From 15129c71af0a5aea630ef858b8a5830135741fc4 Mon Sep 17 00:00:00 2001
From: Tomasz Bartosiak <tomasz.bartosiak@gmail.com>
Date: Thu, 16 Nov 2023 16:12:08 +0100
Subject: [PATCH] Added structured representation of syntax
---
common/management/commands/valunifier_tei.py | 623 ++++++++++++++++++-
1 file changed, 601 insertions(+), 22 deletions(-)
diff --git a/common/management/commands/valunifier_tei.py b/common/management/commands/valunifier_tei.py
index bca4ac1..4c807cc 100644
--- a/common/management/commands/valunifier_tei.py
+++ b/common/management/commands/valunifier_tei.py
@@ -450,7 +450,7 @@ def schema_textual_representation(subentry, schema):
if subentry.inherent_sie.name == 'true':
result += ' się:'
else:
- rerult += ':'
+ result += ':'
opinion = schema.opinion.key
if opinion == 'vul':
@@ -476,7 +476,7 @@ def schema_textual_representation(subentry, schema):
if subentry.predicativity.name == 'true':
result += ' pred:'
else:
- rerult += ':'
+ result += ':'
if subentry.aspect is not None:
result += ' ' + subentry.aspect.name + ':'
@@ -498,15 +498,6 @@ def write_schema(parent, subentry, subentry_id, schema, schema_id):
schema_fs.attrib[etree.QName(XML_NAMESPACE, 'id')] = schema_xml_id
schema_fs.attrib['type'] = 'schema'
- # textual representation @TODO -- not present in the database
- # text_rep_f_elem = etree.SubElement(schema_fs_elem, 'f')
- # text_rep_f_elem.attrib['name'] = 'text_rep'
- # text_rep_string = etree.SubElement(text_rep_f_elem, 'string')
- # text_rep = schema.get_position_spaced_text_rep()
- # if schema.characteristics.filter(type=u'ZWROTNOŚĆ', value__value=u'się').exists():
- # text_rep = ' ' + text_rep
- # text_rep_string.text = lemma.entry_obj.name + text_rep.replace(':',': ')
-
# schema opinion
schema_opinion = schema.opinion.key
if schema.opinion.key is None:
@@ -547,20 +538,22 @@ def write_schema(parent, subentry, subentry_id, schema, schema_id):
# positions
write_positions(schema_fs, subentry_id, schema, schema_id)
-def write_positions(parent, subentry_id, schema, schema_id):
+def write_positions(parent, subentry_id, schema, schema_id, fake_id=False):
positions = schema.positions.all()
positions_f = etree.SubElement(parent, 'f')
positions_f.attrib['name'] = 'positions'
vColl = etree.SubElement(positions_f, 'vColl')
vColl.attrib['org'] = 'set'
for position in positions:
- write_position(vColl, subentry_id, schema_id, position)
+ write_position(vColl, subentry_id, schema_id, position, fake_id)
-def write_position(parent, subentry_id, schema_id, position):
- position_xml_id = u'unif_%d.%d.%d-psn' %(subentry_id, schema_id, position.id)
+def write_position(parent, subentry_id, schema_id, position, fake_id):
+ if not fake_id:
+ position_xml_id = u'unif_%d.%d.%d-psn' %(subentry_id, schema_id, position.id)
position_fs = etree.SubElement(parent, 'fs')
position_fs.attrib['type'] = 'position'
- position_fs.attrib[etree.QName(XML_NAMESPACE, 'id')] = position_xml_id
+ if not fake_id:
+ position_fs.attrib[etree.QName(XML_NAMESPACE, 'id')] = position_xml_id
text_rep_f = etree.SubElement(position_fs, 'f')
text_rep_f.attrib['name'] = 'textual_representation'
@@ -571,7 +564,7 @@ def write_position(parent, subentry_id, schema_id, position):
write_control(position_fs, position)
- write_phrases(position_fs, subentry_id, schema_id, position)
+ write_phrases(position_fs, subentry_id, schema_id, position, fake_id)
def write_function(parent, position):
function = position.function
@@ -598,24 +591,28 @@ def write_control(parent, position):
pred_control_symbol = etree.SubElement(vColl, 'symbol')
pred_control_symbol.attrib['value'] = control
-def write_phrases(parent, subentry_id, schema_id, position):
+def write_phrases(parent, subentry_id, schema_id, position, fake_id):
phrases = position.phrase_types.all()
phrases_f = etree.SubElement(parent, 'f')
phrases_f.attrib['name'] = 'phrases'
vColl = etree.SubElement(phrases_f, 'vColl')
vColl.attrib['org'] = 'set'
for phrase in phrases:
- write_phrase(vColl, subentry_id, schema_id, position, phrase)
+ write_phrase(vColl, subentry_id, schema_id, position, phrase, fake_id)
