Create function check_phone_number in efi parsing

HISTORY.rst

@@ -3,6 +3,12 @@ History
 =======
 
 
+v0.17.35 (2023-02-23)
+
+* Create function check_phone_number in efi parsing.
+* Update bytes read in detect_delimiter function from 4096 to 8192.
+
+
 v0.17.34 (2023-02-16)
 
 * Revise ethnicity check in EFI and increase truncation limit.

aioradio/file_ingestion.py

@@ -418,20 +418,24 @@ def __post_init__(self):
             '0': 'N'
         }
 
-    def check_width(self, value: str, field: str, row_idx: int) -> str:
+    def check_width(self, value: str, field: str, row_idx: int, truncate_backward: bool=False) -> str:
         """Check field value and truncate if it is longer than expected.
 
         Args:
             value (str): Value
             field (str): Column header field value
             row_idx (int): Row index
+            truncate_backward (bool, optional): Truncate from back of value
 
         Returns:
             str: [description]
         """
 
         if len(value) > self.field_to_max_widths[field]:
-            new_value = value[:self.field_to_max_widths[field]].rstrip()
+            if truncate_backward:
+                new_value = value[-self.field_to_max_widths[field]:].lstrip()
+            else:
+                new_value = value[:self.field_to_max_widths[field]].rstrip()
             LOG.warning(f"[{self.filename}] [row:{row_idx}] [{field}] - '{value}' "
                         f"exceeds max width of {self.field_to_max_widths[field]}. Trimming value to {new_value}")
             value = new_value
@@ -456,6 +460,25 @@ def check_name(self, value: str, field: str, row_idx: int) -> str:
 
         return value
 
+    def check_phone_number(self, value: str, field: str, row_idx: int) -> str:
+        """Check Phone Number logic.
+
+        Args:
+            value (str): Phone number value
+            field (str): Column header field value
+            row_idx (int): Row number in file
+
+        Returns:
+            str: Phone number value
+        """
+
+        if value != '':
+            value = value.replace(' ', '')
+            value = "".join(filter(str.isdigit, value))
+            value = self.check_width(value, field, row_idx, truncate_backward=True)
+
+        return value
+
     def check_gender(self, value: str) -> str:
         """Check Gender logic.
 
@@ -1054,6 +1077,18 @@ def check_name_efi(self, records: list[str], field: str, row_idx: int):
         for idx in range(len(records)):
             records[idx] = self.check_name(records[idx], field, row_idx + idx)
 
+    def check_phone_number_efi(self, records: list[str], field: str, row_idx: int):
+        """Check Phone Number logic.
+
+        Args:
+            records (list[str]): List of a specific columns values
+            field (str): Column header field value
+            row_idx (int): Row number in file
+        """
+
+        for idx in range(len(records)):
+            records[idx] = self.check_phone_number(records[idx], field, row_idx + idx)
+
     def check_gender_efi(self, records: list[str]):
         """Check Gender logic.
 
@@ -1821,7 +1856,7 @@ def detect_delimiter(path: str, encoding: str) -> str:
 
     delimiter = ''
     with open(path, newline='', encoding=encoding) as csvfile:
-        data = csvfile.read(4096)
+        data = csvfile.read(8192)
         count = -1
         for item in [',', '\t', '|']:
             char_count = data.count(item)

aioradio/tests/file_ingestion_test.py

@@ -11,7 +11,7 @@
 
 import pytest
 
-from aioradio.file_ingestion import (async_db_wrapper, async_wrapper,
+from aioradio.file_ingestion import (EFIParse, async_db_wrapper, async_wrapper,
                                      delete_ftp_file, establish_ftp_connection,
                                      get_current_datetime_from_timestamp,
                                      list_ftp_objects,
@@ -227,3 +227,14 @@ async def func(**kwargs):
             print(f"Connection name: {name}\tConnection object: {conn}")
 
     await func()
+
+
+def test_check_phone_number():
+    """Test check_phone_number in EFIParse."""
+
+    efi = EFIParse('')
+    number = efi.check_phone_number('+1 (512) 573-5819', 'CellPhoneNumber', 0)
+    assert number == '5125735819'
+
+    number = efi.check_phone_number('215â€"863-79', 'CellPhoneNumber', 0)
+    assert number == '21586379'

setup.py

@@ -7,7 +7,7 @@
     long_description = fileobj.read()
 
 setup(name='aioradio',
-    version='0.17.34',
+    version='0.17.35',
     description='Generic asynchronous i/o python utilities for AWS services (SQS, S3, DynamoDB, Secrets Manager), Redis, MSSQL (pyodbc), JIRA and more',
     long_description=long_description,
     long_description_content_type="text/markdown",