def _set_attributes(ann_obj, ann, attributes, mods, undo_resp={}):
# Find existing attributes (if any)
existing_attr_anns = set((a for a in ann_obj.get_attributes()
if a.target == ann.id))
#log_info('ATTR: %s' %(existing_attr_anns, ))
# Note the existing annotations for undo
undo_resp['attributes'] = json_dumps(dict([(e.type, e.value)
for e in existing_attr_anns]))
for existing_attr_ann in existing_attr_anns:
if existing_attr_ann.type not in attributes:
# Delete attributes that were un-set existed previously
ann_obj.del_annotation(existing_attr_ann)
mods.deletion(existing_attr_ann)
else:
# If the value of the attribute is different, alter it
new_value = attributes[existing_attr_ann.type]
#log_info('ATTR: "%s" "%s"' % (new_value, existing_attr_ann.value))
if existing_attr_ann.value != new_value:
before = str(existing_attr_ann)
existing_attr_ann.value = new_value
mods.change(before, existing_attr_ann)
# The remaining annotations are new and should be created
for attr_type, attr_val in attributes.items():
if attr_type not in set((a.type for a in existing_attr_anns)):
new_attr = AttributeAnnotation(ann.id, ann_obj.get_new_id('A'),
attr_type, '', attr_val)
ann_obj.add_annotation(new_attr)
mods.addition(new_attr)
def _set_attributes(ann_obj, ann, attributes, mods, undo_resp={}):
# Find existing attributes (if any)
existing_attr_anns = set((a for a in ann_obj.get_attributes()
if a.target == ann.id))
#log_info('ATTR: %s' %(existing_attr_anns, ))
# Note the existing annotations for undo
undo_resp['attributes'] = json_dumps(dict([(e.type, e.value)
for e in existing_attr_anns]))
for existing_attr_ann in existing_attr_anns:
if existing_attr_ann.type not in attributes:
# Delete attributes that were un-set existed previously
ann_obj.del_annotation(existing_attr_ann)
mods.deletion(existing_attr_ann)
else:
# If the value of the attribute is different, alter it
new_value = attributes[existing_attr_ann.type]
#log_info('ATTR: "%s" "%s"' % (new_value, existing_attr_ann.value))
if existing_attr_ann.value != new_value:
before = unicode(existing_attr_ann)
existing_attr_ann.value = new_value
mods.change(before, existing_attr_ann)
# The remaining annotations are new and should be created
for attr_type, attr_val in attributes.iteritems():
if attr_type not in set((a.type for a in existing_attr_anns)):
new_attr = AttributeAnnotation(ann.id, ann_obj.get_new_id('A'),
attr_type, '', attr_val)
ann_obj.add_annotation(new_attr)
mods.addition(new_attr)
def create_comment(collection, document, id, comment=None):
