From e735cd33f311d57c91690442e6062b9f98da25d8 Mon Sep 17 00:00:00 2001
From: ShikiSuen <shikisuen@outlook.com>
Date: Thu, 3 Mar 2022 21:21:11 +0800
Subject: [PATCH] DataCompiler // Regex Optimization.

---
 DataCompiler/dataCompiler.swift | 51 ++++++++++++++-------------------
 1 file changed, 21 insertions(+), 30 deletions(-)

diff --git a/DataCompiler/dataCompiler.swift b/DataCompiler/dataCompiler.swift
index 4128f4c6..dea6fa77 100644
--- a/DataCompiler/dataCompiler.swift
+++ b/DataCompiler/dataCompiler.swift
@@ -118,16 +118,13 @@ func rawDictForPhrases(isCHS: Bool) -> [Entry] {
     }
     // 預處理格式
     strRAW = strRAW.replacingOccurrences(of: " #MACOS", with: "") // 去掉 macOS 標記
-    strRAW = strRAW.replacingOccurrences(of: "　", with: " ") // CJKWhiteSpace (\x{3000}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: " ", with: " ") // NonBreakWhiteSpace (\x{A0}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: "\t", with: " ") // Tab to ASCII Space
-    strRAW.regReplace(pattern: "\\f", replaceWith: "\n") // Form Feed to LF
-    strRAW = strRAW.replacingOccurrences(of: "\r", with: "\n") // CR to LF
-    strRAW.regReplace(pattern: " +", replaceWith: " ") // 統整連續空格為一個 ASCII 空格
-    // strRAW.regReplace(pattern: "\\n+", replaceWith: "\n") // 統整連續 LF 為一個 LF
-    // (不需要處理純空行，因為空記錄不會被轉為 Entry)
-    strRAW = strRAW.replacingOccurrences(of: " \n", with: "\n") // 去除行尾空格
-    strRAW = strRAW.replacingOccurrences(of: "\n ", with: "\n") // 去除行首空格
+    // CJKWhiteSpace (\x{3000}) to ASCII Space
+    // NonBreakWhiteSpace (\x{A0}) to ASCII Space
+    // Tab to ASCII Space
+    // 統整連續空格為一個 ASCII 空格
+    strRAW.regReplace(pattern: #"( +|　+| +|\t+)+"#, replaceWith: " ")
+    strRAW.regReplace(pattern: #"(\f+|\r+)+"#, replaceWith: "\n") // CR & Form Feed to LF
+    strRAW.regReplace(pattern: #"(\n+| \n+|\n+ )"#, replaceWith: "\n") // 去除行尾行首空格與重複行
     if strRAW.prefix(1) == " " { // 去除檔案開頭空格
         strRAW.removeFirst()
     }
@@ -195,16 +192,13 @@ func rawDictForKanjis(isCHS: Bool) -> [Entry] {
     }
     // 預處理格式
     strRAW = strRAW.replacingOccurrences(of: " #MACOS", with: "") // 去掉 macOS 標記
-    strRAW = strRAW.replacingOccurrences(of: "　", with: " ") // CJKWhiteSpace (\x{3000}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: " ", with: " ") // NonBreakWhiteSpace (\x{A0}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: "\t", with: " ") // Tab to ASCII Space
-    strRAW.regReplace(pattern: "\\f", replaceWith: "\n") // Form Feed to LF
-    strRAW = strRAW.replacingOccurrences(of: "\r", with: "\n") // CR to LF
-    strRAW.regReplace(pattern: " +", replaceWith: " ") // 統整連續空格為一個 ASCII 空格
-    // strRAW.regReplace(pattern: "\\n+", replaceWith: "\n") // 統整連續 LF 為一個 LF
-    // (不需要處理純空行，因為空記錄不會被轉為 Entry)
-    strRAW = strRAW.replacingOccurrences(of: " \n", with: "\n") // 去除行尾空格
-    strRAW = strRAW.replacingOccurrences(of: "\n ", with: "\n") // 去除行首空格
+    // CJKWhiteSpace (\x{3000}) to ASCII Space
+    // NonBreakWhiteSpace (\x{A0}) to ASCII Space
+    // Tab to ASCII Space
+    // 統整連續空格為一個 ASCII 空格
+    strRAW.regReplace(pattern: #"( +|　+| +|\t+)+"#, replaceWith: " ")
+    strRAW.regReplace(pattern: #"(\f+|\r+)+"#, replaceWith: "\n") // CR & Form Feed to LF
+    strRAW.regReplace(pattern: #"(\n+| \n+|\n+ )"#, replaceWith: "\n") // 去除行尾行首空格與重複行
     if strRAW.prefix(1) == " " { // 去除檔案開頭空格
         strRAW.removeFirst()
     }
@@ -277,16 +271,13 @@ func rawDictForNonKanjis(isCHS: Bool) -> [Entry] {
     }
     // 預處理格式
     strRAW = strRAW.replacingOccurrences(of: " #MACOS", with: "") // 去掉 macOS 標記
-    strRAW = strRAW.replacingOccurrences(of: "　", with: " ") // CJKWhiteSpace (\x{3000}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: " ", with: " ") // NonBreakWhiteSpace (\x{A0}) to ASCII Space
-    strRAW = strRAW.replacingOccurrences(of: "\t", with: " ") // Tab to ASCII Space
-    strRAW.regReplace(pattern: "\\f", replaceWith: "\n") // Form Feed to LF
-    strRAW = strRAW.replacingOccurrences(of: "\r", with: "\n") // CR to LF
-    strRAW.regReplace(pattern: " +", replaceWith: " ") // 統整連續空格為一個 ASCII 空格
-    // strRAW.regReplace(pattern: "\\n+", replaceWith: "\n") // 統整連續 LF 為一個 LF
-    // (不需要處理純空行，因為空記錄不會被轉為 Entry)
-    strRAW = strRAW.replacingOccurrences(of: " \n", with: "\n") // 去除行尾空格
-    strRAW = strRAW.replacingOccurrences(of: "\n ", with: "\n") // 去除行首空格
+    // CJKWhiteSpace (\x{3000}) to ASCII Space
+    // NonBreakWhiteSpace (\x{A0}) to ASCII Space
+    // Tab to ASCII Space
+    // 統整連續空格為一個 ASCII 空格
+    strRAW.regReplace(pattern: #"( +|　+| +|\t+)+"#, replaceWith: " ")
+    strRAW.regReplace(pattern: #"(\f+|\r+)+"#, replaceWith: "\n") // CR & Form Feed to LF
+    strRAW.regReplace(pattern: #"(\n+| \n+|\n+ )"#, replaceWith: "\n") // 去除行尾行首空格與重複行
     if strRAW.prefix(1) == " " { // 去除檔案開頭空格
         strRAW.removeFirst()
     }