ecdict 1.2.alpha5 → 1.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5d3b7af98d6ca07e5b9228c8ed515af5250f194b74ff1b5b8a15cc4a53418d48
-  data.tar.gz: 0ac17588afaa66aaddc89b65a870ccebf7d8318dfdc8526f1d2ba804d4624897
+  metadata.gz: 3b7ecd655d55d7a4fd69f9f508520cc26537c0d97b9bcc411449963e4fc8b868
+  data.tar.gz: cce06613d92a5550fa54fdd677f54ff9f2435597eddebae545bde391edc93a6e
 SHA512:
-  metadata.gz: f8e715abe0412e94f79739a43cbabd063211d50977c4d7d8dbb49851cdedfd51f9ab4847b91d046a817c64c240baff591a37fd00d7ea45ee13c119c9afce61c8
-  data.tar.gz: ab5ec984d53c633ce72f5a557e5f7f7aab91f395cf24d8706562bbeef5a4efd9d3c653879a28170a0b6ea1dc1b05c819022656e29b97c8977e734a5508940d0e
+  metadata.gz: cfc2f765aee2c9e9a9284a6316eb37c548b5da8f3f8b94556a295c0892ed0341a01762bfaf119467397c698bd426e4c4f707d55031a4ed2cc90a82d513f9b81b
+  data.tar.gz: e9bd11e1448f51341bf71f334d6ae210615f04580d3900e9fd971700fa3ee3b7d9eee008b3791a1ac560b5af9f77612698f2d390927cf429e9f1f69f4b26dc82

data/bin/ecdict

@@ -3,7 +3,7 @@
 # File          : ecdict.rb
 # Authors       : ccmywish <ccmywish@qq.com>
 # Created on    : <2020-10-18>
-# Last modified : <2022-03-24>
+# Last modified : <2022-04-29>
 #
 # ecdict:
 #
@@ -27,25 +27,23 @@
 
 require 'sqlite3'
 
-ECDICT_STORAGE   = File.expand_path("~/.local/share/ecdict")
-ECDICT_RAW_DATA  = File.join(ECDICT_STORAGE, 'ecdict-csv.7z')
+ECDict::STORAGE   = File.expand_path("~/.local/share/ecdict")
+ECDict::RAW_DATA  = File.join(ECDict::STORAGE, 'ecdict-csv.7z')
 
-ECDictDB  = File.join(ECDICT_STORAGE, 'ecdict.db')
-ECDictCSV = File.join(ECDICT_STORAGE, 'ecdict.csv')
-
-ECDICT_GEM_VERSION  = "1.2.alpha5"
+ECDict::DB  = File.join(ECDict::STORAGE, 'ecdict.db')
+ECDict::CSV = File.join(ECDict::STORAGE, 'ecdict.csv')
 
 # 这个版本是我从上游下载的CSV后自己压缩的
 # 这个文件的实际创建日期(上游作者创建)是2017/06/04
 # 至2022/03/22，未见更新，估计以后上游维护这个词典的也不会经常更新
 # 而340万的单词量已经足够，因此我们可以足够长的时间一直停留在此词典版本
-ECDICT_DICT_DOWNLOAD_LINK = "https://gitee.com/ccmywish/ecdict-data"
-ECDICT_DICT_SHA256 = "86782a0e5d05486b482be5a2e5fa99b1a9b2ae51240c853ecfe181886133818a"
+ECDict::DICT_DOWNLOAD_LINK = "https://gitee.com/ccmywish/ecdict-data"
+ECDict::DICT_SHA256 = "86782a0e5d05486b482be5a2e5fa99b1a9b2ae51240c853ecfe181886133818a"
 
 
 require 'fileutils'
-FileUtils.mkdir_p(ECDICT_STORAGE)
-$DB = SQLite3::Database.new ECDictDB
+FileUtils.mkdir_p(ECDict::STORAGE)
+$DB = SQLite3::Database.new ECDict::DB
 
 def close_db
   $DB.close if $DB
@@ -84,9 +82,9 @@ def cyan(str)       "\e[36m#{str}\e[0m" end
 