directory = collection
undo_resp = {}
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
# XXX what is this doing here?
# path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
_set_special_comments(ann_obj, id, comment, mods, undo_resp=undo_resp)
mods_json = mods.json_response()
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
def _set_normalizations(ann_obj, ann, normalizations, mods, undo_resp={}):
# Find existing normalizations (if any)
existing_norm_anns = set(
(a for a in ann_obj.get_normalizations() if a.target == ann.id))
# Note the existing annotations for undo
undo_resp['normalizations'] = json_dumps([(n.refdb, n.refid, n.reftext)
for n in existing_norm_anns])
# Organize into dictionaries for easier access
old_norms = dict([((n.refdb, n.refid), n) for n in existing_norm_anns])
new_norms = dict([((n[0], n[1]), n[2]) for n in normalizations])
#Messager.info("Old norms: "+str(old_norms))
#Messager.info("New norms: "+str(new_norms))
# sanity check
for refdb, refid, refstr in normalizations:
# TODO: less aggressive failure
assert refdb is not None and refdb.strip(
) != '', "Error: client sent empty norm DB"
assert refid is not None and refid.strip(
) != '', "Error: client sent empty norm ID"
# (the reference string is allwed to be empty)
# Process deletions and updates of existing normalizations
for old_norm_id, old_norm in list(old_norms.items()):
if old_norm_id not in new_norms:
# Delete IDs that were referenced previously but not anymore
ann_obj.del_annotation(old_norm)
mods.deletion(old_norm)
else:
# If the text value of the normalizations is different, update
# (this shouldn't happen on a stable norm DB, but anyway)
new_reftext = new_norms[old_norm_id]
if old_norm.reftext != new_reftext:
old = str(old_norm)
old_norm.reftext = new_reftext
mods.change(old, old_norm)
# Process new normalizations
for new_norm_id, new_reftext in list(new_norms.items()):
if new_norm_id not in old_norms:
new_id = ann_obj.get_new_id('N')
# TODO: avoid magic string value
norm_type = 'Reference'
new_norm = NormalizationAnnotation(new_id, norm_type, ann.id,
new_norm_id[0], new_norm_id[1],
'\t' + new_reftext)
ann_obj.add_annotation(new_norm)
mods.addition(new_norm)
def _set_normalizations(ann_obj, ann, normalizations, mods, undo_resp={}):
# Find existing normalizations (if any)
existing_norm_anns = set((a for a in ann_obj.get_normalizations()
if a.target == ann.id))
# Note the existing annotations for undo
undo_resp['normalizations'] = json_dumps([(n.refdb, n.refid, n.reftext)
for n in existing_norm_anns])
# Organize into dictionaries for easier access
old_norms = dict([((n.refdb,n.refid),n) for n in existing_norm_anns])
new_norms = dict([((n[0],n[1]), n[2]) for n in normalizations])
#Messager.info("Old norms: "+str(old_norms))
#Messager.info("New norms: "+str(new_norms))
# sanity check
for refdb, refid, refstr in normalizations:
# TODO: less aggressive failure
assert refdb is not None and refdb.strip() != '', "Error: client sent empty norm DB"
assert refid is not None and refid.strip() != '', "Error: client sent empty norm ID"
# (the reference string is allwed to be empty)
# Process deletions and updates of existing normalizations
for old_norm_id, old_norm in old_norms.items():
if old_norm_id not in new_norms:
# Delete IDs that were referenced previously but not anymore
ann_obj.del_annotation(old_norm)
mods.deletion(old_norm)
else:
# If the text value of the normalizations is different, update
# (this shouldn't happen on a stable norm DB, but anyway)
new_reftext = new_norms[old_norm_id]
if old_norm.reftext != new_reftext:
old = unicode(old_norm)
old_norm.reftext = new_reftext
mods.change(old, old_norm)
# Process new normalizations
for new_norm_id, new_reftext in new_norms.items():
if new_norm_id not in old_norms:
new_id = ann_obj.get_new_id('N')
# TODO: avoid magic string value
norm_type = u'Reference'
new_norm = NormalizationAnnotation(new_id, norm_type,
ann.id, new_norm_id[0],
new_norm_id[1],
u'\t'+new_reftext)
ann_obj.add_annotation(new_norm)
mods.addition(new_norm)
def _create_span(collection, document, offsets, _type, attributes=None,
normalizations=None, _id=None, comment=None):
if _offset_overlaps(offsets):
raise SpanOffsetOverlapError(offsets)
directory = collection
undo_resp = {}
_attributes = _parse_attributes(attributes)
_normalizations = _parse_span_normalizations(normalizations)
#log_info('ATTR: %s' %(_attributes, ))
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
working_directory = path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
if _id is not None:
# We are to edit an existing annotation
tb_ann, e_ann = _edit_span(ann_obj, mods, _id, offsets, projectconf,
_attributes, _type, undo_resp=undo_resp)
else:
# We are to create a new annotation
#Messager.info('offsets ' + '_'.join(str(s) for s in offsets[0]) + ' _ann_by_offset ' + str(ann_obj._ann_by_offset.keys()))
#Messager.info('Creating span collection(' + str(collection) + ') document(' + str(document) + ') offsets(' + str(offsets) + ') _type(' + str(_type) + ') exists(' + str(ann_obj.ann_exists_in_offset('_'.join(str(s) for s in offsets[0]))) + ')' )
if ann_obj.ann_exists_in_offset('_'.join(str(s) for s in offsets[0])):
raise SpanOffsetOverlapError([str(ann_obj.get_ann_by_offset('_'.join(str(s) for s in offsets[0]))).split()[-1]])
if ann_obj.other_ann_exists_in_offset(offsets[0]):
raise SpanOffsetOverlapError([], True)
#Messager.info('offsets(' + str(offsets) + ')')
##Messager.info('offsets(' + str(offsets) + ') alloffsets ' + str(ann_obj._ann_offsets))
tb_ann, e_ann = __create_span(ann_obj, mods, _type, offsets, txt_file_path,
projectconf, _attributes)
undo_resp['action'] = 'add_tb'
if e_ann is not None:
undo_resp['id'] = e_ann.id
else:
