ga4-export-fixer 0.4.6 → 0.4.7-dev.0

package/createTable.js

@@ -0,0 +1,58 @@
+const utils = require('./utils.js');
+const preOperations = require('./preOperations.js');
+
+/**
+ * Shared createTable lifecycle for all table modules.
+ *
+ * Merges configuration, validates it, builds the Dataform table config
+ * (name, schema, columns, description), and publishes the table with
+ * pre-operations and the module's SQL query.
+ *
+ * @param {Function} dataformPublish - Dataform publish() function.
+ * @param {Object} userConfig - User-provided configuration.
+ * @param {Object} tableModule - Table module definition conforming to the table module interface:
+ *   @param {Object}   tableModule.defaultConfig - Default config extending baseConfig.
+ *   @param {string}   tableModule.defaultTableName - Default table name (e.g. 'ga4_events_enhanced').
+ *   @param {Function} tableModule.validate - (mergedConfig, options?) => void.
+ *   @param {Function} tableModule.generateSql - (mergedConfig) => string.
+ *   @param {Function} tableModule.getColumnDescriptions - (mergedConfig) => Dataform columns object.
+ *   @param {Function} tableModule.getTableDescription - (mergedConfig) => string.
+ * @returns {Object} The Dataform publish() object for the table.
+ */
+const createTable = (dataformPublish, userConfig, tableModule) => {
+    const mergedConfig = utils.mergeSQLConfigurations(tableModule.defaultConfig, userConfig);
+    tableModule.validate(mergedConfig, { skipDataformContextFields: true });
+
+    const dataset = utils.getDatasetName(mergedConfig.sourceTable);
+
+    // Build dataformTableConfig: static defaults → dynamic fields → user overrides.
+    // Deep-clone defaults to prevent Dataform's publish() from mutating nested objects (e.g. bigquery)
+    // across multiple createTable calls in the same process.
+    const dataformTableConfig = utils.mergeDataformTableConfigurations(
+        {
+            ...JSON.parse(JSON.stringify(tableModule.defaultConfig.dataformTableConfig || {})),
+            name: `${tableModule.defaultTableName}_${dataset.replace('analytics_', '')}`,
+            schema: dataset,
+            columns: tableModule.getColumnDescriptions(mergedConfig),
+        },
+        userConfig.dataformTableConfig
+    );
+
+    // Pass dataformTableConfig to getTableDescription via a new object to avoid mutating mergedConfig
+    // (Dataform's sandboxed runtime may freeze objects returned by mergeSQLConfigurations)
+    const tableDescription = tableModule.getTableDescription({ ...mergedConfig, dataformTableConfig });
+
+    // Set description (user override from the merge wins if provided)
+    if (!dataformTableConfig.description) {
+        dataformTableConfig.description = tableDescription;
+    }
+
+    // Create the table using Dataform publish()
+    return dataformPublish(dataformTableConfig.name, dataformTableConfig).preOps(ctx => {
+        return preOperations.setPreOperations(utils.setDataformContext(ctx, mergedConfig));
+    }).query(ctx => {
+        return tableModule.generateSql(utils.setDataformContext(ctx, mergedConfig));
+    });
+};
+
+module.exports = { createTable };

package/defaultConfig.js

@@ -28,72 +28,6 @@ const baseConfig = {
     },
 };
 
-/*
-The default configuration for the GA4 Events Enhanced table.
-*/
-const ga4EventsEnhancedConfig = {
-    ...baseConfig,
-    sourceTable: undefined,
-    sourceTableType: 'GA4_EXPORT', // used with pre operations to detect if ga4 export specific pre operations are needed
-    // optional but recommended
-    schemaLock: undefined,
-    // only used with js tables
-    dataformTableConfig: {
-        type: 'incremental',
-        bigquery: {
-            partitionBy: 'event_date',
-            clusterBy: ['event_name', 'session_id', 'page_location', 'data_is_final'],
-            labels: {
-                'ga4_export_fixer': 'true'
-            }
-        },
-        onSchemaChange: 'EXTEND',
-        tags: ['ga4_export_fixer'],
-    },
-    // optional
-    includedExportTypes: {
-        daily: true,
-        fresh: false,
-        intraday: true,
-    },
-    timezone: 'Etc/UTC',
-    customTimestampParam: undefined,
-    dataIsFinal: {
-        detectionMethod: 'DAY_THRESHOLD', // 'EXPORT_TYPE' or 'DAY_THRESHOLD'
-        dayThreshold: 3 // only used if detectionMethod is 'DAY_THRESHOLD'
-        // according to GA4 documentation, the data up to 72 hours old is subject to possible changes
-        // in reality, there have been cases where the data has changed even after 72 hours (4 day window would have covered these)
-    },
-    // number of additional days to take in for taking into account sessions that overlap days
-    bufferDays: 1,
-    // these parameters are excluded by default because they've been made available in other columns
-    defaultExcludedEventParams: [
-        'page_location',
-        'ga_session_id',
-        //'custom_event_timestamp', // removed if customTimestampParam is used
-    ],
-    excludedEventParams: [],
-    eventParamsToColumns: [
-        //{name: 'page_location', type: 'string', columnName: 'page_location2'},
-    ],
-    sessionParams: [],
-    defaultExcludedEvents: [],
-    // session_start and first_visit are excluded via the excludedEvents array
-    // this allows the user to include them if needed
-    excludedEvents: [
-        'session_start',
-        'first_visit'
-    ],
-    defaultExcludedColumns: [
-        'event_dimensions', // legacy column, not needed
-        'traffic_source', // renamed to user_traffic_source
-        'session_id'
-    ],
-    // exclude these columns when extracting raw data from the export tables
-    excludedColumns: [],
-};
-
 module.exports = {
     baseConfig,
-    ga4EventsEnhancedConfig,
 };