chupa-text 1.3.4 → 1.3.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d81031d80654a357700972d1443aea5994ba4e3f4a1684e670a19b7041b12073
-  data.tar.gz: 17d66e6f6660066fd8c4fd83baf01eee8c2296a2a7a777ab7c0db3bbf9917d56
+  metadata.gz: 111930c1ba73f6eaae79fb538e34a51e72fd29768bca71ec89d43f041c34d960
+  data.tar.gz: 4051f7bc52f057a2e06cd7c7d23d3d8ce1755aae47b2352cc8d9e9e2dddfdfde
 SHA512:
-  metadata.gz: e3ab55e7923da653b4c277491712e6564db633b664830ad2153b60b0ada7d09de69576074a0a3592406b03041f09990fa7a0daf0fd3f99d798543d22d5416a2c
-  data.tar.gz: ab8e04142b49119e51c9d7da40776efc674e75034148670c7f921a06320896426f7aba99a42ff467a97d07bd89c776bc4e04ddfa788ce826d51481baf8890470
+  metadata.gz: 9ef8a9c17bf65b1d1a7e26112390964738985c3c9b761dc37c157292a324045eba7565576523817fe738f6158aa1d0362260cf13d28360baec2aa8e9942f2fdc
+  data.tar.gz: 75f63640af7ad7a55d0925dcd37ee23e7bd64c83291d80f0796d6e18fff86f2b76340170c7fa1be572070a4bdee9502a8aa9dfbf7e9946fb4dbd06e6fdb1edbf

data/Rakefile

@@ -1,6 +1,6 @@
-# -*- mode: ruby; coding: utf-8 -*-
+# -*- ruby -*-
 #
-# Copyright (C) 2013  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2013-2025  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -46,3 +46,13 @@ desc "Run tests"
 task :test do
   ruby("test/run-test.rb")
 end
+
+release_task = Rake.application["release"]
+# We use Trusted Publishing.
+release_task.prerequisites.delete("build")
+release_task.prerequisites.delete("release:rubygem_push")
+release_task_comment = release_task.comment
+if release_task_comment
+  release_task.clear_comments
+  release_task.comment = release_task_comment.gsub(/ and build.*$/, "")
+end

data/doc/text/news.md

@@ -1,5 +1,23 @@
 # News
 
+## 1.3.6: 2025-01-10
+
+### Fixes
+
+  * xlsx: Fixed a bug that wrong text is extracted with complex shared
+    strings.
+    * Reported by Tomohisa Kusukawa.
+
+### Thanks
+
+  * Tomohisa Kusukawa
+
+## 1.3.5: 2024-09-22
+
+### Improvements
+
+  * Added support for REXML 3.3.2 or later.
+
 ## 1.3.4: 2024-09-22
 
 ### Improvements

data/lib/chupa-text/decomposers/csv.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2013-2017  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2013-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -36,7 +36,7 @@ module ChupaText
       end
 
       def decompose(data)
-        text = ""
+        text = +""
         data.open do |input|
           begin
             csv = ::CSV.new(input, liberal_parsing: true)
@@ -68,7 +68,7 @@ module ChupaText
         width, height = data.expected_screenshot_size
         max_n_lines = 10
         font_size = height / max_n_lines
-        target_text = ""
+        target_text = +""
         text.each_line.with_index do |line, i|
           break if i == max_n_lines
           target_text << line

data/lib/chupa-text/decomposers/office-open-xml-document.rb

@@ -41,7 +41,7 @@ module ChupaText
 
       private
       def start_decompose(context)
-        context[:text] = ""
+        context[:text] = +""
       end
 
       def process_entry(entry, context)

data/lib/chupa-text/decomposers/office-open-xml-presentation.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -56,7 +56,7 @@ module ChupaText
         case entry.zip_path
         when /\Appt\/slides\/slide(\d+)\.xml/
           nth_slide = Integer($1, 10)
-          slide_text = ""
+          slide_text = +""
           extract_text(entry, slide_text)
           context[:slides] << [nth_slide, slide_text]
         end

data/lib/chupa-text/decomposers/office-open-xml-workbook.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019-2022  Sutou Kouhei <kou@clear-code.com>
+# Copyright (C) 2019-2025  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -49,7 +49,7 @@ module ChupaText
       def process_entry(entry, context)
         case entry.zip_path
         when "xl/sharedStrings.xml"
-          extract_text(entry, context[:shared_strings])
+          extract_shared_strings(entry, context[:shared_strings])
         when "xl/workbook.xml"
           listener = WorkbookListener.new(context[:sheet_names])
           parse(entry.file_data, listener)
@@ -73,7 +73,7 @@ module ChupaText
         sheets = context[:sheets].sort_by(&:first).collect(&:last)
         sheet_names = context[:sheet_names]
         sheets.each_with_index do |sheet, i|
-          sheet_text = ""
+          sheet_text = +""
           sheet.each do |row|
             row_texts = row.collect do |cell|
               case cell

data/lib/chupa-text/decomposers/office-open-xml.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2025  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -70,6 +70,11 @@ module ChupaText
         parse(entry.file_data, listener)
       end
 
