tapsoob 0.3.27 → 0.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b7b43f7fe0ec4c4f90c32f00e0ca57ab252c78cdf5f13dfe1c689f8d7a031d32
-  data.tar.gz: 3d27c0eb39663d9766d1104df903045be1ddbd1580cfcf854e0ea6a98eb64f70
+  metadata.gz: 469c560664c5ba21ac179589064ae693ec3cc158d5a2d9ad2a9a0117a3657c71
+  data.tar.gz: b05907cdcf4df4ffeb66fcf3cd086f8083b84b1189f559aca4cb1ef1661fd7ca
 SHA512:
-  metadata.gz: be0a84840b6036dfd5ee03a4c659978eb38b30353961f2797bdb03a6f5ea6fddad407ff3442252a87f0619968cafe57a7c52f7c3102de8757b70ffaf2babd0d4
-  data.tar.gz: c415ccc0c1d035ece72938c443ff2deba691432f4f9baa6560cba29eac0537bd7c48579a5640727af5f2a42091b2ffdffb63fd2c88bb90f7cf4c266a86449990
+  metadata.gz: cfe9c901740ae209bf8abf7cc534edfeba3e6a209da81acc2ae6776fa4a79ec28b7d1425c9ae4825aff07ae4e4fe9a5b0a1df1769c57ea22a513bb0013a54d9d
+  data.tar.gz: 5768ef7b85b8188215a82d9f00177d51443558e62baf4085ea4ec007116a844587e145cd1d4a6b507aea93e1066a8c5f60caa95630665dacf0fb77ca70908107

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
@@ -222,12 +223,24 @@ 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
@@ -237,9 +250,9 @@ module Tapsoob
         %w(date datetime time).each do |type|
           if rows[:types].include?(type)
             type_indices = rows[:types].each_index.select { |idx| rows[:types][idx] == type }
-            rows[:data].each_index do |idx|
+            data.each_index do |idx|
               type_indices.each do |ti|
-                rows[:data][idx][ti] = Sequel.send("string_to_#{type}".to_sym, rows[:data][idx][ti])
+                data[idx][ti] = Sequel.send("string_to_#{type}".to_sym, data[idx][ti]) unless data[idx][ti].nil?
               end
             end
           end
@@ -247,11 +260,11 @@ module Tapsoob
       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
@@ -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 = "0.3.27".freeze
+  VERSION = "0.4.0".freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tapsoob
 version: !ruby/object:Gem::Version
-  version: 0.3.27
+  version: 0.4.0
 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: 2021-06-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ripl