Documentation Index
Fetch the complete documentation index at: https://mintlify.com/acdeveloper-sci/forti4d/llms.txt
Use this file to discover all available pages before exploring further.
structure_analysis.py and cross_analysis.py form the architectural classification tier of the pipeline. structure_analysis.py uses the call graph coupling data to assign each source file a structural role — identifying which files are indispensable shared libraries, which orchestrate many dependencies, and which are completely isolated. cross_analysis.py then crosses statement density profiles with coupling metrics to compute two composite indices per unit and recommend a concrete migration strategy for each. Together with executive_summary.py, these three scripts close the Tier 2 pipeline and produce the primary outputs that guide migration planning.
structure_analysis.py — Architectural Role Classification
structure_analysis.py is step 5 of the pipeline.
<FORT_OUT>/dep_03_impact_matrix.csv<FORT_OUT>/inventory_report.csv
Output: report_structure_analysis.csv
One row per source file, sorted by category priority.
| Column | Description |
|---|
File | Source file name |
Category | Architectural role (see below) |
Fan_In_Max | Maximum Fan-In across all units in the file |
Unit_Max_In | Unit with the highest Fan-In |
Fan_Out_Max | Maximum Fan-Out across all units in the file |
Unit_Max_Out | Unit with the highest Fan-Out |
Total_Units | Number of units in the file |
Has_Main | YES if the file contains a PROGRAM or IMPLICIT-MAIN unit |
Detail | Human-readable explanation of the classification |
Architectural Roles
Categories are assigned in priority order — the first matching rule wins.
| Category | Condition | Meaning |
|---|
ENTRY_POINT | File contains an IMPLICIT-MAIN unit | Compiles to an executable |
ISLAND | Fan-In = 0 and Fan-Out = 0 | No connections — potential dead code |
CRITICAL_NODE | Max Fan-In ≥ 10 | High reuse — core library unit |
ORCHESTRATOR | Max Fan-Out ≥ 10 | Coordinates many dependencies |
WORKER | Connected but Fan-Out < 5 | Service or calculation routine |
MIXED | All other cases | Standard functionality |
Files present in inventory_report.csv but absent from dep_03_impact_matrix.csv (no call graph entries) are automatically classified as ISLAND. The threshold of 10 for CRITICAL_NODE and ORCHESTRATOR is empirically calibrated; it is a candidate for a future --critical-threshold CLI flag.
cross_analysis.py — Migration Strategy Assignment
cross_analysis.py is step 6 of the pipeline.
<FORT_OUT>/report_density.csv<FORT_OUT>/dep_03_impact_matrix.csv
Optional (used when present, silently skipped otherwise):
<FORT_OUT>/report_reachability.csv — enables the confirmed dead-code rule<FORT_OUT>/symbol_implicit.csv — enables the E4 ICM penalty<FORT_OUT>/equivalences.csv — enables the E4 ICM penalty
Output: report_migration_strategy.csv
One row per unit, sorted by Priority_Num ascending (most urgent first), then by IVC descending within the same priority.
| Column | Description |
|---|
Priority_Num | Numeric priority (1 = most urgent) |
Strategy | Recommended action (see below) |
File | Source file name |
Unit | Unit name |
Type | Unit type |
ICM | Migration Complexity Index (base + E4 penalty) |
IVC | Calculation Value Index (= Pct_Calc) |
Pct_Calc | % calculation statements |
Pct_Control | % control-flow statements |
Pct_Legacy | % legacy statements |
Fan_In | Number of callers |
Fan_Out | Number of callees |
Reachability_Status | Reachability status (empty if CSV not present) |
Explanation | Human-readable reason for the strategy |
Composite Indices
IVC (Calculation Value Index): equals Pct_Calc. Measures how much of the unit is pure algorithmic computation.
ICM (Migration Complexity Index):
ICM_base = 0.15 × Pct_Control
+ 0.45 × min(Pct_Legacy × 4, 100)
+ 0.20 × min(Fan_Out × 5, 100)
+ 0.20 × min(Fan_In × 5, 100)
E4_penalty = 7.0 × (0.70 × no_IMPLICIT_NONE + 0.30 × has_EQUIVALENCE)
ICM = ICM_base + E4_penalty
IMPLICIT NONE (no_IMPLICIT_NONE = 1) and/or have at least one EQUIVALENCE aliasing group (has_EQUIVALENCE = 1). The penalty is applied only when symbol_implicit.csv and/or equivalences.csv are present.
Migration Strategy Rule Engine
Rules are evaluated in order; the first match wins.
| Priority | Strategy | Condition |
|---|
| — | ELIMINATE | Reachability_Status = UNREACHABLE — confirmed dead code (requires reachability CSV) |
| 1 | ANALYZE_UTILITY / ELIMINATE | Fan-In = 0 (proxy dead-code check when reachability not available) |
| 2 | DIRECT_MIGRATION | IVC > 50 and ICM < 30 — pure algorithm, low coupling |
| 3 | REPLACE_LIB | Pct_IO > 30 or Pct_Decl > 40, and IVC < 20 — infrastructure/boilerplate |
| 4 | REFACTOR_CORE | ICM > 25 and Fan-In > 5 — high-risk, high-dependency knot |
| 5 | REWRITE_ISOLATED | ICM > 20 — complex but low systemic impact |
| 6 | STANDARD_MIGRATION | All other connected units |
PROGRAM, IMPLICIT-MAIN, MODULE, or BLOCK DATA are exempt from the Fan-In = 0 dead-code rule.
In a standard full pipeline run, cross_analysis executes at step 6 — before symbols (step 10), equivalences (step 12), and reachability (step 13). The three optional inputs are therefore never available in a standard run; the E4 penalty and the UNREACHABLE → ELIMINATE rule are silently skipped. To use the full rule engine, re-run cross_analysis manually after the pipeline completes:forti4d --from cross_analysis --only cross_analysis
cross_analysis to after reachability in the pipeline. executive_summary.py — High-Level Project Summary
executive_summary.py is step 7 of the pipeline.
executive_summary.py produces a Markdown overview of the entire corpus and a per-file statistics CSV.
Required inputs: inventory_report.csv, dep_03_impact_matrix.csv
Optional inputs (enable the Scope Health section): symbol_implicit.csv, equivalences.csv, common_usage.csv, symbol_variables.csv
Outputs:
<FORT_OUT>/PROJECT_SUMMARY.md — global metrics, unit type distribution, top 10 largest files, legacy/I/O health indicators, optional Scope Health (E4) section, and top critical/orchestrating units<FORT_OUT>/file_statistics.csv — one row per source file with LOC, unit count, legacy/IO flags, and unit types present
Like cross_analysis, executive_summary runs at step 7 — before the E4 scripts complete. The Scope Health section of PROJECT_SUMMARY.md is therefore absent in a standard full pipeline run. Re-run it after the pipeline to include E4 data:forti4d --from executive_summary --only executive_summary