-def write_phrase(parent, subentry_id, schema_id, position, phrase):
- phrase_xml_id = u'unif_%d.%d.%d.%d-phr' %(subentry_id, schema_id, position.id, phrase.id)
+def write_phrase(parent, subentry_id, schema_id, position, phrase, fake_id):
+ if not fake_id:
+ phrase_xml_id = u'unif_%d.%d.%d.%d-phr' %(subentry_id, schema_id, position.id, phrase.id)
phrase_fs = etree.SubElement(parent, 'fs')
- phrase_fs.attrib[etree.QName(XML_NAMESPACE, 'id')] = phrase_xml_id
+ if not fake_id:
+ phrase_fs.attrib[etree.QName(XML_NAMESPACE, 'id')] = phrase_xml_id
phrase_fs.attrib['type'] = phrase.main_type.name
# @TODO -- currently no expansions file
# if phrase.realizations.exists() and write_expansions_id:
# write_expansions_link(phrase_fs_elem, phrase)
write_phrase_textrep(phrase_fs, phrase)
+
+ write_phrase_structure(phrase_fs, phrase)
def write_phrase_textrep(parent, phrase):
text_rep = phrase.text_rep
@@ -623,4 +620,586 @@ def write_phrase_textrep(parent, phrase):
textrep_f.attrib['name'] = 'textual_representation'
textrep_string = etree.SubElement(textrep_f, 'string')
textrep_string.text = text_rep
+
+def write_phrase_structure(parent, phrase):
+ phrase_type = phrase.main_type
+ phrase_type_fs = etree.SubElement(parent, 'fs')
+ phrase_type_fs.attrib['type'] = phrase_type.name
+ attributes = phrase.attributes
+
+ if phrase_type.name == 'np':
+ write_np_attributes(phrase_type_fs, attributes.npattributes)
+ elif phrase_type.name == 'adjp':
+ write_adjp_attributes(phrase_type_fs, attributes.adjpattributes)
+ elif phrase_type.name == 'possp':
+ write_possp_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'compar':
+ write_compar_attributes(phrase_type_fs, attributes.comparattributes)
+ elif phrase_type.name == 'prepnp':
+ write_prepnp_attributes(phrase_type_fs, attributes.prepnpattributes)
+ elif phrase_type.name == 'prepadjp':
+ write_prepadjp_attributes(phrase_type_fs, attributes.prepadjpattributes)
+ elif phrase_type.name == 'comprepnp':
+ write_comprepnp_attributes(phrase_type_fs, attributes.comprepnpattributes)
+ elif phrase_type.name == 'cp':
+ write_cp_attributes(phrase_type_fs, attributes.cpattributes)
+ elif phrase_type.name == 'ncp':
+ write_ncp_attributes(phrase_type_fs, attributes.ncpattributes)
+ elif phrase_type.name == 'prepncp':
+ write_prepncp_attributes(phrase_type_fs, attributes.prepncpattributes)
+ elif phrase_type.name == 'nonch':
+ write_nonch_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'infp':
+ write_infp_attributes(phrase_type_fs, attributes.infpattributes)
+ elif phrase_type.name == 'advp':
+ try: # due to database error
+ a = attributes.advpattributes
+ except:
+ a = attributes.xpattributes
+ write_advp_attributes(phrase_type_fs, a)
+ elif phrase_type.name == 'xp':
+ write_xp_attributes(phrase_type_fs, attributes.xpattributes)
+ elif phrase_type.name == 'E':
+ write_e_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'or':
+ write_or_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'refl':
+ write_refl_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'recip':
+ write_recip_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'distrp':
+ write_distrp_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'nump':
+ write_nump_attributes(phrase_type_fs, attributes.numpattributes)
+ elif phrase_type.name == 'prepnump':
+ write_prepnump_attributes(phrase_type_fs, attributes.prepnumpattributes)
+ elif phrase_type.name == 'pactp':
+ write_pactp_attributes(phrase_type_fs, attributes.pactpattributes)
+ elif phrase_type.name == 'preppactp':
+ write_preppactp_attributes(phrase_type_fs, attributes.preppactpattributes)
+ elif phrase_type.name == 'ppasp':
+ write_ppasp_attributes(phrase_type_fs, attributes.ppaspattributes)
+ elif phrase_type.name == 'prepppasp':
+ write_prepppasp_attributes(phrase_type_fs, attributes.prepppaspattributes)
+ elif phrase_type.name == 'gerp':
+ write_gerp_attributes(phrase_type_fs, attributes.gerpattributes)
+ elif phrase_type.name == 'prepgerp':
+ write_prepgerp_attributes(phrase_type_fs, attributes.prepgerpattributes)
