def test_complete_pcc_relnodes_have_ns_children():
"""All relation nodes in all PCC files have only nucleii and/or satellites
as children.
"""
debug = False
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile, debug=debug)
if relnodes_have_ns_children(rst_tree, debug=debug):
okay += 1
else:
fail += 1
logging.log(
logging.WARN, "File '{}' has bad relname children".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of parsed PCC files have only relname->N/S parent/child relations ({1} of {2})".format(success_rate, okay, okay+fail)
debug = True
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile, debug=debug)
if relnodes_have_ns_children(rst_tree, debug=debug):
okay += 1
else:
relnodes_have_ns_children(rst_tree, debug=debug)
fail += 1
logging.log(
logging.WARN,
"File '{}' has bad relname children in debug-mode".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of debug-parsed PCC files have only relname->N/S parent/child relations ({1} of {2})".format(success_rate, okay, okay+fail)
def test_complete_pcc_no_double_ns():
"""There is no PCC file in which a nucleus/satellite has a nucleus or
satellite as a child.
"""
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile)
if no_double_ns(rst_tree.tree, filename):
okay += 1
else:
fail += 1
logging.log(
logging.WARN, "File '{}' has N/S->N/S parent/child".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of parsed PCC files have no N/S->N/S parent/child ({1} of {2})".format(success_rate, okay, okay+fail)
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile)
if no_double_ns(rst_tree.tree, filename, debug=True):
okay += 1
else:
fail += 1
logging.log(
logging.WARN,
"File '{}' has N/S->N/S parent/child in debug mode".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of debug-parsed PCC files have no N/S->N/S parent/child ({1} of {2})".format(success_rate, okay, okay+fail)
def test_complete_pcc_no_span_nodes():
"""There are no 'span' nodes in any PCC file."""
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile)
if no_span_nodes(rst_tree.tree, debug=rst_tree.debug):
okay += 1
else:
fail += 1
logging.log(
logging.WARN, "File '{}' has bad span nodes".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of parsed PCC files have no bad span nodes ({1} of {2})".format(success_rate, okay, okay+fail)
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
filename = os.path.basename(rfile)
try:
rst_tree = rstc.RSTTree(rfile, debug=True)
if no_span_nodes(rst_tree.tree, debug=rst_tree.debug):
okay += 1
else:
fail += 1
logging.log(
logging.WARN,
"File '{}' has bad span nodes in debug mode".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of debug-parsed PCC files have no bad span nodes ({1} of {2})".format(success_rate, okay, okay+fail)
def test_parse_complete_pcc():
"""All *.rs3 files can be parsed into RSTTree instances."""
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
try:
x = rstc.RSTTree(rfile)
okay += 1
except Exception as e:
logging.log(
logging.WARN,
"File '{0}' can't be parsed into an RSTTree: {1}".format(
os.path.basename(rfile), e))
fail += 1
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"{0}% of PCC files could be parsed ({1} of {2})".format(success_rate, okay, okay+fail)
def test_complete_pcc_edu_order():
"""The order of EDUs in all generated RSTTrees is the same as in the
original *.rs3 files.
"""
okay = 0.0
fail = 0.0
for i, rfile in enumerate(glob.glob(os.path.join(PCC_RS3_DIR, '*.rs3'))):
try:
rst_tree = rstc.RSTTree(rfile)
if rst_tree.edu_strings == rst_tree.tree.leaves():
okay += 1
else:
fail += 1
logging.log(
logging.WARN, "EDU order in file '{}' is wrong!".format(
os.path.basename(rfile)))
except TooManyChildrenError as e:
pass
success_rate = okay / (okay + fail) * 100
assert success_rate == 100, \
"\n{0}% of parsed PCC files have correct EDU order ({1} of {2})".format(success_rate, okay, okay+fail)