From e20a6dcd6c074dabf513f6a82cd93f85822a13eb Mon Sep 17 00:00:00 2001
From: dcz <dcz@ipipan.waw.pl>
Date: Thu, 9 Jun 2022 16:36:44 +0200
Subject: [PATCH] Importing data with original primary keys
---
data/unification/unify_example_small.xml | 18 +++++++++---------
.../templates/unification_edit_display.html | 4 ++--
importer/Argument.py | 1 +
importer/Entry.py | 3 ++-
importer/Example.py | 6 ++++--
importer/Frame.py | 1 +
importer/unification/UnifiedFrameImport.py | 2 +-
.../management/commands/import_plWordnet.py | 2 +-
unifier/models.py | 4 +---
9 files changed, 22 insertions(+), 19 deletions(-)
diff --git a/data/unification/unify_example_small.xml b/data/unification/unify_example_small.xml
index 3276082..4dbecd2 100644
--- a/data/unification/unify_example_small.xml
+++ b/data/unification/unify_example_small.xml
@@ -33,22 +33,22 @@
</roles>
</argument>
<connections>
- <slowal_frame id="1">
+ <slowal_frame id="104274">
<arguments_connections>
- <arguments_connection unifier_argument_id="1" slowal_id="1"/>
- <arguments_connection unifier_argument_id="2" slowal_id="2"/>
- <arguments_connection unifier_argument_id="3" slowal_id="3"/>
+ <arguments_connection unifier_argument_id="1" slowal_id="141952"/>
+ <arguments_connection unifier_argument_id="2" slowal_id="141953"/>
+ <arguments_connection unifier_argument_id="3" slowal_id="143669"/>
</arguments_connections>
</slowal_frame>
- <slowal_frame id="2">
+ <slowal_frame id="104232">
<arguments_connections>
- <arguments_connection unifier_argument_id="1" slowal_id="5"/>
- <arguments_connection unifier_argument_id="2" slowal_id="4"/>
+ <arguments_connection unifier_argument_id="1" slowal_id="141913"/>
+ <arguments_connection unifier_argument_id="2" slowal_id="141914"/>
</arguments_connections>
</slowal_frame>
- <slowal_frame id="3">
+ <slowal_frame id="104238">
<arguments_connections>
- <arguments_connection unifier_argument_id="3" slowal_id="6"/>
+ <arguments_connection unifier_argument_id="3" slowal_id="141916"/>
</arguments_connections>
</slowal_frame>
</connections>
diff --git a/entries/templates/unification_edit_display.html b/entries/templates/unification_edit_display.html
index bae37f4..af4a9b0 100644
--- a/entries/templates/unification_edit_display.html
+++ b/entries/templates/unification_edit_display.html
@@ -51,7 +51,7 @@
Błędna
</td>
<td id="extract-frame" style="padding: 10px 15px 10px 15px; color: #000000;"
- onclick="extract_frames_to_new_frame(1, [1,2], null)">
+ onclick="extract_frames_to_new_frame(1, [104274,104238], null)">
Rozdziel
</td>
<td id="hide-slowal-frame" style="padding: 10px 15px 10px 15px; color: #000000;">
@@ -69,7 +69,7 @@
NiepasujÄ…ca
</td>
<td id="move-slowal-frame" style="padding: 10px 15px 10px 15px; color: #000000;"
- onclick="extract_frames_to_new_frame(7, [1], 1)">
+ onclick="extract_frames_to_new_frame(3, [104274], 1)">
Przerzuć
</td>
<td id="show-slowal-frame" style="padding: 10px 15px 10px 15px; color: #000000;">
diff --git a/importer/Argument.py b/importer/Argument.py
index 956d304..7cf48ff 100644
--- a/importer/Argument.py
+++ b/importer/Argument.py
@@ -158,6 +158,7 @@ class Argument:
argument = semantics.models.Argument(role=role,
frame=frame,
preferences_count=len(self._selectional_preferences))
+ argument.pk = int(self._id.split(".")[1].split("-")[0])
argument.save()
self._db_id = argument.id
return argument
diff --git a/importer/Entry.py b/importer/Entry.py
index ad552b7..9bd173e 100644
--- a/importer/Entry.py
+++ b/importer/Entry.py
@@ -12,6 +12,7 @@ import connections.models
class Entry:
def __init__(self, entry_tree, entry_meanings, meanings, frames, examples_in_data, examples_out_file, misconnected_out_file):
+ self._id = int(entry_tree._attrs['xml:id'].split("_")[1].split("-")[0])
self._base = entry_tree._children[0]._children[0]._content
self._pos = entry_tree._children[0]._children[1]._content
print("processing: " + self._base)
@@ -29,7 +30,7 @@ class Entry:
pos = POS.objects.get(tag=self._pos)
status = Status.objects.get(key=self._status)
