lib: Introduce the PlainExporter

This class exports things into plain text. It follows a close format to
that of the one for UOC, but without any formatting.

Signed-off-by: Miquel Sabaté Solà <mikisabate@gmail.com>

7 files changed, 117 insertions, 33 deletions

diff --git a/app/controllers/exports_controller.rb b/app/controllers/exports_controller.rb
index 4303ea6..812cf39 100644
--- a/app/controllers/exports_controller.rb
+++ b/app/controllers/exports_controller.rb
@@ -8,6 +8,12 @@ class ExportsController < ApplicationController
     respond_to do |format|
       format.csv { set_file_name!(ext: 'csv') }
       format.uoc { set_file_name!(ext: 'tex') }
+      format.text do
+        send_data PlainExporter.new(things: @results).export,
+                  type:        'text/plain',
+                  disposition: 'attachment',
+                  filename:    "#{@search.name}.txt"
+      end
     end
   end
 
diff --git a/app/views/exports/new.html.erb b/app/views/exports/new.html.erb
index 631da20..fa52360 100644
--- a/app/views/exports/new.html.erb
+++ b/app/views/exports/new.html.erb
@@ -11,6 +11,7 @@
             <select id="export-select-format" data-action="export#change">
                 <option selected="selected" value="csv">CSV</option>
                 <option value="uoc">UOC</option>
+                <option value="txt">Plain text</option>
             </select>
 
             <a id="export-format-button" class="button" href="#"><%= I18n.t('searches.export.action').capitalize %></a>
diff --git a/lib/base_exporter.rb b/lib/base_exporter.rb
index b040592..f64301b 100644
--- a/lib/base_exporter.rb
+++ b/lib/base_exporter.rb
@@ -5,6 +5,8 @@ class BaseExporter
     @things = things
   end
 
+  protected
+
   # Returns a string which forces a new TeX line.
   def tex_new_line
     '\\\\~\\\\'
@@ -26,4 +28,44 @@ class BaseExporter
 
     [a[0..-2].join(' '), a.last]
   end
+
+  # Parse the string in `thing.authors` and put each author followed by
+  # semicolon and the right order for first and last name. It will also cover
+  # whenever editors are involved.
+  def parse_authors(thing:)
+    str = thing.authors.split(',').map do |author|
+      first, last = parse_author(author: author.strip)
+      last.blank? ? upper_case(text: first).to_s : "#{upper_case(text: last)}, #{first}"
+    end.join('; ')
+
+    str += ' (eds.)' if thing.editors
+    str
+  end
+
+  # Returns an "Inside of" string where `sub` is the substring contained in it.
+  def inside_of(sub:)
+    str = if sub.downcase.start_with?(*%w[a e i o u])
+            I18n.t('general.insideof-vowel')
+          else
+            I18n.t('general.insideof')
+          end
+
+    str + " #{emph(text: sub)}"
+  end
+
+  # Returns the head of a formatted "thing", containing authors, year, title
+  # and, if available, the "inside of" information. This is shared across some
+  # exporters, thus placed in here.
+  #
+  # NOTE: the subclass *must* implement `parse_title`.
+  def head(thing:)
+    str = "#{parse_authors(thing:)} (#{thing.year}). #{parse_title(title: thing.title)}."
+
+    if thing.insideof.present?
+      str += " #{inside_of(sub: thing.insideof)}"
+      str += thing.pages.present? ? ", #{thing.pages}." : '.'
+    end
+
+    str
+  end
 end
diff --git a/lib/plain_exporter.rb b/lib/plain_exporter.rb
new file mode 100644
index 0000000..514d243
--- /dev/null
+++ b/lib/plain_exporter.rb
@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+# Plain text exporter.
+class PlainExporter < BaseExporter
+  def export
+    @things.map { |thing| "#{head(thing:)} #{body(thing:)}".strip }.join("\n")
+  end
+
+  protected
+
+  # Returns the same text as the plain exporter does not have formatting.
+  def emph(text:)
+    text
+  end
+
+  # Returns the same text as the plain exporter does not have formatting.
+  def upper_case(text:)
+    text
+  end
+
+  # Returns the same text with underscores stripped, as the plain exporter does
+  # not have formatting.
+  def parse_title(title:)
+    title.delete('_')
+  end
+
+  # Returns text with information on the publisher, notes, year, address, etc;
+  # if available.
+  def body(thing:)
+    str = ''
+
+    if thing.publisher.present?
+      str += thing.note.present? ? "#{thing.note}. #{thing.publisher}" : thing.publisher
+    end
+
+    if thing.address.present?
+      str + ", #{thing.address}: #{thing.year}."
+    else
+      "#{str}."
+    end
+  end
+end
diff --git a/lib/uoc_exporter.rb b/lib/uoc_exporter.rb
index e8dcc62..9deffc5 100644
--- a/lib/uoc_exporter.rb
+++ b/lib/uoc_exporter.rb
@@ -13,17 +13,14 @@ class UocExporter < BaseExporter
 
