add script to substitute emails from backups retrieved from prod

Athos
1 parent a7978199
Showing 1 changed file with 61 additions and 0 deletions Show diff stats
utils/remove_backup_emails.rb
@@ -0,0 +1,61 @@
+#!/bin/ruby
+
+require 'fileutils'
+require 'find'
+
+$gitlab_tag = Find.find('.').grep(/_gitlab_backup.tar/)[0].match(/[0-9]+/)
+def decompress
+  system 'mkdir -p gitlab'
+  system 'mkdir -p noosfero'
+  system 'mkdir -p mailman'
+  system "tar -xpf #{$gitlab_tag}_gitlab_backup.tar -C gitlab"
+  system 'tar -xzf noosfero_backup.tar.gz -C noosfero'
+  system 'tar -xzf mailman_backup.tar.gz -C mailman'
+end
+
+def recompress
+  system "pushd $PWD && cd gitlab && tar -cpf #{$gitlab_tag}_gitlab_backup.tar * &&  mv #{$gitlab_tag}_gitlab_backup.tar ../ && popd && rm -rf gitlab"
+  system 'pushd $PWD && cd noosfero && tar -czpf noosfero_backup.tar.gz * &&  mv noosfero_backup.tar.gz ../ && popd && rm -rf noosfero'
+  system 'pushd $PWD && cd mailman && tar -czpf mailman_backup.tar.gz * &&  mv mailman_backup.tar.gz ../ && popd && rm -rf mailman'
+end
+
+$email_hash = Hash.new
+$n=1
+
+def create_hashes_from_file file
+  feed_fd = File.open(file)
+  feed_fd.each_line do |line|
+    emails = line.scan(/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,4}/)
+    emails.each do |email|
+      $email_hash[email] = "email#{$n}@example.com" unless $email_hash.has_key?(email)
+      $n+=1
+    end
+  end
+  feed_fd.close
+end
+
+def substitute_hashes_in_file file
+  tmp_file = 'tmp_file'
+  File.open(file) { |source_file|
+    contents = source_file.read
+    $email_hash.each do |key,value|
+      contents.gsub!(key, value)
+    end
+    File.open(tmp_file, "w+") { |f| f.write(contents) }
+  } 
+FileUtils.mv(tmp_file, file)
+end
+
+decompress
+
+noosfero_sql = File.basename(Find.find('./noosfero/tmp/').grep(/\.sql/)[0])
+noosfero_sql = Find.find('./noosfero/tmp/').grep(/\.sql/)[0]
+deEmailedFiles = ['colab.dump', 'gitlab/db/database.sql', noosfero_sql]
+deEmailedFiles += Find.find('./mailman/lists/').grep(/config\.pck/)
+
+deEmailedFiles.each do |deEmailedFile|
+  create_hashes_from_file(deEmailedFile)
+  substitute_hashes_in_file(deEmailedFile)
+end
+recompress
+
...	...	@@ -0,0 +1,61 @@
	1	+#!/bin/ruby
	2	+
	3	+require 'fileutils'
	4	+require 'find'
	5	+
	6	+$gitlab_tag = Find.find('.').grep(/_gitlab_backup.tar/)[0].match(/[0-9]+/)
	7	+def decompress
	8	+ system 'mkdir -p gitlab'
	9	+ system 'mkdir -p noosfero'
	10	+ system 'mkdir -p mailman'
	11	+ system "tar -xpf #{$gitlab_tag}_gitlab_backup.tar -C gitlab"
	12	+ system 'tar -xzf noosfero_backup.tar.gz -C noosfero'
	13	+ system 'tar -xzf mailman_backup.tar.gz -C mailman'
	14	+end
	15	+
	16	+def recompress
	17	+ system "pushd $PWD && cd gitlab && tar -cpf #{$gitlab_tag}_gitlab_backup.tar * && mv #{$gitlab_tag}_gitlab_backup.tar ../ && popd && rm -rf gitlab"
	18	+ system 'pushd $PWD && cd noosfero && tar -czpf noosfero_backup.tar.gz * && mv noosfero_backup.tar.gz ../ && popd && rm -rf noosfero'
	19	+ system 'pushd $PWD && cd mailman && tar -czpf mailman_backup.tar.gz * && mv mailman_backup.tar.gz ../ && popd && rm -rf mailman'
	20	+end
	21	+
	22	+$email_hash = Hash.new
	23	+$n=1
	24	+
	25	+def create_hashes_from_file file
	26	+ feed_fd = File.open(file)
	27	+ feed_fd.each_line do \|line\|
	28	+ emails = line.scan(/\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,4}/)
	29	+ emails.each do \|email\|
	30	+ $email_hash[email] = "email#{$n}@example.com" unless $email_hash.has_key?(email)
	31	+ $n+=1
	32	+ end
	33	+ end
	34	+ feed_fd.close
	35	+end
	36	+
	37	+def substitute_hashes_in_file file
	38	+ tmp_file = 'tmp_file'
	39	+ File.open(file) { \|source_file\|
	40	+ contents = source_file.read
	41	+ $email_hash.each do \|key,value\|
	42	+ contents.gsub!(key, value)
	43	+ end
	44	+ File.open(tmp_file, "w+") { \|f\| f.write(contents) }
	45	+ }
	46	+FileUtils.mv(tmp_file, file)
	47	+end
	48	+
	49	+decompress
	50	+
	51	+noosfero_sql = File.basename(Find.find('./noosfero/tmp/').grep(/\.sql/)[0])
	52	+noosfero_sql = Find.find('./noosfero/tmp/').grep(/\.sql/)[0]
	53	+deEmailedFiles = ['colab.dump', 'gitlab/db/database.sql', noosfero_sql]
	54	+deEmailedFiles += Find.find('./mailman/lists/').grep(/config\.pck/)
	55	+
	56	+deEmailedFiles.each do \|deEmailedFile\|
	57	+ create_hashes_from_file(deEmailedFile)
	58	+ substitute_hashes_in_file(deEmailedFile)
	59	+end
	60	+recompress
	61	+
...	...