From 20a57d7ca160140ca12ecf80a09bf6a302e26a37 Mon Sep 17 00:00:00 2001
From: Taha Yassine Kraiem <tahayk2@gmail.com>
Date: Wed, 26 Mar 2025 13:27:06 +0100
Subject: [PATCH] feat(chalice): initial lexicon for events & properties

---
 .../core/product_analytics/events.py          | 23 ++++++++++++
 .../core/product_analytics/properties.py      | 35 +++++++++++++++++++
 api/routers/subs/product_analytics.py         | 12 +++++++
 3 files changed, 70 insertions(+)

diff --git a/api/chalicelib/core/product_analytics/events.py b/api/chalicelib/core/product_analytics/events.py
index 37a4c5ebb..9dc157203 100644
--- a/api/chalicelib/core/product_analytics/events.py
+++ b/api/chalicelib/core/product_analytics/events.py
@@ -111,3 +111,26 @@ def search_events(project_id: int, data: schemas.EventsSearchPayloadSchema):
         for r in rows:
             r.pop("total")
         return {"total": total, "rows": rows, "src": 2}
+
+
+def get_lexicon(project_id: int, page: schemas.PaginatedSchema):
+    with ClickHouseClient() as ch_client:
+        r = ch_client.format(
+            """SELECT COUNT(1) OVER () AS total,
+                            all_events.event_name AS name,
+                            *
+                      FROM product_analytics.all_events 
+                      WHERE project_id=%(project_id)s
+                      ORDER BY display_name
+                      LIMIT %(limit)s OFFSET %(offset)s;""",
+            parameters={"project_id": project_id, "limit": page.limit, "offset": (page.page - 1) * page.limit})
+        rows = ch_client.execute(r)
+    if len(rows) == 0:
+        return {"total": 0, "list": []}
+    total = rows[0]["total"]
+    for i, row in enumerate(rows):
+        row["id"] = f"event_{i}"
+        row["icon"] = None
+        row["possibleTypes"] = ["String"]
+        row.pop("total")
+    return {"total": total, "list": helper.list_to_camel_case(rows)}
diff --git a/api/chalicelib/core/product_analytics/properties.py b/api/chalicelib/core/product_analytics/properties.py
index a62f243f2..e88f51de1 100644
--- a/api/chalicelib/core/product_analytics/properties.py
+++ b/api/chalicelib/core/product_analytics/properties.py
@@ -44,3 +44,38 @@ def get_event_properties(project_id: int, event_name):
         properties = ch_client.execute(r)
 
         return helper.list_to_camel_case(properties)
+
+
+def get_lexicon(project_id: int, page: schemas.PaginatedSchema):
+    with ClickHouseClient() as ch_client:
+        r = ch_client.format(
+            """SELECT COUNT(1) OVER ()       AS total,
+                           all_properties.property_name AS name,
+                           all_properties.*,
+                           possible_types.values  AS possible_types,
+                           possible_values.values AS sample_values
+                    FROM product_analytics.all_properties
+                             LEFT JOIN (SELECT project_id, property_name, array_agg(DISTINCT value_type) AS values
+                                        FROM product_analytics.event_properties
+                                        WHERE project_id=%(project_id)s
+                                        GROUP BY 1, 2) AS possible_types
+                                       USING (project_id, property_name)
+                             LEFT JOIN (SELECT project_id, property_name, array_agg(DISTINCT value) AS values
+                                        FROM product_analytics.property_values_samples
+                                        WHERE project_id=%(project_id)s
+                                        GROUP BY 1, 2) AS possible_values USING (project_id, property_name)
+                    WHERE project_id=%(project_id)s
+                    ORDER BY display_name
+                    LIMIT %(limit)s OFFSET %(offset)s;""",
+            parameters={"project_id": project_id,
+                        "limit": page.limit,
+                        "offset": (page.page - 1) * page.limit})
+        properties = ch_client.execute(r)
+        if len(properties) == 0:
+            return {"total": 0, "list": []}
+        total = properties[0]["total"]
+        for i, p in enumerate(properties):
+            p["id"] = f"prop_{i}"
+            p["icon"] = None
+            p.pop("total")
+        return {"total": total, "list": helper.list_to_camel_case(properties)}
diff --git a/api/routers/subs/product_analytics.py b/api/routers/subs/product_analytics.py
index aaebf788b..5b18ca93e 100644
--- a/api/routers/subs/product_analytics.py
+++ b/api/routers/subs/product_analytics.py
@@ -41,3 +41,15 @@ def get_event_properties(projectId: int, event_name: str = None,
 def search_events(projectId: int, data: schemas.EventsSearchPayloadSchema = Body(...),
                   context: schemas.CurrentContext = Depends(OR_context)):
     return {"data": events.search_events(project_id=projectId, data=data)}
+
+
+@app.get('/{projectId}/lexicon/events', tags=["product_analytics", "lexicon"])
+def get_all_lexicon_events(projectId: int, filter_query: Annotated[schemas.PaginatedSchema, Query()],
+                           context: schemas.CurrentContext = Depends(OR_context)):
+    return {"data": events.get_lexicon(project_id=projectId, page=filter_query)}
+
+
+@app.get('/{projectId}/lexicon/properties', tags=["product_analytics", "lexicon"])
+def get_all_lexicon_properties(projectId: int, filter_query: Annotated[schemas.PaginatedSchema, Query()],
+                               context: schemas.CurrentContext = Depends(OR_context)):
+    return {"data": properties.get_lexicon(project_id=projectId, page=filter_query)}