Handles most common French and Spanish formats

kle-roy · kle-roy · commit f27009b78e2c · 2016-03-11T16:31:57.000+01:00
diff --git a/lib/griddler/email_parser.rb b/lib/griddler/email_parser.rb
@@ -84,7 +84,9 @@ def self.regex_split_points
       /^On.*<\n?.*>.*\n?wrote:$/,
       /On.*wrote:/,
       /\*?From:.*$/i,
-      /^[[:space:]]*\d{4}[-\/]\d{1,2}[-\/]\d{1,2}[[:space:]].*[[:space:]]<.*>?$/i
+      /^[[:space:]]*\d{4}[-\/]\d{1,2}[-\/]\d{1,2}[[:space:]].*[[:space:]]<.*>?$/i,
+      /^[[:space:]]*\>?[[:space:]]*Le.*<\n?.*>.*\n?a[[:space:]]?\n?&eacute;crit :$/, # French
+      /^[[:space:]]*\>?[[:space:]]*El.*<\n?.*>.*\n?escribi&oacute;:$/ # Spanish
     ]
   end
 
diff --git a/spec/griddler/email_spec.rb b/spec/griddler/email_spec.rb
@@ -157,6 +157,73 @@
     expect(body_from_email(text: body)).to eq 'Hello.'
   end
 
+  it 'handles french format: "Le [date], [soandso] <email@example.com> a &eacute;crit :"' do
+    body = <<-EOF.strip_heredoc
+      Hello.
+
+      Le 11 mars 2016, at 18:00, Tristan <email@example.com> a &eacute;crit :
+      > Check out this report.
+      >
+      > It's pretty cool.
+      >
+      > Thanks, Tristan
+      >
+    EOF
+
+    expect(body_from_email(text: body)).to eq 'Hello.'
+  end
+
+  it 'handles LONG french format: "Le [date], [soandso] <email@example.com> a &eacute;crit :"' do
+    body = <<-EOF.strip_heredoc
+      Hello.
+
+      Le 11 mars 2016, at 18:00, Tristan With A Really Really Long Name <
+      tristanhasasuperlongemailthatforcesanewline@candidates.welcomekit.co> a
+      &eacute;crit :
+      > Check out this report.
+      >
+      > It's pretty cool.
+      >
+      > Thanks, Tristan
+      >
+    EOF
+
+    expect(body_from_email(text: body)).to eq 'Hello.'
+  end
+
+  it 'handles spanish format: "El [date], [soandso] <email@example.com> escribi&oacute;:"' do
+    body = <<-EOF.strip_heredoc
+      Hello.
+
+      El 11/03/2016 11:34, Pedro P&eacute;rez <email@example.com> escribi&oacute;:
+      > Check out this report.
+      >
+      > It's pretty cool.
+      >
+      > Thanks, Pedro
+      >
+    EOF
+
+    expect(body_from_email(text: body)).to eq 'Hello.'
+  end
+
+  it 'handles LONG spanish format: "El [date], [soandso] <email@example.com> escribi&oacute;:"' do
+    body = <<-EOF.strip_heredoc
+      Hello.
+
+      El 11/03/2016 11:34, Pedro P&eacute;rez <
+      pedrohasasuperlongemailthatforcesanewline@example.com> escribi&oacute;:
+      > Check out this report.
+      >
+      > It's pretty cool.
+      >
+      > Thanks, Pedro
+      >
+    EOF
+
+    expect(body_from_email(text: body)).to eq 'Hello.'
+  end
+
   it 'handles "From: email@email.com" format' do
     body = <<-EOF
       Hello.
