Some fieldmapper.js adjustments

Author: ddooley

lib/Toolbar.js

@@ -558,7 +558,7 @@ class Toolbar {
       "schema": "https://example.com/folder",
       "version": "x.y.z",
       "in_language": "en",
-      "container": {(class table name): [{row 1 object}, {row 2 object}, ...], ...}  
+      "container": {(class table name): [{row 1 object}, {row 2 object}, ...], ...}
     }
   */
   async openJSONDataFile(file) {
@@ -621,27 +621,27 @@ class Toolbar {
       }
 
       /**
-       * The JSON array of row objects sparse, so only keys with values are
-       * included. Search through a table's row objects to accumulate a full
-       * list of their keys. The col_index provides a ROUGH idea of which 
-       * column to claim for a given key, though a sparse array may vary from
-       * row to row depending on required/optional fields.
+       * JSON rows are sparse objects (only keys with values are included).
+       * Collect all unique field names across all rows to build the header,
+       * then convert each row-object to a value-array aligned to that header.
+       * This makes JSON data compatible with the same tabular loading pipeline.
        */
       const header = [];
-      Object.entries(data).forEach(([row_index, row]) => {
-        Object.entries(row).forEach(([key, col_index]) => {
-          if (!(header.includes(key))) {
-            header.splice(col_index, 0, key);
-          }
+      data.forEach((row) => {
+        Object.keys(row).forEach((key) => {
+          if (!header.includes(key)) header.push(key);
         });
       });
 
+      // Convert each sparse row-object to a dense array of values.
+      const data_matrix = data.map((row) => header.map((field) => row[field] ?? null));
+
       // JSON data SHOULD USE slot_names BUT some test files have slot_titles.
-      const matches = header.filter((element, index) => dh.slot_names.includes(element) || dh.slot_titles.includes(element)).length;
-      // raw_data only contains templates that match incomming data tables.  
+      const matches = header.filter((element) => dh.slot_names.includes(element) || dh.slot_titles.includes(element)).length;
+      // raw_data only contains templates that match incoming data tables.
       raw_data[template_name] = {
         header: header,
-        data: data,
+        data: data_matrix,
         matches: matches,
         table_name: dh.container_name
       }
@@ -752,12 +752,12 @@ class Toolbar {
      */
     Object.entries(this.context.dhs).forEach(([template_name, dh]) => {
       let data = raw_data[template_name];
+      if (!data) return;
       // 29, 34 total, 173 fields total.
       //console.log("MATCHES",data.matches,data.header.length,dh.slots.length);
-      if (data && (data.matches != data.header.length) && (data.matches != dh.slots.length)) { 
-        // Have user try to correct any missing mappings in a class/template. 
-        fieldMapper.appendFieldMappingModal(dh);
-      } 
+      const needs_modal = (data.matches != data.header.length) && (data.matches != dh.slots.length);
+      // Always compute slot_to_data_col_matches; only build modal HTML when needed.
+      fieldMapper.appendFieldMappingModal(dh, needs_modal);
     });
 
     if (fieldMapper.field_mapping_html) {
@@ -891,7 +891,7 @@ class Toolbar {
       const schema = this.context.template.default.schema;
       const container_table_list = Object.keys(schema.classes?.Container?.attributes || {}).join(';');
       const xlsx_table = [
-        ["id", "name", "version", "in_language", "container"], 
+        ["id", "name", "version", "in_language", "container"],
         [schema.id, schema.name, schema.version, schema.in_language, container_table_list]
       ];
       const ws = XlsxUtils.aoa_to_sheet(xlsx_table);