Support underscore in the exporters on titles

Underscores on titles are meant to be replaced by something else on each
exporter (e.g. removed on CSV, or surrounded by \emph on UOC).

Signed-off-by: Miquel Sabaté Solà <msabate@suse.com>

4 files changed, 58 insertions, 4 deletions

diff --git a/lib/csv_exporter.rb b/lib/csv_exporter.rb
index a44815d..0463ba6 100644
--- a/lib/csv_exporter.rb
+++ b/lib/csv_exporter.rb
@@ -12,10 +12,18 @@ class CsvExporter < BaseExporter
     str
   end
 
+  protected
+
+  # Generate a new CSV line from the given `thing`.
   def thing_to_csv(thing:)
-    CSV.generate_line([thing.target, thing.authors, thing.editors, thing.year, thing.title,
-                       thing.note, thing.insideof, thing.url, thing.publisher, thing.kind,
-                       thing.access, thing.bought_at],
+    CSV.generate_line([thing.target, thing.authors, thing.editors, thing.year,
+                       clean_title(title: thing.title), thing.note, thing.insideof,
+                       thing.url, thing.publisher, thing.kind, thing.access, thing.bought_at],
                       quote_empty: false)
   end
+
+  # Clean the given title from underscores and other such characters.
+  def clean_title(title:)
+    title.delete('_')
+  end
 end
diff --git a/lib/uoc_exporter.rb b/lib/uoc_exporter.rb
index 9a83d7c..e8dcc62 100644
--- a/lib/uoc_exporter.rb
+++ b/lib/uoc_exporter.rb
@@ -16,7 +16,7 @@ class UocExporter < BaseExporter
   # 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}). \\emph{#{thing.title}}."
+    str = "#{parse_authors(thing:)} (#{thing.year}). #{parse_title(title: thing.title)}."
 
     if thing.insideof.present?
       str += " #{inside_of(thing.insideof)}"
@@ -26,6 +26,25 @@ class UocExporter < BaseExporter
     str
   end
 
+  # Returns the title without underscores and with the expected formatting.
+  def parse_title(title:)
+    str = ''
+    seen = false
+
+    # I'm sure there is a clever (and comboluted) regexp that can also achieve
+    # this, but after some tries I decided to cut the crap.
+    title.each_char do |c|
+      if c == '_'
+        str += seen ? '\emph{' : '}'
+        seen = !seen
+      else
+        str += c
+      end
+    end
+
+    "\\emph{#{str}}"
+  end
+
   # Parse the authors as delivered by `thing.authors` and render them in the
   # proper format.
   def parse_authors(thing:)
diff --git a/test/controllers/exports_controller_test.rb b/test/controllers/exports_controller_test.rb
index 9e69df8..eefff6f 100644
--- a/test/controllers/exports_controller_test.rb
+++ b/test/controllers/exports_controller_test.rb
@@ -15,10 +15,28 @@ class ExportsControllerTest < ActionDispatch::IntegrationTest
     assert_equal 'text/csv', @response.media_type
   end
 
+  test 'csv: underscore is removed for the title' do
+    things(:thing1).update!(title: 'This is a _title_ with _many_ underscores')
+
+    get search_exports_url(0, format: 'csv')
+
+    body = @response.body.split("\n")
+
+    assert_equal 'This is a title with many underscores', body.last.split(',')[7]
+  end
+
   test 'uoc: works' do
     get search_exports_url(0, format: 'uoc')
 
     assert_equal @response.body, file_fixture('all.uoc.tex').read
     assert_equal 'application/x-tex', @response.media_type
   end
+
+  test 'uoc: underscore is supported for the title' do
+    things(:thing1).update!(title: 'This is a _title_ with _many_ underscores')
+
+    get search_exports_url(0, format: 'uoc')
+
+    assert_equal @response.body, file_fixture('underscores.uoc.tex').read
+  end
 end
diff --git a/test/fixtures/files/underscores.uoc.tex b/test/fixtures/files/underscores.uoc.tex
new file mode 100644
index 0000000..0c86890
--- /dev/null
+++ b/test/fixtures/files/underscores.uoc.tex
@@ -0,0 +1,9 @@
+\chapter{Bibliografia}
+
+{\setlength{\parskip}{-0.3cm}
+
+\hypertarget{target2}{\textsc{Sabaté}, Miquel; \textsc{Nom} (eds.) (2024). \emph{This is a 2}. Dins de \emph{Some other thing}, 22-24. This is a note 2. Publisher 2, Address 2: 2024.}
+
+\\~\\\hypertarget{target1}{\textsc{Sabaté}, Miquel; \textsc{Smith}, John; \textsc{cognom2}, Nom cognom1; \textsc{cognom3 cognom4}, Nom (2024). \emph{This is a }title\emph{ with }many\emph{ underscores}. Dins de \emph{Some other thing}, 22-24. This is a note 1. Publisher 1. [\url{http://example.com}]. [Última data de consulta: 24 de febrer de 2024].}
+
+}