def _get_field_elements(self, field_name, prefix, kind):
prefix = '{0}{1}'.format(prefix, cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(
prefix, field_name, gl.get_type_code_elements,
[field_name, self.codebase, kind, True, FieldElement])
return code_elements
def _link_all_references(self, unknown_refs, ucount, progress_monitor):
skipped = 0
class_tuples = []
progress_monitor.start('Parsing all unknown refs', ucount)
for reference in unknown_refs:
content = su.safe_strip(reference.content)
if content is None or content == '':
progress_monitor.info('Empty {0}'.format(reference.pk))
progress_monitor.work('Empty {0}'.format(reference.pk), 1)
skipped += 1
(simple, fqn) = je.clean_java_name(je.get_clean_name(content))
prefix = '{0}{1}'.format(PREFIX_GENERIC_LINKER,
cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(prefix, simple,
gl.get_any_code_element,
[simple, self.codebase])
classified_elements = self._classify_code_elements(code_elements)
class_tuples.append((reference, simple, fqn) + classified_elements)
count = self._process_tuples(class_tuples, progress_monitor)
progress_monitor.info(
'Associated {0} elements, Skipped {1} elements'.format(
count, skipped))
progress_monitor.done()
def reclassify_java(code_element, scode_reference):
reclassified = False
if scode_reference.snippet is not None or code_element is not None:
# We assume that references from snippet are always correctly
# classified. References that were linked to a code element
# do not need to be reclassified.
return reclassified
automatic_reclass = set(['method', 'field', 'annotation field',
'enumeration value', 'annotation', 'enumeration'])
unknown_kind = cu.get_value(PREFIX_UNKNOWN, UNKNOWN_KEY,
gl.get_unknown_kind, None)
if scode_reference.kind_hint.kind in automatic_reclass:
scode_reference.kind_hint = unknown_kind
reclassified = True
elif scode_reference.child_references.count() == 0:
# This was a single class reference, not mixed with a field or a
# method (in which case, the field/method will be reclassified if
# there is a need to). Maybe it was a reference to a method or a
# field. The simple name will be compared with all code elements
scode_reference.kind_hint = unknown_kind
reclassified = True
if reclassified:
scode_reference.save()
return reclassified
def _link_enumerations(self, enum_refs, ecount, progress_monitor):
count = 0
progress_monitor.start('Parsing enumerations', ecount)
log = gl.LinkerLog(self, self.enum_kind.kind)
for scode_reference in enum_refs:
if scode_reference.declaration:
progress_monitor.work('Skipped declaration', 1)
continue
(simple, fqn) = je.get_annotation_name(scode_reference.content,
scode_reference.snippet is not None)
if simple is not None:
prefix = '{0}{1}{2}'.format(PREFIX_ENUMERATION_LINKER, EXACT,
cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(
prefix,
simple,
gl.get_type_code_elements,
[simple, self.codebase, self.enum_kind])
(code_element, potentials) = self.get_code_element(
scode_reference, code_elements, simple, fqn, log)
count += gl.save_link(scode_reference, code_element,
potentials, self)
if not log.custom_filtered:
reclassify_java(code_element, scode_reference)
progress_monitor.work('Processed enumeration', 1)
log.close()
progress_monitor.done()
print('Associated {0} enumerations'.format(count))
def compute_project_code_words(codebases):
code_words = set()
for codebase in codebases:
code_words.update(
get_value(PREFIX_CODEBASE_CODE_WORDS, get_codebase_key(codebase),
compute_code_words, [codebase]))
return code_words
def reclassify_java(code_element, scode_reference):
reclassified = False
if scode_reference.snippet is not None or code_element is not None:
