Update app.py

Change `get_query_values` output for better readability for LLMs and therefore changing `update_topic_params_and_link` due to different output + correcting descriptions

app.py

@@ -95,14 +95,15 @@ def get_query_values(topic_params: list[str] = []) -> List[Dict]:
             - report ID.
             These codes you can get from the function get_titles.
     Returns:
-        list: The list of the dictionaries, where dictionary's 'code' value you should use as an argument name and 'values' value(s) as an argument values.
+        Dict: A dictionary where each key is a query parameter code and the value is another dictionary mapping possible values to their descriptive texts.
     Examples:
         >>> # First get report topic parameters from get_titles
         ... print(get_titles('DS'))
         {'Dzīvi un nedzīvi dzimušo skaits pēc dzimuma 1920 - 2020': ['POP', 'ID', 'IDS', 'IDS010'],...
         >>> # Then use these values to get possible query values
         ... print(get_query_values(['POP', 'ID', 'IDS', 'IDS010']))
-        [{'code': 'SEX_NEWBORN', 'values': ['T', 'M', 'F'], 'valueTexts': ['Pavisam', 'Vīrieši', 'Sievietes']}, {'code': 'TIME', 'values': ['1920', '1921', '1922', '1923',...
+        {'SEX_NEWBORN': {'T': 'Pavisam', 'M': 'Vīrieši', 'F': 'Sievietes'}, 'ContentsCode': {'IDS010': 'Dzīvi dzimuši', 'IDS0101': 'Nedzīvi dzimuši', 'IDS0102': 'Nedzīvi dzimuši uz 1000 dzīvi dzimušiem'}, 
+        'TIME': {'1920': '1920' ... '2024': '2024'}}
     """
     base_url = 'https://data.stat.gov.lv/api/v1/lv/OSP_PUB/START/'
     url = base_url + '/'.join(topic_params)
@@ -112,10 +113,18 @@ def get_query_values(topic_params: list[str] = []) -> List[Dict]:
         data = response.json()
         if 'variables' not in data:
             raise ValueError("Unexpected JSON structure: 'variables' key missing")
-        return [{'code': i.get('code', ''),
-                 'values': i.get('values', ''),
-                 'valueTexts': i.get('valueTexts', '')}
-                for i in data['variables']]
+        
+        result = {}
+
+        for var in data['variables']:
+            code = var.get('code', '')
+            values = var.get('values', [])
+            value_texts = var.get('valueTexts', [])
+            # build dict mapping value -> valueText
+            mapping = dict(zip(values, value_texts))
+            result[code] = mapping
+        
+        return result
     except requests.exceptions.RequestException as e:
         raise RuntimeError(f"Request failed: {e}")
     except ValueError as ve:
@@ -128,7 +137,7 @@ def construct_csp_link(params: list[str]) -> str:
     return f"{base_url}{mid_path}/{last_part}/"
 
 def get_csp_data(lang: str = 'en', topic_params: list[str] = [], **kwargs) -> List[Dict]:
-    """Get statistics from Official Statistics Portal of Latvia (CSP or Centrālā statistikas pārvalde). Always cite 'Source URL' in the output.
+    """Get statistics from Official Statistics Portal of Latvia (CSP or Centrālā statistikas pārvalde). Use 'Source URL' from the Returns to cite the data source.
     
     Args:
         lang (str): Language. Default value 'en'.
@@ -140,16 +149,19 @@ def get_csp_data(lang: str = 'en', topic_params: list[str] = [], **kwargs) -> Li
             - report ID.
             These codes you can get from the function get_titles.
         kwargs: Keyword arguments for query configuration.
-            Possible query argument names ('code') and their possible values ('values')
+            Possible query argument names and their possible values
             can be obtained using the function get_query_values.
-            'valueTexts' from get_query_values could help if 'values' is unclear.
     Returns:
         list: The list of the dictionaries, where dictionary's key 'key' contains query parameters and key 'values' contains values. First list element is data source (URL) from CSB.
     Examples:
         >>> topics = ['POP', 'IR', 'IRE', 'IRE010']
         >>> query_args = get_query_values(topics)
         >>> print(query_args)
-        [{'code': 'ETHNICITY', 'values': ['TOTAL', 'E_LAT',...], 'valueTexts': ['Pavisam', 'Latvieši',...]}, {'code': 'TIME', 'values': ['1935', '1959',..., '2025'],...}]
+        {'ETHNICITY': {'TOTAL': 'Pavisam', 'E_LAT': 'Latvieši', 'E_ABZ': 'Abāzi', 'E_ABK': 'Abhāzi',...
+        'E_SWE': 'Zviedri', 'OTH': 'Cita tautība', 'UNK_NSP': 'Nezināma, neizvēlēta'}, 'ContentsCode': {'IRE010': 'Skaits'}, 
+        'TIME': {'1935': '1935', ... '2025': '2025'}}
+        >>> # Then use these codes and values to get data for example Latvians for years 2024 and 2025.
+        >>> # Value text 'Latvieši' explains what 'E_LAT' means.
         >>> data = get_csp_data(
         ...     lang='en',
         ...     topic_params=topics,
@@ -272,11 +284,16 @@ with gr.Blocks() as demo:
 
         try:
             q = get_query_values(title_value)
+            timeval = q.get('TIME','')
             query = {}
-            for i in q:
-                if i.get('code', '') == 'TIME':
-                    query = {'TIME': i['values'][-3:]}
-                    break
+
+            if len(timeval) > 0:
+                query = {'TIME': list(timeval.keys())[-3:]}
+                
+            # for i in q:
+            #     if i.get('code', '') == 'TIME':
+            #         query = {'TIME': i['values'][-3:]}
+            #         break
             query_str = str(query)
         except Exception as e:
             query_str = '{}'
@@ -302,8 +319,7 @@ with gr.Blocks() as demo:
         topic_params_str (str): string representation of a list that is needed for data extraction. Arguments in the list should be in the following order: topic code, topic content code, topic sub-content code, report ID.\
             These codes you can get from the function get_titles.
         query_kwargs_str (str): string representation of a dictionary - keyword arguments for query configuration.\
-            Possible query argument names ('code') and their possible values ('values') can be obtained using the function get_query_values.\
-            'valueTexts' from get_query_values could help if 'values' is unclear.
+            Possible query argument names and their possible values can be obtained using the function get_query_values.
     Returns:
         list: The list of the dictionaries, where dictionary's key 'key' contains query parameters and key 'values' contains values.
     Examples:
@@ -321,7 +337,7 @@ with gr.Blocks() as demo:
         >>> # Get 'code' and 'values' for filtering data
         ... query_args = get_query_values(report)
         >>> print(query_args)
-        [{'code': 'ETHNICITY', 'values': ['TOTAL', 'E_LAT',...], 'valueTexts': ['Pavisam', 'Latvieši',...]}, {'code': 'TIME', 'values': ['1935', '1959',..., '2025'],...}]
+        {'ETHNICITY': {'TOTAL': 'Pavisam', 'E_LAT': 'Latvieši',...}, 'ContentsCode': {'IRE010': 'Skaits'}, 'TIME': {'1935': '1935', ..., '2025': '2025'}}
         >>> # Get final result
         ... data = run_get_csp_data(
         ...        lang='en',