample LocalChange instances
+    added = [
+        LocalChange(path="file1.gpkg", checksum="abc123", size=1024, mtime=datetime.now()),
+        LocalChange(path="file2.gpkg", checksum="xyz789", size=2048, mtime=datetime.now(), diff={"path": "diff1"}),
+    ]
+    updated = [
+        LocalChange(path="file3.gpkg", checksum="lmn456", size=5120, mtime=datetime.now()),
+        LocalChange(path="file4.txt", checksum="opq123", size=1024, mtime=datetime.now()),
+    ]
+
+    # Initialize LocalChanges
+    local_changes = LocalChanges(added=added, updated=updated)
+
+    # Call get_gpgk_upload_size
+    gpkg_size = local_changes.get_gpgk_upload_size()
+
+    # Assertions
+    assert gpkg_size == 6144  # Only GPKG files without diffs are included
+
+
+def test_local_changes_post_init():
+    """Test the __post_init__ method of LocalChanges."""
+    # Create more than MAX_UPLOAD_CHANGES changes
+    added = [LocalChange(path=f"file{i}.txt", checksum="abc123", size=1024, mtime=datetime.now()) for i in range(80)]
+    updated = [LocalChange(path=f"file{i}.txt", checksum="xyz789", size=2048, mtime=datetime.now()) for i in range(21)]
+
+    # Initialize LocalChanges
+    local_changes = LocalChanges(added=added, updated=updated)
+
+    # Assertions
+    assert len(local_changes.added) == 80  # All 80 added changes are included
+    assert len(local_changes.updated) == 20  # Only 20 updated changes are included to respect the limit
+    assert len(local_changes.added) + len(local_changes.updated) == 100  # Total is limited to MAX_UPLOAD_CHANGES