chore: update analytics to May 2026 #4857

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: MillenniumFalconMechanic

analytics/analytics_package/analytics/static_site/charts.py

@@ -20,21 +20,21 @@ def make_event_charts(entity_label, entity_path, chart_start):
                 "title": "Export to Terra",
                 "series": [
                     {"label": f"Single {entity_label}", "event_key": "dataset_analyze_in_terra_requested", "event_name": "dataset_analyze_in_terra_requested"},
-                    {"label": "Cross-Dataset", "event_key": "index_analyze_in_terra_requested", "event_name": "index_analyze_in_terra_requested"},
+                    {"label": f"Cross-{entity_label}", "event_key": "index_analyze_in_terra_requested", "event_name": "index_analyze_in_terra_requested"},
                 ],
             },
             {
                 "title": "curl Command",
                 "series": [
                     {"label": f"Single {entity_label}", "event_key": "dataset_bulk_download_requested", "event_name": "bulk_download_requested", "page_path_regex": rf"^/{path_segment}/[0-9a-f-]+"},
-                    {"label": "Cross-Dataset", "event_key": "index_bulk_download_requested", "event_name": "bulk_download_requested", "page_path_regex": rf"^(?!/{path_segment}/[0-9a-f-]+)"},
+                    {"label": f"Cross-{entity_label}", "event_key": "index_bulk_download_requested", "event_name": "bulk_download_requested", "page_path_regex": rf"^(?!/{path_segment}/[0-9a-f-]+)"},
                 ],
             },
             {
                 "title": "File Manifest",
                 "series": [
                     {"label": f"Single {entity_label}", "event_key": "dataset_file_manifest_requested", "event_name": "dataset_file_manifest_requested"},
-                    {"label": "Cross-Dataset", "event_key": "index_file_manifest_requested", "event_name": "index_file_manifest_requested"},
+                    {"label": f"Cross-{entity_label}", "event_key": "index_file_manifest_requested", "event_name": "index_file_manifest_requested"},
                 ],
             },
         ],

analytics/analytics_package/analytics/static_site/export.py

@@ -148,11 +148,14 @@ def export_data(data, config, current_month, analytics_start, custom_events, out
     for event in custom_events:
         key = event_key(event)
         stats = data.get(f"event_{key}", {"current": 0, "prior": 0, "change": None})
-        events_output.append({
+        event_entry = {
             "event_name": key,
             "label": event["label"],
             **stats,
-        })
+        }
+        if event.get("detail_file_column"):
+            event_entry["detail_file_column"] = event["detail_file_column"]
+        events_output.append(event_entry)
 
     with open(os.path.join(output_dir, "custom_events.json"), "w") as f:
         json.dump(events_output, f, indent=2)
@@ -211,7 +214,6 @@ def export_data(data, config, current_month, analytics_start, custom_events, out
         "prior_month_end": dates.get("end_prior", ""),
         "analytics_start": analytics_start,
         "sessions": data.get("sessions", {}),
-        "engaged_sessions": data.get("engaged_sessions", {}),
         "engagement_rate": data.get("engagement_rate", {}),
     }
 

analytics/analytics_package/analytics/static_site/fetch.py

@@ -22,11 +22,6 @@
     "alias": "Engagement Rate",
 }
 
-METRIC_ENGAGED_SESSIONS = {
-    "id": "engagedSessions",
-    "alias": "Engaged Sessions",
-}
-
 # Regex matching page paths that are clearly not real pages (bot probes,
 # broken markdown links, asset requests, etc.).
 SUSPICIOUS_PAGE_PATH_RE = re.compile(
@@ -49,11 +44,13 @@ def event_key(event):
     return event.get("key", event["event_name"])
 
