Create a DataTable for displaying transform logs (#75)

Why these changes are being introduced:
* These changes are required in order to display the
individual log files retrieved from each container.

How this addresses that need:
* Add a new route that renders a transform log in plain text
* Update run route to retrieve tranform log filenames from new
'logs' directory
* Update run.html

Side effects of this change:
* None

Relevant ticket(s):
* https://mitlibraries.atlassian.net/browse/TIMX-368

Author: jonavellecuerdo

abdiff/webapp/app.py

@@ -82,15 +82,15 @@ def run(run_timestamp: str) -> str:
         run_directory = get_run_directory(run_timestamp)
         run_data = read_run_json(run_directory)
 
-        # load transform logs
-        try:
-            with open(Path(run_directory) / "transformed/logs.txt") as f:
-                transform_logs = f.read()
-        except FileNotFoundError:
-            transform_logs = "'logs.txt' not found for transform logs"
+        # get filenames of transform logs
+        transform_logs = [
+            os.path.basename(log_filepath)
+            for log_filepath in glob.glob(f"{run_directory}/logs/*")
+        ]
 
         return render_template(
             "run.html",
+            run_timestamp=run_timestamp,
             run_data=run_data,
             run_json=json.dumps(run_data),
             transform_logs=transform_logs,
@@ -99,6 +99,17 @@ def run(run_timestamp: str) -> str:
             modified_fields=sorted(run_data["metrics"]["summary"]["fields_with_diffs"]),
         )
 
+    @app.route("/run/<run_timestamp>/log/<log_filename>")
+    def transform_log(run_timestamp: str, log_filename: str) -> Response:
+        run_directory = get_run_directory(run_timestamp)
+
+        try:
+            with open(Path(run_directory) / "logs" / log_filename) as f:
+                transform_logs = f.read()
+        except FileNotFoundError:
+            transform_logs = f"Log file '{log_filename}' not found."
+        return Response(transform_logs, mimetype="text/plain")
+
     @app.route("/run/<run_timestamp>/records/data", methods=["POST"])
     def records_data(run_timestamp: str) -> Response:
         """Endpoint to provide data for Records table in Run view.

abdiff/webapp/templates/run.html

@@ -21,10 +21,27 @@ <h3>Details</h3>
     </div>
 
     <div class="row">
-        <h3 data-bs-toggle="collapse" data-bs-target="#collapseLogs" role="button" aria-expanded="false" aria-controls="collapseLogs">Transform Logs ⤵</h3>
-        <div class="collapse" id="collapseLogs">
-            <pre><code>{{transform_logs}}</code></pre>
-        </div>
+        <h3>Transform Logs</h3>
+        {% if transform_logs %}
+            <table id="transformLogsTable" class="table">
+                <thead>
+                    <tr>
+                        <th>Transform Log Filename</th>
+                        <th>Actions</th>
+                    </tr>
+                </thead>
+                <tbody>
+                    {% for log_filename in transform_logs %}
+                    <tr>
+                        <td>{{log_filename}}</td>
+                        <td><a href="/run/{{run_timestamp}}/log/{{log_filename}}" target="blank">View Log</a></td>
+                    </tr>
+                    {% endfor %}
+                </tbody>
+            </table>
+        {% else %}
+            <span>No transform logs were found.</span>
+        {% endif %}
     </div>
 
     <div class="row">
@@ -90,6 +107,9 @@ <h3>Records</h3>
     </script>
     <script src="{{ url_for('static', filename='jquery-3.6.0.min.js') }}"></script>
     <script src="{{ url_for('static', filename='jquery.dataTables.min.js') }}"></script>
+    <script>
+        new DataTable('#transformLogsTable');
+    </script>
     <script>
     $(document).ready(function() {
         var table = $('#recordsTable').DataTable( {

tests/fixtures/jobs/example-job-1/runs/2024-01-01_12-00-00/logs/alma-2024-08-29-daily-aaaaaa123456-logs.txt

@@ -0,0 +1 @@
+This text file represents a transform log file.

tests/fixtures/jobs/example-job-1/runs/2024-01-01_12-00-00/logs/alma-2024-08-29-daily-bbbbbb123456-logs.txt

@@ -0,0 +1 @@
+This text file represents a transform log file.

tests/fixtures/jobs/example-job-1/runs/2024-01-01_12-00-00/logs/dspace-2024-10-14-daily-cccccc123456-logs.txt

@@ -0,0 +1 @@
+This text file represents a transform log file.

tests/fixtures/jobs/example-job-1/runs/2024-01-01_12-00-00/logs/dspace-2024-10-14-daily-dddddd123456-logs.txt

@@ -0,0 +1 @@
+This text file represents a transform log file.

tests/fixtures/jobs/example-job-1/runs/2024-01-01_12-00-00/transformed/logs.txt

@@ -56,16 +56,12 @@ def test_run_route_load_transformation_logs(
         _response = webapp_client.get(f"/run/{webapp_run_timestamp}")
     args, kwargs = mock_render.call_args
 
-    assert (
-        "docker_image: transmogrifier-best-job-008e20c:latest | source: alma | "
-        "input_file: s3://timdex-extract-prod-300442551476/alma/alma-2024-08-29-daily"
-        "-extracted-records-to-index.xml" in kwargs["transform_logs"]
-    )
-    assert (
-        "docker_image: transmogrifier-best-job-395e612:latest | source: alma | "
-        "input_file: s3://timdex-extract-prod-300442551476/alma/alma-2024-08-29-daily"
-        "-extracted-records-to-index.xml" in kwargs["transform_logs"]
-    )
+    assert set(kwargs["transform_logs"]) == {
+        "alma-2024-08-29-daily-aaaaaa123456-logs.txt",
+        "alma-2024-08-29-daily-bbbbbb123456-logs.txt",
+        "dspace-2024-10-14-daily-cccccc123456-logs.txt",
+        "dspace-2024-10-14-daily-dddddd123456-logs.txt",
+    }
 
 
 def test_shutdown_route_success(caplog, webapp_client):