+ elif phrase_type.name == 'qub':
+ write_qub_attributes(phrase_type_fs, attributes.emptyattributes)
+ elif phrase_type.name == 'lex':
+ write_lex(phrase_type_fs, phrase)
+ elif phrase_type.name == 'fixed':
+ write_fixed_attributes(phrase_type_fs, attributes.fixedattributes)
+ else:
+ print(phrase_type.name)
+ raise PhraseTypeError()
+
+def write_np_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_adjp_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_possp_attributes(parent, attributes):
+ # no expansions file
+ pass
+
+def write_compar_attributes(parent, attributes):
+ cat_f = etree.SubElement(parent, 'f')
+ cat_f.attrib['name'] = 'compar_category'
+ cat_symbol = etree.SubElement(cat_f,'symbol')
+ cat_symbol.attrib['value'] = attributes.comparcat.name
+
+def write_prepnp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_prepadjp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_comprepnp_attributes(parent, attributes):
+ # expansions file does not exist
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'complex_preposition'
+ prep_string = etree.SubElement(prep_f,'string')
+ prep_string.text = attributes.comprep.name
+
+def write_cp_attributes(parent, attributes):
+ type_f = etree.SubElement(parent, 'f')
+ type_f.attrib['name'] = 'type'
+ typedef_fs = etree.SubElement(type_f, 'fs')
+ typedef_fs.attrib['type'] = 'type_def'
+ conj_f = etree.SubElement(typedef_fs, 'f')
+ conj_f.attrib['name'] = 'conjunction'
+ conj_symbol = etree.SubElement(conj_f, 'symbol')
+ conj_symbol.attrib['value'] = attributes.cptype.name
+
+ if len(attributes.cpreals.all()) > 0:
+ constr_f = etree.SubElement(typedef_fs, 'f')
+ constr_f.attrib['name'] = 'constraints'
+ vColl = etree.SubElement(constr_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for constraint in attributes.cpreals.all():
+ constraint_symbol = etree.SubElement(parent, 'symbol')
+ constraint_symbol.attrib['value'] = constraint.name
+
+def write_ncp_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+ type_f = etree.SubElement(parent, 'f')
+ type_f.attrib['name'] = 'type'
+ typedef_fs = etree.SubElement(type_f, 'fs')
+ typedef_fs.attrib['type'] = 'type_def'
+ conj_f = etree.SubElement(typedef_fs, 'f')
+ conj_f.attrib['name'] = 'conjunction'
+ conj_symbol = etree.SubElement(conj_f, 'symbol')
+ conj_symbol.attrib['value'] = attributes.cptype.name
+
+ if len(attributes.cpreals.all()) > 0:
+ constr_f = etree.SubElement(typedef_fs, 'f')
+ constr_f.attrib['name'] = 'constraints'
+ vColl = etree.SubElement(constr_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for constraint in attributes.cpreals.all():
+ constraint_symbol = etree.SubElement(parent, 'symbol')
+ constraint_symbol.attrib['value'] = constraint.name
+
+def write_prepncp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+ type_f = etree.SubElement(parent, 'f')
+ type_f.attrib['name'] = 'type'
+ typedef_fs = etree.SubElement(type_f, 'fs')
+ typedef_fs.attrib['type'] = 'type_def'
+ conj_f = etree.SubElement(typedef_fs, 'f')
+ conj_f.attrib['name'] = 'conjunction'
+ conj_symbol = etree.SubElement(conj_f, 'symbol')
+ conj_symbol.attrib['value'] = attributes.cptype.name
+
+ if len(attributes.cpreals.all()) > 0:
+ constr_f = etree.SubElement(typedef_fs, 'f')
+ constr_f.attrib['name'] = 'constraints'
+ vColl = etree.SubElement(constr_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for constraint in attributes.cpreals.all():
+ constraint_symbol = etree.SubElement(parent, 'symbol')
+ constraint_symbol.attrib['value'] = constraint.name
+
+def write_nonch_attributes(parent, attributes):
+ pass
+
+def write_infp_attributes(parent, attributes):
+ asp_f = etree.SubElement(parent, 'f')
+ asp_f.attrib['name'] = 'aspect'
+ asp_symbol = etree.SubElement(asp_f,'symbol')
+ asp_symbol.attrib['value'] = attributes.aspect.name
+
+def write_advp_attributes(parent, attributes):
+ cat_f = etree.SubElement(parent, 'f')
+ cat_f.attrib['name'] = 'category'
+ cat_symbol = etree.SubElement(cat_f,'symbol')
+ cat_symbol.attrib['value'] = attributes.advcat.name
+
+def write_xp_attributes(parent, attributes):