 def check_download_integrity
   require 'digest'
-  return false if !File.exists?(ECDICT_RAW_DATA)
-  sha256 = Digest::SHA256.file(ECDICT_RAW_DATA).hexdigest
-  if ECDICT_DICT_SHA256 == sha256
+  return false if !File.exists?(ECDict::RAW_DATA)
+  sha256 = Digest::SHA256.file(ECDict::RAW_DATA).hexdigest
+  if ECDict::DICT_SHA256 == sha256
     true
   else
     false
@@ -96,40 +94,40 @@ end
 
 def download_ecdict_raw_data
 
-  if File.exists?(ECDICT_RAW_DATA) && check_download_integrity
-    puts "=> 已存在原始数据文件#{ECDICT_RAW_DATA}, 无需再次下载"
+  if File.exists?(ECDict::RAW_DATA) && check_download_integrity
+    puts "=> 已存在原始数据文件#{ECDict::RAW_DATA}, 无需再次下载"
   else
 
-    if File.exists?(ECDICT_RAW_DATA)
+    if File.exists?(ECDict::RAW_DATA)
       puts "=> 删除已存在但不完整的原始数据文件"
-      FileUtils.rm ECDICT_RAW_DATA
+      FileUtils.rm ECDict::RAW_DATA
     end
 
     # 若已经有拉取的仓库，先检查里面的数据是否可用
-    raw_data_git_dir = File.join(ECDICT_STORAGE, 'ecdict-data')
+    raw_data_git_dir = File.join(ECDict::STORAGE, 'ecdict-data')
     raw_data_git_dir_data = File.join(raw_data_git_dir, 'ecdict-csv.7z')
 
     if Dir.exists?(raw_data_git_dir)
       if File.exists?(raw_data_git_dir_data)
-        FileUtils.cp(raw_data_git_dir_data, ECDICT_STORAGE)
+        FileUtils.cp(raw_data_git_dir_data, ECDict::STORAGE)
         if check_download_integrity
           puts "=> 从已经Git pull到的仓库中获取原始数据文件"
           return true
         else
-          FileUtils.rm ECDICT_RAW_DATA
+          FileUtils.rm ECDict::RAW_DATA
         end
       end
       FileUtils.rm_rf(raw_data_git_dir)
     end
 
     begin
-      puts "=> 使用Git从#{ECDICT_DICT_DOWNLOAD_LINK}获取原始数据库文件"
-      ret = system("git -C #{ECDICT_STORAGE} clone #{ECDICT_DICT_DOWNLOAD_LINK} ")
-      FileUtils.cp(File.join(ECDICT_STORAGE, 'ecdict-data', 'ecdict-csv.7z'), ECDICT_STORAGE) rescue nil
+      puts "=> 使用Git从#{ECDict::DICT_DOWNLOAD_LINK}获取原始数据库文件"
+      ret = system("git -C #{ECDict::STORAGE} clone #{ECDict::DICT_DOWNLOAD_LINK} ")
+      FileUtils.cp(File.join(ECDict::STORAGE, 'ecdict-data', 'ecdict-csv.7z'), ECDict::STORAGE) rescue nil
       raise "Git拉取仓库过程失败，原始数据文件不完整!" if (ret != true  || !check_download_integrity)
     rescue StandardError => e
       puts "=> #{e.message}"
-      FileUtils.rm(ECDICT_RAW_DATA) rescue nil
+      FileUtils.rm(ECDict::RAW_DATA) rescue nil
       return false
     else
       puts "=> 下载完成!"
@@ -143,10 +141,10 @@ def decompress_7z_to_csv()
   require 'seven_zip_ruby'
 
   # 正常解压出来的CSV应当至少200MB以上
-  if File.exists?(ECDictCSV)
-    if File.size(ECDictCSV) <= 200*1024*1024
+  if File.exists?(ECDict::CSV)
+    if File.size(ECDict::CSV) <= 200*1024*1024
       puts "=> 删除旧有的不完全词典数据"
-      FileUtils.rm(ECDictCSV)
+      FileUtils.rm(ECDict::CSV)
     else
       puts "=> CSV数据已存在，无需再次解压，直接使用"
       return true
@@ -154,12 +152,12 @@ def decompress_7z_to_csv()
   end
 