# @TODO: nie ma frequency w xml-u
- entry = connections.models.Entry(name=self._base, pos=pos, status=status, frequency_1M=0, frequency_300M=0)
+ entry = connections.models.Entry(pk=self._id, name=self._base, pos=pos, status=status, frequency_1M=0, frequency_300M=0)
entry.save()
try:
self._syntax.store(entry, stored_positions)
diff --git a/importer/Example.py b/importer/Example.py
index bd93adb..4623bd1 100644
--- a/importer/Example.py
+++ b/importer/Example.py
@@ -14,7 +14,8 @@ def clean_sentence(sentence):
class Example:
- def __init__(self, sentence, source, opinion, note, illustrated_syntax, meaning, illustrated_semantics):
+ def __init__(self, eid, sentence, source, opinion, note, illustrated_syntax, meaning, illustrated_semantics):
+ self._eid = eid
self._sentence = sentence
self._source = source
self._opinion = opinion
@@ -66,7 +67,7 @@ class Example:
print(example_tree)
raise UnknownError()
- return cls(sentence, source, opinion, note, illustrated_syntax, meaning, illustrated_semantics)
+ return cls(eid, sentence, source, opinion, note, illustrated_syntax, meaning, illustrated_semantics)
def store(self, entry, meanings):
# self._phrases = illustrated_syntax
@@ -79,6 +80,7 @@ class Example:
opinion=opinion,
source=source,
note=self._note)
+ example.pk = self._eid
example.save()
if not self._phrases.exists():
print(' STORING EXAMPLE:', self._sentence)
diff --git a/importer/Frame.py b/importer/Frame.py
index d728626..ee6f94b 100644
--- a/importer/Frame.py
+++ b/importer/Frame.py
@@ -52,6 +52,7 @@ class Frame:
if len(frames) == 0 and self._base is not None:
opinion = FrameOpinion.objects.get(key=self._opinion)
frame = semantics.models.Frame(opinion=opinion)
+ frame.pk = int(self._id.split("_")[1].split("-")[0])
frame.save()
self._db_id = frame.id
arguments = {}
diff --git a/importer/unification/UnifiedFrameImport.py b/importer/unification/UnifiedFrameImport.py
index 7774f8f..cef6d69 100644
--- a/importer/unification/UnifiedFrameImport.py
+++ b/importer/unification/UnifiedFrameImport.py
@@ -35,7 +35,7 @@ class UnifiedFrameImport:
role_type_obj = UnifiedFrameImport.storeAndGetRoleType(role_type)
role_type_obj.save()
- argument = UnifiedFrameArgument(id=id,
+ argument = UnifiedFrameArgument(pk=int(id),
role_type=role_type_obj,
unified_frame=unifiedFrame)
argument.proposed_roles.set([])
diff --git a/meanings/management/commands/import_plWordnet.py b/meanings/management/commands/import_plWordnet.py
index 4cf34a6..9449a47 100644
--- a/meanings/management/commands/import_plWordnet.py
+++ b/meanings/management/commands/import_plWordnet.py
@@ -72,7 +72,7 @@ class PlWNHandler(handler.ContentHandler):
if desc == 'brak danych':
desc = ''
pos = POS_MAP[pos]
- lu = LexicalUnit(luid=luid, base=lubase, sense=lusense, pos=pos, synset=s, gloss='', definition=desc, text_rep='{}-{}'.format(lubase, lusense))
+ lu = LexicalUnit(pk=luid, luid=luid, base=lubase, sense=lusense, pos=pos, synset=s, gloss='', definition=desc, text_rep='{}-{}'.format(lubase, lusense))
# print luid, lubase, lusense
self._lexical_units_to_base.append(lu)
self._unit = False
diff --git a/unifier/models.py b/unifier/models.py
index 98efe6d..4962b3e 100644
--- a/unifier/models.py
+++ b/unifier/models.py
@@ -43,7 +43,7 @@ class UnifiedFrame(models.Model):
cnt = UnifiedFrameArgument.objects.count()+1
old_2_new_argument_mapping = {}
for unified_frame_argument in unified_frame_arguments:
- new_unified_frame_argument = UnifiedFrameArgument.objects.create(id=cnt,
+ new_unified_frame_argument = UnifiedFrameArgument.objects.create(pk=cnt,
role_type=unified_frame_argument.role_type,
role=unified_frame_argument.role,
unified_frame=new_frame)
@@ -79,8 +79,6 @@ class UnifiedFrame(models.Model):
class UnifiedFrameArgument(models.Model):
- id = models.CharField(max_length=20, primary_key=True)
-
role_type = models.ForeignKey(RoleType, on_delete=models.PROTECT)
#rola - wybrana przez użytkownika
role = models.ForeignKey(ArgumentRole, on_delete=models.PROTECT, default=None, blank=True, null=True)
--
GitLab