   protected
 
-  # Returns the head of a line, containing authors, year, title and, if
-  # available, the "inside of" information.
-  def head(thing:)
-    str = "#{parse_authors(thing:)} (#{thing.year}). #{parse_title(title: thing.title)}."
-
-    if thing.insideof.present?
-      str += " #{inside_of(thing.insideof)}"
-      str += thing.pages.present? ? ", #{thing.pages}." : '.'
-    end
+  # Returns the same text with LaTeX \emph applied.
+  def emph(text:)
+    "\\emph{#{text}}"
+  end
 
-    str
+  # Returns the same text with LaTeX \textsc applied.
+  def upper_case(text:)
+    "\\textsc{#{text}}"
   end
 
   # Returns the title without underscores and with the expected formatting.
@@ -45,29 +42,6 @@ class UocExporter < BaseExporter
     "\\emph{#{str}}"
   end
 
-  # Parse the authors as delivered by `thing.authors` and render them in the
-  # proper format.
-  def parse_authors(thing:)
-    str = thing.authors.split(',').map do |author|
-      first, last = parse_author(author: author.strip)
-      last.blank? ? "\\textsc{#{first}}" : "\\textsc{#{last}}, #{first}"
-    end.join('; ')
-
-    str += ' (eds.)' if thing.editors
-    str
-  end
-
-  # Returns the proper 'inside of' translated text.
-  def inside_of(sub)
-    str = if sub.downcase.start_with?(*%w[a e i o u])
-            I18n.t('general.insideof-vowel')
-          else
-            I18n.t('general.insideof')
-          end
-
-    str + " \\emph{#{sub}}"
-  end
-
   # Returns the 'body' of the line, which is the rest of it (i.e. note,
   # publisher, year).
   def body(thing:)
diff --git a/test/controllers/exports_controller_test.rb b/test/controllers/exports_controller_test.rb
index fe94362..0b0ba1d 100644
--- a/test/controllers/exports_controller_test.rb
+++ b/test/controllers/exports_controller_test.rb
@@ -42,4 +42,21 @@ class ExportsControllerTest < ActionDispatch::IntegrationTest
 
     assert_equal @response.body, file_fixture('underscores.uoc.tex').read
   end
+
+  test 'plain: works' do
+    get search_exports_url(0, format: 'txt')
+
+    assert_equal @response.body, file_fixture('all.txt').read.strip
+    assert_equal 'text/plain', @response.media_type
+  end
+
+  test 'plain: underscore is removed for the title' do
+    things(:thing1).update!(title: 'This is a _title_ with _many_ underscores')
+
+    get search_exports_url(0, format: 'txt')
+
+    body = @response.body.split("\n")
+
+    assert_includes body.last, 'This is a title with many underscores'
+  end
 end
diff --git a/test/fixtures/files/all.txt b/test/fixtures/files/all.txt
new file mode 100644
index 0000000..49205a9
--- /dev/null
+++ b/test/fixtures/files/all.txt
@@ -0,0 +1,2 @@
+Sabaté, Miquel; Nom (eds.) (2024). This is a 2. Dins de Some other thing, 22-24. This is a note 2. Publisher 2, Address 2: 2024.
+Sabaté, Miquel; Smith, John; cognom2, Nom cognom1; cognom3 cognom4, Nom (2024). This is a title 1. Dins de Some other thing, 22-24. This is a note 1. Publisher 1.