 
-def _count_events(event_name, params, page_path_regex=None):
-    """Fetch event count, optionally filtered by page path regex."""
+def _count_events(event_name, params, page_path_regex=None, click_url_regex=None):
+    """Fetch event count, optionally filtered by page path and/or click URL regex."""
     dimensions = [DIMENSION_EVENT_NAME]
     if page_path_regex:
         dimensions.append(DIMENSION_PAGE_PATH)
+    if click_url_regex:
+        dimensions.append(DIMENSION_CUSTOM_URL)
 
     df = get_data_df_from_fields(
         [METRIC_EVENT_COUNT],
@@ -66,27 +63,29 @@ def _count_events(event_name, params, page_path_regex=None):
         return 0
 
     if page_path_regex:
-        pattern = re.compile(page_path_regex)
-        mask = df[DIMENSION_PAGE_PATH["alias"]].str.match(pattern, na=False)
-        return int(df.loc[mask, METRIC_EVENT_COUNT["alias"]].sum())
+        df = df[df[DIMENSION_PAGE_PATH["alias"]].str.match(page_path_regex, na=False)]
+
+    if click_url_regex:
+        df = df[df[DIMENSION_CUSTOM_URL["alias"]].str.contains(click_url_regex, na=False)]
 
     return int(df[METRIC_EVENT_COUNT["alias"]].sum())
 
 
-def get_custom_event_change(event_name, params_current, params_prior, page_path_regex=None):
+def get_custom_event_change(event_name, params_current, params_prior, page_path_regex=None, click_url_regex=None):
     """Fetch a custom event count with month-over-month change.
 
     Args:
         event_name: GA4 event name (e.g., "chat_submitted").
         params_current: Analytics params for the current period.
         params_prior: Analytics params for the prior period.
         page_path_regex: Optional regex to filter by page path.
+        click_url_regex: Optional regex to filter by click URL.
 
     Returns:
         Dict with "current", "prior", and "change" keys.
     """
-    current_count = _count_events(event_name, params_current, page_path_regex)
-    prior_count = _count_events(event_name, params_prior, page_path_regex)
+    current_count = _count_events(event_name, params_current, page_path_regex, click_url_regex)
+    prior_count = _count_events(event_name, params_prior, page_path_regex, click_url_regex)
 
     change = None
     if prior_count > 0:
@@ -95,21 +94,26 @@ def get_custom_event_change(event_name, params_current, params_prior, page_path_
     return {"current": current_count, "prior": prior_count, "change": change}
 
 
-def get_event_detail_table(event_name, params, page_path_regex=None):
+def get_event_detail_table(event_name, params, page_path_regex=None, click_url_regex=None):
     """Fetch event details broken down by page path and entity name.
 
     Args:
         event_name: GA4 event name.
         params: Analytics params for the period.
         page_path_regex: Optional regex to filter by page path.
+        click_url_regex: Optional regex to filter by click URL.
 
     Returns:
         List of dicts with "page_path", "entity_name", and "count" keys,
-        sorted by count descending.
+        sorted by count descending. Includes "click_url" when click_url_regex is used.
     """
+    dimensions = [DIMENSION_EVENT_NAME, DIMENSION_PAGE_PATH, DIMENSION_ENTITY_NAME]
+    if click_url_regex:
+        dimensions.append(DIMENSION_CUSTOM_URL)
+
     df = get_data_df_from_fields(
         [METRIC_EVENT_COUNT],
-        [DIMENSION_EVENT_NAME, DIMENSION_PAGE_PATH, DIMENSION_ENTITY_NAME],
+        dimensions,
         dimension_filter=f"eventName=={event_name}",
         **params,
     )
@@ -118,14 +122,25 @@ def get_event_detail_table(event_name, params, page_path_regex=None):
         return []
 
     if page_path_regex:
-        pattern = re.compile(page_path_regex)
-        df = df[df[DIMENSION_PAGE_PATH["alias"]].str.match(pattern, na=False)]
+        df = df[df[DIMENSION_PAGE_PATH["alias"]].str.match(page_path_regex, na=False)]
+
+    if click_url_regex:
+        df = df[df[DIMENSION_CUSTOM_URL["alias"]].str.contains(click_url_regex, na=False)]
 
     if len(df) == 0:
         return []
 
