• There are 164 files with units, with 52,912 lines of code.
• 7 very complex files (McCabe index sum 251+; 8,560 lines of code)
• 11 complex files (McCabe index sum 126-250; 16,415 lines of code)
• 26 medium complex files (McCabe index sum 51-125; 12,175 lines of code)
• 36 slightly complex files (McCabe index sum 26-50; 8,858 lines of code)
• 84 simple files (McCabe index sum 1-25; 6,904 lines of code)
• 302 main files have no units (e.g. languages without unit analysis, or files without functions) and are not classified.

The McCabe index of a file grows with its size: per 100 lines tells a large but plain file from a dense one, and most complex unit whether the complexity sits in one unit or is spread over many.

File	# lines	# units	McCabe index (sum)	per 100 lines	most complex unit
svm.cpp in libvmaf/src	2677	82	494	18	43
train_test_model.py in python/vmaf/core	1416	105	356	25	27
routine.py in python/vmaf	962	19	314	33	171
libvmaf.c in libvmaf/src	1075	40	292	27	34
asset.py in python/vmaf/core	716	81	271	38	17
executor.py in python/vmaf/core	675	60	267	40	21
quality_runner.py in python/vmaf/core	1039	95	264	25	62
integer_adm.c in libvmaf/src/feature	2656	22	241	9	36
cambi.c in libvmaf/src/feature	1177	44	238	20	42
speed.c in libvmaf/src/feature	1356	54	228	17	20
adm_tools.c in libvmaf/src/feature	1074	13	211	20	113
adm_avx2.c in libvmaf/src/feature/x86	3358	16	207	6	37
adm_avx512.c in libvmaf/src/feature/x86	3358	14	203	6	37
read_json_model.c in libvmaf/src	611	20	180	29	44
feature_extractor.py in python/vmaf/core	756	29	176	23	47
pdjson.c in libvmaf/src	823	32	134	16	30
cli_parse.c in libvmaf/tools	682	26	127	19	45
vif_tools.c in libvmaf/src/feature	564	31	126	22	10
y4m_input.c in libvmaf/tools	618	13	116	19	24
scanf.py in python/vmaf/tools	460	63	115	25	6
misc.py in python/vmaf/tools	345	45	113	33	14
edges.c in matlab/strred/matlabPyrTools/MEX	466	11	113	24	28
adm_neon.c in libvmaf/src/feature/arm64	763	23	106	14	18
integer_adm_cuda.c in libvmaf/src/feature/cuda	971	27	101	10	17
result.py in python/vmaf/core	335	32	100	30	16
picture.c in libvmaf/src	480	24	100	21	18
predict.c in libvmaf/src	433	13	100	23	22
output.c in libvmaf/src	323	6	98	30	45
vmaf.c in libvmaf/tools	403	4	96	24	80
feature_collector.c in libvmaf/src/feature	364	16	92	25	14
feature_extractor.c in libvmaf/src/feature	411	12	90	22	13
nn_train_test_model.py in python/vmaf/core	401	24	84	21	10
float_motion.c in libvmaf/src/feature	409	4	84	21	43
cambi_avx2.c in libvmaf/src/feature/x86	358	12	76	21	20
edges-orig.c in matlab/strred/matlabPyrTools/MEX	318	10	73	23	15
__init__.py in python/vmaf	219	10	70	32	27
integer_vif.c in libvmaf/src/feature	700	12	67	10	11
getopt.c in libvmaf/src/compat/getopt	260	9	61	23	25
perf_metric.py in python/vmaf/core	307	19	57	19	15
cross_validation.py in python/vmaf/core	245	9	57	23	16
vif_avx2.c in libvmaf/src/feature/x86	1283	4	57	4	19
ciede.c in libvmaf/src/feature	304	18	57	19	8
vif_neon.c in libvmaf/src/feature/arm64	810	4	54	7	16
convolve.c in libvmaf/src/feature/iqa	189	7	52	28	16
matlab_feature_extractor.py in python/vmaf/core	302	16	50	17	8
common.c in libvmaf/src/cuda	187	13	50	27	7
reader.py in python/vmaf/tools	152	18	48	32	23
dict.c in libvmaf/src	177	9	47	27	21
barten_csf_tools.h in libvmaf/src/feature	185	6	46	25	19
convolution_avx.c in libvmaf/src/feature/common	174	7	46	26	10

• File complexity is the sum of the McCabe indexes (conditional complexity) of the units (methods, functions...) in a file.
• Files are classified in five categories based on that sum: 1-25 (simple files), 26-50 (slightly complex files), 51-125 (medium complex files), 126-250 (complex files), 251+ (very complex files). The default bands are five times the unit complexity bands, as a file is a group of units.
• Only main code files with units are classified. Files without units (languages without unit analysis, or files without functions) have no complexity to measure and are left out rather than counted as simple.
• A very complex file holds many decisions, which often means it has too many responsibilities: a candidate to split, and a place where changes need more testing.