ṛʰ romanisation rdh -> rh (hi NAT) / dh (all other NAT and PSA)

PiperOrigin-RevId: 673463527
google-research · Sep 11, 2024 · 1493ff3 · 1493ff3
1 parent 0069fc4
commit 1493ff3
Show file tree

Hide file tree

Showing 6 changed files with 24 additions and 1 deletion.
diff --git a/nisaba/scripts/natural_translit/brahmic/romanizer.py b/nisaba/scripts/natural_translit/brahmic/romanizer.py
@@ -108,6 +108,8 @@ def _transliterate_vocalic(
 
 TT_TR = rw.reassign(gr.RR + gr.RR, ph.T + ph.T, tr.T + tr.T + tr.R)
 RD_R = rw.reassign(gr.RD, ph.RD, tr.R)
+RDH_DH = rw.reassign(gr.RDH, ph.RD + ph.ASP, tr.D + tr.H)
+RDH_RH = rw.reassign(gr.RDH, ph.RD + ph.ASP, tr.R + tr.H)
 
 NY_N = rw.rewrite(ph.NY, tr.N, following=ph.CONSONANT)
 NY_GN = rw.reassign(gr.NY, ph.NY, tr.G + tr.N, ph.VOWEL, ph.VOWEL)
@@ -157,7 +159,7 @@ def _transliterate_vocalic(
 
 
 # Compose common rules for romanization
-TXN_TO_PSA_COMMON = (DIPHTHONG_GR @ NON_LABIAL_ANUSVARA)
+TXN_TO_PSA_COMMON = (DIPHTHONG_GR @ NON_LABIAL_ANUSVARA @ RDH_DH)
 
 # Convert txn to PSAF and outputs only translit strings.
 TXN_TO_PSAF = (

diff --git a/nisaba/scripts/natural_translit/language_params/hi.py b/nisaba/scripts/natural_translit/language_params/hi.py
@@ -110,6 +110,7 @@ def iso_to_nat() -> fl.FstList:
       _TXN_OPS,
       romanizer.SIBV_TO_SIBW,
       romanizer.AA_WI,
+      romanizer.RDH_RH,
       romanizer.TXN_TO_PSA_COMMON,
       romanizer.IGNORE_LONG,
       romanizer.TRANSLIT_BY_PSA,

diff --git a/nisaba/scripts/natural_translit/romanization/testdata/hi_iso_nat.textproto b/nisaba/scripts/natural_translit/romanization/testdata/hi_iso_nat.textproto
@@ -75,3 +75,8 @@ rewrite {
   input: "akailā"
   output: "akaila"
 }
+rewrite {
+  rule: "ISO_TO_NAT"
+  input: "bāṛʰa"
+  output: "barh"
+}
diff --git a/nisaba/scripts/natural_translit/romanization/testdata/hi_iso_psac.textproto b/nisaba/scripts/natural_translit/romanization/testdata/hi_iso_psac.textproto
@@ -74,3 +74,8 @@ rewrite {
   input: "akailā"
   output: "akaila"
 }
+rewrite {
+  rule: "ISO_TO_PSAC"
+  input: "bāṛʰa"
+  output: "badh"
+}
diff --git a/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_nat.textproto b/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_nat.textproto
@@ -60,3 +60,8 @@ rewrite {
   input: "siddʰēgavhāṇa"
   output: "siddhegavhan"
 }
+rewrite {
+  rule: "ISO_TO_NAT"
+  input: "māṛʰā"
+  output: "madha"
+}
diff --git a/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_psac.textproto b/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_psac.textproto
@@ -60,3 +60,8 @@ rewrite {
   input: "siddʰēgavhāṇa"
   output: "sidhegavhan"
 }
+rewrite {
+  rule: "ISO_TO_PSAC"
+  input: "māṛʰā"
+  output: "madha"
+}