Add an endpoint to query a dataset's git history from it's id.

* gn3/api/metadata.py: Import get_history.
(view_history): New endpoint.
* gn3/db/datasets.py (get_history): New function.

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>

2 files changed, 53 insertions, 1 deletions

diff --git a/gn3/api/metadata.py b/gn3/api/metadata.py
index fe7e903..a6a88fc 100644
--- a/gn3/api/metadata.py
+++ b/gn3/api/metadata.py
@@ -7,7 +7,8 @@ from flask import request
 from flask import current_app
 
 from gn3.db.datasets import (retrieve_metadata,
-                             save_dataset_metadata)
+                             save_metadata,
+                             get_history)
 from gn3.db.rdf import RDF_PREFIXES
 from gn3.db.rdf import (query_frame_and_compact,
                         query_and_compact,
@@ -263,6 +264,26 @@ CONSTRUCT {
     )
 
 
+@metadata.route("/datasets/<id_>/history")
+def view_history(id_):
+    history = get_history(
+        git_dir=Path(current_app.config.get("DATA_DIR"),
+                     "gn-docs"),
+        name=id_,
+    ).either(
+        lambda error: {
+            "error": "Unable to fetch history",
+            "error_description": error,
+        },
+        lambda history: {
+            "id": id_,
+            "history": history,
+        })
+    if history.get("error"):
+        raise Exception(history.get("error_description"))
+    return history
+
+
 @metadata.route("/datasets/search/<term>", methods=["GET"])
 def search_datasets(term):
     """Search datasets"""
diff --git a/gn3/db/datasets.py b/gn3/db/datasets.py
index fef01a5..f3b4f9f 100644
--- a/gn3/db/datasets.py
+++ b/gn3/db/datasets.py
@@ -408,3 +408,34 @@ def save_metadata(
             ]))
         .then(lambda _: monadic_run_cmd(f"git -C {git_dir} \
 push origin master --dry-run".split(" "))))
+
+
+def get_history(git_dir: str, name: str) -> Either:
+    """Get the git history for a given dataset.  This history is
+    returned as a HTML formatted text. Example of the formatted
+    output:
+
+    ```
+    <tr>
+      <td><i>3 Weeks Ago</i></td>
+      <td>
+          <a style='color:green;' href='..id=some-id' target='_blank'>
+             some-id
+          </a>
+      </td>
+      <td>commit header</td>
+      <td>Author Name</td>
+    </tr>
+    ```
+
+    """
+    pretty_format_str = "<tr><td><i>%cr</i></td>\
+<td>\
+<a style='color:green;' href='https://git.genenetwork.org/gn-docs/commit/general?id=%H' \
+target='_blank'>%h</a></td>\
+<td>%s</td><td>%an</td></tr>"
+    return monadic_run_cmd([
+        "git", "-C",
+        str(Path(git_dir) / "general/datasets/" / name),
+        "log", f"--pretty=format:{pretty_format_str}",
+    ])