Century implementations

Signed-off-by: Namrata Gachchi <ngachchi@nvidia.com>

Author: ngachchi

nemo_text_processing/text_normalization/hi/taggers/date.py

@@ -74,6 +74,10 @@ def __init__(self, cardinal: GraphFst):
 
         range_graph = pynini.cross("-", "से")
 
+        # Graph for century
+        century_number = pynini.compose(pynini.closure(NEMO_HI_DIGIT, 1), cardinal.final_graph) + pynini.accep("वीं")
+        century_text = pynutil.insert("text: \"") + century_number + pynutil.insert("\"") + insert_space
+
         graph_dd_mm_yyyy = (
             days_graph + (delete_dash | delete_slash) + months_graph + (delete_dash | delete_slash) + years_graph
         )
@@ -109,6 +113,7 @@ def __init__(self, cardinal: GraphFst):
             | pynutil.add_weight(graph_mm_yyyy, -0.2)
             | pynutil.add_weight(graph_year_suffix, -0.001)
             | pynutil.add_weight(graph_range, -0.005)
+            | pynutil.add_weight(century_text, -0.001)
         )
 
         self.final_graph = final_graph.optimize()

tests/nemo_text_processing/hi/data_text_normalization/test_cases_date.txt

@@ -19,3 +19,6 @@
 २०२४~दो हज़ार चौबीस
 १२० ई. पू.~एक सौ बीस ईसा पूर्व
 २९७-२७२ ई. पू.~दो सौ सत्तानबे से दो सौ बहत्तर ईसा पूर्व
+३२७वीं सदी~तीन सौ सत्ताईसवीं सदी
+१८वीं शताब्दी~अठारहवीं शताब्दी
+१९वीं दशक~उन्नीसवीं दशक