adjust upload

Author: emptymalei

dataherb/cmd/sync_git.py

@@ -1,9 +1,72 @@
+import sys
+import json
 import click
+import git
+from pathlib import Path
 
 
-def upload_dataset_to_git(source, target):
+def is_git_repo(path):
+    try:
+        _ = git.Repo(path).git_dir
+        return True
+    except git.exc.InvalidGitRepositoryError:
+        return False
+
+
+def get_dataherb(source):
+
+    if not (Path(source) / "dataherb.json").exists():
+        click.echo(
+            f"No dataherb.json found in {source}. Please run `dataherb create` first."
+        )
+        sys.exit()
+
+    with open(Path(source) / "dataherb.json", "r") as f:
+        data = json.load(f)
+
+    return data
+
+
+def upload_dataset_to_git(source, target, experimental=False):
+    """uploads local folder to remote"""
+
+    is_git_initialized = is_git_repo(source)
+
+    if not experimental:
+        text = (
+            f"git sync is still a WIP.\n"
+            f"Please go to {source} and sync your git repository to {target} manually.\n"
+        )
+        if is_git_initialized:
+            text += (
+                f"Note: simply add, commit and push."
+            )
+        else:
+            text += (
+                f"Note: git init your repo, commit, add remote {target}, and push."
+            )
+        click.echo(text)
+    else:
+        if is_git_initialized:
+            repo = git.Repo(source)
+            repo.index.add(["*"])
+            repo.index.commit("created datset: added dataherb.json")
+
+            if len(repo.remotes) == 0:
+                origin = repo.create_remote('origin', target)
+                assert origin.exists()
+                origin.fetch()
+                repo.create_head('master', origin.refs.master).set_tracking_branch(origin.refs.master).checkout()
+                origin.push()
+            else:
+                repo.git.push()
+        else:
+            repo = git.Repo.init(source)
+            repo.git.add(["*"])
+            repo.index.commit("initial commit")
+            origin = repo.create_remote('origin', target)
+            assert origin.exists()
+            origin.fetch()
+            repo.create_head('master', origin.refs.master).set_tracking_branch(origin.refs.master).checkout()
+            origin.push()
 
-    click.echo(
-        f"git sync has not yet been implemented.\n"
-        f"Please go to {source} and sync your git repository to {target} manually."
-    )

dataherb/command.py

@@ -287,7 +287,8 @@ def remove(flora, herb_id):
     "All contents in this folder will be uploaded.\n"
     "Are you sure this is the correct path?"
 )
-def upload():
+@click.option("--experimental", "-e", default=False)
+def upload(experimental):
     """
     upload dataset in the current folder to the remote destination
     """
@@ -312,7 +313,7 @@ def upload():
         if md.metadata.get("source") == "s3":
             upload_dataset_to_s3(__CWD__, md_uri)
         elif md.metadata.get("source") == "git":
-            upload_dataset_to_git(__CWD__, md_uri)
+            upload_dataset_to_git(__CWD__, md_uri, experimental=experimental)
 
 
 @dataherb.command()