Fix date ranges

Author: ameliereymond

nemo_text_processing/text_normalization/fr/taggers/date.py

@@ -76,10 +76,16 @@ def __init__(self, cardinal: GraphFst, deterministic: bool = True):
 
         # Accepts date ranges, "17-18-19 juin"  -> date { day: "17" day: "18": day: "19"}
         for separator in ["-", "/"]: 
+            day_range_graph = (
+                pynutil.insert("day: \"")
+                + pynini.closure(digit_to_day + pynutil.delete(separator) + pynutil.insert(" "), 1)
+                + digit_to_day
+                + pynutil.insert("\"")
+            )
+
             self.fst |= (
                 pynutil.insert("date { ")
-                + pynini.closure(self.day_graph + pynutil.delete(separator) + pynutil.insert(" "), 1)
-                + self.day_graph
+                + day_range_graph
                 + pynini.accep(" ")
                 + month_name_graph
                 + pynini.closure(pynini.accep(" ") + self.year_graph, 0, 1)
@@ -92,7 +98,8 @@ def __init__(self, cardinal: GraphFst, deterministic: bool = True):
 
 def apply_fst(text, fst):
     try:
-        print(text, "-->", pynini.shortestpath(text @ fst).string())
+        output = pynini.shortestpath(text @ fst).string()
+        print(f"'{text}' --> '{output}'")
     except pynini.FstOpError:
         print(f"Error: No valid output with given input: '{text}'")
 

nemo_text_processing/text_normalization/fr/verbalizers/date.py

@@ -41,7 +41,7 @@ def __init__(self, deterministic: bool = True):
         year = pynutil.delete("year: \"") + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete("\"")
         decade = pynutil.delete("decade: \"") + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete("\"")
 
-        graph_dmy = pynini.closure(day + NEMO_SPACE, 1) + month + pynini.closure(NEMO_SPACE + year, 0, 1) + delete_preserve_order
+        graph_dmy = day + NEMO_SPACE + month + pynini.closure(NEMO_SPACE + year, 0, 1) + delete_preserve_order
         graph_my = month + NEMO_SPACE + year + delete_preserve_order
         graph_decade = decade + delete_preserve_order
 
@@ -50,15 +50,15 @@ def __init__(self, deterministic: bool = True):
         delete_tokens = self.delete_tokens(self.graph)
         self.fst = delete_tokens.optimize()
 
-
 def apply_fst(text, fst):
     try:
-        print(text, "-->", pynini.shortestpath(text @ fst).string())
+        output = pynini.shortestpath(text @ fst).string()
+        print(f"'{text}' --> '{output}'")
     except pynini.FstOpError:
         print(f"Error: No valid output with given input: '{text}'")
 
 if __name__ == "__main__":
     fst = DateFst()
 
     # tagger output for "les 17/18/19 juin"
-    apply_fst('date { day: "les dix-sept" day: "dix-huit" day: "dix-neuf" month: "juin" preserve_order: true }', fst.fst)
+    apply_fst('date { day: "les dix-sept dix-huit dix-neuf" month: "juin" preserve_order: true }', fst.fst)

tests/nemo_text_processing/fr/data_text_normalization/test_cases_date.txt

@@ -8,6 +8,6 @@ le 02.03.2003~le deux mars deux mille trois
 le 10 mars 2023~le dix mars deux mille vingt-trois
 les 80s~les eighties
 les 17/18 juin~les dix-sept dix-huit juin
-les 17/18/19 mars~les ldix-sept dix-huit dix-neuf mars
+les 17/18/19 mars~les dix-sept dix-huit dix-neuf mars
 les 17-18-19 juin~les dix-sept dix-huit dix-neuf juin
 les 17-18-19 juin 2025~les dix-sept dix-huit dix-neuf juin deux mille vingt-cinq