tapsoob 0.3.25 → 0.4.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cfc8db7ed412920a9e55e43d5ee66c57b0dd09a86e322c0b6f37bd90762ef261
-  data.tar.gz: 46dba518c294c033bf7c12c06f79278700be456d00554ad30d9149decaf8b62e
+  metadata.gz: 661f0fd554e2cd6781c9a174c4051331024879fb1abecb9f48f19b9e365c1e46
+  data.tar.gz: 10535d8376543d13e16a35b8d7e098cdb4c4757b518e75068d6a83884196cd98
 SHA512:
-  metadata.gz: c2ef008de5be1f5e169055026c65b04518163e7987e34f2e1be5c554d7114be96d99f4fe7ae3284798f4e11225c1554f461db8c9cc487d1befb09ef654e8d0ad
-  data.tar.gz: b532c43909abc6afda5314e55da082ddbceb3a3ed285b5ba133f024f969120ec0d17b3d6f75c0423588b3fb61c015763038219f7b6c025d224c340416f7abf6c
+  metadata.gz: 171152e81bcb75da25c9865b8b8211256e7fce176bca88f576b4eb9cf159d67a66badacc51344b3c3a4efa137f16dd57384f606780f0b35984ecfb1d6bfbe96c
+  data.tar.gz: a9a59c2a045419d2b580b9d3932055f012f9b192daf6feef8e35602546b0b8dcbd77fc87e56ffa728c1818c6de5004b8027cc0051c9e44d313acc1c5673bc862

data/lib/tapsoob/cli/data_stream.rb

@@ -28,6 +28,9 @@ module Tapsoob
       option :tables, desc: "Shortcut to filter on a list of tables", type: :array, aliases: "-t"
       option :"exclude-tables", desc: "Shortcut to exclude a list of tables", type: :array, aliases: "-e"
       option :progress, desc: "Show progress", default: true, type: :boolean, aliases: "-p"
+      option :purge, desc: "Purge data in tables prior to performing the import", default: false, type: :boolean, aliases: "-p"
+      option :"skip-duplicates", desc: "Remove duplicates when loading data", default: false, type: :boolean
+      option :"discard-identity", desc: "Remove identity when pushing data (may result in creating duplicates)", default: false, type: :boolean
       option :debug, desc: "Enable debug messages", default: false, type: :boolean, aliases: "-d"
       def push(database_url, dump_path = nil)
         # instantiate stuff
@@ -46,9 +49,8 @@ module Tapsoob
         data.each do |table|
           stream = Tapsoob::DataStream.factory(db(database_url, opts), {
             table_name: table[:table_name],
-            chunksize: opts[:default_chunksize],
-            debug: opts[:debug]
-          })
+            chunksize: opts[:default_chunksize]
+          }, { :"discard-identity" => opts[:"discard-identity"] || false, :purge => opts[:purge] || false, :debug => opts[:debug] })
 
           begin
             stream.import_rows(table)
@@ -66,6 +68,11 @@ module Tapsoob
             debug: options[:debug]
           }
 
+          # Push only options
+          opts[:purge] = options[:purge] if options.key?(:purge)
+          opts[:"skip-duplicates"] = options[:"skip-duplicates"] if options.key?(:"skip-duplicates")
+          opts[:"discard-identity"] = options[:"discard-identity"] if options.key?(:"discard-identity")
+
           # Default chunksize
           if options[:chunksize]
             opts[:default_chunksize] = (options[:chunksize] < 10 ? 10 : options[:chunksize])

data/lib/tapsoob/cli/root.rb

@@ -39,6 +39,7 @@ module Tapsoob
       option :tables, desc: "Shortcut to filter on a list of tables", type: :array, aliases: "-t"
       option :"exclude-tables", desc: "Shortcut to exclude a list of tables", type: :array, aliases: "-e"
       option :purge, desc: "Purge data in tables prior to performing the import", default: false, type: :boolean, aliases: "-p"
+      option :"skip-duplicates", desc: "Remove duplicates when loading data", default: false, type: :boolean
       option :"discard-identity", desc: "Remove identity when pushing data (may result in creating duplicates)", default: false, type: :boolean
       option :debug, desc: "Enable debug messages", default: false, type: :boolean, aliases: "-d"
       def push(dump_path, database_url)
@@ -74,6 +75,7 @@ module Tapsoob
 
           # Push only options
           opts[:purge] = options[:purge] if options.key?(:purge)
+          opts[:"skip-duplicates"] = options[:"skip-duplicates"] if options.key?(:"skip-duplicates")
           opts[:"discard-identity"] = options[:"discard-identity"] if options.key?(:"discard-identity")
 
           # Resume

data/lib/tapsoob/data_stream.rb

@@ -91,7 +91,8 @@ module Tapsoob
       rows = {
         :table_name => ds["table_name"],
         :header     => ds["header"],
-        :data       => ds["data"][state[:offset], (state[:offset] + state[:chunksize])] || [ ]
+        :data       => ((@options[:"skip-duplicates"] ? ds["data"].uniq : ds["data"])[state[:offset], (state[:offset] + state[:chunksize])] || [ ]),
+        :types      => ds["types"]
       }
       update_chunksize_stats
       rows
@@ -222,22 +223,48 @@ module Tapsoob
       columns = rows[:header]
       data    = rows[:data]
 
