From d7530b4cdfe248a768dc5c81ddea10d38d2007f7 Mon Sep 17 00:00:00 2001
From: dmitry-weirdo <dmitry.weirdo@gmail.com>
Date: Wed, 27 Nov 2024 01:35:57 +0100
Subject: [PATCH] feat: #98 Validate that there are no dictionary code
 duplicates in the json file.

---
 .../common/NonStandardDictionary.kt           |  2 ++
 .../NonStandardDictionariesCache.kt           | 29 ++++++++++++++++-
 .../non-standard-dictionaries.json            | 32 +++++++++++++++++++
 3 files changed, 62 insertions(+), 1 deletion(-)

diff --git a/kgCommon/src/main/kotlin/ru/klavogonki/common/NonStandardDictionary.kt b/kgCommon/src/main/kotlin/ru/klavogonki/common/NonStandardDictionary.kt
index e837300a..1bb3b874 100644
--- a/kgCommon/src/main/kotlin/ru/klavogonki/common/NonStandardDictionary.kt
+++ b/kgCommon/src/main/kotlin/ru/klavogonki/common/NonStandardDictionary.kt
@@ -21,6 +21,8 @@ enum class NonStandardDictionary(
     MINI_MARATHON(6018, "Мини-марафон, 800 знаков", "Мини-марафоне, 800 знаков"),
     CYBERTEXT(6562, "Кибертекст", "Кибертексте"),
     CYBERTEXT_2(115606, "Кибертекст 2.0", "Кибертексте 2.0"),
+
+    // Пальцы
     PINKIES_PLUS(3714, "Мизинцы+", "Мизинцах+"),
     TRAINING_INDEX_FINGERS(226, "Тренируем указательные", "Тренируем указательные"),
     RING_FINGERS(8223, "Безымянные", "Безымянных"),
diff --git a/kgstatsSrv/src/main/kotlin/ru/klavogonki/statistics/dictionaries/NonStandardDictionariesCache.kt b/kgstatsSrv/src/main/kotlin/ru/klavogonki/statistics/dictionaries/NonStandardDictionariesCache.kt
index 0bc2ebee..2c5ecf25 100644
--- a/kgstatsSrv/src/main/kotlin/ru/klavogonki/statistics/dictionaries/NonStandardDictionariesCache.kt
+++ b/kgstatsSrv/src/main/kotlin/ru/klavogonki/statistics/dictionaries/NonStandardDictionariesCache.kt
@@ -20,13 +20,40 @@ object NonStandardDictionariesCache : Logging {
 
         val nonStandardDictionaries = JacksonUtils.parseNonStandardDictionaryData(fileName)
 
+        logger.info(
+            "Total non-standard dictionaries read from json file \"$NON_STANDARD_DICTIONARIES_JSON_RESOURCE\": " +
+                "${nonStandardDictionaries.size}."
+        )
+
+        // validate no duplicate ids
+        validate(nonStandardDictionaries)
+
         return nonStandardDictionaries
     }
 
+    private fun validate(dictionaries: List<NonStandardDictionaryData>) {
+        val repeatCodeToCount: Map<Int, Int> = dictionaries
+            .groupingBy { it.code }
+            .eachCount()
+            .filter { it.value > 1 }
+
+        if (repeatCodeToCount.isNotEmpty()) {
+            repeatCodeToCount.forEach {
+                logger.error("Duplicate dictionary code ${it.key}: used ${it.value} times.")
+            }
+        }
+
+        check(repeatCodeToCount.isEmpty()) {
+            "Duplicate dictionary codes: $repeatCodeToCount"
+        }
+
+        logger.info("No dictionary code duplicates.")
+    }
+
     @JvmStatic
     fun main(args: Array<String>) {
         val nonStandardDictionaries = parse()
 
-        logger.info("Total non-standard dictionaries read from json file: ${nonStandardDictionaries.size}.")
+        logger.info("[main]: Total non-standard dictionaries read from json file: ${nonStandardDictionaries.size}.")
     }
 }
diff --git a/kgstatsSrv/src/main/resources/dictionaries/non-standard-dictionaries.json b/kgstatsSrv/src/main/resources/dictionaries/non-standard-dictionaries.json
index 2ac3c6a7..8df6a6e8 100644
--- a/kgstatsSrv/src/main/resources/dictionaries/non-standard-dictionaries.json
+++ b/kgstatsSrv/src/main/resources/dictionaries/non-standard-dictionaries.json
@@ -13,6 +13,28 @@
     }
   },
 
+  {
+    "code": 192,
+    "displayName": "Частотный словарь",
+    "displayNamePrepositional": "Частотном словаре",
+    "tags": [
+      "FREQUENT"
+    ],
+    "top": {
+      "loggerName": "FrequencyVocabulary",
+      "minRacesCount": 200,
+
+      //      "topByBestSpeedExcelSheetName": "Топ по рекорду в «Частотном словаре»", // 36 chars
+      "topByBestSpeedExcelSheetName": "Топ по рекорду в «Частотном»", // 28 chars
+
+      //      "topByRacesCountExcelSheetName": "Топ по пробегу в «Частотном словаре»", // 36 chars
+      "topByRacesCountExcelSheetName": "Топ по пробегу в «Частотном»", // 28 chars
+
+      // "topByHaulExcelSheetName": "Топ по времени в «Частотном словаре»", // 36 chars
+      "topByHaulExcelSheetName": "Топ по времени в «Частотном»" // 28 chars
+    }
+  },
+
   {
     "code": 5539,
     "displayName": "Обычный in English",
@@ -65,5 +87,15 @@
       "CYBERTEXT",
       "FREQUENT"
     ]
+  },
+
+  { // without generated top
+    "code": 115606,
+    "displayName": "Кибертекст 2.0",
+    "displayNamePrepositional": "Кибертексте 2.0",
+    "tags": [
+      "CYBERTEXT",
+      "FREQUENT"
+    ]
   }
 ]
\ No newline at end of file