 
-  File.open(ECDICT_RAW_DATA, "rb") do |file|
+  File.open(ECDict::RAW_DATA, "rb") do |file|
     puts "=> 正在解压ecdict-csv.7z(#{ '%.2f' % (file.size/1024.0/1024) }MB)"
-    SevenZipRuby::Reader.extract_all(file, ECDICT_STORAGE)
+    SevenZipRuby::Reader.extract_all(file, ECDict::STORAGE)
   end
 
-  puts "=> 已成功解压出ecdict.csv(#{ '%.2f' % (File.size(ECDictCSV)/1024.0/1024) }MB)"
+  puts "=> 已成功解压出ecdict.csv(#{ '%.2f' % (File.size(ECDict::CSV)/1024.0/1024) }MB)"
   true
 end
 
@@ -170,7 +168,7 @@ end
 #
 def check_db_integrity
   # 正常从CSV生成出来的数据库应当至少400MB以上
-  if File.size(ECDictDB) <= 400*1024*1024
+  if File.size(ECDict::DB) <= 400*1024*1024
     false
   else
     true
@@ -191,13 +189,13 @@ end
 def generate_sqlite_db()
 
   # SQLite3::Database.new已经生成了该文件,所以需要提前判断到底是否存在
-  if File.exist?(ECDictDB)
+  if File.exist?(ECDict::DB)
 
     if !check_db_integrity
       puts "=> 删除旧有的不完整数据库"
       $DB.close
-      FileUtils.rm(ECDictDB)
-      $DB = SQLite3::Database.new ECDictDB
+      FileUtils.rm(ECDict::DB)
+      $DB = SQLite3::Database.new ECDict::DB
     else
       puts "=> 完整(可能的)数据库已存在，无需再次从CSV文件生成，直接使用"
       return true
@@ -249,12 +247,12 @@ EOF
 
   begin
     $DB.execute sql
-    puts "=> 创建数据库文件#{ECDictDB}，并生成表`ecdict`"
+    puts "=> 创建数据库文件#{ECDict::DB}，并生成表`ecdict`"
   rescue Exception => e
     puts "=> #{e.message}"
-    puts "=> 创建数据库文件#{ECDictDB}失败，或表`ecdict`生成失败"
+    puts "=> 创建数据库文件#{ECDict::DB}失败，或表`ecdict`生成失败"
     close_db
-    File.delete(ECDictDB)
+    File.delete(ECDict::DB)
     return false
   end
 
@@ -276,7 +274,7 @@ EOF
     puts <<~EOC
 
     #{blue("gem_name  = 'ecdict'")}
-    #{blue("version   = '#{ECDICT_GEM_VERSION}'")}
+    #{blue("version   = '#{ECDict::VERSION}'")}
     #{blue("author    = 'ccmywish'")}
     #{blue("bug_track = [ 'https://gitee.com/ccmywish/ecdict/issues'
               'https://github.com/ccmywish/ecdict/issues ]")}
@@ -301,7 +299,7 @@ EOF
     require 'csv'
     num = 0
     $DB.transaction
-    CSV.foreach(ECDictCSV) do |row|
+    CSV.foreach(ECDict::CSV) do |row|
       num = num + 1
       next if num == 1     # headers不加入
       row.map! do |n|
@@ -319,11 +317,11 @@ EOF
     puts "=> #{e.message}"
     puts "=> 数据库插入信息失败"
     close_db
-    File.delete(ECDictDB)
+    File.delete(ECDict::DB)
     return false
   end
   puts
-  puts "=> 数据库#{ECDictDB}已完整生成"
+  puts "=> 数据库#{ECDict::DB}已完整生成"
   close_db
   return true
 end
@@ -350,7 +348,7 @@ end
 #
 # Error code -1: No db or db not intact
 # Error code  1: No input
-# Error code  2: Don't find a result
+# Error code  2: Doesn't find a result
 #
 
