Textualize / rich · Architecture

At the top level the tree holds 553 analyzed files, and the largest directories are benchmarks (182 files), rich (101), tests (68) and docs (69). The analyzer treats benchmarks, docs, rich and tests as candidate subsystems. Smaller top-level locations include examples with 37 files, a 39-file root, imgs with 22, .github with 11, questions with 11, tools with 8, assets with 3 and .faq with 2. The inventory counts 68 test-like files, so tests appears to hold the test suite. The shipping package appears to be the rich directory, matching the package name passed to setuptools.setup in setup.py. Repository metadata describes the project as a Python library for rich text and terminal formatting.

The architecture reconstruction counts 63 entrypoints and 612 bounded execution paths. Listed entrypoint records include setup.py, whose __main__ guard calls setuptools.setup with the name 'rich'. Package modules also carry demo guards: the records for rich/abc.py, rich/align.py, rich/console.py and rich/table.py show __main__ guards marked '# pragma: no cover'. All ten listed bounded execution paths start at examples/attrs.py:__main__, and each consists of a single recorded step. Five of them end at a local leaf — Model, Triangle or Point3D in examples/attrs.py. The other five end at external or unresolved targets — print (twice), SystemExit, attr.Factory and Console — with terminal reason 'unresolved_boundary'. No listed path steps into the rich package, and none of the ten is truncated or marked as containing a cycle. The packet separately reports one bounded path with a detected static call cycle and marks that finding unresolved; it is not among the ten listed.

These role labels are the analyzer's inference from static resolved call degree and entrypoint membership, not from runtime behavior. rich/text.py carries the highest incoming degree among the listed roles: 707 incoming and 44 outgoing calls, labeled an entry/orchestration candidate. rich/style.py (474 incoming, 26 outgoing) is labeled a service/core candidate, and rich/console.py (416 incoming, 78 outgoing) an entry/orchestration candidate. rich/segment.py (337 incoming, 7 outgoing) and rich/pretty.py (149 incoming, 71 outgoing) are also labeled entry/orchestration candidates. rich/default_styles.py shows the largest outgoing degree among the listed rich modules: 154 outgoing calls against 1 incoming. Test modules appear as orchestration candidates; for example tests/test_text.py records 0 incoming and 329 outgoing calls.

Of 11,248 recorded static relations, 6,703 remain external or unresolved and 4,545 resolve to local targets. A step marked external_or_unresolved records a call the analyzer could not tie to analyzed code. In the listed paths, the five unresolved-boundary paths stop at Console, print, SystemExit or attr.Factory reached from examples/attrs.py. Past those targets the paths record no further steps, so control flow beyond the boundary is not reconstructed.

The parsed dependency records are four runtime-scoped entries from docs/requirements.txt: alabaster ==1.0.0, Sphinx ==8.2.3, sphinx-rtd-theme ==3.0.2 and sphinx-copybutton ==0.5.2. Their names suggest the docs subsystem carries Sphinx-based documentation tooling. Dependency coverage is partial: the analyzer parses only requirements-style manifests, and pyproject.toml dependency tables are not parsed. Across the analyzed set the resolver recorded 9,630 call relations and 1,618 import relations. Because 6,703 relations stay external or unresolved, the picture of which parts depend on which is incomplete.

The 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. README claim extraction is line-based and may capture code lines instead of prose, so README-derived statements are author_claimed at best. Bounded analysis caps relation extraction at 2,500 relations per file and 150,000 per repository, and files over 400 kB, minified bundles and vendored directories are inventoried but not parsed. The manifest's capability fields describe the analyzer itself, not this repository, and are excluded from repository evidence. The tree includes 68 test-like files, but the analysis does not execute them, so behavior encoded in tests is not established.

README.md, pyproject.toml, setup.py and LICENSE are classified as project-level important files. The suggested path is to start with manifests and important symbols, trace the bounded execution paths, and inspect unresolved boundaries before changing code; the analyzer flags README.md, pyproject.toml and setup.py as early reads. Within the package, the role table points to rich/text.py, rich/style.py and rich/console.py as the most connected modules to read first.