feat(intermediate): architect transactional silver layer firewalls with multi-locale parsing and defensive metrics bounding

models/intermediate/Employees/_int_employees_models.yml

@@ -1,5 +1,3 @@
-
-
 version: 2
 models:
 
@@ -125,6 +123,10 @@ models:
           the source value into a standard `INT`. If the raw entry is non-numeric (fails parsing) OR 
           is a negative integer (physically impossible for our enterprise sequence), the model coerces 
           the result to `NULL`. This completely eliminates downstream foreign key join failures.
+        tests:
+          - relationships:
+              to: ref('int_store_profile')
+              field: store_id
 
       - name: store_name
         description: |

models/intermediate/Inventory/_int_inventory_models.yml

@@ -35,6 +35,9 @@ models:
           or missing values are safely coerced to NULL to prevent downstream join failures.
         tests:
           - not_null
+          - relationships:
+              to: ref('int_product_attributes')
+              field: product_id
 
       - name: product_name
         description: |

models/intermediate/Returns/_int_returns_models.yml

@@ -0,0 +1,150 @@
+version: 2
+
+models:
+####################################################################################
+####################### Model: int_return_processing ###############################
+####################################################################################
+  - name: int_return_processing
+    description: |
+      Creates a trusted return processing operational dataset by standardizing categorical metrics, 
+      normalizing return fulfillment channels, and cleansing workflow status identifiers. The model 
+      applies Title Case to reason fields and handles alphanumeric casting exceptions for logistics personnel.
+
+    columns:
+      - name: return_id
+        description: Primary key uniquely identifying the individual return log.
+        tests:
+          - unique
+          - not_null
+
+      - name: return_reason
+        description: |
+          The documented customer reason for returning the product.
+          **Transformation Logic:** Standardized using whitespace trimming and converted to Title Case. Missing or empty string inputs fallback to 'Unknown'.
+        tests:
+          - not_null
+
+      - name: return_channel
+        description: |
+          The operational channel through which the return was initiated.
+          **Transformation Logic:** Groups highly fragmented text inputs into specific corporate categories:
+          - 'Mobile App' (Mapped from: 'app', 'mobile app', 'mobile')
+          - 'In Store' (Mapped from: 'in store', 'in-store', 'store')
+          - 'Online' (Mapped from: 'online', 'web')
+          - 'Phone Call' (Mapped from: 'phone')
+          - 'Catalog' (Mapped from: 'catalog')
+          - Unmapped formats default to 'Unknown'.
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values: ['Mobile App', 'In Store', 'Online', 'Phone Call', 'Catalog', 'Unknown']
+
+      - name: restocked
+        description: |
+          Operational flag indicating whether the item was returned to store inventory shelf space.
+          **Transformation Logic:** Standardized from multi-type strings ('yes', 'y', '1' to 'Yes' / 'no', 'n', '0' to 'No'). Outliers resolve to 'Unknown'.
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values: ['Yes', 'No', 'Unknown']
+
+      - name: return_status
+        description: |
+          Current state of the return processing lifecycle. Standardized to Title Case; missing source entries default to 'Unknown'.
+        tests:
+          - not_null
+
+      - name: handled_by_emp_id
+        description: |
+          The unique identifier of the store employee processing the return ticket. Converted to an integer data type, with missing or invalid fields coerced to NULL.
+
+####################################################################################
+####################### Model: int_return_refund ###################################
+####################################################################################
+  - name: int_return_refund
+    description: |
+      Extracts and sanitizes financial reimbursement attributes for returned orders. The model programmatically 
+      scrubs text currency strings to isolate decimals and normalizes payment methods to support precise gross margin 
+      reconciliation tracking.
+
+    columns:
+      - name: return_id
+        description: Unique primary key mapping to the corresponding return entry.
+        tests:
+          - unique
+          - not_null
+
+      - name: refund_amount
+        description: |
+          The total dollar value reimbursed back to the customer, cast to decimal(10,2) precision.
+          **Transformation Logic:** Strips formatting artifacts including currency symbols ('$') and thousands separators (',') before typecasting. Negative entries or un-castable formats default to NULL.
+
+      - name: refund_method
+        description: |
+          The transaction payment type used for the reimbursement (e.g., Credit Card, Cash). Standardized using Title Case, defaulting blank rows to 'Unknown'.
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values:
+                  - Cash
+                  - Original Payment
+                  - Store Credit
+
+####################################################################################
+####################### Model: int_return_transaction ##############################
+####################################################################################
+  - name: int_return_transaction
+    description: |
+      Normalizes and links return vouchers back to their core original transaction entities. The model enforces 
+      prefix patterns for tracking codes, tokenizes associated consumer profile fields into Title Case, and 
+      executes a multi-locale temporal handling sequence to cast multi-format text values into clean ISO dates.
+
+    columns:
+      - name: return_id
+        description: Cleaned primary identification key. Values below 1 or non-numeric elements resolve to NULL.
+        tests:
+          - unique
+          - not_null
+
+      - name: original_txn_id
+        description: |
+          The alphanumeric string of the original purchasing transaction header.
+          **Transformation Logic:** Enforces strict systemic syntax verification: must contain the 
+          corporate prefix 'TXN-' and maintain a minimum character length of 10. Failing rows are coerced to NULL.
+
+      - name: original_order_id
+        description: The integer identifier tracking the master order entry. Values below 1 or invalid formats resolve to NULL.
+
+      - name: customer_id
+        description: Unique identifier linking the transaction to a customer account profile. Invalid formats resolve to NULL.
+
+      - name: customer_name
+        description: Full name of the purchasing client. Standardized into Title Case; blank inputs fallback to 'Unknown'.
+        tests:
+          - not_null
+
+      - name: product_id
+        description: Unique identifier tracking the specific physical item returned. Invalid formats resolve to NULL.
+
+      - name: product_name
+        description: Official description name of the item. Standardized into Title Case; blanks default to 'Unknown'.
+        tests:
+          - not_null
+
+      - name: quantity_returned
+        description: |
+          Total count of physical units returned. Enforces business rule threshold :
+          values below 1 or non-numeric elements are coerced to NULL.
+
+      - name: return_date
+        description: |
+          The calendar date the return was officially recorded by the system.
+          **Transformation Logic:** Evaluates multiple text structures into an ISO-standard DATE format via an explicit CASE validation tree:
+          - Textual month lines and standard hyphen/slash paths are natively parsed.
+          - Multi-locale ambiguity is handled defensively: if the middle token exceeds 12, it is treated as US format (Style 101);
+           if the leading segment exceeds 12, it is handled as International format (Style 103).
+        tests:
+          - not_null

models/intermediate/Reviews/_int_reviews_models.yml

@@ -0,0 +1,122 @@
+version: 2
+
+models:
+####################################################################################
+####################### Model: int_review_feedback #################################
+####################################################################################
+  - name: int_review_feedback
+    description: |
+      Creates a trusted product review metadata dataset by standardizing submission channels 
+      and cleansing text field values. The model normalizes user engagement attributes, 
+      strips carriage returns and line breaks from titles using scalar operations, and maps 
+      highly fragmented input strings into canonical omni-channel tracking vectors.
+
+    columns:
+      - name: review_id
+        description: Primary key uniquely identifying the consumer review record.
+        tests:
+          - unique
+          - not_null
+
+      - name: helpful_votes
+        description: The total volume of user votes marking the review as informative.
+
+      - name: review_channel
+        description: |
+          The unified front-end channel where the submission was published.
+          **Transformation Logic:** Groups variations and shortcuts into a Title Case business taxonomy:
+          - 'Mobile App' (From: 'app', 'mobile app', 'mobile')
+          - 'In Store' (From: 'in store', 'in-store', 'store')
+          - 'Online' (From: 'online', 'web')
+          - 'Phone Call' (From: 'phone')
+          - 'Catalog' (From: 'catalog')
+          - Missing or unmapped structures safely default to 'Unknown'.
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values: ['Mobile App', 'In Store', 'Online', 'Phone Call', 'Catalog', 'Unknown']
+
+      - name: review_title
+        description: |
+          The sanitized summary title text.
+          **Transformation Logic:** Applies whitespace trimming, Title Case conversions, and strips 
+          hidden carriage returns `CHAR(13)` and line feeds `CHAR(10)`. Empty fields default to 'Unknown'.
+        tests:
+          - not_null
+
+####################################################################################
+####################### Model: int_review_rating ###################################
+####################################################################################
+  - name: int_review_rating
+    description: |
+      Processes customer review scores by implementing defensive type-casting and explicit boundary validation. 
+      The model protects downstream aggregate scoring matrices (such as Average Rating Score) by 
+      filtering out out-of-bounds ratings and corrupted text responses.
+
+    columns:
+      - name: review_id
+        description: Primary key matching back to the master feedback dimension.
+        tests:
+          - unique
+          - not_null
+
+      - name: rating
+        description: |
+          The validated numerical assessment score assigned by the consumer.
+          **Transformation Logic:** Casts the source string to an integer. Enforces a strict business boundary: 
+          values MUST sit between 1 and 5 inclusive. Any values outside this target scale default to NULL.
+        tests:
+          - accepted_values:
+              arguments:
+                values: [1, 2, 3, 4, 5]
+
+      - name: rating_text
+        description: The raw textual comment or contextual feedback submitted along with the rating score.
+
+####################################################################################
+####################### Model: int_review_transaction ##############################
+####################################################################################
+  - name: int_review_transaction
+    description: |
+      Normalizes and connects product review headers back to their original purchasing transactions, 
+      consumer accounts, and items. The model processes temporal values through a multi-locale validation 
+      ladder to convert multi-format text values into immutable ISO-standard dates.
+
+    columns:
+      - name: review_id
+        description: Primary key uniquely identifying the review transaction bridge.
+        tests:
+          - unique
+          - not_null
+
+      - name: txn_id
+        description: Foreign key mapping the review back to the source point-of-sale transaction record.
+        tests:
+          - not_null
+
+      - name: customer_id
+        description: Foreign key linking to the corresponding customer entity profile.
+        tests:
+          - not_null
+
+      - name: customer_name
+        description: The full identity string of the purchasing customer.
+
+      - name: product_id
+        description: Foreign key linking to the corresponding product master record.
+        tests:
+          - not_null
+
+      - name: product_name
+        description: Cleaned retail name nomenclature of the item reviewed.
+
+      - name: review_date
+        description: |
+          The standardized calendar logging date when the review was written.
+          **Transformation Logic:** Converts multi-format text entries into a safe DATE data type via an explicit CASE validation tree:
+          - Abbreviated or full textual month variations and standard ISO hyphen/slash paths are natively parsed.
+          - Date ambiguity is handled defensively: if the middle segment exceeds 12, it is treated as US format (Style 101/110).
+          If the leading segment exceeds 12, it is handled as International format (Style 103/105).
+        tests:
+          - not_null