 #
@@ -423,7 +421,7 @@ end
 #
 # Search Chinese word to find English words
 #
-def chinese_search(cn_word, support_phrase: false)
+def search_chinese(cn_word, support_phrase: false)
 
   if cn_word.empty?
     puts "ecdict: 请输入要查询的中文，或使用`ecdict -h`查看帮助"
@@ -435,7 +433,7 @@ def chinese_search(cn_word, support_phrase: false)
   end
 
   if support_phrase
-    puts "ecdict: 搜索加强，支持短语"
+    puts "ecdict: 搜索加强，包含短语"
   end
 
   tables = []
@@ -451,7 +449,7 @@ def chinese_search(cn_word, support_phrase: false)
     end
   end
   if rows.empty?
-    puts "ecdict: 未找到相关英文,请不要输入中文短语或短句，仅支持词的搜索"
+    puts "ecdict: 抱歉，未找到与之相关的英文"
     close_db
     return
   else
@@ -467,12 +465,14 @@ def chinese_search(cn_word, support_phrase: false)
       next if trans.include?("\r\n")
       next if trans.include?("\n")
 
-      # 不要搜索词组
+
       if !support_phrase
+        # 不要搜索词组
         next if en_word.include?(' ')
+        # 不要搜索连字词
+        next if en_word.include?('-')
       end
-      # 不要搜索连字词
-      next if en_word.include?('-')
+
 
       # filter
       # "[网络] 微软，认证专家；微软认证产品专家；微软专家认证"
@@ -506,8 +506,14 @@ def chinese_search(cn_word, support_phrase: false)
     if found_a_word
       puts
     else
-      puts "ecdict: 抱歉，为了防止显示内容过多，已将可能的答案过滤了"
-      puts "ecdict: 您可尝试使用`ecdict -cp`来搜索短语，扩大搜索范围"
+
+      if !support_phrase
+        puts "ecdict: 扩大搜索范围，再次尝试搜索..."
+        search_chinese(cn_word, support_phrase: true)
+      else
+        puts "ecdict: 抱歉，未找到与之相关的英文"
+      end
+
     end
   # end of else
   end
@@ -592,7 +598,9 @@ def start_ecrepl
   require 'reline'
   Reline.completion_proc = lambda do |word|
     if word.strip.empty?
-      return %w{1.输入单词并回车查询含义 2.输入单词时按一次tab键反馈搜索建议 3.输入exit或按Ctrl-C或Ctrl-D退出}
+    return %w[1.输入单词并回车查询含义
+              2.输入单词时按一次tab键反馈搜索建议
+              3.输入exit或按Ctrl-C或Ctrl-D退出]
     end
 
     max_len = word.length + 4
@@ -600,30 +608,54 @@ def start_ecrepl
 
     # $DB.execute "SELECT DISTINCT sw FROM ecdict WHERE sw LIKE '#{word}%' AND length(sw)<#{max_len} LIMIT 12" { |row| puts row }
 
-    # 以word开头的单词
-    ret = $DB.execute "SELECT DISTINCT sw FROM ecdict WHERE sw LIKE '#{word}%'
-    AND length(sw)<#{max_len} LIMIT 64"
-    # [["baba"], ["babe"], ["babn"], ["baby"]]
-    ret = ret.to_a.flatten
 
-    if word.length <= 7
-      LsTable.ls(ret) { puts blue(_1) }
-    else
-      LsTable.ls(ret, cell_len: 14, cell_num: 6) { puts blue(_1) }
+    # Display suggestions
+    #
+    # @params word [String] The word we search
+    # @params ret  [Array]  The suggestions array returned
+    def _display_suggestions(word, ret)
+      return if ret.empty?
+      if word.length <= 7
+        LsTable.ls(ret) { puts blue(_1) }
+      else
+       LsTable.ls(ret, cell_len: 14, cell_num: 6) { puts blue(_1) }
+      end
+      puts
     end
 
