def test_preprocess_taxons():
dbf = ProgramFilter(db)
names = ["X/S/M.*", "O(?!/C).*", "Y/E", "non_matching_pattern"]
taxons = dbf.preprocess_taxons(names)
print(sorted(taxons))
assert taxons == [
"O",
"O/J",
"O/N",
"O/N/P",
"X/S/M",
"X/S/M/L",
"X/S/M/L/R",
"X/S/M/L/R/D",
"X/S/M/L/R/D/A",
"X/S/M/L/V",
"Y/E",
]
names = [
("never mind", "O/J", "Y/E"),
("whatever", r"X/S/M/L/R/D.*", r"O/N.*"),
("who cares", r"X/S/M/L/R.*", r"O/N"),
]
taxons = dbf.preprocess_taxons(names)
print(sorted(taxons))
assert taxons == [
("never mind", "O/J", "Y/E"),
("whatever", "X/S/M/L/R/D", "O/N"),
("whatever", "X/S/M/L/R/D", "O/N/P"),
("whatever", "X/S/M/L/R/D/A", "O/N"),
("whatever", "X/S/M/L/R/D/A", "O/N/P"),
("who cares", "X/S/M/L/R", "O/N"),
("who cares", "X/S/M/L/R/D", "O/N"),
("who cares", "X/S/M/L/R/D/A", "O/N"),
]
def test_include_taxons():
# The taxon "variable/assignment/single" is featured by assignment.py and collatz_print.py. In
# collatz_print.py, it appears after a taxon "io/standard/print". Consequently, it should be included
# in the results, but not the programs which import it: fizzbuzz.py and is_even.py.
dbf = ProgramFilter(db)
dbf.include_taxons({("after", "variable/assignment/single", "io/standard/print")})
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == {"collatz_print.py"}
# "operator/arithmetic/modulo" and "type/number/integer/literal" are both featured on
# the same line in all programs except assignment.py
dbf = ProgramFilter(db)
dbf.include_taxons({("equals", "operator/arithmetic/modulo", "type/number/integer/literal")})
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == {"collatz_print.py", "is_even.py", "fizzbuzz.py"}
# The same with "x == y" instead of "equals"
dbf = ProgramFilter(db)
dbf.include_taxons({("equals", "operator/arithmetic/modulo", "type/number/integer/literal")})
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == {"collatz_print.py", "is_even.py", "fizzbuzz.py"}
# "test/equality" is inside "subroutine/function" in is_even.py, which is not imported anywhere.
dbf = ProgramFilter(db)
dbf.include_taxons({("inside", "test/equality", "subroutine/function")})
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == {"is_even.py"}
# "test/equality" is inside "subroutine/function" in is_even.py and inside
# "subroutine/procedure" in collatz_print.py. Both will be included.
dbf = ProgramFilter(db)
taxons = dbf.preprocess_taxons([("inside", "test/equality", "subroutine/.*")])
print(taxons)
assert taxons == [
("inside", "test/equality", "subroutine/argument/arg"),
("inside", "test/equality", "subroutine/function"),
("inside", "test/equality", "subroutine/predicate"),
("inside", "test/equality", "subroutine/procedure"),
]
dbf.include_taxons(taxons)
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == {"collatz_print.py", "is_even.py"}
# "call/function/builtin/range" is not inside "flow/conditional" anywhere.
dbf = ProgramFilter(db)
dbf.include_taxons([("inside", "call/function/builtin/range", "flow/conditional")])
print(set(dbf.selected_programs.keys()))
assert set(dbf.selected_programs.keys()) == set()