+      def extract_shared_strings(entry, shared_strings)
+        listener = SharedStringsListener.new(shared_strings, @namespace_uri)
+        parse(entry.file_data, listener)
+      end
+
       def log_tag
         "[decomposer][office-open-xml]"
       end
@@ -90,26 +95,74 @@ module ChupaText
         end
 
         def end_element(uri, local_name, qname)
+          if uri == @target_uri
+            case local_name
+            when "p", "br"
+              @output << "\n"
+            when "t"
+              @in_target = false
+            end
+          end
+        end
+
+        def characters(text)
+          add_text(text)
+        end
+
+        def cdata(content)
+          add_text(content)
+        end
+
+        private
+        def add_text(text)
+          return unless @in_target
+          @output << text
+        end
+      end
+
+      class SharedStringsListener < SAXListener
+        def initialize(output, target_uri)
+          @output = output
+          @target_uri = target_uri
+          @tag_stack = []
           @in_target = false
+          @current_text = +""
+        end
+
+        def start_element(uri, local_name, qname, attributes)
+          @tag_stack << local_name
+
+          return unless uri == @target_uri
+          case local_name
+          when "t"
+            @in_target = true
+            @current_text = +""
+          end
+        end
 
+        def end_element(uri, local_name, qname)
           return unless uri == @target_uri
           case local_name
-          when "p", "br"
-            @output << "\n"
+          when "t"
+            add_text(@current_text)
+            @in_target = false
           end
+        ensure
+          @tag_stack.pop
         end
 
         def characters(text)
-          add_text(text)
+          @current_text << text if @in_target
         end
 
         def cdata(content)
-          add_text(content)
+          @current_text << content if @in_target
         end
 
         private
         def add_text(text)
-          return unless @in_target
+          parent_tag = @tag_stack[-2]
+          return unless parent_tag == "si"
           @output << text
         end
       end

data/lib/chupa-text/decomposers/opendocument-presentation.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -72,7 +72,7 @@ module ChupaText
           when DRAW_URI
             case local_name
             when "page"
-              @slides << {text: ""}
+              @slides << {text: +""}
             end
           end
         end

data/lib/chupa-text/decomposers/opendocument-spreadsheet.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -96,9 +96,9 @@ module ChupaText
             when "table-row"
               @sheets.last[:rows] << []
             when "table-cell"
-              @sheets.last[:rows].last << {text: ""}
+              @sheets.last[:rows].last << {text: +""}
             when "covered-table-cell"
-              @sheets.last[:rows].last << {text: ""}
+              @sheets.last[:rows].last << {text: +""}
             when "shapes"
               @in_shapes = true
             end
@@ -116,7 +116,7 @@ module ChupaText
             case local_name
             when "table"
               sheet = @sheets.last
-              text = ""
+              text = +""
               shape_texts = sheet[:shape_texts]
               unless shape_texts.empty?
                 text << shape_texts.join("\n") << "\n"

data/lib/chupa-text/decomposers/opendocument-text.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -29,7 +29,7 @@ module ChupaText
 
       private
       def process_content(entry, context, &block)
-        context[:text] = ""
+        context[:text] = +""
         listener = TextListener.new(context[:text])
         parse(entry.file_data, listener)
       end

data/lib/chupa-text/decomposers/xml.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2013-2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2013-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -29,7 +29,7 @@ module ChupaText
       end
 
       def decompose(data)
-        text = ""
+        text = +""
         listener = Listener.new(text)
         data.open do |input|
           begin

data/lib/chupa-text/sax-parser.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -16,6 +16,7 @@
 
 require "cgi/util"
 require "rexml/parsers/sax2parser"
+require "rexml/rexml"
 require "rexml/sax2listener"
 
 begin
@@ -156,12 +157,22 @@ module ChupaText
           @listener.end_element(*args)
         end
 
-        def characters(text)
-          @listener.characters(CGI.unescapeHTML(text))
-        end
+        if (REXML::VERSION <=> "3.3.2") >= 0
+          def characters(text)
+            @listener.characters(text)
+          end
+
+          def cdata(content)
+            @listener.cdata(content)
+          end
+        else
+          def characters(text)
+            @listener.characters(CGI.unescapeHTML(text))
+          end
 
-        def cdata(content)
-          @listener.cdata(CGI.unescapeHTML(content))
+          def cdata(content)
+            @listener.cdata(CGI.unescapeHTML(content))
+          end
         end
       end
     end

data/lib/chupa-text/version.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2013-2022  Sutou Kouhei <kou@clear-code.com>
+# Copyright (C) 2013-2024  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -15,5 +15,5 @@
 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 
 module ChupaText
-  VERSION = "1.3.4"
+  VERSION = "1.3.6"
 end

data/test/decomposers/test-office-open-xml-workbook.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2019  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2019-2025  Sutou Kouhei <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -153,6 +153,81 @@ class TestDecomposersOfficeOpenXMLWorkbook < Test::Unit::TestCase
       end
     end
 
