start of the actual mapping code

compendium_v2/conversion/conversion.py

@@ -9,6 +9,7 @@ it can be used to prefill the 2023 survey.
 """
 import logging
 import click
+import json
 
 from sqlalchemy import delete, text, select
 
@@ -20,6 +21,8 @@ from compendium_v2.survey_db import model as survey_model
 from compendium_v2.db.model import NREN
 from compendium_v2.db.survey_model import Survey, SurveyResponse
 
+from compendium_v2.conversion.mapping import id_to_name
+
 setup_logging()
 
 logger = logging.getLogger('conversion')
@@ -111,6 +114,49 @@ def query_nren(nren_id: int):
     return answers
 
 
+def convert_answers(answers):
+    data = {}
+
+    for id, question_name in id_to_name.items():
+        if not id in answers:
+            continue
+
+        if '{' in question_name:
+            continue
+
+        answer = answers[id]
+        if len(answer) > 1 and answer[0] == '"' and answer[-1] == '"':
+            answer = answer[1:-1]
+        if len(answer) > 1 and answer[0] == '[' and answer[-1] == ']':
+            answer = json.loads(answer)
+
+        question_names = question_name.split(":")
+        subdict = data
+        for name in question_names[0:-1]:
+            if name[-1] == "]":
+                index = name[-2]
+                if index == "[":
+                    subdict = subdict.setdefault(name[:-2], [])
+                    break  # special case where json list is mapped to a list of dicts
+                sublist = subdict.setdefault(name[:-3], [])
+                index = int(index)
+                while len(sublist) <= index:
+                    sublist.append({})
+                subdict = sublist[index]
+            else:
+                subdict = subdict.setdefault(name, {})
+
+        # print(question_names, answer)
+
+        if type(subdict) == list:  # special case where json list is mapped to a list of dicts
+            for answer_entry in answer:
+                subdict.append({question_names[-1]: answer_entry})
+        else:
+            subdict[question_names[-1]] = answer
+
+
+    return {"data": data}
+
 def _cli(app):
     with app.app_context():
 
@@ -132,12 +178,13 @@ def _cli(app):
         db.session.add(survey)
 
         for nren, answers in nren_surveys.items():
+            survey_dict = convert_answers(answers)
             response = SurveyResponse(
                 nren=nren,
                 nren_id=nren.id,
                 survey_year=2022,
                 survey=survey,
-                answers=answers  # TODO structure should be different
+                answers=survey_dict
             )
             db.session.add(response)