-    result = df[[DIMENSION_PAGE_PATH["alias"], DIMENSION_ENTITY_NAME["alias"], METRIC_EVENT_COUNT["alias"]]].copy()
-    result.columns = ["page_path", "entity_name", "count"]
+    export_cols = [DIMENSION_PAGE_PATH["alias"], DIMENSION_ENTITY_NAME["alias"]]
+    output_names = ["page_path", "entity_name"]
+    if click_url_regex:
+        export_cols.append(DIMENSION_CUSTOM_URL["alias"])
+        output_names.append("click_url")
+    export_cols.append(METRIC_EVENT_COUNT["alias"])
+    output_names.append("count")
+
+    result = df[export_cols].copy()
+    result.columns = output_names
+    result["count"] = result["count"].astype(int)
     result = result.sort_values("count", ascending=False)
     return result.to_dict(orient="records")
 
@@ -396,15 +411,13 @@ def fetch_data(
 
     print("Fetching sessions and engagement data...")
     df_sessions_current = get_data_df_from_fields(
-        [METRIC_SESSIONS, METRIC_ENGAGED_SESSIONS, METRIC_ENGAGEMENT_RATE], [], **params,
+        [METRIC_SESSIONS, METRIC_ENGAGEMENT_RATE], [], **params,
     )
     df_sessions_prior = get_data_df_from_fields(
-        [METRIC_SESSIONS, METRIC_ENGAGED_SESSIONS, METRIC_ENGAGEMENT_RATE], [], **params_prior,
+        [METRIC_SESSIONS, METRIC_ENGAGEMENT_RATE], [], **params_prior,
     )
     sessions_current = int(df_sessions_current[METRIC_SESSIONS["alias"]].sum()) if len(df_sessions_current) > 0 else 0
     sessions_prior = int(df_sessions_prior[METRIC_SESSIONS["alias"]].sum()) if len(df_sessions_prior) > 0 else 0
-    engaged_sessions_current = int(df_sessions_current[METRIC_ENGAGED_SESSIONS["alias"]].sum()) if len(df_sessions_current) > 0 else 0
-    engaged_sessions_prior = int(df_sessions_prior[METRIC_ENGAGED_SESSIONS["alias"]].sum()) if len(df_sessions_prior) > 0 else 0
     engagement_current = float(df_sessions_current[METRIC_ENGAGEMENT_RATE["alias"]].mean()) if len(df_sessions_current) > 0 else 0
     engagement_prior = float(df_sessions_prior[METRIC_ENGAGEMENT_RATE["alias"]].mean()) if len(df_sessions_prior) > 0 else 0
 
@@ -440,10 +453,6 @@ def fetch_data(
             "current": sessions_current,
             "prior": sessions_prior,
         },
-        "engaged_sessions": {
-            "current": engaged_sessions_current,
-            "prior": engaged_sessions_prior,
-        },
         "engagement_rate": {
             "current": engagement_current,
             "prior": engagement_prior,
@@ -470,12 +479,14 @@ def fetch_data(
         data[f"event_{key}"] = get_custom_event_change(
             event["event_name"], params, params_prior,
             page_path_regex=event.get("page_path_regex"),
+            click_url_regex=event.get("click_url_regex"),
         )
         if event.get("detail_table"):
             print(f"Fetching {event['label']} detail table...")
             data[f"event_{key}_detail"] = get_event_detail_table(
                 event["event_name"], params,
                 page_path_regex=event.get("page_path_regex"),
+                click_url_regex=event.get("click_url_regex"),
             )
 
     if event_charts:

analytics/analytics_package/analytics/static_site/template/index.html

@@ -489,11 +489,10 @@ <h3 class="fui-card-header-title fui-heading-xsmall">Filter Selections</h3>
             const previous = traffic[1] || {};
 
             const sessions = meta.sessions || {};
-            const engagedSessions = meta.engaged_sessions || {};
             const engagement = meta.engagement_rate || {};
 
             const usersChange = pctChange(latest.users, previous.users);
-            const engagedSessionsChange = pctChange(engagedSessions.current, engagedSessions.prior);
+            const sessionsChange = pctChange(sessions.current, sessions.prior);
             const pageviewsChange = pctChange(latest.pageviews, previous.pageviews);
             const engagementChange = pctChange(engagement.current, engagement.prior);
 
