simonw / sqlite-utils

GitHub metadata describes this repository as a 'Python CLI utility and library for manipulating SQLite databases'. Its listed homepage is https://sqlite-utils.datasette.io. Platform metadata records Python as the primary language, with 935,041 Python bytes, 1,547 C bytes, and 955 Just bytes. The license is Apache-2.0, and a LICENSE file exists at the project root. The analyzer recorded release 4.2.1 (published 2026-08-13) and 2,168 stargazers (2026-09-14). Topics include cli, click, sqlite, and datasette-tool. The code lives in a top-level sqlite_utils package directory. The analyzer's summary describes a Python project of 107 analyzed files organized around docs, sqlite_utils, and tests.

sqlite_utils/__main__.py contains a __main__ execution guard whose excerpt reads if __name__ == "__main__": cli(). The analyzer separately records sqlite_utils/cli.py as a likely executable entrypoint by filename heuristic. A bounded static execution path runs from __main__.py to the function cli in sqlite_utils/cli.py as a local call. From cli.py, recorded paths reach click.option through unresolved external boundaries, and cli.py calls maximize_csv_field_size_limit in sqlite_utils/utils.py, which reaches csv.field_size_limit and the builtin int at unresolved boundaries. One further path from cli.py terminates at an unresolved external target named strip. Most recorded paths therefore stop at unresolved boundaries after one or two hops rather than tracing deep local call chains.

Three top-level directories are recorded as subsystems: docs, sqlite_utils, and tests. The repository contains 59 test files or test-like files, anchored at tests/conftest.py. The packet records 8,672 static relations — 8,263 calls and 409 imports — of which 1,312 resolved to local targets and 7,360 stayed external or unresolved. By static call degree, the analyzer classifies sqlite_utils/db.py as a service/core candidate with the packet's highest incoming degree: 381 incoming and 317 outgoing calls. sqlite_utils/cli.py is an entry/orchestration candidate (127 in, 194 out), while sqlite_utils/__init__.py is a leaf/data-boundary candidate with 312 incoming and zero outgoing calls. sqlite_utils/create_table_parser.py (165 in, 122 out) and sqlite_utils/utils.py (91 in, 16 out) are also service/core candidates. These degrees are static counts, and the role labels are the analyzer's inferences; they suggest db.py is the module most called into, not a measured runtime property.

The repository contains exactly one build/dependency manifest: pyproject.toml. The analyzer parses dependency records only from requirements-style manifests and does not parse pyproject.toml dependency tables. Consequently, this packet contains no parsed third-party dependency records, and the project's dependency list cannot be recovered from this evidence alone. Click appears here only as a repository topic and as unresolved click.option call boundaries from cli.py, not as a parsed dependency. The test surface counts 59 test files or test-like files, anchored at tests/conftest.py. Named test modules include tests/test_cli.py, tests/test_cli_insert.py, tests/test_transform.py, tests/test_create_table_parser.py, and tests/test_mutator_transactions.py. Example tests: test_help and test_tables in tests/test_cli.py; test_insert_simple and test_insert_from_stdin in tests/test_cli_insert.py.

The analyzer flags README.md, pyproject.toml, LICENSE, sqlite_utils/__main__.py, and sqlite_utils/cli.py as important files. A workable order: README.md first; then pyproject.toml and LICENSE; then sqlite_utils/__main__.py and sqlite_utils/cli.py, starting at its cli function; then sqlite_utils/db.py, beginning at quote_identifier; then sqlite_utils/utils.py at maximize_csv_field_size_limit; then tests/test_cli.py for concrete CLI behavior. This ordering is a suggestion built from the analyzer's flags and symbol records, not a fact the repository states.

This analysis is static only: reflection, runtime dependency injection, dynamic imports, monkey-patching, generated code, framework runtime wiring, and dynamic dispatch are not resolved. Install, run, and build inference is partial, and installation commands are not verified. Dependency parsing covers only requirements-style manifests, so pyproject.toml dependency tables are missing from the evidence. README-derived text is extracted line by line and may capture code lines instead of prose, so README claims count as author claims at best. Of the 8,672 recorded static relations, 7,360 are external or unresolved. Because dynamic imports and dynamic dispatch are unresolved, plugin-style or dynamically loaded code paths cannot be traced by this analysis. Runtime behavior of the CLI and library is not established by this packet.

• sqlite_utils/cli.py: cli (lines 180-181), CaseInsensitiveChoice (96-101), output_options (104-151), tables (213-287), views (313-355), optimize (367-384), rebuild_fts (395-410), analyze (420-437).
• sqlite_utils/db.py: quote_identifier (93-99), fold_identifier_case (147-152), resolve_casing (155-167), ForeignKey (251-304), TransformError (325-326), AlterError (435-436).
• sqlite_utils/utils.py: maximize_csv_field_size_limit (87-99), find_spatialite (102-127), column_affinity (170-188), decode_base64_values (191-204), Format (244-248).
• sqlite_utils/create_table_parser.py: Check (16-26), Unique (43-50), ParseError (53-54), _Token (58-65), _lex (233-293), _split_spans (322-349).