tapsoob 0.3.24-java → 0.4.1-java

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5f2ae263c3b94ba8a2d39cdcae3c4ed22462e8b0350a48b865300c5800988ac4
-  data.tar.gz: ebb482c1449af92696f4ef1cfcaae8336c22d018a90cc3f1e9c0726820c79e3b
+  metadata.gz: 4c06335f502637ae12395ff3c5d183297a64c650cfebe4df4f7383e45e87ebaa
+  data.tar.gz: f12f5540fdbc10d947ca5d224b816f055a5f4f543c26be47b15c9fed8476ec96
 SHA512:
-  metadata.gz: da3ff1a49face2aec964a6848c994decece8606c8b5681dd065bb4dbb9f7f456734511fb2791c383dc35500fb2aa59b7c262ccde288f28eb9783392723fdaf04
-  data.tar.gz: fe0a13d721abbdf54145a6ce875334b96668edd505159395ad65043105e7a716f8feabe43c35bebc0ddcc5baf30a273b4d3431718346f56044f5dfbe83118fa4
+  metadata.gz: c04ab5686c8bd94b517b023b71bcb5910072b4e271ac822d2d4866bb03a376a5d358dfce9b32cb1eb37f047de9eb676d404488c9313240a55714cf038a13c713
+  data.tar.gz: e65ac3b651435e35c4abbfcb0318e30d32310f3e58274d0b36ffe0ea14eaf3aa52995cef440589c4d4cd4016201c665baee161c3e4020c53ae37f791f9017383

data/lib/tapsoob/cli/data_stream.rb

@@ -28,6 +28,8 @@ 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 :"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 +48,8 @@ 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 +67,10 @@ module Tapsoob
             debug: options[:debug]
           }
 
+          # Push only options
+          opts[:purge] = options[:purge] if options.key?(:purge)
+          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/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       => (ds["data"][state[:offset], (state[:offset] + state[:chunksize])] || [ ]),
+        :types      => ds["types"]
       }
       update_chunksize_stats
       rows
@@ -219,20 +220,52 @@ module Tapsoob
     end
 
     def import_rows(rows)
+      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
-      columns = ((@options[:"discard-identity"] && rows[:header].include?("id")) ? rows[:header] - ["id"] : rows[:header])
-      
-      table.import(columns, rows[:data], :commit_every => 100)
+      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
       case ex.message

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
@@ -393,10 +393,10 @@ module Tapsoob
 
       tables.each do |table_name, count|
         next unless File.exists?(File.join(dump_path, "data", "#{table_name}.json"))
-        db[table_name.to_sym].truncate if @opts[:purge]
+        db[table.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 }, { :"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 +429,8 @@ 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 +482,7 @@ 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/version.rb

@@ -1,4 +1,4 @@
 # -*- encoding : utf-8 -*-
 module Tapsoob
-  VERSION = "0.3.24".freeze
+  VERSION = "0.4.1".freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tapsoob
 version: !ruby/object:Gem::Version
-  version: 0.3.24
+  version: 0.4.1
 platform: java
 authors:
 - Félix Bellanger
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-06-11 00:00:00.000000000 Z
+date: 2021-06-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement