Travis Vasceannie
e0bfb7a2f2
* feat: enhance coverage reporting and improve tool configuration - Added support for JSON coverage reports in pyproject.toml. - Updated .gitignore to include coverage.json and task files for better management. - Introduced a new Type Safety Audit Report to document findings and recommendations for type safety improvements. - Created a comprehensive coverage configuration guide to assist in understanding coverage reporting setup. - Refactored tools configuration to utilize environment variables for concurrent scraping settings. These changes improve the project's testing and reporting capabilities while enhancing overall code quality and maintainability. * feat: enhance configuration handling and improve error logging - Introduced a new utility function `_get_env_int` for robust environment variable integer retrieval with validation. - Updated `WebToolsConfig` and `ToolsConfigModel` to utilize the new utility for environment variable defaults. - Enhanced logging in `CircuitBreaker` to provide detailed state transition information. - Improved URL handling in `url_analyzer.py` for better file extension extraction and normalization. - Added type validation and logging in `SecureInputMixin` to ensure input sanitization and validation consistency. These changes improve the reliability and maintainability of configuration management and error handling across the codebase. * refactor: update imports and enhance .gitignore for improved organization - Updated import paths in various example scripts to reflect the new structure under `biz_bud`. - Enhanced .gitignore to include clearer formatting for task files. - Removed obsolete function calls and improved error handling in several scripts. - Added public alias for backward compatibility in `upload_r2r.py`. These changes improve code organization, maintainability, and compatibility across the project. * refactor: update graph paths in langgraph.json for improved organization - Changed paths for research, catalog, paperless, and url_to_r2r graphs to reflect new directory structure. - Added new entries for analysis and scraping graphs to enhance functionality. These changes improve the organization and maintainability of the graph configurations. * fix: enhance validation and error handling in date range and scraping functions - Updated date validation in UserFiltersModel to ensure date values are strings. - Improved error messages in create_scraped_content_dict to clarify conditions for success and failure. - Enhanced test coverage for date validation and scraping content creation to ensure robustness. These changes improve input validation and error handling across the application, enhancing overall reliability. * refactor: streamline graph creation and enhance type annotations in examples - Simplified graph creation in `catalog_ingredient_research_example.py` and `catalog_tech_components_example.py` by directly compiling the graph. - Updated type annotations in `catalog_intel_with_config.py` for improved clarity and consistency. - Enhanced error handling in catalog data processing to ensure robustness against unexpected data types. These changes improve code readability, maintainability, and error resilience across example scripts. * Update src/biz_bud/nodes/extraction/extractors.py Co-authored-by: sourcery-ai[bot] <58596630+sourcery-ai[bot]@users.noreply.github.com> * Update src/biz_bud/core/validation/pydantic_models.py Co-authored-by: sourcery-ai[bot] <58596630+sourcery-ai[bot]@users.noreply.github.com> * refactor: migrate Jina and Tavily clients to use ServiceFactory dependency injection * refactor: migrate URL processing to provider-based architecture with improved error handling * feat: add FirecrawlApp compatibility classes and mock implementations * fix: add thread-safe locking to LazyLoader factory management * feat: implement service restart and refactor cache decorator helpers * refactor: move r2r_direct_api_call to tools.clients.r2r_utils and improve HTTP service error handling * chore: update Sonar task IDs in report configuration --------- Co-authored-by: sourcery-ai[bot] <58596630+sourcery-ai[bot]@users.noreply.github.com>
131 lines
5.2 KiB
SQL
131 lines
5.2 KiB
SQL
-- public.rpt_master_products definition
|
|
|
|
-- Drop table
|
|
|
|
-- DROP TABLE public.rpt_master_products;
|
|
|
|
CREATE TABLE public.rpt_master_products (
|
|
id serial4 NOT NULL,
|
|
canonical_name varchar(500) NOT NULL,
|
|
canonical_description text NULL,
|
|
category varchar(100) NULL,
|
|
unit_of_measure varchar(50) NULL,
|
|
estimated_unit_price numeric(10, 2) NULL,
|
|
is_active bool DEFAULT true NULL,
|
|
first_seen_date timestamptz DEFAULT now() NULL,
|
|
last_updated timestamptz DEFAULT now() NULL,
|
|
total_occurrences int4 DEFAULT 1 NULL,
|
|
CONSTRAINT rpt_master_products_pkey PRIMARY KEY (id)
|
|
);
|
|
CREATE INDEX idx_rpt_master_products_canonical_name ON public.rpt_master_products USING btree (canonical_name);
|
|