# We assume that references from snippet are always correctly
# classified. References that were linked to a code element
# do not need to be reclassified.
return reclassified
automatic_reclass = set([
'method', 'field', 'annotation field', 'enumeration value',
'annotation', 'enumeration'
])
unknown_kind = cu.get_value(PREFIX_UNKNOWN, UNKNOWN_KEY,
gl.get_unknown_kind, None)
if scode_reference.kind_hint.kind in automatic_reclass:
scode_reference.kind_hint = unknown_kind
reclassified = True
elif scode_reference.child_references.count() == 0:
# This was a single class reference, not mixed with a field or a
# method (in which case, the field/method will be reclassified if
# there is a need to). Maybe it was a reference to a method or a
# field. The simple name will be compared with all code elements
scode_reference.kind_hint = unknown_kind
reclassified = True
if reclassified:
scode_reference.save()
return reclassified
def _link_all_references(self, unknown_refs, ucount, progress_monitor):
skipped = 0
class_tuples = []
progress_monitor.start('Parsing all unknown refs', ucount)
for reference in unknown_refs:
content = su.safe_strip(reference.content)
if content is None or content == '':
progress_monitor.info('Empty {0}'.format(reference.pk))
progress_monitor.work('Empty {0}'.format(reference.pk), 1)
skipped += 1
(simple, fqn) = je.clean_java_name(je.get_clean_name(content))
prefix = '{0}{1}'.format(PREFIX_GENERIC_LINKER,
cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(
prefix,
simple,
gl.get_any_code_element,
[simple, self.codebase])
classified_elements = self._classify_code_elements(code_elements)
class_tuples.append((reference, simple, fqn) + classified_elements)
count = self._process_tuples(class_tuples, progress_monitor)
progress_monitor.info('Associated {0} elements, Skipped {1} elements'
.format(count, skipped))
progress_monitor.done()
def get_project_code_words(project):
codebases = CodeBase.objects.filter(project_release__project=project).all()
return get_value(
PREFIX_PROJECT_CODE_WORDS,
project.pk,
compute_project_code_words,
[codebases]
)
def _get_field_elements(self, field_name, prefix, kind):
prefix = '{0}{1}'.format(prefix, cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(
prefix,
field_name,
gl.get_type_code_elements,
[field_name, self.codebase, kind, True,
FieldElement])
return code_elements
def compute_project_code_words(codebases):
code_words = set()
for codebase in codebases:
code_words.update(
get_value(PREFIX_CODEBASE_CODE_WORDS,
get_codebase_key(codebase),
compute_code_words,
[codebase])
)
return code_words
def _get_method_elements(self, method_info):
prefix = '{0}{1}'.format(PREFIX_METHOD_LINKER,
cu.get_codebase_key(self.codebase))
method_name = method_info.method_name
code_elements = cu.get_value(prefix, method_name,
gl.get_type_code_elements, [
method_name, self.codebase,
self.method_kind, True, MethodElement
])
return code_elements
def _get_method_elements(self, method_info):
prefix = '{0}{1}'.format(PREFIX_METHOD_LINKER,
cu.get_codebase_key(self.codebase))
method_name = method_info.method_name
code_elements = cu.get_value(
prefix,
method_name,
gl.get_type_code_elements,
[method_name, self.codebase, self.method_kind, True,
MethodElement])
return code_elements
def test_cache_hit(self):
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
self.assertEqual(2, cu.cache_miss)
self.assertEqual(6, cu.cache_total)
def test_cache_hit(self):
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
self.assertEqual(2, cu.cache_miss)
self.assertEqual(6, cu.cache_total)
def test_cache_clear(self):
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
cu.clear_cache()
self.assertEqual(3, cu.get_value('p', 'k', func1, None))
