2501 intake add rd and u validations to new workbook vals (#2561)

* #2501 Updated ALN pattern * #2501 Added logic to check ALN pattern and to check if additional award identification is present * Apply suggestions from code review * Update backend/audit/intakelib/checks/runners.py --------- Co-authored-by: Dan Swick <[email protected]>
GSA-TTS · Oct 19, 2023 · 87e9c62 · 87e9c62
1 parent e9d5bfe
commit 87e9c62
Show file tree

Hide file tree

Showing 7 changed files with 92 additions and 3 deletions.
diff --git a/...ixtures/workbooks/should_fail/federal-awards/missing-additional-award-identification.xlsx b/...ixtures/workbooks/should_fail/federal-awards/missing-additional-award-identification.xlsx
diff --git a/backend/audit/intakelib/checks/check_additional_award_identification_present.py b/backend/audit/intakelib/checks/check_additional_award_identification_present.py
@@ -0,0 +1,34 @@
+import logging
+import re
+from audit.intakelib.intermediate_representation import (
+    get_range_values_by_name,
+    get_range_by_name,
+)
+from .check_aln_three_digit_extension_pattern import (
+    REGEX_RD_EXTENSION,
+    REGEX_U_EXTENSION,
+)
+from .util import get_message, build_cell_error_tuple
+
+logger = logging.getLogger(__name__)
+
+
+# DESCRIPTION
+# Additional award identification should be present if the ALN three digit extension is RD#, or U##
+def additional_award_identification(ir):
+    extension = get_range_values_by_name(ir, "three_digit_extension")
+    additional = get_range_values_by_name(ir, "additional_award_identification")
+    errors = []
+    patterns = [REGEX_RD_EXTENSION, REGEX_U_EXTENSION]
+    for index, (ext, add) in enumerate(zip(extension, additional)):
+        if any(re.match(pattern, ext) for pattern in patterns) and not add:
+            errors.append(
+                build_cell_error_tuple(
+                    ir,
+                    get_range_by_name(ir, "additional_award_identification"),
+                    index,
+                    get_message("check_additional_award_identification_present"),
+                )
+            )
+
+    return errors
diff --git a/backend/audit/intakelib/checks/check_aln_three_digit_extension_pattern.py b/backend/audit/intakelib/checks/check_aln_three_digit_extension_pattern.py
@@ -0,0 +1,45 @@
+import logging
+import re
+from audit.intakelib.intermediate_representation import (
+    get_range_values_by_name,
+    get_range_by_name,
+)
+from .util import get_message, build_cell_error_tuple
+
+logger = logging.getLogger(__name__)
+
+# A version of these regexes also exists in Base.libsonnet
+REGEX_RD_EXTENSION = r"^RD[0-9]?$"
+REGEX_THREE_DIGIT_EXTENSION = r"^[0-9]{3}[A-Za-z]{0,1}$"
+REGEX_U_EXTENSION = r"^U[0-9]{2}$"
+
+
+# DESCRIPTION
+# The three digit extension should follow one of these formats: ###, RD#, or U##, where # represents a number
+def aln_three_digit_extension(ir):
+    extension = get_range_values_by_name(ir, "three_digit_extension")
+    errors = []
+    # Define regex patterns
+    patterns = [REGEX_RD_EXTENSION, REGEX_THREE_DIGIT_EXTENSION, REGEX_U_EXTENSION]
+    for index, ext in enumerate(extension):
+        # Check if ext is None or does not match any of the regex patterns
+        if not ext:
+            errors.append(
+                build_cell_error_tuple(
+                    ir,
+                    get_range_by_name(ir, "three_digit_extension"),
+                    index,
+                    get_message("check_aln_three_digit_extension_missing"),
+                )
+            )
+        elif not any(re.match(pattern, ext) for pattern in patterns):
+            errors.append(
+                build_cell_error_tuple(
+                    ir,
+                    get_range_by_name(ir, "three_digit_extension"),
+                    index,
+                    get_message("check_aln_three_digit_extension_invalid"),
+                )
+            )
+
+    return errors
diff --git a/backend/audit/intakelib/checks/error_messages.py b/backend/audit/intakelib/checks/error_messages.py
@@ -26,6 +26,9 @@
     "check_passthrough_name_when_no_direct_n_and_empty_number": "When the award is direct, passthrough number must be empty",
     "check_findings_grid_validation": "The combination of findings <b>{}</b> is not a valid combination under Uniform Guidance",
     "check_eins_are_not_empty": "EIN cannot be empty",
+    "check_aln_three_digit_extension_missing": "Missing ALN (CFDA) three digit extension",
+    "check_aln_three_digit_extension_invalid": "The three digit extension should follow one of these formats: ###, RD#, or U##, where # represents a number",
+    "check_additional_award_identification_present": "Missing additional award identification",
     "check_federal_program_total": "Federal program total is {}, but should be {}",
     "check_cluster_total": "This cluster total is {}, but should be {}",
     "check_total_amount_expended": "Total amount expended is {}, but should be {}",

diff --git a/backend/audit/intakelib/checks/runners.py b/backend/audit/intakelib/checks/runners.py
@@ -29,6 +29,10 @@
 from .check_federal_award_passed_always_present import (
     federal_award_passed_always_present,
 )
+from .check_aln_three_digit_extension_pattern import aln_three_digit_extension
+from .check_additional_award_identification_present import (
+    additional_award_identification,
+)
 from .check_federal_program_total import federal_program_total_is_correct
 from .check_cluster_total import cluster_total_is_correct
 from .check_total_amount_expended import total_amount_expended_is_correct
@@ -69,6 +73,8 @@
     no_major_program_no_type,
     all_unique_award_numbers,
     sequential_award_numbers,
+    aln_three_digit_extension,
+    additional_award_identification,
     federal_program_total_is_correct,
     cluster_total_is_correct,
     total_amount_expended_is_correct,

diff --git a/backend/schemas/output/sections/FederalAwards.schema.json b/backend/schemas/output/sections/FederalAwards.schema.json
@@ -373,7 +373,7 @@
                               "if": {
                                  "properties": {
                                     "three_digit_extension": {
-                                       "pattern": "^(RD|U[0-9]{2})$",
+                                       "pattern": "^(RD[0-9]?|U[0-9]{2})$",
                                        "type": "string"
                                     }
                                  }
@@ -438,7 +438,7 @@
                               "type": "string"
                            },
                            "three_digit_extension": {
-                              "pattern": "^(RD|[0-9]{3}[A-Za-z]{0,1}|U[0-9]{2})$",
+                              "pattern": "^(RD[0-9]?|[0-9]{3}[A-Za-z]{0,1}|U[0-9]{2})$",
                               "type": "string"
                            }
                         },

diff --git a/backend/schemas/source/base/Base.libsonnet b/backend/schemas/source/base/Base.libsonnet
@@ -66,7 +66,8 @@ local Meta = {
 };
 
 local REGEX_ALN_PREFIX = '^([0-9]{2})$';
-local REGEX_RD_EXTENSION = 'RD';
+# A python version of these regexes also exists in intakelib
+local REGEX_RD_EXTENSION = 'RD[0-9]?';
 local REGEX_THREE_DIGIT_EXTENSION = '[0-9]{3}[A-Za-z]{0,1}';
 local REGEX_U_EXTENSION = 'U[0-9]{2}';
 local REGEX_NUMBER = '[0-9]+';