undo_resp['id'] = tb_ann.id
# Determine which annotation attributes, normalizations,
# comments etc. should be attached to. If there's an event,
# attach to that; otherwise attach to the textbound.
if e_ann is not None:
# Assign to the event, not the trigger
target_ann = e_ann
else:
target_ann = tb_ann
# Set attributes
_set_attributes(ann_obj, target_ann, _attributes, mods,
undo_resp=undo_resp)
# Set normalizations
_set_normalizations(ann_obj, target_ann, _normalizations, mods,
undo_resp=undo_resp)
# Set comments
if tb_ann is not None:
_set_comments(ann_obj, target_ann, comment, mods,
undo_resp=undo_resp)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, probably we had a new-line in the span
mods_json = {}
Messager.error('Text span contained new-line, rejected', duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
_set_normalizations(ann_obj, target_ann, _normalizations, mods,
undo_resp=undo_resp)
# Set comments
if tb_ann is not None:
_set_comments(ann_obj, target_ann, comment, mods,
undo_resp=undo_resp)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, probably we had a new-line in the span
mods_json = {}
Messager.error('Text span contained new-line, rejected', duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
from annotation import BinaryRelationAnnotation
def _create_equiv(ann_obj, projectconf, mods, origin, target, type, attributes,
old_type, old_target):
# due to legacy representation choices for Equivs (i.e. no
# unique ID), support for attributes for Equivs would need
# some extra work. Getting the easy non-Equiv case first.
if attributes is not None:
Messager.warning('_create_equiv: attributes for Equiv annotation not supported yet, please tell the devs if you need this feature (mention "issue #799").')
attributes = None
def _create_span(collection, document, offsets, _type, attributes=None,
normalizations=None, _id=None, comment=None):
if _offset_overlaps(offsets):
raise SpanOffsetOverlapError(offsets)
directory = collection
undo_resp = {}
_attributes = _parse_attributes(attributes)
_normalizations = _parse_span_normalizations(normalizations)
#log_info('ATTR: %s' %(_attributes, ))
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
if _id is not None:
# We are to edit an existing annotation
tb_ann, e_ann = _edit_span(ann_obj, mods, _id, offsets, projectconf,
_attributes, _type, undo_resp=undo_resp)
else:
# We are to create a new annotation
tb_ann, e_ann = __create_span(
ann_obj, mods, _type, offsets, txt_file_path, projectconf, _attributes)
undo_resp['action'] = 'add_tb'
if e_ann is not None:
undo_resp['id'] = e_ann.id
else:
