Merge pull request #59 from CentreForDigitalHumanities/feature/derived-problems

Feature/derived problems

Author: XanderVertegaal

backend/problem/management/commands/import_fracas.py

@@ -43,6 +43,9 @@ def _annotate_section_subsections(tree: ET.ElementTree) -> None:
 
         root = tree.getroot()
 
+        if root is None:
+            raise ValueError("The XML file is empty or malformed.")
+
         for element in root:
             if element.tag == "comment" and element.attrib.get("class") == "section":
                 current_section = element.text.strip() if element.text else None
@@ -107,15 +110,15 @@ def import_fracas_problems(self, fracas_path: str) -> None:
                 )
                 skipped += 1
                 continue
-            hypothesis = Sentence.objects.create(
+            hypothesis = Sentence.objects.update_or_create(
                 text=FracasData._text_from_element(hypothesis_node)
-            )
+            )[0]
 
             premises = []
             premise_nodes = problem.findall("p")
             for node in premise_nodes:
                 if node.text:
-                    premises.append(Sentence.objects.create(text=node.text.strip()))
+                    premises.append(Sentence.objects.update_or_create(text=node.text.strip())[0])
 
 
             entailment_label = self.ENTAILMENT_LABELS.get(

backend/problem/management/commands/import_sick.py

@@ -57,9 +57,8 @@ def import_sick_problems(self, sick_path: str) -> None:
 
                 extra_data = SickData.import_data(problem)
 
-                premise = Sentence.objects.create(text=problem["sentence_A"])
-
-                hypothesis = Sentence.objects.create(text=problem["sentence_B"])
+                premise = Sentence.objects.update_or_create(text=problem["sentence_A"])[0]
+                hypothesis = Sentence.objects.update_or_create(text=problem["sentence_B"])[0]
 
                 problem = Problem.objects.create(
                     dataset=Problem.Dataset.SICK,

backend/problem/management/commands/import_snli.py

@@ -1,4 +1,5 @@
 import csv
+from typing import Literal
 
 from django.core.management.base import BaseCommand
 from tqdm import tqdm
@@ -53,7 +54,9 @@ def handle(self, *args, **options):
 
         self.import_snli_problems(snli_paths)
 
-    def import_snli_problems(self, snli_paths: list[tuple[str, str]]) -> None:
+    def import_snli_problems(
+        self, snli_paths: list[tuple[Literal["dev", "train", "test"], str]]
+    ) -> None:
         """
         Import SNLI 1.0 problems from a list of SNLI TSV files and enter them into the database.
         """
@@ -99,9 +102,13 @@ def import_snli_problems(self, snli_paths: list[tuple[str, str]]) -> None:
 
                         extra_data = SNLIData.import_data(problem, subset)
 
-                        premise = Sentence.objects.create(text=problem["sentence1"])
+                        premise = Sentence.objects.update_or_create(
+                            text=problem["sentence1"]
+                        )[0]
 
-                        hypothesis = Sentence.objects.create(text=problem["sentence2"])
+                        hypothesis = Sentence.objects.update_or_create(
+                            text=problem["sentence2"]
+                        )[0]
 
                         new_problem = Problem.objects.create(
                             dataset=Problem.Dataset.SNLI,

backend/problem/migrations/0005_problem_base.py

@@ -0,0 +1,26 @@
+# Generated by Django 4.2.20 on 2025-09-22 11:47
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("problem", "0004_sentence_remove_problem_premises_knowledgebase_and_more"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="problem",
+            name="base",
+            field=models.ForeignKey(
+                blank=True,
+                help_text="The base problem from which this problem was derived, if any.",
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="derived_problems",
+                to="problem.problem",
+            ),
+        ),
+    ]

backend/problem/models.py

@@ -28,6 +28,15 @@ class EntailmentLabel(models.TextChoices):
         default=Dataset.USER,
     )
 
