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.
dependencies.py is step 2 of the pipeline. It reads the unit inventory and all Fortran source files to resolve every CALL, USE, and inline function-call reference into a directed edge in the corpus call graph. The result is seven output files covering the full spectrum of dependency information: raw relationships, resolved graph edges, per-unit coupling metrics, ambiguous definitions, external orphans, file-level summaries, and INCLUDE cross-references. All seven files are always written — even when empty — so downstream scripts can reliably distinguish “no data” from “step never ran.”
dependencies.py is step 2 of the pipeline. It requires inventory_report.csv from step 1.
<FORT_OUT>/inventory_report.csv- Fortran source files in
FORT_SRC
Output Files
dep_00_ambiguities.csv
Units with the same name defined in more than one source file. Always written (headers only if no ambiguities exist).
| Column | Description |
|---|
Unit_Name | Unit name |
Type | Unit type |
Count | Number of files where this name is defined |
File_List | Semicolon-separated list of those files |
dep_01_master_data.csv
All detected call/use relationships before resolution. Always written (headers only when no dependencies are found).
dep_02_unit_graph.csv
The resolved call graph — one row per directed edge. Edge types are CALL, USE, and FUNC_CALL. Always written (headers only when no resolved edges exist, e.g. a pure library corpus with no cross-file dependencies).
| Column | Description |
|---|
Source_Unit | Calling unit. IMPLICIT-MAIN units appear as MAIN__<filename> |
Source_Type | Type of the calling unit |
Target_Unit | Called or used unit |
Target_Type | Type of the called unit |
Dep_Type | Edge type: CALL, USE, or FUNC_CALL |
Target_File | Source file(s) of the destination unit |
Weight | Number of times this dependency appears in the calling unit |
dep_03_impact_matrix.csv
Fan-In and Fan-Out per unit. Always written (headers only when no resolved dependencies exist).
| Column | Description |
|---|
Unit | Unit name |
Type | Unit type |
File | Source file |
Fan_In | Number of units that call or use this unit |
Fan_Out | Number of distinct units this unit calls or uses |
dep_04_external_orphans.csv
References found in source code with no matching definition in the corpus — external library calls, unresolved symbols. Always written (headers only when all references are resolved internally).
dep_05_file_dependencies.csv
Aggregated dependency summary at the file level. Always written (headers only when all units reside in a single file or have no cross-file dependencies).
dep_06_include_files.csv
Explicit Fortran INCLUDE file references — one row per unique (Source_File, Source_Unit, Included_File) triple. Duplicate references from the same triple at different lines are deduplicated. Always written (headers only when no INCLUDE directives are found).
| Column | Description |
|---|
Source_File | Source file containing the INCLUDE statement |
Source_Unit | Unit (scope) in which the INCLUDE appears |
Included_File | Filename referenced by the INCLUDE directive |
Status | PRESENT if the file exists in FORT_SRC; MISSING otherwise |
Edge Types
| Edge Type | Triggered by |
|---|
CALL | CALL subroutine_name(...) statement |
USE | USE module_name statement |
FUNC_CALL | Inline function call within an expression |
Notes
MAIN__<file> naming convention: dependencies.py represents IMPLICIT-MAIN entry points as MAIN__<filename> (e.g. MAIN__solver.f90) in dep_02_unit_graph.csv. reachability.py and visual_graph.py resolve this convention back to the inventory unit name for display.
Ambiguous targets: When a called name matches multiple definitions in dep_00_ambiguities.csv, the edge in dep_02_unit_graph.csv may point to multiple candidate files (semicolon-separated in the Target_File column).
All seven dep_*.csv files are always written — even when they contain only headers. This guarantees that downstream scripts can tell the difference between “dependencies ran and found nothing” and “dependencies never ran.”