self.assertEqual('6', cu.get_value('p', 'k2', func2, [1, 5]))
self.assertEqual(4, cu.cache_miss)
self.assertEqual(6, cu.cache_total)
def test_cache_clear(self):
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
cu.clear_cache()
self.assertEqual(3, cu.get_value("p", "k", func1, None))
self.assertEqual("6", cu.get_value("p", "k2", func2, [1, 5]))
self.assertEqual(4, cu.cache_miss)
self.assertEqual(6, cu.cache_total)
def _link_enumerations(self, enum_refs, ecount, progress_monitor):
count = 0
progress_monitor.start('Parsing enumerations', ecount)
log = gl.LinkerLog(self, self.enum_kind.kind)
for scode_reference in enum_refs:
if scode_reference.declaration:
progress_monitor.work('Skipped declaration', 1)
continue
(simple,
fqn) = je.get_annotation_name(scode_reference.content,
scode_reference.snippet is not None)
if simple is not None:
prefix = '{0}{1}{2}'.format(PREFIX_ENUMERATION_LINKER, EXACT,
cu.get_codebase_key(self.codebase))
code_elements = cu.get_value(
prefix, simple, gl.get_type_code_elements,
[simple, self.codebase, self.enum_kind])
(code_element,
potentials) = self.get_code_element(scode_reference,
code_elements, simple,
fqn, log)
count += gl.save_link(scode_reference, code_element,
potentials, self)
if not log.custom_filtered:
reclassify_java(code_element, scode_reference)
progress_monitor.work('Processed enumeration', 1)
log.close()
progress_monitor.done()
print('Associated {0} enumerations'.format(count))
def get_container(code_element):
return cu.get_value(PREFIX_GETCONTAINER, code_element.pk,
get_container_value, [code_element])
def get_hierarchy(code_element):
return cu.get_value(PREFIX_GETHIERARCHY, code_element.pk,
get_hierarchy_value, [code_element])
def link_references(self,
progress_monitor=NullProgressMonitor(),
local_object_id=None):
# Get link to filter.
linksets = ReleaseLinkSet.objects\
.filter(project_release=self.prelease)\
.filter(first_link__code_element__kind__is_type=True)
count = linksets.count()
progress_monitor.start('Post-Processing Classes', count)
log = gl.LinkerLog(self, 'type')
for linkset in queryset_iterator(linksets):
log.reset_variables()
scode_reference = linkset.code_reference
lcount = linkset.links.count()
if lcount <= 1:
log.log_type('', '', scode_reference,
linkset.first_link.code_element,
[linkset.first_link], lcount, 'onlyone')
progress_monitor.work('Skipped a linkset.', 1)
continue
local_ctx_id = scode_reference.local_object_id
source = scode_reference.source
prefix = '{0}{1}{2}'.format(PREFIX_CLASS_POST_LINKER,
cu.get_codebase_key(self.codebase),
source)
package_freq = cu.get_value(prefix, local_ctx_id, get_package_freq,
[local_ctx_id, source, self.prelease])
# Find package with highest frequency
code_element = self._find_package_by_freq(linkset, package_freq)
rationale = 'highest_frequency'
# Heuristic
if code_element is None:
code_element = self._find_package_by_depth(linkset)
rationale = 'heuristic_depth'
# Update links
if code_element is not None:
linkset.first_link.code_element = code_element
linkset.first_link.linker_name = self.name
linkset.first_link.rationale = rationale
linkset.first_link.save()
for link in linkset.links.all()[1:]:
link.delete()
log.log_type('', '', scode_reference,
linkset.first_link.code_element,
[linkset.first_link], lcount, rationale)
progress_monitor.info(
'FOUND Best Package {0} because {1}'.format(
code_element.fqn, rationale))
else:
potentials =\
[link.code_element for link in linkset.links.all()]
log.log_type('', '', scode_reference,
linkset.first_link.code_element, potentials,
lcount, 'nomatch')
progress_monitor.work('Processed a linkset.', 1)
progress_monitor.done()
def _link_classes(self, class_refs, ccount, progress_monitor):
count = 0