+    base = models.ForeignKey(
+        "self",
+        on_delete=models.SET_NULL,
+        null=True,
+        blank=True,
+        related_name="derived_problems",
+        help_text="The base problem from which this problem was derived, if any.",
+    )
+
     premises = models.ManyToManyField(
         Sentence,
         related_name="premise_problems",
@@ -56,8 +65,6 @@ def get_index(self, qs: QuerySet) -> int | None:
         except Exception as e:
             logger.exception(f"Error getting index for problem {self.pk}: {e}")
             return None
-
-
 class KnowledgeBase(models.Model):
     class Relationship(models.TextChoices):
         EQUAL = "equal", "Equal"

backend/problem/serializers.py

@@ -62,6 +62,7 @@ class Meta:
             "entailmentLabel",
             "extraData",
             "kbItems",
+            "base",
         ]
 
     def get_premises(self, problem):
@@ -104,6 +105,7 @@ def create(self, validated_data: dict) -> Problem:
         )[0]
 
         problem = Problem.objects.create(
+            base_id=validated_data.get("base", None),
             hypothesis=hypothesis_sentence,
             dataset=Problem.Dataset.USER,
             # TODO: Determine entailment label based on LangPro parser output.
@@ -132,6 +134,19 @@ def update(self, instance: Problem, validated_data: dict) -> Problem:
         instance.hypothesis = Sentence.objects.get_or_create(
             text=validated_data["hypothesis"],
         )[0]
+
+        validated_base_id = validated_data.get("base", None)
+        if validated_base_id is None:
+            instance.base = None
+        else:
+            try:
+                base_problem = Problem.objects.get(id=validated_base_id)
+            except Problem.DoesNotExist:
+                raise serializers.ValidationError(
+                    f"Base problem with ID {validated_base_id} does not exist."
+                )
+            instance.base = base_problem # type: ignore
+
         instance.save()
 
         premise_sentences = [
@@ -188,6 +203,8 @@ class ProblemInputSerializer(serializers.Serializer):
         many=True, allow_empty=True, help_text="List of knowledge base items"
     )
 
+    base = serializers.IntegerField(required=False, allow_null=True)
+
     def validate_id(self, value):
         """Validate that the Problem ID, if provided, exists and belongs to a user-created problem."""
         if value is not None:
@@ -198,3 +215,12 @@ def validate_id(self, value):
                     f"Problem with ID {value} does not exist."
                 )
         return value
+
+    def validate_base(self, value):
+        """Validate that the base problem ID exists if provided."""
+        if value is not None:
+            if not Problem.objects.filter(id=value).exists():
+                raise serializers.ValidationError(
+                    f"Base problem with ID {value} does not exist."
+                )
+        return value

backend/problem/views/problem.py

@@ -55,7 +55,7 @@ def _get_problem_response(self, request: Request, pk: int | None) -> Response:
         qs = self.get_queryset()
 
         if filters is not None:
-            qs = qs.filter(filters)
+            qs = qs.filter(filters).distinct()
 
         problem = None
         if pk is not None:
@@ -111,7 +111,7 @@ def _handle_update_create_problem(
         problem_serializer = ProblemSerializer()
 
         if problem_id is None:
-            problem = problem_serializer.create(validated_input)
+            problem = problem_serializer.create(validated_input) # type: ignore
             status = HTTP_201_CREATED
         else:
             problem_instance = get_object_or_404(

frontend/src/app/annotate/annotate.component.html

@@ -4,52 +4,39 @@
 @let adding = appMode === 'add';
 @let editing = appMode === 'edit';
 