+ # no expansions file
+ cat_f = etree.SubElement(parent, 'f')
+ cat_f.attrib['name'] = 'category'
+ catdef_fs = etree.SubElement(cat_f, 'fs')
+ catdef_fs.attrib['type'] = 'category_def'
+ name_f = etree.SubElement(catdef_fs, 'f')
+ name_f.attrib['name'] = 'name'
+ name_symbol = etree.SubElement(name_f, 'symbol')
+ name_symbol.attrib['value'] = attributes.advcat.name
+
+ if len(attributes.reals.all()) > 0:
+ constr_f = etree.SubElement(catdef_fs, 'f')
+ constr_f.attrib['name'] = 'constraints'
+ vColl = etree.SubElement(constr_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for constraint in attributes.reals.all():
+ write_phrase_structure(vColl, constraint)
+
+def write_e_attributes(parent, attributes):
+ pass
+
+def write_or_attributes(parent, attributes):
+ pass
+
+def write_refl_attributes(parent, attributes):
+ pass
+
+def write_recip_attributes(parent, attributes):
+ pass
+
+def write_distrp_attributes(parent, attributes):
+ # no expansions file
+ pass
+
+def write_nump_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_prepnump_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_pactp_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_preppactp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_ppasp_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_prepppasp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_gerp_attributes(parent, attributes):
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_prepgerp_attributes(parent, attributes):
+ prep_f = etree.SubElement(parent, 'f')
+ prep_f.attrib['name'] = 'preposition'
+ prep_symbol = etree.SubElement(prep_f,'symbol')
+ prep_symbol.attrib['value'] = attributes.prep.name
+ case_f = etree.SubElement(parent, 'f')
+ case_f.attrib['name'] = 'case'
+ case_symbol = etree.SubElement(case_f,'symbol')
+ case_symbol.attrib['value'] = attributes.case.name
+
+def write_qub_attributes(parent, attributes):
+ pass
+
+def write_fixed_attributes(parent, attributes):
+ arg_f = etree.SubElement(parent, 'f')
+ arg_f.attrib['name'] = 'argument'
+ write_phrase_structure(arg_f, attributes.phrase)
+ fixed_f = etree.SubElement(parent, 'f')
+ fixed_f.attrib['name'] = 'string'
+ fixed_string = etree.SubElement(fixed_f, 'string')
+ fixed_string.text = attributes.text
+
+def write_lex(parent, phrase_type):
+ write_lex_argument(parent, phrase_type)
+ write_lex_parameters(parent, phrase_type)
+ write_lex_lemma(parent, phrase_type)
+ write_lex_modification(parent, phrase_type)
+
+def write_lex_argument(parent, phrase_type):
+ arg_f = etree.SubElement(parent, 'f')
+ arg_f.attrib['name'] = 'argument'
+ write_phrase_structure(arg_f, phrase_type.lexicalized_phrase)
+
+def write_lex_parameters(parent, phrase_type):
+ if phrase_type.lexicalized_phrase.main_type.name == 'np':
+ write_lexnp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexnpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'adjp':
+ write_lexadjp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexadjpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'possp':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'compar':
+ write_lexcompar_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexcomparattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepnp':
+ write_lexprepnp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexprepnpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepadjp':
+ write_lexprepadjp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexprepadjpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'comprepnp':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'cp':
+ write_lexcp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexcpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'ncp':
+ write_lexncp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexncpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepncp':
+ # not present in the database
+ write_lexprepncp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexprepncpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'nonch':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'infp':
