From ff11c2c0091d38135d2c2d7a2e124d4f7536523e Mon Sep 17 00:00:00 2001
From: Camila Maia <cmaiacd@gmail.com>
Date: Sat, 1 Apr 2023 13:26:13 +0200
Subject: [PATCH] feat: add parse_cpf

---
 README.md           |  3 ++-
 README_PT_BR.md     |  3 ++-
 brutils/__init__.py |  2 +-
 brutils/cpf.py      | 11 +++++++++++
 tests/test_cpf.py   |  9 +++++++++
 5 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 62563a81..34ab7901 100644
--- a/README.md
+++ b/README.md
@@ -39,7 +39,8 @@ False
 
 ### What if my string has formatting symbols in it?
 ```
->>> cpf.sieve('000.111.222-33')
+>>> from brutils import parse_cpf
+>>> parse_cpf('000.111.222-33')
 '00011122233'
 >>> cnpj.sieve('00.111.222/0001-00')
 '00111222000100'
diff --git a/README_PT_BR.md b/README_PT_BR.md
index 5b852f30..dead564c 100644
--- a/README_PT_BR.md
+++ b/README_PT_BR.md
@@ -41,7 +41,8 @@ False
 
 ### E se a minha string estiver formatada com simbolos?
 ```
->>> cpf.sieve('000.111.222-33')
+>>> from brutils import parse_cpf
+>>> parse_cpf('000.111.222-33')
 '00011122233'
 >>> cnpj.sieve('00.111.222/0001-00')
 '00111222000100'
diff --git a/brutils/__init__.py b/brutils/__init__.py
index b6f839ae..6df05236 100644
--- a/brutils/__init__.py
+++ b/brutils/__init__.py
@@ -1,2 +1,2 @@
-from brutils.cpf import is_valid as is_valid_cpf, format_cpf
+from brutils.cpf import is_valid as is_valid_cpf, format_cpf, parse as parse_cpf
 from brutils.cnpj import is_valid as is_valid_cnpj, format_cnpj
diff --git a/brutils/cpf.py b/brutils/cpf.py
index 5013f8c5..20df3374 100644
--- a/brutils/cpf.py
+++ b/brutils/cpf.py
@@ -16,6 +16,17 @@ def sieve(dirty):  # type: (str) -> str
     return "".join(filter(lambda char: char not in ".-", dirty))
 
 
+def parse(dirty):  # type: (str) -> str
+    """
+    Filters out CPF formatting symbols. Symbols that are not used
+    in the CPF formatting are left unfiltered on purpose so that
+    if fails other tests, because their presence indicate that the
+    input was somehow corrupted.
+    """
+
+    return sieve(dirty)
+
+
 def display(cpf):  # type: (str) -> str
     """
     Will format an adequately formatted numbers-only CPF string,
diff --git a/tests/test_cpf.py b/tests/test_cpf.py
index ddb53e7a..03ed793a 100644
--- a/tests/test_cpf.py
+++ b/tests/test_cpf.py
@@ -17,6 +17,7 @@
     generate,
     is_valid,
     format_cpf,
+    parse,
 )
 from unittest import TestCase, main
 
@@ -30,6 +31,14 @@ def test_sieve(self):
         assert sieve("ab.c1.--.2-309.-1-.6-.*.-!*&#") == "abc1230916*!*&#"
         assert sieve("...---...") == ""
 
+    def test_parse(self):
+        assert parse("00000000000") == "00000000000"
+        assert parse("123.456.789-10") == "12345678910"
+        assert parse("134..2435.-1892.-") == "13424351892"
+        assert parse("abc1230916*!*&#") == "abc1230916*!*&#"
+        assert parse("ab.c1.--.2-309.-1-.6-.*.-!*&#") == "abc1230916*!*&#"
+        assert parse("...---...") == ""
+
     def test_display(self):
         assert display("00000000011") == "000.000.000-11"
         assert display("00000000000") is None