-
 <div class="row mb-4">
     <div class="d-flex flex-column col-4 gap-2">
         @if (browsing) {
         <div class="card text-bg-light p-3">
             <la-navigator />
             <la-search />
         </div>
-        } @else if (adding) {
+        } @else if (adding || editing) {
         <section class="d-flex flex-column gap-2">
+            @if (adding) {
             <p class="text-muted" i18n>
                 You are currently adding a new problem.
             </p>
-            <p class="text-muted" i18n>
-                Hit 'Start Parse' to send your problem data to the parser and
-                inspect the results.
-            </p>
-            <p class="text-muted" i18n>
-                If you are happy with the changes, save the problem by clicking
-                the 'Save' button.
-            </p>
-            <p class="text-muted" i18n>
-                This will add the problem to the database.
-            </p>
-        </section>
-        } @else if (editing) {
-        <section>
+            } @else {
             <p class="text-muted" i18n>
                 You are currently editing an existing problem.
             </p>
+            }
             <p class="text-muted" i18n>
                 Hit 'Start Parse' to send your problem data to the parser and
                 inspect the results.
             </p>
             <p class="text-muted" i18n>
-                If you are happy with the changes, save the changes by clicking
+                If you are happy with the changes, save the problem by clicking
                 the 'Save problem' button.
             </p>
-            @if (isUserProblem$ | async) {
+            @if (adding) {
             <p class="text-muted" i18n>
-                This will update this problem in the database.
+                This will add the problem to the database.
             </p>
             } @else {
             <p class="text-muted" i18n>
-                This will create a new problem in the database, leaving the
-                original problem unchanged.
+                This will update this problem in the database.
             </p>
             }
         </section>

frontend/src/app/annotate/annotate.component.ts

@@ -11,6 +11,7 @@ import { combineLatest, distinctUntilChanged, map } from "rxjs";
 import { CommonModule } from "@angular/common";
 import { Dataset } from "@/types";
 import { takeUntilDestroyed } from "@angular/core/rxjs-interop";
+import areParamsEqual from "@/shared/areParamsEqual";
 
 @Component({
     selector: "la-annotate",
@@ -54,6 +55,7 @@ export class AnnotateComponent implements OnInit {
             editParam$
         ])
             .pipe(
+                distinctUntilChanged((oldParams, newParams) => areParamsEqual(oldParams, newParams)),
                 takeUntilDestroyed(this.destroyRef)
             )
             .subscribe(([params, queryParams, edit]) => {

frontend/src/app/annotate/annotation-input/annotation-input.component.html

@@ -12,32 +12,35 @@
             <la-knowledge-base-form [form]="form" />
         </div>
         <div class="d-flex justify-content-between mt-3">
-            <button
-                type="button"
-                class="btn btn-success d-flex align-items-center justify-content-center"
+            <la-icon-button
+                buttonStyle="success"
+                [icon]="faTree"
+                label="Start parse"
                 (click)="startParse()"
+            />
+            @if (appMode === 'browse') {
+            <a
+                class="btn btn-secondary d-flex align-items-center justify-content-center"
+                [routerLink]="['/', 'annotate', 'new']"
+                [queryParams]="{base: form.getRawValue().id }"
+
             >
                 <fa-icon
-                    [icon]="faTree"
+                    [icon]="faCopy"
                     class="fa-md"
                     aria-hidden="true"
                     focusable="false"
                 />
-                <span class="ms-2" i18n>Start parse</span>
-            </button>
+                <span class="ms-2" i18n>Copy problem</span>
+            </a>
+            }
             @if ((appMode === 'edit' || appMode === 'add') && userProblem) {
-            <button
-                type="button"
-                class="btn btn-primary d-flex align-items-center justify-content-center"
+            <la-icon-button
+                buttonStyle="primary"
+                [icon]="faFloppyDisk"
+                label="Save problem"
                 (click)="saveProblem()"
-            >
-                <fa-icon
-                    [icon]="faFloppyDisk"
-                    aria-hidden="true"
-                    focusable="false"
-                ></fa-icon>
-                <span class="ms-3" i18n>Save problem</span>
-            </button>
+            />
             } @else if (userProblem) {
             <a
                 class="btn btn-secondary d-flex align-items-center justify-content-center"