-    puts
+    # Generate suggestions
+    # 1. the words beginning with our search word
+    # 2. the words including our search word
+    #
+    # @params word    [String]  The word we search
+    # @params max_len [Integer] Suggestion's max length
+    def _gen_suggestion_1(word, max_len)
+      ret = $DB.execute <<-SQL
+        SELECT DISTINCT sw FROM ecdict WHERE sw LIKE '#{word}%'
+        AND length(sw)<#{max_len} LIMIT 64
+      SQL
+      # [["baba"], ["babe"], ["babn"], ["baby"]]
+      ret = ret.to_a.flatten
+    end
 
-    # 前面至少有一个字符
-    ret = $DB.execute "SELECT DISTINCT sw FROM ecdict WHERE sw LIKE '_%#{word}%'
-    AND length(sw)<#{max_len} LIMIT 64"
-    ret = ret.to_a.flatten
-    if word.length <= 7
-      LsTable.ls(ret) { puts blue(_1) }
-    else
-      LsTable.ls(ret, cell_len: 14, cell_num: 6) { puts blue(_1) }
+    def _gen_suggestion_2(word, max_len)
+      ret = $DB.execute <<-SQL
+        SELECT DISTINCT sw FROM ecdict WHERE sw LIKE '_%#{word}%'
+        AND length(sw)<#{max_len} LIMIT 64
+      SQL
+      ret = ret.to_a.flatten
     end
 
+    suggestions = [
+      Thread.new {
+        _display_suggestions word, _gen_suggestion_1(word, max_len)
+      },
+      Thread.new {
+        _display_suggestions word, _gen_suggestion_2(word, max_len)
+      }
+    ].each(&:join)
+
+    ""
   end
 
   #
@@ -654,12 +686,12 @@ end
 
 
 #
-# remove everything in #{ECDICT_STORAGE}
+# remove everything in #{ECDict::STORAGE}
 #
 def delete_cache
   close_db
   begin
-    FileUtils.rm_rf(ECDICT_STORAGE)
+    FileUtils.rm_rf(ECDict::STORAGE)
   rescue => e
     puts "ecdict: #{e.message}"
     puts "ecdict: 清空词典数据失败"
@@ -676,7 +708,7 @@ end
 
 def print_version
   puts <<EOH
-ecdict (v#{ECDICT_GEM_VERSION}): A courteous cli translator.
+ecdict (v#{ECDict::VERSION}): A courteous cli translator.
 EOH
 
 end
@@ -684,14 +716,13 @@ end
 
 def help
   puts <<EOH
-ecdict (v#{ECDICT_GEM_VERSION}): A courteous cli translator.
+ecdict (v#{ECDict::VERSION}): A courteous cli translator.
 
 usage:
 
   ecdict word      => 查询单词word
   ecdict -r        => 启动ecdict repl交互式查询，输入exit或Ctrl-C/D退出
-  ecdict -c  中文  => 搜索中文单词对应的英文单词
-  ecdict -cp 中文  => 搜索中文短语对应的英文短语
+  ecdict -c  中文  => 搜索中文单词对应的英文单词/短语
   ecdict -i        => 下载ecdict词典数据并安装
   ecdict -v        => 打印此Gem版本号
   ecdict -h        => 打印此帮助
@@ -722,11 +753,10 @@ when "-h"  then help
 when "-i"  then download_and_install_ecdict_data
 when "-r"  then start_ecrepl
 when "-d"  then delete_cache
-when "-c"  then chinese_search(ARGV.join)
-when "-cp" then chinese_search(ARGV.join, support_phrase: true)
+when "-c"  then search_chinese(ARGV.join)
 else
   reply_once(query)
 end
 
 # ensure close db
-close_db
+close_db

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ecdict
 version: !ruby/object:Gem::Version
-  version: 1.2.alpha5
+  version: 1.3.0
 platform: ruby
 authors:
 - ccmywish
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-03-23 00:00:00.000000000 Z
+date: 2022-04-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: sqlite3
@@ -94,11 +94,11 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">"
+  - - ">="
     - !ruby/object:Gem::Version
-      version: 1.3.1
+      version: '0'
 requirements: []
-rubygems_version: 3.3.7
+rubygems_version: 3.3.12
 signing_key: 
 specification_version: 4
 summary: 'ecdict: English-to-Chinese dictionary on the cli.'