Module: Hdfs

Includes:
Java
Defined in:
lib/hdfs_jruby.rb,
lib/hdfs_jruby/file.rb,
lib/hdfs_jruby/version.rb

Defined Under Namespace

Classes: Configuration, File, FileSystem, FsPermission, Path

Constant Summary collapse

JAR_PATTERN_0_20 = 
"hadoop-core-*.jar"
HADOOP_HOME = 

      
        
VERSION =
          
        

        
"0.0.12"


      
    

  







  
    

      Class Method Summary
      collapse
    


    
  


  

  

    
Class Method Details


    
      

  

  
    ._conv(stat)  ⇒ Object 
  

  

  




  

    
      


249
250
251
252
253
254
255
256
257
258
259

    		
    
      
# File 'lib/hdfs_jruby.rb', line 249

def _conv(stat)
  file_info = {}
  file_info['path'] = stat.getPath.to_s
  file_info['length'] = stat.getLen.to_i
  file_info['modificationTime'] = stat.getModificationTime.to_i
  file_info['owner'] = stat.getOwner.to_s
  file_info['group'] = stat.getGroup.to_s
  file_info['permission'] = stat.getPermission.toShort.to_i
  file_info['type'] = !stat.isDir ? 'FILE': 'DIRECTORY'
  return file_info
end


    		
  





    
      

  

  
    ._path(path)  ⇒ Object 
  

  

  




  

    
      


241
242
243
244
245
246

    		
    
      
# File 'lib/hdfs_jruby.rb', line 241

def _path(path)
  if path.nil?
    raise "path is nil"
  end
  Path.new(path)
end


    		
  





    
      

  

  
    .connectAsUser(user)  ⇒ Object 
  

  

  




  

    
      


54
55
56
57
58

    		
    
      
# File 'lib/hdfs_jruby.rb', line 54

def connectAsUser(user)
  uri =  Hdfs::FileSystem.getDefaultUri(@conf)
  @fs.close if ! @fs.nil?
  @fs = Hdfs::FileSystem.get(uri, @conf, user)
end


    		
  





    
      

  

  
    .delete(path, r = false)  ⇒ Object 
  

  

  




  

    

delete



  





  



  

    
      


146
147
148

    		
    
      
# File 'lib/hdfs_jruby.rb', line 146

def delete(path, r=false)
  @fs.delete(_path(path), r)
end


    		
  





    
      

  

  
    .directory?(path)  ⇒ Boolean 
  

  

  




  

    
      


156
157
158

    		
    
      
# File 'lib/hdfs_jruby.rb', line 156

def directory?(path)
  @fs.isDirectory(_path(path))
end


    		
  





    
      

  

  
    .exists?(path)  ⇒ Boolean 
  

  

  




  

    
      


132
133
134

    		
    
      
# File 'lib/hdfs_jruby.rb', line 132

def exists?(path)
  @fs.exists(_path(path))
end


    		
  





    
      

  

  
    .file?(path)  ⇒ Boolean 
  

  

  




  

    
      


151
152
153

    		
    
      
# File 'lib/hdfs_jruby.rb', line 151

def file?(path)
  @fs.isFile(_path(path))
end


    		
  





    
      

  

  
    .get(remote, local)  ⇒ Object 
  

  

  




  

    

get file or directory from hdfs



  





  



  

    
      


181
182
183

    		
    
      
# File 'lib/hdfs_jruby.rb', line 181

def get(remote, local)
  @fs.copyToLocalFile(Path.new(remote), Path.new(local))
end


    		
  





    
      

  

  
    .get_fs  ⇒ Object 
  

  

  




  

    
      


215
216
217

    		
    
      
# File 'lib/hdfs_jruby.rb', line 215

def get_fs
  @fs
end


    		
  





    
      

  

  
    .get_home_directory  ⇒ Object 
  

  

  




  

    

get home directory



  





  



  

    
      


186
187
188

    		
    
      
# File 'lib/hdfs_jruby.rb', line 186

def get_home_directory()
  @fs.getHomeDirectory()
end


    		
  





    
      

  

  
    .get_working_directory  ⇒ Object 
  

  

