Added pre-processing

1 files changed, 72 insertions, 0 deletions

diff --git a/src/lib/prononciation.mly b/src/lib/prononciation.mly
new file mode 100644
index 0000000..f865abd
--- /dev/null
+++ b/src/lib/prononciation.mly
@@ -0,0 +1,72 @@
+%{
+
+    (** This module transform the words into a semi-prononciation elements. 
+
+    The letters should be quite close with the associate phonem, but some
+    modification can occur.
+
+     *)
+
+    open Tokens
+%}
+
+%start<Tokens.token list> main
+%%
+
+initial_voyel:
+  | A { A }
+  | E { E }
+  | I { I }
+  | O { O }
+  | U { U }
+  | E_ACUTE { E_ACUTE }
+
+voyel: 
+  | initial_voyel { $1 }
+  
+
+letters:
+  |                 { [] }
+  | letters voyel  { $2 :: $1 }
+  | letters Space   { Space :: $1 }
+  | letters Sep     { Sep :: $1 }
+
+  | letters B       { B :: $1 }
+  | letters C  %prec Low     { K :: $1 }
+  | letters C H     { X :: $1 }
+  | letters C I     { I :: S :: $1 }
+  | letters C E     { E :: S :: $1 }
+  | letters C U I   { I :: K :: $1 }
+  | letters C U E   { E :: K :: $1 }
+  | letters D       { D :: $1 }
+  | letters F       { F :: $1 }
+  | letters G %prec Low      { G :: $1 }
+  | letters G I     { I :: J :: $1 }
+  | letters G E     { E :: J :: $1 }
+  | letters G U I   { I :: G :: $1 }
+  | letters G U E   { E :: G :: $1 }
+
+  | letters J       { J :: $1 }
+  | letters K       { K :: $1 }
+  | letters L       { L :: $1 }
+  | letters M       { M :: $1 }
+  | letters N       { N :: $1 }
+  | letters P       { P :: $1 }
+  | letters P H     { F :: $1 }
+
+  | letters Q U     { K :: $1 }
+
+  | letters R       { R :: $1 }
+  | letters S S     { S :: $1 }
+  | letters S       { SZ :: $1 }
+  | letters T       { T :: $1 }
+
+  | letters V       { V :: $1 }
+  | letters W       { V :: $1 }
+  | letters X       { S :: K :: $1 }
+  | letters Y       { I :: $1 }
+
+  | letters Z       { Z :: $1 }
+
+main: 
+    letters EOL { EOL::$1 }