undo_resp['id'] = tb_ann.id
# Determine which annotation attributes, normalizations,
# comments etc. should be attached to. If there's an event,
# attach to that; otherwise attach to the textbound.
if e_ann is not None:
# Assign to the event, not the trigger
target_ann = e_ann
else:
target_ann = tb_ann
# Set attributes
_set_attributes(ann_obj, target_ann, _attributes, mods,
undo_resp=undo_resp)
# Set normalizations
_set_normalizations(ann_obj, target_ann, _normalizations, mods,
undo_resp=undo_resp)
# Set comments
if tb_ann is not None:
_set_comments(ann_obj, target_ann, comment, mods,
undo_resp=undo_resp)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, probably we had a new-line in the span
mods_json = {}
Messager.error(
'Text span contained new-line, rejected',
duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
def _create_span(collection, document, start, end, _type, attributes=None,
normalizations=None, _id=None, comment=None):
directory = collection
undo_resp = {}
_attributes = _parse_attributes(attributes)
_normalizations = _parse_span_normalizations(normalizations)
#log_info('ATTR: %s' %(_attributes, ))
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
working_directory = path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
if _id is not None:
# We are to edit an existing annotation
tb_ann, e_ann = _edit_span(ann_obj, mods, _id, start, end, projectconf,
_attributes, _type, undo_resp=undo_resp)
else:
# We are to create a new annotation
tb_ann, e_ann = __create_span(ann_obj, mods, _type, start, end, txt_file_path,
projectconf, _attributes)
undo_resp['action'] = 'add_tb'
if e_ann is not None:
undo_resp['id'] = e_ann.id
else:
undo_resp['id'] = tb_ann.id
# Determine which annotation attributes, normalizations,
# comments etc. should be attached to. If there's an event,
# attach to that; otherwise attach to the textbound.
if e_ann is not None:
target_ann = e_ann
else:
target_ann = tb_ann
# Set annotation attributes
_set_attributes(ann_obj, target_ann, _attributes, mods,
undo_resp=undo_resp)
# Set normalizations
_set_normalizations(ann_obj, target_ann, _normalizations, mods,
undo_resp=undo_resp)
# Handle annotation comments
if tb_ann is not None:
_set_comments(ann_obj, target_ann, comment, mods,
undo_resp=undo_resp)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, probably we had a new-line in the span
mods_json = {}
Messager.error('Text span contained new-line, rejected', duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
def _create_span(collection, document, start, end, _type, attributes=None,
normalizations=None, _id=None, comment=None):
directory = collection
undo_resp = {}
_attributes = _parse_span_attributes(attributes)
_normalizations = _parse_span_normalizations(normalizations)
#log_info('ATTR: %s' %(_attributes, ))
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
working_directory = path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
if _id is not None:
# We are to edit an existing annotation
tb_ann, e_ann = _edit_span(ann_obj, mods, _id, start, end, projectconf,
_attributes, _type, undo_resp=undo_resp)
else:
# We are to create a new annotation
tb_ann, e_ann = __create_span(ann_obj, mods, _type, start, end, txt_file_path,
projectconf, _attributes)
undo_resp['action'] = 'add_tb'
if e_ann is not None:
undo_resp['id'] = e_ann.id
else:
undo_resp['id'] = tb_ann.id
# Determine which annotation attributes, normalizations,
# comments etc. should be attached to. If there's an event,
# attach to that; otherwise attach to the textbound.
if e_ann is not None:
target_ann = e_ann
else:
target_ann = tb_ann
# Set annotation attributes
_set_attributes(ann_obj, target_ann, _attributes, mods,
undo_resp=undo_resp)
# Set normalizations
_set_normalizations(ann_obj, target_ann, _normalizations, mods,
undo_resp=undo_resp)
# Handle annotation comments
if tb_ann is not None:
# We are only interested in id;ed comments
try:
target_ann.id
has_id = True
except AttributeError:
has_id = False
if has_id:
# Check if there is already an annotation comment
for com_ann in ann_obj.get_oneline_comments():
if (com_ann.type == 'AnnotatorNotes'
and com_ann.target == target_ann.id):
found = com_ann
# Note the comment in the undo
undo_resp['comment'] = found.tail[1:]
break
else:
found = None
if comment:
if found is not None:
# Change the comment
# XXX: Note the ugly tab, it is for parsing the tail
before = unicode(found)
found.tail = u'\t' + comment
mods.change(before, found)
else:
# Create a new comment
new_comment = OnelineCommentAnnotation(
target_ann.id, ann_obj.get_new_id('#'),
# XXX: Note the ugly tab
u'AnnotatorNotes', u'\t' + comment)
ann_obj.add_annotation(new_comment)
mods.addition(new_comment)
else:
# We are to erase the annotation
if found is not None:
ann_obj.del_annotation(found)
mods.deletion(found)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, we had a new-line in the span
mods_json = {}
Messager.error('Text span contained new-line, rejected', duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
def _create_span(collection, document, start, end, _type, attributes=None,
_id=None, comment=None):
directory = collection
undo_resp = {}
if attributes is None:
_attributes = {}
else:
try:
_attributes = json_loads(attributes)
except ValueError:
# Failed to parse attributes, warn the client
Messager.warning((u'Unable to parse attributes string "%s" for '
u'"createSpan", ignoring attributes for request and '
u'assuming no attributes set') % (attributes, ))