@@ -511,9 +510,9 @@ <h3 class="fui-card-header-title fui-heading-xsmall">Filter Selections</h3>
                     `Total number of unique individuals who visited your site during ${monthName}.`
                 ) +
                 statCard(
-                    formatNumber(engagedSessions.current != null ? engagedSessions.current : (sessions.current || 0)), 'Engaged Sessions',
-                    engagedSessionsChange,
-                    `Sessions where users actively engaged with your site during ${monthName} (stayed 10+ seconds, viewed 2+ pages, or triggered a conversion). <a href="https://support.google.com/analytics/answer/11109416" target="_blank" rel="noopener noreferrer">Learn more</a>.`
+                    formatNumber(sessions.current || 0), 'User Sessions',
+                    sessionsChange,
+                    `Total number of visits to your site during ${monthName}. A single user can have multiple sessions.`
                 ) +
                 statCard(
                     formatNumber(latest.pageviews), 'Pageviews',
@@ -580,7 +579,8 @@ <h3 class="fui-card-header-title fui-heading-xsmall">Filter Selections</h3>
                     if (chartDataByKey[card.event_key] && chartDataByKey[card.event_key].length > 0) {
                         const canvasId = `event-trend-chart-${chartIndex++}`;
                         const parts = card.label.split('\n');
-                        const chartTitle = escapeHtml(rowLabel) + ' Over Time' + (parts[1] ? ` (${escapeHtml(parts[1].replace(/[()]/g, ''))})` : '');
+                        const cardLabel = parts[0];
+                        const chartTitle = escapeHtml(cardLabel) + ' Over Time' + (parts[1] ? ` (${escapeHtml(parts[1].replace(/[()]/g, ''))})` : '');
                         html += `
                             <div class="fui-card fui-grid-item-6">
                                 <div class="fui-card-header">
@@ -662,6 +662,7 @@ <h3 class="fui-card-header-title fui-heading-xsmall">${chartTitle}</h3>
                 if (!detail) continue;
 
                 const hasDatasetTitle = detail.some(r => r.dataset_title);
+                const hasClickUrl = detail.some(r => r.click_url);
                 const isEmpty = detail.length === 0;
 
                 let tableContent;
@@ -672,9 +673,12 @@ <h3 class="fui-card-header-title fui-heading-xsmall">${chartTitle}</h3>
                         const label = r.dataset_title || r.entity_name || r.page_path || '-';
                         const link = r.page_path ? `${siteBase}${r.page_path.replace(/\/export.*/, '')}` : null;
                         const cell = link ? `<a href="${safeHref(link)}" target="_blank" rel="noopener noreferrer">${escapeHtml(label)}</a>` : escapeHtml(label);
-                        return `<tr><td>${cell}</td><td class="col-number">${formatNumber(r.count)}</td></tr>`;
+                        const fileTd = hasClickUrl ? `<td class="truncate">${escapeHtml(r.file_label || fileNameFromUrl(r.click_url))}</td>` : '';
+                        return `<tr><td>${cell}</td>${fileTd}<td class="col-number">${formatNumber(r.count)}</td></tr>`;
                     }).join('');
-                    tableContent = `<table class="detail-table collapsible-table"><thead><tr><th>${escapeHtml(entityLabel)}</th><th class="col-number">Count</th></tr></thead><tbody>${rows}</tbody></table>`;
+                    const fileColLabel = event.detail_file_column || 'File';
+                    const fileHeader = hasClickUrl ? `<th>${escapeHtml(fileColLabel)}</th>` : '';
+                    tableContent = `<table class="detail-table collapsible-table"><thead><tr><th>${escapeHtml(entityLabel)}</th>${fileHeader}<th class="col-number">Count</th></tr></thead><tbody>${rows}</tbody></table>`;
                 } else {
                     const rows = detail.map(r => {
                         const raw = r.entity_name && r.entity_name !== '(not set)' ? r.entity_name : null;
@@ -945,6 +949,12 @@ <h3 class="fui-card-header-title fui-heading-xsmall">${escapeHtml(event.label)}
             }
         }
 