+ write_lexinfp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexinfpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'advp':
+ write_lexadvp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexadvpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'xp':
+ write_lex_parameters(parent, phrase_type.attributes.lexphrasetypeattributes.lexxpattributes.lex)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'E':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'or':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'refl':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'recip':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'distrp':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'nump':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepnump':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'pactp':
+ write_lexpactp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexpactpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'preppactp':
+ # not present in database
+ write_lexpreppactp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexpreppactpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'ppasp':
+ write_lexppasp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexppaspattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepppasp':
+ write_lexprepppasp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexprepppaspattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'gerp':
+ # not present in database
+ write_lexgerp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexgerpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'prepgerp':
+ write_lexprepgerp_attributes(parent, phrase_type.attributes.lexphrasetypeattributes.lexprepgerpattributes)
+ elif phrase_type.lexicalized_phrase.main_type.name == 'qub':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'lex':
+ pass
+ elif phrase_type.lexicalized_phrase.main_type.name == 'fixed':
+ pass
+ else:
+ print(phrase_type.name)
+ raise LexicalisedPhraseTypeError()
+
+def write_lex_lemma(parent, phrase_type):
+ if phrase_type.lexicalized_phrase.main_type.name == 'compar':
+ return
+ pt = phrase_type
+ if phrase_type.lexicalized_phrase.main_type.name == 'xp':
+ pt = phrase_type.attributes.lexphrasetypeattributes.lexxpattributes.lex
+ print(phrase_type.id)
+ lemma_f = etree.SubElement(parent, 'f')
+ lemma_f.attrib['name'] = 'lemma'
+ lemma_fs = etree.SubElement(lemma_f, 'fs')
+ lemma_fs.attrib['type'] = 'lemma_def'
+
+ select_f = etree.SubElement(lemma_fs, 'f')
+ select_f.attrib['name'] = 'selection_mode'
+ select_symbol = etree.SubElement(select_f, 'symbol')
+ select_symbol.attrib['value'] = pt.lemma_operator.name
+
+ cooc_f = etree.SubElement(lemma_fs, 'f')
+ cooc_f.attrib['name'] = 'selection_mode'
+ cooc_symbol = etree.SubElement(cooc_f, 'symbol')
+ cooc_symbol.attrib['value'] = pt.lemma_cooccur.name
+
+ lemmas_f = etree.SubElement(lemma_fs, 'f')
+ lemmas_f.attrib['name'] = 'lemmas'
+ vColl = etree.SubElement(lemmas_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for lemma in pt.lemmata.all():
+ lemma_string = etree.SubElement(vColl, 'string')
+ lemma_string.text = lemma.name
+
+def write_lex_modification(parent, phrase_type):
+ if phrase_type.lexicalized_phrase.main_type.name == 'compar':
+ return
+ pt = phrase_type
+ if phrase_type.lexicalized_phrase.main_type.name == 'xp':
+ pt = phrase_type.attributes.lexphrasetypeattributes.lexxpattributes.lex
+ mod_f = etree.SubElement(parent, 'f')
+ mod_f.attrib['name'] = 'modification'
+ mod_fs = etree.SubElement(mod_f, 'fs')
+ mod_fs.attrib['type'] = 'modification_def'
+ type_f = etree.SubElement(mod_fs, 'f')
+ type_f.attrib['name'] = 'type'
+ type_symbol = etree.SubElement(type_f, 'symbol')
+ type_symbol.attrib['value'] = pt.modification.mod_type.name
+
+ if len(pt.modification.positions.all()) > 0:
+ positions = pt.modification.positions.all()
+ positions_f = etree.SubElement(parent, 'f')
+ positions_f.attrib['name'] = 'positions'
+ vColl = etree.SubElement(positions_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for position in positions:
