tag_remover 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3f1f0436eab08ba2850e4b36e2450e59154ddfe7
-  data.tar.gz: 872a58291e7ff4973a1436d5b2699e7c06328ea9
+  metadata.gz: e5782c529dc1431ea66b7f8c1f23b986b7898c24
+  data.tar.gz: 151094b85405708da5f121e724329742c2b2c44b
 SHA512:
-  metadata.gz: d5fbf8487101e5668118050531c040a44174de61c055386250d5aabf696ed947be0078713673fd8407f87716e8a1c203b342a15180c23a69e286a78e90dca037
-  data.tar.gz: a48ed26ae553d43d973975ffe35e7e090a92c4a38fff206d9d3bb87c57d9221a2026531a4726f3078a60bb0c224d0ccb66660b3ce4bb4c3dd7639306cabad04c
+  metadata.gz: 061fae588623198bd1be710c32d65e4ea983934d2d7ef0b2a7d698f29e0b95074501071c98cc128a591dd35db69eb48feebb7c891a3e028f8597a19526a4196b
+  data.tar.gz: 63afd88c539c6c24bbce653c79ada29977cf18340ced56de75baef26940d1dfb51e0a74dc5957c90933a3405ce43016a5d1912c84b17ae3fbc0281968396f3c9

data/lib/tag_remover.rb

@@ -2,43 +2,97 @@ require "tag_remover/version"
 
 module TagRemover
 
-  def self.process input, output, opts = {}
-    tags_to_remove = opts[:remove_tags] || []
+  class Worker
+    def initialize input, output, opts
+      @input = input
+      @output = output
+      @opts = opts
+    end
 
-    in_tag = nil
-    depth = 0
+    def perform
+      @tags_to_remove = @opts[:remove_tags] || []
 
-    input.each_line.each do |line|
-      if in_tag
-        in_tag_str = tags_to_remove[in_tag]
+      @in_tag = nil
+      @depth = 0
+
+      each_tag do |tag, type|
+        process_tag tag, type
+      end
+
+      if @opts[:close_streams]
+        @input.close
+        @output.close
+      end
+    end
 
-        if line =~ /<#{in_tag_str}(\s|(\s+.+?=(".+?"|'.+?')))*?>/
-          depth += 1
-        elsif line =~ /<\/#{in_tag_str}\s*>/
-          depth -= 1
-          in_tag = nil if depth == 0
+    private
+      def process_tag tag, type
+        if @in_tag
+          in_tag_str = @tags_to_remove[@in_tag]
+
+          if tag =~ opening_tag(in_tag_str)
+            @depth += 1
+          elsif tag =~ closing_tag(in_tag_str)
+            @depth -= 1
+            @in_tag = nil if @depth == 0
+          end
+        else
+          found_tag = false
+          @tags_to_remove.each_with_index do |tag_str,index|
+            if tag =~ opening_tag(tag_str)
+              @in_tag = index
+              @depth = 1
+              found_tag = true
+              break
+            elsif tag =~ single_tag(tag_str)
+              found_tag = true
+              break
+            end
+          end
+
+          @output.write "#{tag}\n" unless found_tag || tag.empty?
         end
-      else
-        found_tag = false
-        tags_to_remove.each_with_index do |tag,index|
-          if line =~ /<#{tag}(\s|(\s+.+?=(".+?"|'.+?')))*?>/
-            in_tag = index
-            depth = 1
-            found_tag = true
-            break
-          elsif line =~ /<#{tag}(\s|(\s+.+?=(".+?"|'.+?')))*?\/\s*>/
-            found_tag = true
-            break
+      end
+
+      def opening_tag tag
+        /<#{tag}(\s|(\s+.+?=(".+?"|'.+?')))*?>/
+      end
+
+      def closing_tag tag
+        /<\/#{tag}\s*>/
+      end
+
+      def single_tag tag
+        /<#{tag}(\s|(\s+.+?=(".+?"|'.+?')))*?\/\s*>/
+      end
+
+      def each_tag &block
+        acc = ""
+        type = nil
+
+        @input.each_char do |c|
+          case c
+          when '<'
+            yield acc.strip, type if type == :text
+            type = :tag
+            acc = c
+          when '>'
+            acc += c
+            yield acc.strip, type
+            type = nil
+            acc = ""
+          else
+            type ||= :text
+            acc += c
           end
         end
-        output.write line unless found_tag
+
+        yield acc.strip, type if type
       end
-    end
+  end
 