  




  

    

get working directory



  





  



  

    
      


191
192
193

    		
    
      
# File 'lib/hdfs_jruby.rb', line 191

def get_working_directory()
  @fs.getWorkingDirectory()
end


    		
  





    
      

  

  
    .list(path, opts = {})  ⇒ Object 
  

  

  




  

    
      


106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129

    		
    
      
# File 'lib/hdfs_jruby.rb', line 106

def list(path, opts={})
  use_glob = opts[:glob] ? true : false
  p = _path(path)

  list = nil
  if use_glob
    list = @fs.globStatus(p)
  else
    list = @fs.listStatus(p)
  end
  return [] if list.nil?
    
  if ! block_given?
    ret_list = []
    list.each do | stat |
      ret_list << _conv(stat)
    end
    return ret_list
  else
    list.each do | stat |
      yield _conv(stat)
    end
  end
end


    		
  





    
      

  

  
    .ls(path)  ⇒ Array 
  

  

  




  

    
  

    Note:
    


file status: path length modificationTime owner group permission type



  




ls



  





  
  

    
Examples:

    
      
      
Hdfs.ls("hoge/").each do | stat |
  p stat
end


    
  





  

    
      


76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103

    		
    
      
# File 'lib/hdfs_jruby.rb', line 76

def ls(path)
  p = _path(path)
  list = @fs.globStatus(p)
  return [] if list.nil?

  ret_list = []
  list.each do |stat|
    if stat.isDir
      sub_list = @fs.listStatus(stat.getPath)
      next if sub_list.nil?
      
      sub_list.each do | s |
        if block_given?
          yield _conv(s)
        else
          ret_list << _conv(s)
        end
      end
    else
      if block_given?
        yield _conv(stat)
      else
        ret_list << _conv(stat)
      end
    end
  end
  ret_list if ! block_given?
end


    		
  





    
      

  

  
    .mkdir(path)  ⇒ Object 
  

  

  




  

    

make directory



  





  



  

    
      


167
168
169

    		
    
      
# File 'lib/hdfs_jruby.rb', line 167

def mkdir(path)
  @fs.mkdirs(_path(path))
end


    		
  





    
      

  

  
    .move(src, dst)  ⇒ Object 
  

  

  




  

    
      


138
139
140

    		
    
      
# File 'lib/hdfs_jruby.rb', line 138

def move(src, dst)
  @fs.rename(Path.new(src), Path.new(dst))
end


    		
  





    
      

  

  
    .put(local, remote)  ⇒ Object 
  

  

  




  

    

put file or directory to hdfs



  





  



  

    
      


174
175
176

    		
    
      
# File 'lib/hdfs_jruby.rb', line 174

def put(local, remote)
  @fs.copyFromLocalFile(Path.new(local), Path.new(remote))
end


    		
  





    
      

  

  
    .set_owner(path, owner, group)  ⇒ Object 
  

  

  




  

    

set owner & group



  





  



  

    
      


211
212
213

    		
    
      
# File 'lib/hdfs_jruby.rb', line 211

def set_owner(path, owner, group)
  @fs.setOwner(_path(path), owner, group)
end


    		
  





    
      

  

  
    .set_permission(path, perm)  ⇒ Object 
  

  

  




  

    

set permission



  





  



  

    
      


203
204
205

    		
    
      
# File 'lib/hdfs_jruby.rb', line 203

def set_permission(path, perm)
  @fs.setPermission(_path(path), org.apache.hadoop.fs.permission.FsPermission.new(perm))
end


    		
  





    
      

  

  
    .set_working_directory(path)  ⇒ Object 
  

  

  




  

    

set working directory



  





  



  

    
      


196
197
198

    		
    
      
# File 'lib/hdfs_jruby.rb', line 196

def set_working_directory(path)
  @fs.setWorkingDirectory(_path())
end


    		
  





    
      

  

  
    .size(path)  ⇒ Integer 
  

  

  




  

    
      


161
162
163

    		
    
      
# File 'lib/hdfs_jruby.rb', line 161

def size(path)
  @fs.getFileStatus(_path(path)).getLen()
end