progress_monitor.start('Parsing classes', ccount)
log = gl.LinkerLog(self, self.class_kind.kind)
for scode_reference in class_refs:
if scode_reference.declaration:
progress_monitor.work('Skipped declaration', 1)
continue
(simple, fqn) = je.get_annotation_name(scode_reference.content,
scode_reference.snippet is not None)
case_insensitive = scode_reference.snippet == None and\
self.source != 'd'
if case_insensitive:
exact = False
exact_prefix = IEXACT
else:
exact = True
exact_prefix = EXACT
if simple is not None:
prefix = '{0}{1}{2}'.format(PREFIX_CLASS_LINKER, exact_prefix,
cu.get_codebase_key(self.codebase))
code_elements = []
code_elements.extend(cu.get_value(
prefix,
simple,
gl.get_type_code_elements,
[simple, self.codebase, self.class_kind, exact]))
prefix = '{0}{1}{2}'.format(PREFIX_ANNOTATION_LINKER,
exact_prefix, cu.get_codebase_key(self.codebase))
code_elements.extend(cu.get_value(
prefix,
simple,
gl.get_type_code_elements,
[simple, self.codebase, self.ann_kind, exact]))
prefix = '{0}{1}{2}'.format(PREFIX_ENUMERATION_LINKER,
exact_prefix, cu.get_codebase_key(self.codebase))
code_elements.extend(cu.get_value(
prefix,
simple,
gl.get_type_code_elements,
[simple, self.codebase, self.enum_kind, exact]))
(code_element, potentials) = self.get_code_element(
scode_reference, code_elements, simple, fqn, log,
not exact)
count += gl.save_link(scode_reference, code_element,
potentials, self)
if not log.custom_filtered:
reclassify_java(code_element, scode_reference)
progress_monitor.work('Processed class', 1)
log.close()
progress_monitor.done()
print('Associated {0} classes'.format(count))
def get_hierarchy(code_element):
return cu.get_value(PREFIX_GETHIERARCHY, code_element.pk,
get_hierarchy_value, [code_element])
def get_context_types_hierarchy(context_id, source, filter_func, codebase,
context_level):
return cu.get_value(
PREFIX_GETCONTEXT + HIERARCHY + source + context_level, context_id,
get_context_types_hier_value,
[context_id, source, filter_func, codebase, context_level])
def get_filters(codebase):
return get_value(PREFIX_CODEBASE_FILTERS,
get_codebase_key(codebase),
compute_filters,
[codebase])
def get_container(code_element):
return cu.get_value(PREFIX_GETCONTAINER, code_element.pk,
get_container_value, [code_element])
def get_ancestors(code_element):
return cu.get_value(PREFIX_GETANCESTORS, code_element.pk,
get_ancestors_value, [code_element])
def get_descendants(code_element):
return cu.get_value(PREFIX_GETDESCENDANTS, code_element.pk,
get_descendants_value, [code_element])
def link_references(self, progress_monitor=NullProgressMonitor(),
local_object_id=None):
# Get link to filter.
linksets = ReleaseLinkSet.objects\
.filter(project_release=self.prelease)\
.filter(first_link__code_element__kind__is_type=True)
count = linksets.count()
progress_monitor.start('Post-Processing Classes', count)
log = gl.LinkerLog(self, 'type')
for linkset in queryset_iterator(linksets):
log.reset_variables()
scode_reference = linkset.code_reference
lcount = linkset.links.count()
if lcount <= 1:
log.log_type('', '', scode_reference,
linkset.first_link.code_element,
[linkset.first_link], lcount, 'onlyone')
progress_monitor.work('Skipped a linkset.', 1)
continue
local_ctx_id = scode_reference.local_object_id
source = scode_reference.source
prefix = '{0}{1}{2}'.format(PREFIX_CLASS_POST_LINKER,
cu.get_codebase_key(self.codebase), source)
package_freq = cu.get_value(
prefix,
local_ctx_id,
get_package_freq,
[local_ctx_id, source, self.prelease])
# Find package with highest frequency
code_element = self._find_package_by_freq(linkset, package_freq)
rationale = 'highest_frequency'
# Heuristic
if code_element is None:
code_element = self._find_package_by_depth(linkset)
rationale = 'heuristic_depth'