-    if opts[:close_streams]
-      input.close
-      output.close
-    end
+  def self.process input, output, opts = {}
+    Worker.new(input, output, opts).perform
   end
 
 end

data/lib/tag_remover/version.rb

@@ -1,3 +1,3 @@
 module TagRemover
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

data/spec/{tag_helper_spec.rb → tag_remover_spec.rb}

@@ -3,72 +3,89 @@ require 'spec_helper'
 describe TagRemover do
   describe ".process" do
     it "removes elements" do
-      input = StringIO.new """
-      <root>
-        <remove>
-          Some contents
-        </remove>
-        <remove >
-        </remove >
-      </root>
-      """
+      input = """
+<root>
+  <remove>
+    Some contents
+  </remove>
+  <remove >
+  </remove >
+</root>"""
+
       output = StringIO.new
       tags_to_remove = ['remove']
 
       TagRemover.process input, output, remove_tags: tags_to_remove
 
-      expect(output.string).to eq """
-      <root>
-      </root>
-      """
+      expect(output.string).to eq """<root>
+</root>
+"""
     end
 
     it "removes single tags" do
-      input = StringIO.new """
-      <root>
-        <remove/>
-        <remove />
-        <remove/ >
-        <remove / >
-      </root>
-      """
+      input = StringIO.new """<root>
+  <remove/>
+  <remove />
+  <remove/ >
+  <remove / >
+</root>
+"""
       output = StringIO.new
       tags_to_remove = ['remove']
 
       TagRemover.process input, output, remove_tags: tags_to_remove
 
-      expect(output.string).to eq """
-      <root>
-      </root>
-      """
+      expect(output.string).to eq """<root>
+</root>
+"""
     end
 
     it "removes tags with attributes" do
-      input = StringIO.new '''
-      <root>
-        <remove x="y" a="b" >
-        </remove>
-        <remove x="y" a="b" />
-      </root>
-      '''
+      input = StringIO.new '''<root>
+  <remove x="y" a="b" >
+  </remove>
+  <remove x="y" a="b" />
+</root>
+'''
+      output = StringIO.new
+      tags_to_remove = ['remove']
+
+      TagRemover.process input, output, remove_tags: tags_to_remove
+
+      expect(output.string).to eq """<root>
+</root>
+"""
+    end
+
+    it "deals with multiple tags on one line" do
+      input = StringIO.new '''<root><remove>
+  </remove>
+  <remove/><keep>
+  </keep>
+  <remove>Bad Stuff!
+  </remove>Happy Stuff :)<keep/>
+</root>
+'''
       output = StringIO.new
       tags_to_remove = ['remove']
 
       TagRemover.process input, output, remove_tags: tags_to_remove
 
-      expect(output.string).to eq """
-      <root>
-      </root>
-      """
+      expect(output.string).to eq """<root>
+<keep>
+</keep>
+Happy Stuff :)
+<keep/>
+</root>
+"""
     end
 
     it "keeps elements" do
-      input = StringIO.new """
-      <root>
-        <keep>
-        </keep>
-      </root>
-      """
+      input = StringIO.new """<root>
+<keep>
+</keep>
+</root>
+"""
       tags_to_remove = ['remove']
 
       output = StringIO.new
@@ -79,10 +96,9 @@ describe TagRemover do
     end
 
     it "is ok with doing nothing" do
-      input = StringIO.new """
-      <root>
-      </root>
-      """
+      input = StringIO.new """<root>
+</root>
+"""
       output = StringIO.new
 
       TagRemover.process input, output
@@ -91,23 +107,21 @@ describe TagRemover do
     end
 
     it "removes nested tags" do
-      input = StringIO.new """
-      <root>
-        <remove>
-          <remove>
-          </remove>
-        </remove>
-      </root>
-      """
+      input = StringIO.new """<root>
+  <remove>
+    <remove>
+    </remove>
+  </remove>
+</root>
+"""
       tags_to_remove = ['remove']
       output = StringIO.new
 
       TagRemover.process input, output, remove_tags: tags_to_remove
 
-      expect(output.string).to eq """
-      <root>
-      </root>
-      """
+      expect(output.string).to eq """<root>
+</root>
+"""
     end
 
     it "closes the streams" do

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: tag_remover
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Daniel Smith
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-10-27 00:00:00.000000000 Z
+date: 2014-10-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -72,7 +72,7 @@ files:
 - lib/tag_remover.rb
 - lib/tag_remover/version.rb
 - spec/spec_helper.rb
-- spec/tag_helper_spec.rb
+- spec/tag_remover_spec.rb
 - tag_remover.gemspec
 homepage: https://github.com/jellymann/tag_remover
 licenses:
@@ -100,4 +100,4 @@ specification_version: 4
 summary: Remove elements from large XML documents.
 test_files:
 - spec/spec_helper.rb
-- spec/tag_helper_spec.rb
+- spec/tag_remover_spec.rb