+ write_position(vColl, None, None, position, True)
+
+def write_lexnp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+
+def write_lexadjp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+ gen_f = etree.SubElement(parent, 'f')
+ gen_f.attrib['name'] = 'gender'
+ gen_symbol = etree.SubElement(gen_f, 'symbol')
+ gen_symbol.attrib['value'] = attributes.gend.name
+ deg_f = etree.SubElement(parent, 'f')
+ deg_f.attrib['name'] = 'degree'
+ deg_symbol = etree.SubElement(deg_f, 'symbol')
+ deg_symbol.attrib['value'] = attributes.deg.name
+
+def write_lexcompar_attributes(parent, attributes):
+ arg_f = etree.SubElement(parent, 'f')
+ arg_f.attrib['argument'] = 'arguments'
+ vColl = etree.SubElement(arg_f, 'vColl')
+ vColl.attrib['org'] = 'set'
+ for lex in attributes.lexes.all():
+ write_phrase_structure(vColl, lex)
+def write_lexprepnp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+
+def write_lexprepadjp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+ gen_f = etree.SubElement(parent, 'f')
+ gen_f.attrib['name'] = 'gender'
+ gen_symbol = etree.SubElement(gen_f, 'symbol')
+ gen_symbol.attrib['value'] = attributes.gend.name
+ deg_f = etree.SubElement(parent, 'f')
+ deg_f.attrib['name'] = 'degree'
+ deg_symbol = etree.SubElement(deg_f, 'symbol')
+ deg_symbol.attrib['value'] = attributes.deg.name
+
+def write_lexcp_attributes(parent, attributes):
+ neg_f = etree.SubElement(parent, 'f')
+ neg_f.attrib['name'] = 'negativity'
+ neg_symbol = etree.SubElement(neg_f, 'symbol')
+ neg_symbol.attrib['value'] = attributes.num.name
+ inh_f = etree.SubElement(parent, 'f')
+ inh_f.attrib['name'] = 'inherent_sie'
+ inh_symbol = etree.SubElement(inh_f, 'symbol')
+ inh_symbol.attrib['value'] = attributes.inhsie.name
+
+def write_lexncp_attributes(parent, attributes):
+ neg_f = etree.SubElement(parent, 'f')
+ neg_f.attrib['name'] = 'negativity'
+ neg_symbol = etree.SubElement(neg_f, 'symbol')
+ neg_symbol.attrib['value'] = attributes.num.name
+ inh_f = etree.SubElement(parent, 'f')
+ inh_f.attrib['name'] = 'inherent_sie'
+ inh_symbol = etree.SubElement(inh_f, 'symbol')
+ inh_symbol.attrib['value'] = attributes.inhsie.name
+
+def write_lexprepncp_attributes(parent, attributes):
+ # not present in database
+ pass
+
+def write_lexinfp_attributes(parent, attributes):
+ neg_f = etree.SubElement(parent, 'f')
+ neg_f.attrib['name'] = 'negativity'
+ neg_symbol = etree.SubElement(neg_f, 'symbol')
+ neg_symbol.attrib['value'] = attributes.num.name
+ inh_f = etree.SubElement(parent, 'f')
+ inh_f.attrib['name'] = 'inherent_sie'
+ inh_symbol = etree.SubElement(inh_f, 'symbol')
+ inh_symbol.attrib['value'] = attributes.inhsie.name
+
+def write_lexadvp_attributes(parent, attributes):
+ deg_f = etree.SubElement(parent, 'f')
+ deg_f.attrib['name'] = 'degree'
+ deg_symbol = etree.SubElement(deg_f, 'symbol')
+ deg_symbol.attrib['value'] = attributes.deg.name
+
+def write_lexpactp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+ gen_f = etree.SubElement(parent, 'f')
+ gen_f.attrib['name'] = 'gender'
+ gen_symbol = etree.SubElement(gen_f, 'symbol')
+ gen_symbol.attrib['value'] = attributes.gend.name
+
+def write_lexpreppactp_attributes(parent, attributes):
+ # not present in database
+ pass
+
+def write_lexppasp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+ gen_f = etree.SubElement(parent, 'f')
+ gen_f.attrib['name'] = 'gender'
+ gen_symbol = etree.SubElement(gen_f, 'symbol')
+ gen_symbol.attrib['value'] = attributes.gend.name
+
+def write_lexprepppasp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
+ gen_f = etree.SubElement(parent, 'f')
+ gen_f.attrib['name'] = 'gender'
+ gen_symbol = etree.SubElement(gen_f, 'symbol')
+ gen_symbol.attrib['value'] = attributes.gend.name
+
+def write_lexgerp_attributes(parent, attributes):
+ # not present in database
+ pass
+
+def write_lexprepgerp_attributes(parent, attributes):
+ num_f = etree.SubElement(parent, 'f')
+ num_f.attrib['name'] = 'number'
+ num_symbol = etree.SubElement(num_f, 'symbol')
+ num_symbol.attrib['value'] = attributes.num.name
--
GitLab