_attributes = {}
### XXX: Hack since the client is sending back False and True as values...
# These are __not__ to be sent, they violate the protocol
for _del in [k for k, v in _attributes.items() if v == False]:
del _attributes[_del]
# These are to be old-style modifiers without values
for _revalue in [k for k, v in _attributes.items() if v == True]:
_attributes[_revalue] = True
###
#log_info('ATTR: %s' %(_attributes, ))
real_dir = real_directory(directory)
document = path_join(real_dir, document)
projectconf = ProjectConfiguration(real_dir)
txt_file_path = document + '.' + TEXT_FILE_SUFFIX
working_directory = path_split(document)[0]
with TextAnnotations(document) as ann_obj:
# bail as quick as possible if read-only
if ann_obj._read_only:
raise AnnotationsIsReadOnlyError(ann_obj.get_document())
mods = ModificationTracker()
if _id is not None:
# We are to edit an existing annotation
tb_ann, e_ann = _edit_span(ann_obj, mods, _id, start, end, projectconf,
_attributes, _type, undo_resp=undo_resp)
else:
# We are to create a new annotation
tb_ann, e_ann = __create_span(ann_obj, mods, _type, start, end, txt_file_path,
projectconf, _attributes)
undo_resp['action'] = 'add_tb'
if e_ann is not None:
undo_resp['id'] = e_ann.id
else:
undo_resp['id'] = tb_ann.id
# Set annotation attributes
if e_ann is not None:
# Assign attributes to the event, not the trigger
_set_attributes(ann_obj, e_ann, _attributes, mods,
undo_resp=undo_resp)
else:
_set_attributes(ann_obj, tb_ann, _attributes, mods,
undo_resp=undo_resp)
# Handle annotation comments
if tb_ann is not None:
# If this is an event, we want to attach the comment to it
if e_ann is not None:
comment_on = e_ann
else:
comment_on = tb_ann
# We are only interested in id;ed comments
try:
comment_on.id
has_id = True
except AttributeError:
has_id = False
if has_id:
# Check if there is already an annotation comment
for com_ann in ann_obj.get_oneline_comments():
if (com_ann.type == 'AnnotatorNotes'
and com_ann.target == comment_on.id):
found = com_ann
# Note the comment in the undo
undo_resp['comment'] = found.tail[1:]
break
else:
found = None
if comment:
if found is not None:
# Change the comment
# XXX: Note the ugly tab, it is for parsing the tail
before = unicode(found)
found.tail = u'\t' + comment
mods.change(before, found)
else:
# Create a new comment
new_comment = OnelineCommentAnnotation(
comment_on.id, ann_obj.get_new_id('#'),
# XXX: Note the ugly tab
u'AnnotatorNotes', u'\t' + comment)
ann_obj.add_annotation(new_comment)
mods.addition(new_comment)
else:
# We are to erase the annotation
if found is not None:
ann_obj.del_annotation(found)
mods.deletion(found)
if tb_ann is not None:
mods_json = mods.json_response()
else:
# Hack, we had a new-line in the span
mods_json = {}
Messager.error('Text span contained new-line, rejected', duration=3)
if undo_resp:
mods_json['undo'] = json_dumps(undo_resp)
mods_json['annotations'] = _json_from_ann(ann_obj)
return mods_json