+      # Only import existing columns
+      if table.columns.size != columns.size
+        existing_columns        = table.columns.map(&:to_s)
+        additional_columns      = columns - existing_columns
+        additional_columns_idxs = additional_columns.map { |c| columns.index(c) }
+        additional_columns_idxs.reverse.each do |idx|
+          columns.delete_at(idx)
+          rows[:types].delete_at(idx)
+        end
+        data.each_index { |didx| additional_columns_idxs.reverse.each { |idx| data[didx].delete_at(idx) } }
+      end
+
       # Decode blobs
       if rows.has_key?(:types) && rows[:types].include?("blob")
         blob_indices = rows[:types].each_index.select { |idx| rows[:types][idx] == "blob" }
-        rows[:data].each_index do |idx|
+        data.each_index do |idx|
           blob_indices.each do |bi|
-            rows[:data][idx][bi] = Sequel::SQL::Blob.new(Tapsoob::Utils.base64decode(rows[:data][idx][bi])) unless rows[:data][idx][bi].nil?
+            data[idx][bi] = Sequel::SQL::Blob.new(Tapsoob::Utils.base64decode(data[idx][bi])) unless data[idx][bi].nil?
+          end
+        end
+      end
+
+      # Parse date/datetime/time columns
+      if rows.has_key?(:types)
+        %w(date datetime time).each do |type|
+          if rows[:types].include?(type)
+            type_indices = rows[:types].each_index.select { |idx| rows[:types][idx] == type }
+            data.each_index do |idx|
+              type_indices.each do |ti|
+                data[idx][ti] = Sequel.send("string_to_#{type}".to_sym, data[idx][ti]) unless data[idx][ti].nil?
+              end
+            end
           end
         end
       end
 
       # Remove id column
-      if @options[:"discard-identity"]
+      if @options[:"discard-identity"] && rows[:header].include?("id")
         columns = rows[:header] - ["id"]
         data    = data.map { |d| d[1..-1] }
       end
-      
+
       table.import(columns, data, :commit_every => 100)
       state[:offset] += rows[:data].size
     rescue Exception => ex

data/lib/tapsoob/operation.rb

@@ -198,7 +198,7 @@ module Tapsoob
         stream   = Tapsoob::DataStream.factory(db, {
           :chunksize  => default_chunksize,
           :table_name => table_name
-        })
+        }, { :debug => opts[:debug] })
         pull_data_from_table(stream, progress)
       end
     end
@@ -396,7 +396,13 @@ module Tapsoob
         db[table_name.to_sym].truncate if @opts[:purge]
         stream = Tapsoob::DataStream.factory(db, {
           :table_name => table_name,
-          :chunksize => default_chunksize }, { :"discard-identity" => @opts[:"discard-identity"] || false })
+          :chunksize => default_chunksize
+        }, {
+          :"skip-duplicates" => opts[:"skip-duplicates"] || false,
+          :"discard-identity" => opts[:"discard-identity"] || false,
+          :purge => opts[:purge] || false,
+          :debug => opts[:debug]
+        })
         progress = ProgressBar.new(table_name.to_s, count)
         push_data_from_file(stream, progress)
       end
@@ -429,7 +435,8 @@ module Tapsoob
               }
             end
 
-            size = stream.fetch_data_in_database({ :encoded_data => encoded_data, :checksum => data[:checksum] })
+            row_size = stream.fetch_data_in_database({ :encoded_data => encoded_data, :checksum => data[:checksum] })
+            log.debug "row size: #{row_size}"
             self.stream_state = stream.to_hash
 
             c.idle_secs = (d1 + d2)
@@ -481,7 +488,7 @@ module Tapsoob
       tbls.each do |table|
         if File.exists?(File.join(dump_path, "data", "#{table}.json"))
           data = JSON.parse(File.read(File.join(dump_path, "data", "#{table}.json")))
-          tables_with_counts[table] = data.size
+          tables_with_counts[table] = data["data"].size
         else
           tables_with_counts[table] = 0
         end

data/lib/tapsoob/progress_bar.rb

@@ -22,7 +22,7 @@ class ProgressBar
     @current = 0
     @previous = 0
     @finished_p = false
-    @start_time = Time.now
+    @start_time = ::Time.now
     @previous_time = @start_time
     @title_width = 14
     @format = "%-#{@title_width}s %3d%% %s %s"
@@ -76,7 +76,7 @@ class ProgressBar
   end
 
   def transfer_rate
-    bytes_per_second = @current.to_f / (Time.now - @start_time)
+    bytes_per_second = @current.to_f / (::Time.now - @start_time)
     sprintf("%s/s", convert_bytes(bytes_per_second))
   end
 
@@ -97,14 +97,14 @@ class ProgressBar
     if @current == 0
       "ETA:  --:--:--"
     else
-      elapsed = Time.now - @start_time
+      elapsed = ::Time.now - @start_time
       eta = elapsed * @total / @current - elapsed;
       sprintf("ETA:  %s", format_time(eta))
     end
   end
 
   def elapsed
-    elapsed = Time.now - @start_time
+    elapsed = ::Time.now - @start_time
     sprintf("Time: %s", format_time(elapsed))
   end
 
@@ -155,7 +155,7 @@ class ProgressBar
       @terminal_width += width - line.length + 1
       show
     end
-    @previous_time = Time.now
+    @previous_time = ::Time.now
   end
 
   def show_if_needed
@@ -169,7 +169,7 @@ class ProgressBar
 
     # Use "!=" instead of ">" to support negative changes
     if cur_percentage != prev_percentage ||
-        Time.now - @previous_time >= 1 || @finished_p
+        ::Time.now - @previous_time >= 1 || @finished_p
       show
     end
   end

data/lib/tapsoob/version.rb

@@ -1,4 +1,4 @@
 # -*- encoding : utf-8 -*-
 module Tapsoob
-  VERSION = "0.3.25".freeze
+  VERSION = "0.4.2".freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tapsoob
 version: !ruby/object:Gem::Version
-  version: 0.3.25
+  version: 0.4.2
 platform: ruby
 authors:
 - Félix Bellanger
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-06-11 00:00:00.000000000 Z
+date: 2021-06-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ripl