Module csv 

CSV emitters. Headers match code-maat exactly for golden-test parity.

Writers are grouped into per-domain submodules and re-exported here so callers keep the flat output::csv::write_* paths. Shared helpers live in this module.

Functions

write_abs_churn_csv

write_arch_violations_csv

architecture-violations CSV emitter — one row per import edge that crosses a forbidden layer boundary per .codelore-arch-rules.toml.

write_architecture_metrics_csv

write_architecture_roles_csv

architecture-roles CSV emitter — per-file Core/Shared/Control/ Periphery role + visibility fan-in/out.

write_architecture_trend_csv

architecture-trend CSV emitter — structural-health metrics at sampled historical revs.

write_author_churn_csv

write_authors_csv

Modern columns for the authors analysis (per-entity author breakdown — Bird et al. 2011 risk indicator). Code-maat’s parity columns entity,n-authors,n-revs are emitted under --code-maat-compat.

write_bus_factor_csv

bus-factor CSV emitter.

write_centrality_csv

write_clone_coupling_csv

write_clones_csv

write_code_age_csv

write_code_familiarity_csv

write_code_health_csv

write_communication_csv

Under --code-maat-compat the header uses code-maat’s author,peer column names (matching communication.clj’s output); CodeLore’s modern default uses the symmetric author-a,author-b pair which is clearer about the equality of roles.

write_communities_csv

write_coordination_needs_csv

write_coupling_csv

DEEP-1, DEEP-2, DEEP-3: under --code-maat-compat, emit code-maat’s verbose 7-column shape with truncated-integer degree and the legacy column-name set (entity, coupled, first-entity-revisions, second-entity-revisions, shared-revisions). The Fisher exact p-value is a CodeLore extension with no code-maat equivalent and gets dropped under compat — migrating tools wouldn’t know how to interpret it anyway. CodeLore’s modern default keeps the 8-column shape with float degree and the always-verbose paired columns.

write_crossing_csv

crossing CSV emitter — structural “X” files that co-change in both directions (DV8 Crossing).

write_cycle_health_csv

cycle-health CSV emitter — per-SCC heat, verdict, extraction candidate, and predicted propagation-cost drop.

write_cycle_origins_csv

architecture-metrics CSV emitter — repo-level (metric, value) rows. cycle-origins CSV emitter — the commit each HEAD dependency cycle first formed at.

write_defect_validation_csv

defect-validation CSV emitter — flat (metric, value) evidence rows from a defect-calibration artifact. Header metric,value; zero rows (no artifact configured) still emit the header.

write_delivery_friction_csv

delivery-friction CSV emitter.

write_delivery_metrics_csv

write_dependency_cycles_csv

dependency-cycles CSV emitter — one row per (cycle, member file); rows sharing a cycle_id form one tangle.

write_effort_exposure_csv

write_entity_churn_csv

write_entity_effort_csv

write_entity_ownership_csv

write_finding_hotspot_overlap_csv

write_function_coupling_csv

write_function_hotspots_csv

function-hotspots CSV emitter — repo-wide function-level hotspot ranking (revs × cognitive, the hotspots scoring shape at function granularity).

write_function_xray_csv

write_god_classes_csv

god-classes CSV emitter — fan_in / fan_out / cognitive intersection ranking per file.

write_health_trend_csv

health-trend CSV emitter — repo health timeline across sampled revs.

write_hotspot_velocity_csv

hotspot-velocity CSV emitter — recent vs baseline churn rate + acceleration (heating up / cooling down).

write_hotspots_csv

write_instability_csv

instability CSV emitter — Martin Ca/Ce/Instability per file.

write_knowledge_islands_csv

T8: per-file knowledge-loss risk (knowledge-islands analysis). No code-maat equivalent — strict CodeLore extension; no compat-mode header variant needed.

write_lead_time_csv

lead-time CSV emitter.

write_main_dev_by_deletions_csv

write_main_dev_by_revs_csv

write_main_dev_csv

write_marginal_owner_risk_csv

write_messages_csv

write_modularity_violations_csv

modularity-violations CSV emitter — Fisher-significant co-change pairs with no structural import edge (the structure×history fusion).

write_ownership_csv

Under --code-maat-compat, emit code-maat’s exact 3-column fragmentation output (entity,fractal-value,total-revs — note the column order; code-maat sorts the columns differently from CodeLore’s natural path / main_author / total_revs / fractal_value shape). CodeLore’s modern default surfaces main_author as a context column for triage (single-author file? long-tail file? the operator sees without re-running ownership).

write_pair_programming_csv

pair-programming CSV emitter.

write_refactoring_targets_csv

CSV emitter for the refactoring-targets analysis.

write_release_cadence_csv

write_revisions_csv

write_soc_csv

write_stale_code_csv

stale-code CSV emitter.

write_summary_csv

code-maat’s summary emits statistic,value; CodeLore uses the slightly clearer metric,value. Under --code-maat-compat, emit the legacy header so downstream tooling (code-maat-targeted dashboards) keeps parsing.

write_team_composition_csv

team-composition CSV emitter.

write_top_committers_csv

Per-author commit leaderboard — the previous behaviour of the authors analysis, now exposed as a distinct first-class analysis.

write_unstable_interface_csv

unstable-interface CSV emitter — heavily-imported files that change often and co-change with their dependents.