From d9cb9a56f93d3fa6c1a90d42a7a6f88a3d383fc6 Mon Sep 17 00:00:00 2001
From: Dale Wahl <dalewahl@gmail.com>
Date: Mon, 3 Mar 2025 15:00:12 +0100
Subject: [PATCH] collect example dataset to map keys

---
 common/lib/helpers.py                | 11 +++++++++++
 webtool/templates/data-overview.html | 12 ++++++++++++
 webtool/views/views_misc.py          |  8 +++++++-
 3 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/common/lib/helpers.py b/common/lib/helpers.py
index a4d1bc312..7c63766ba 100644
--- a/common/lib/helpers.py
+++ b/common/lib/helpers.py
@@ -46,6 +46,17 @@ def init_datasource(database, logger, queue, name):
     """
     pass
 
+def get_datasource_example_keys(db, modules, dataset_type):
+    """
+    Get example keys for a datasource
+    """
+    from common.lib.dataset import DataSet
+    example_dataset_key = db.fetchone("SELECT key from datasets WHERE type = %s and is_finished = True and num_rows > 0 LIMIT 1", (dataset_type,))
+    if example_dataset_key:
+        example_dataset = DataSet(db=db, key=example_dataset_key["key"], modules=modules)
+        return example_dataset.get_columns()
+    return []
+
 def strip_tags(html, convert_newlines=True):
     """
     Strip HTML from a string
diff --git a/webtool/templates/data-overview.html b/webtool/templates/data-overview.html
index f31f2178d..8b273d6ab 100644
--- a/webtool/templates/data-overview.html
+++ b/webtool/templates/data-overview.html
@@ -95,6 +95,18 @@ <h3>Metadata</h3>
                         </dd>
                     </div>
                     {% endif %}
+                    {% if example_keys %}
+                    <div class="fullwidth">
+                        <dt>DataSet keys</dt>
+                        <dd>
+                            <ul class="nobullet">
+                            {% for key in example_keys %}
+                                <span class="inline-label property-badge">{{ key }}</span>
+                            {% endfor %}
+                            </ul>
+                        </dd>
+                    </div>
+                    {% endif %}
                 </dl>
             </div>
             {% else %}
diff --git a/webtool/views/views_misc.py b/webtool/views/views_misc.py
index e179085c2..fc9251f48 100644
--- a/webtool/views/views_misc.py
+++ b/webtool/views/views_misc.py
@@ -17,6 +17,7 @@
 from webtool.views.views_dataset import create_dataset, show_results
 
 from common.config_manager import ConfigWrapper
+from common.lib.helpers import get_datasource_example_keys
 config = ConfigWrapper(config, user=current_user, request=request)
 
 csv.field_size_limit(1024 * 1024 * 1024)
@@ -129,6 +130,7 @@ def data_overview(datasource=None):
     daily_counts = None
     references = None
     labels = None
+    example_keys = None
 
     if datasource:
 
@@ -156,6 +158,10 @@ def data_overview(datasource=None):
         if hasattr(worker_class, "is_from_zeeschuimer"):
             labels.append("zeeschuimer")
 
+        # Get example keys for the datasource
+        if datasource_id not in ["upload"]: # ignore upload as keys are variable
+            example_keys = get_datasource_example_keys(db=db, modules=fourcat_modules, dataset_type=datasource_id + "-search")
+
         # Get daily post counts for local datasource to display in a graph
         if is_local == "local":
 
@@ -185,7 +191,7 @@ def data_overview(datasource=None):
 
         references = worker_class.references if hasattr(worker_class, "references") else None        
 
-    return render_template('data-overview.html', datasources=datasources, datasource_id=datasource_id, description=description, labels=labels, total_counts=total_counts, daily_counts=daily_counts, github_url=github_url, references=references)
+    return render_template('data-overview.html', datasources=datasources, datasource_id=datasource_id, description=description, labels=labels, total_counts=total_counts, daily_counts=daily_counts, github_url=github_url, references=references, example_keys=example_keys)
 
 @app.route('/get-boards/<string:datasource>/')
 @login_required