# Update links
if code_element is not None:
linkset.first_link.code_element = code_element
linkset.first_link.linker_name = self.name
linkset.first_link.rationale = rationale
linkset.first_link.save()
for link in linkset.links.all()[1:]:
link.delete()
log.log_type('', '', scode_reference,
linkset.first_link.code_element,
[linkset.first_link], lcount, rationale)
progress_monitor.info('FOUND Best Package {0} because {1}'
.format(code_element.fqn, rationale))
else:
potentials =\
[link.code_element for link in linkset.links.all()]
log.log_type('', '', scode_reference,
linkset.first_link.code_element,
potentials, lcount, 'nomatch')
progress_monitor.work('Processed a linkset.', 1)
progress_monitor.done()
def get_context_types(context_id, source, filter_func, codebase,
context_level):
return cu.get_value(PREFIX_GETCONTEXT + source + context_level, context_id,
get_context_types_value,
[context_id, source, filter_func, codebase])
def get_ancestors(code_element):
return cu.get_value(PREFIX_GETANCESTORS, code_element.pk,
get_ancestors_value, [code_element])
def get_filters(codebase):
return get_value(PREFIX_CODEBASE_FILTERS, get_codebase_key(codebase),
compute_filters, [codebase])
def get_descendants(code_element):
return cu.get_value(PREFIX_GETDESCENDANTS, code_element.pk,
get_descendants_value, [code_element])
def get_project_code_words(project):
codebases = CodeBase.objects.filter(project_release__project=project).all()
return get_value(PREFIX_PROJECT_CODE_WORDS, project.pk,
compute_project_code_words, [codebases])
def get_context_types_hierarchy(context_id, source, filter_func, codebase,
context_level):
return cu.get_value(PREFIX_GETCONTEXT + HIERARCHY + source + context_level,
context_id, get_context_types_hier_value, [context_id, source,
filter_func, codebase, context_level])
def get_context_types(context_id, source, filter_func, codebase,
context_level):
return cu.get_value(PREFIX_GETCONTEXT + source + context_level,
context_id, get_context_types_value, [context_id, source,
filter_func, codebase])
def _link_classes(self, class_refs, ccount, progress_monitor):
count = 0
progress_monitor.start('Parsing classes', ccount)
log = gl.LinkerLog(self, self.class_kind.kind)
for scode_reference in class_refs:
if scode_reference.declaration:
progress_monitor.work('Skipped declaration', 1)
continue
(simple,
fqn) = je.get_annotation_name(scode_reference.content,
scode_reference.snippet is not None)
case_insensitive = scode_reference.snippet == None and\
self.source != 'd'
if case_insensitive:
exact = False
exact_prefix = IEXACT
else:
exact = True
exact_prefix = EXACT
if simple is not None:
prefix = '{0}{1}{2}'.format(PREFIX_CLASS_LINKER, exact_prefix,
cu.get_codebase_key(self.codebase))
code_elements = []
code_elements.extend(
cu.get_value(
prefix, simple, gl.get_type_code_elements,
[simple, self.codebase, self.class_kind, exact]))
prefix = '{0}{1}{2}'.format(PREFIX_ANNOTATION_LINKER,
exact_prefix,
cu.get_codebase_key(self.codebase))
code_elements.extend(
cu.get_value(
prefix, simple, gl.get_type_code_elements,
[simple, self.codebase, self.ann_kind, exact]))
prefix = '{0}{1}{2}'.format(PREFIX_ENUMERATION_LINKER,
exact_prefix,
cu.get_codebase_key(self.codebase))
code_elements.extend(
cu.get_value(
prefix, simple, gl.get_type_code_elements,
[simple, self.codebase, self.enum_kind, exact]))
(code_element,
potentials) = self.get_code_element(scode_reference,
code_elements, simple,
fqn, log, not exact)
count += gl.save_link(scode_reference, code_element,
potentials, self)
if not log.custom_filtered:
reclassify_java(code_element, scode_reference)
progress_monitor.work('Processed class', 1)
log.close()
progress_monitor.done()
print('Associated {0} classes'.format(count))