1 files changed, 57 insertions, 0 deletions
diff --git a/scripts/numspell.py b/scripts/numspell.py
new file mode 100644
index 0000000..2e416e3
--- /dev/null
+++ b/scripts/numspell.py
@@ -0,0 +1,57 @@
+# -*- coding: utf-8 -*-
+"""Pasa los numeros de una transcripcion a palabras, para poder compararla
+con el guion (que escribe todo en letras)."""
+import re
+
+U = ["cero", "uno", "dos", "tres", "cuatro", "cinco", "seis", "siete", "ocho", "nueve",
+     "diez", "once", "doce", "trece", "catorce", "quince", "dieciseis", "diecisiete",
+     "dieciocho", "diecinueve", "veinte", "veintiuno", "veintidos", "veintitres",
+     "veinticuatro", "veinticinco", "veintiseis", "veintisiete", "veintiocho", "veintinueve"]
+D = {3: "treinta", 4: "cuarenta", 5: "cincuenta", 6: "sesenta",
+     7: "setenta", 8: "ochenta", 9: "noventa"}
+C = {1: "ciento", 2: "doscientos", 3: "trescientos", 4: "cuatrocientos", 5: "quinientos",
+     6: "seiscientos", 7: "setecientos", 8: "ochocientos", 9: "novecientos"}
+
+
+def _lt1000(n):
+    if n < 30:
+        return U[n]
+    if n < 100:
+        d, u = divmod(n, 10)
+        return D[d] + (" y " + U[u] if u else "")
+    if n == 100:
+        return "cien"
+    c, r = divmod(n, 100)
+    return C[c] + (" " + _lt1000(r) if r else "")
+
+
+def spell_int(n):
+    if n < 1000:
+        return _lt1000(n)
+    if n < 1000000:
+        m, r = divmod(n, 1000)
+        pre = "mil" if m == 1 else _lt1000(m) + " mil"
+        return pre + (" " + _lt1000(r) if r else "")
+    m, r = divmod(n, 1000000)
+    pre = "un millon" if m == 1 else _lt1000(m) + " millones"
+    return pre + (" " + spell_int(r) if r else "")
+
+
+def _tok(m):
+    s = m.group(0)
+    s = re.sub(r"\.(?=\d{3}\b)", "", s)          # 384.000 -> 384000
+    if "," in s:
+        a, b = s.split(",", 1)
+        ent = spell_int(int(a or 0))
+        dec = " ".join(U[int(c)] for c in b if c.isdigit())
+        return f"{ent} coma {dec}"
+    if "." in s:                                  # 0.5 estilo ingles
+        a, b = s.split(".", 1)
+        return f"{spell_int(int(a or 0))} coma " + " ".join(U[int(c)] for c in b if c.isdigit())
+    return spell_int(int(s))
+
+
+def expand(text):
+    t = text.replace("%", " por ciento ").replace("º", "")
+    t = re.sub(r"\d+(?:[.,]\d+)*", _tok, t)
+    return re.sub(r"\s+", " ", t).strip()