+        function fileNameFromUrl(url) {
+            if (!url) return '-';
+            const parts = url.split('/');
+            return parts[parts.length - 1] || url;
+        }
+
         function formatChange(change) {
             if (change == null) return '-';
             return (change >= 0 ? '+' : '') + (change * 100).toFixed(1) + '%';

analytics/anvil-explorer-sheets/constants.py

@@ -1,8 +1,8 @@
 # CHANGE THESE VALUES TO GENERATE NEW REPORTS
 # The date of the current month to report on (yyyy-mm)
-CURRENT_MONTH = "2026-04"
+CURRENT_MONTH = "2026-05"
 # The name of the folder in which to save the report
-PARENT_FOLDER_NAME = "April 2026"
+PARENT_FOLDER_NAME = "May 2026"
 
 # The name of the spreadsheet with the report
 SHEET_NAME = "AnVIL Explorer"

analytics/hca-explorer-sheets/constants.py

@@ -1,8 +1,8 @@
 # CHANGE THESE VALUES TO GENERATE NEW REPORTS
 # The date of the current month to report on (yyyy-mm)
-CURRENT_MONTH = "2026-04"
+CURRENT_MONTH = "2026-05"
 # The name of the folder in which to save the report
-PARENT_FOLDER_NAME = "April 2026"
+PARENT_FOLDER_NAME = "May 2026"
 
 # The name of the spreadsheet with the report
 SHEET_NAME = "HCA Explorer"

analytics/hca-explorer-sheets/generate_static_site.py

@@ -48,11 +48,11 @@ def resolve_project_titles(data):
         "file_downloads_position": 3,
         "event_counts": [
             {"label": "Export to Terra\n(Single Project)", "event_key": "dataset_analyze_in_terra_requested"},
-            {"label": "Export to Terra\n(Cross-Dataset)", "event_key": "index_analyze_in_terra_requested"},
+            {"label": "Export to Terra\n(Cross-Project)", "event_key": "index_analyze_in_terra_requested"},
             {"label": "curl Command\n(Single Project)", "event_key": "dataset_bulk_download_requested"},
-            {"label": "curl Command\n(Cross-Dataset)", "event_key": "index_bulk_download_requested"},
+            {"label": "curl Command\n(Cross-Project)", "event_key": "index_bulk_download_requested"},
             {"label": "File Manifest\n(Single Project)", "event_key": "dataset_file_manifest_requested"},
-            {"label": "File Manifest\n(Cross-Dataset)", "event_key": "index_file_manifest_requested"},
+            {"label": "File Manifest\n(Cross-Project)", "event_key": "index_file_manifest_requested"},
         ],
     },
     property_id=HCA_ID,

analytics/lungmap-analytics/sheets/constants.py

@@ -1,8 +1,8 @@
 # CHANGE THESE VALUES TO GENERATE NEW REPORTS
 # The date of the current month to report on (yyyy-mm)
-CURRENT_MONTH = "2026-04"
+CURRENT_MONTH = "2026-05"
 # The name of the folder in which to save the report
-PARENT_FOLDER_NAME = "April 2026"
+PARENT_FOLDER_NAME = "May 2026"
 
 # The name of the spreadsheet with the report
 SHEET_NAME = "Lungmap Data Browser"

analytics/lungmap-analytics/sheets/generate_static_site.py

@@ -48,11 +48,11 @@ def resolve_project_titles(data):
         "file_downloads_position": 3,
         "event_counts": [
             {"label": "Export to Terra\n(Single Project)", "event_key": "dataset_analyze_in_terra_requested"},
-            {"label": "Export to Terra\n(Cross-Dataset)", "event_key": "index_analyze_in_terra_requested"},
+            {"label": "Export to Terra\n(Cross-Project)", "event_key": "index_analyze_in_terra_requested"},
             {"label": "curl Command\n(Single Project)", "event_key": "dataset_bulk_download_requested"},
-            {"label": "curl Command\n(Cross-Dataset)", "event_key": "index_bulk_download_requested"},
+            {"label": "curl Command\n(Cross-Project)", "event_key": "index_bulk_download_requested"},
             {"label": "File Manifest\n(Single Project)", "event_key": "dataset_file_manifest_requested"},
-            {"label": "File Manifest\n(Cross-Dataset)", "event_key": "index_file_manifest_requested"},
+            {"label": "File Manifest\n(Cross-Project)", "event_key": "index_file_manifest_requested"},
         ],
     },
     property_id=LUNGMAP_ID,