CREATE INDEX idx_rpt_master_products_category ON public.rpt_master_products USING btree (category);
|
|
|
|
-- public.rpt_product_variations definition
|
|
|
|
-- Drop table
|
|
|
|
-- DROP TABLE public.rpt_product_variations;
|
|
|
|
CREATE TABLE public.rpt_product_variations (
|
|
id serial4 NOT NULL,
|
|
master_product_id int4 NOT NULL,
|
|
original_description varchar(500) NOT NULL,
|
|
verified_description varchar(500) NULL,
|
|
confidence_score numeric(3, 2) NULL,
|
|
verification_source varchar(100) NULL,
|
|
search_variations_used _text NULL,
|
|
successful_variation varchar(500) NULL,
|
|
verification_notes text NULL,
|
|
created_at timestamptz DEFAULT now() NULL,
|
|
occurrence_count int4 DEFAULT 1 NULL,
|
|
CONSTRAINT rpt_product_variations_pkey PRIMARY KEY (id),
|
|
CONSTRAINT rpt_product_variations_master_product_fkey FOREIGN KEY (master_product_id) REFERENCES public.rpt_master_products(id)
|
|
);
|
|
CREATE INDEX idx_rpt_product_variations_master_id ON public.rpt_product_variations USING btree (master_product_id);
|
|
CREATE INDEX idx_rpt_product_variations_original ON public.rpt_product_variations USING btree (original_description);
|
|
|
|
-- public.rpt_receipt_line_items definition
|
|
|
|
-- Drop table
|
|
|
|
-- DROP TABLE public.rpt_receipt_line_items;
|
|
|
|
CREATE TABLE public.rpt_receipt_line_items (
|
|
id int4 DEFAULT nextval('receipt_line_items_id_seq'::regclass) NOT NULL,
|
|
receipt_id numeric NOT NULL,
|
|
line_number int4 NULL,
|
|
product_name varchar(500) NOT NULL,
|
|
product_code varchar(100) NULL,
|
|
quantity numeric(10, 3) NULL,
|
|
unit_of_measure varchar(50) NULL,
|
|
unit_price numeric(10, 2) NULL,
|
|
total_price numeric(10, 2) NULL,
|
|
category varchar(100) NULL,
|
|
created_at timestamptz DEFAULT now() NULL,
|
|
master_product_id int4 NULL,
|
|
product_variation_id int4 NULL,
|
|
original_ocr_text varchar(500) NULL,
|
|
reconciliation_status varchar(50) DEFAULT 'pending'::character varying NULL,
|
|
reconciliation_confidence numeric(3, 2) NULL,
|
|
needs_review bool DEFAULT false NULL,
|
|
CONSTRAINT receipt_line_items_pkey PRIMARY KEY (id),
|
|
CONSTRAINT rpt_receipt_line_items_master_product_fkey FOREIGN KEY (master_product_id) REFERENCES public.rpt_master_products(id),
|
|
CONSTRAINT rpt_receipt_line_items_receipt_id_fkey FOREIGN KEY (receipt_id) REFERENCES public.rpt_receipts(id),
|
|
CONSTRAINT rpt_receipt_line_items_variation_fkey FOREIGN KEY (product_variation_id) REFERENCES public.rpt_product_variations(id)
|
|
);
|
|
CREATE INDEX idx_rpt_receipt_line_items_master_product ON public.rpt_receipt_line_items USING btree (master_product_id);
|
|
CREATE INDEX idx_rpt_receipt_line_items_receipt_id ON public.rpt_receipt_line_items USING btree (receipt_id);
|
|
CREATE INDEX idx_rpt_receipt_line_items_reconciliation_status ON public.rpt_receipt_line_items USING btree (reconciliation_status);
|
|
|
|
-- public.rpt_receipts definition
|
|
|
|
-- Drop table
|
|
|
|
-- DROP TABLE public.rpt_receipts;
|
|
|
|
CREATE TABLE public.rpt_receipts (
|
|
id numeric DEFAULT nextval('receipts_id_seq'::regclass) NOT NULL,
|
|
vendor_name varchar(255) NOT NULL,
|
|
vendor_address text NULL,
|
|
transaction_date date NULL,
|
|
transaction_time time NULL,
|
|
receipt_number varchar(100) NULL,
|
|
customer_info text NULL,
|
|
subtotal numeric(10, 2) NULL,
|
|
tax_amount numeric(10, 2) NULL,
|
|
final_total numeric(10, 2) NULL,
|
|
total_items int4 NULL,
|
|
payment_method varchar(50) NULL,
|
|
card_last_four bpchar(4) NULL,
|
|
card_type varchar(20) NULL,
|
|
raw_receipt_text text NULL,
|
|
created_at timestamptz DEFAULT now() NULL,
|
|
updated_at timestamptz DEFAULT now() NULL,
|
|
CONSTRAINT receipts_pkey PRIMARY KEY (id)
|
|
);
|
|
CREATE INDEX idx_rpt_receipts_payment_method ON public.rpt_receipts USING btree (payment_method);
|
|
CREATE INDEX idx_rpt_receipts_transaction_date ON public.rpt_receipts USING btree (transaction_date);
|
|
CREATE INDEX idx_rpt_receipts_vendor_name ON public.rpt_receipts USING btree (vendor_name);
|
|
|
|
-- public.rpt_reconciliation_log definition
|
|
|
|
-- Drop table
|
|
|
|
-- DROP TABLE public.rpt_reconciliation_log;
|
|
|
|
CREATE TABLE public.rpt_reconciliation_log (
|
|
id serial4 NOT NULL,
|
|
receipt_line_item_id int4 NOT NULL,
|
|
original_description varchar(500) NULL,
|
|
suggested_master_product_id int4 NULL,
|
|
action_taken varchar(50) NULL,
|
|
confidence_score numeric(3, 2) NULL,
|
|
reconciled_by varchar(100) NULL,
|
|
reconciled_at timestamptz DEFAULT now() NULL,
|
|
notes text NULL,
|
|
CONSTRAINT rpt_reconciliation_log_pkey PRIMARY KEY (id),
|
|
CONSTRAINT rpt_reconciliation_log_line_item_fkey FOREIGN KEY (receipt_line_item_id) REFERENCES public.rpt_receipt_line_items(id),
|
|
CONSTRAINT rpt_reconciliation_log_master_product_fkey FOREIGN KEY (suggested_master_product_id) REFERENCES public.rpt_master_products(id)
|
|
);
|
|
CREATE INDEX idx_rpt_reconciliation_log_receipt_item ON public.rpt_reconciliation_log USING btree (receipt_line_item_id);
|