+    def test_complex_shared_strings
+      path = fixture_path("xlsx", "complex-shared-strings.xlsx")
+      actual = decompose(path).collect do |data|
+        [
+          data["index"],
+          data["name"],
+          data.body,
+        ]
+      end
+      assert_equal([
+                     [nil, nil, ""],
+                     [
+                       0,
+                       "新規",
+                       "No\t案件番号\t開始日\t期日\tステータス\t備考\n" +
+                       "1\t-\t45664\t45672\t対応中\n" +
+                       "2\t-\t45664\t45672\t対応中\n" +
+                       "3\t-\t45664\t45672\t対応中\n" +
+                       "4\t-\t45664\t45666\t対応中\n" +
+                       "5\t-\t45664\t45666\t対応中\n" +
+                       "6\t-\t45663\t45665\t承認待ち\n" +
+                       "7\t-\t45660\t45665\t承認待ち\n" +
+                       "8\t-\t45653\t45663\t承認待ち\n" +
+                       "9\t-\t45653\t45663\t承認待ち\n" +
+                       "10\tPSR2401770\t45652\t45666\t対応中\n",
+                     ],
+                     [
+                       1,
+                       "全体",
+                       "No\t案件番号\t開始日\t期日\tステータス\n" +
+                       "1\tPSR2401564\t45617\t45726\t対応中\n" +
+                       "2\tPSR2401194\t45553\t45716\t対応中\n" +
+                       "3\t-\t45664\t45672\t対応中\n" +
+                       "4\t-\t45664\t45672\t対応中\n" +
+                       "5\t-\t45664\t45672\t対応中\n" +
+                       "6\t-\t45645\t45672\t対応中\n" +
+                       "7\tPSR2401746\t45649\t45671\t対応中\n" +
+                       "8\t-\t45640\t45667\t対応中\n" +
+                       "9\t-\t45635\t45667\t対応中\n" +
+                       "10\tPSR2401605\t45623\t45667\t対応中\n" +
+                       "11\t-\t45664\t45666\t対応中\n" +
+                       "12\t-\t45664\t45666\t対応中\n" +
+                       "13\tPSR2401770\t45652\t45666\t対応中\n" +
+                       "14\t-\t45645\t45665\t対応中\n" +
+                       "15\tPSR2401609\t45624\t45666\t対応中\n",
+                     ],
+                     [
+                       2,
+                       "案件",
+                       "No\t案件番号\t開始日\t対応完了時期想定\n" +
+                       "1\tPSR2401244\t45561.40347222222\t45744\n" +
+                       "2\tPSR2401592\t45621.598611111112\t45698\n" +
+                       "3\tPSR2401682\t45638.40902777778\t45688\n" +
+                       "4\tPSR2401706\t45643.383333333331\t45671\n" +
+                       "5\tPSR2401779\t45653.490277777775\t45671\n" +
+                       "6\tPSR2401805\t45664.436805555553\t調整中\n" +
+                       "7\tPSR2400677\t45455.588194444441\t45651\t完了\n" +
+                       "8\tPSR2401666\t45636.405555555553\t45653\t完了\n" +
+                       "9\tPSR2401714\t45644.630555555559\t45652\t完了\n",
+                     ],
+                     [
+                       3,
+                       "障害恒久対応・改善対応",
+                       "No\t案件番号\t分類\t開始日\t対応完了時期想定\n" +
+                       "1\tPSR2401334\t改善対応\t45576.411805555559\t45688\n" +
+                       "2\tPSR2401335\t改善対応\t45576.415277777778\t45688\n" +
+                       "3\tPSR2401410\t改善対応\t45588.428472222222\t調整中\n" +
+                       "4\tPSR2401411\t改善対応\t45588.432638888888\t調整中\n" +
+                       "5\tPSR2401718\t障害恒久対応\t45645.386111111111\t45695\n" +
+                       "6\tPSR2401807\t障害恒久対応\t45664.546527777777\t調整中\t1/16リリースで調整中\n",
+                     ],
+                   ],
+                   actual)
+    end
+
     sub_test_case("invalid") do
       def test_empty
         messages = capture_log do

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: chupa-text
 version: !ruby/object:Gem::Version
-  version: 1.3.4
+  version: 1.3.6
 platform: ruby
 authors:
 - Sutou Kouhei
 bindir: bin
 cert_chain: []
-date: 2024-09-22 00:00:00.000000000 Z
+date: 2025-01-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: archive-zip
@@ -189,6 +189,7 @@ files:
 - test/fixture/tar/top-level.tar
 - test/fixture/tar/utf-8.tar
 - test/fixture/xlsx/attributes.xlsx
+- test/fixture/xlsx/complex-shared-strings.xlsx
 - test/fixture/xlsx/empty.xlsx
 - test/fixture/xlsx/multi-sheets.xlsx
 - test/fixture/xlsx/not-shared-cell.xlsx
@@ -232,7 +233,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.0.dev
+rubygems_version: 3.6.2
 specification_version: 4
 summary: ChupaText is an extensible text extractor. You can plug your custom text
   extractor in ChupaText. You can write your plugin by Ruby.