Merge pull request #591 from tira-io/development

Development

application/Makefile

@@ -1,7 +1,7 @@
 .PHONY: help setup run-develop build-docker clean
 
-VERSION_APPLICATION=0.0.96
-VERSION_GRPC=0.0.96
+VERSION_APPLICATION=0.0.97
+VERSION_GRPC=0.0.97
 
 .DEFAULT: help
 help:

application/src/tira/endpoints/data_api.py

@@ -1,5 +1,9 @@
 import logging
 import json
+import textwrap
+
+from django.core.exceptions import BadRequest
+
 from tira.forms import *
 import tira.tira_model as model
 from tira.checks import check_permissions, check_resources_exist, check_conditional_permissions
@@ -434,36 +438,142 @@ def add_registration(request, context, task_id, vm_id):
         return JsonResponse({'status': 0, "message": f"Encountered an exception: {e}"}, status=HTTPStatus.INTERNAL_SERVER_ERROR)
 
 
+def expand_links(component):
+    links = [*component.get('links', [])]
+    ir_datasets_id = component.get('ir_datasets_id', None)
+    if ir_datasets_id:
+        if '/' in ir_datasets_id:
+            base = ir_datasets_id.split('/')[0]
+            fragment = f'#{ir_datasets_id}'
+        else:
+            base = ir_datasets_id
+            fragment = ''
+
+        links.append({
+            'display_name': 'ir_datasets',
+            'href': f'https://ir-datasets.com/{base}.html{fragment}',
+            'target': '_blank',
+        })
+
+    tirex_submission_id = component.get('tirex_submission_id', None)
+    if tirex_submission_id:
+        links.append({
+            'display_name': 'Submission in TIREx',
+            'href': f'/submissions/{tirex_submission_id}',
+        })
+
+    if links:
+        component['links'] = links
+
+    return component
+
+
+def flatten_components(components):
+    flattened_components = []
+    for identifier, data in components.items():
+        component = {'identifier': identifier, **data}
+
+        if 'components' in component:
+            component['components'] = flatten_components(data['components'])
+
+        if 'tirex_submission_id' in data:
+            component['tirex_submission_id'] = data['tirex_submission_id']
+
+        flattened_components.append(expand_links(component))
+
+    return flattened_components
+
+
 @add_context
 def tirex_components(request, context):
-    context['tirex_components'] = settings.TIREX_COMPONENTS
+    context['tirex_components'] = flatten_components(settings.TIREX_COMPONENTS)
     return JsonResponse({'status': 0, 'context': context})
 
-def get_snippet_to_run_components(request):
-    all_components = settings.TIREX_COMPONENTS
-    component_ids = request.GET.get('components', 'false')
+def flatten_tirex_components_to_id(obj, t=None):
+    ret = {}
 
-    # All links with display_name == "Submission in TIREx" have the ID of the component in their link, its a bit ugly, but at the moment we need to extract the ID from there.
-    # E.g., the ID from the URL "/submissions/ir-benchmarks/ows/query-segmentation-hyp-a" would be ir-benchmarks/ows/query-segmentation-hyp-a
+    if type(obj) != dict:
+        return ret
 
-    # Also Ugly: we need to determine which type of processor (query processor, document processor, etc) something is by using its top-level category, e.g., "Query Processing".
+    if 'tirex_submission_id' in obj:
+        assert obj['tirex_submission_id'] not in ret
+        obj['type'] = t
+        ret[obj['tirex_submission_id']] = obj
 
-    # I think it makes sense to build a small method that uses the settings.TIREX_COMPONENTS as input and produces a mapping form component ID (the thing below "Submission in TIREx") to the properties, e.g., query processor true or false, etc.
+    for k, v in obj.items():
+        for i, j in flatten_tirex_components_to_id(v, t if t else k).items():
+            ret[i] = j  
 
-    # I think we can hard code everything against ROBUST04, we can switch this later.
-    dataset_initialization = 'dataset = pt.get_dataset("irds:disks45/nocr/trec-robust-2004")\n'
+    return ret
 
-    additional_variables = ''
 
-    # If we have a query processor, we need to add an additional variable "topics"
-    # just for this hard coded example:
-    current_component_is_query_processor = True
-    if current_component_is_query_processor:
-        additional_variables += "topics = dataset.get_topics(variant='title')\n"
+TIREX_ID_TO_COMPONENT = flatten_tirex_components_to_id(settings.TIREX_COMPONENTS)
+
+
+def get_snippet_to_run_components(request):
+    component_key = request.GET.get('component')
+
+    if component_key not in TIREX_ID_TO_COMPONENT:
+        return JsonResponse({'status': 1, 'message': f'Component "{component_key}" not found.'})
+
+    component = TIREX_ID_TO_COMPONENT[component_key]
+    component_type = component['type']
+    dataset_initialization = textwrap.dedent('''
+    # You can replace Robust04 with other datasets
+    dataset = pt.get_dataset("irds:disks45/nocr/trec-robust-2004")
+    ''').strip()
+    snippet = ''
+
+    if component_type == 'dataset':
+        dataset_initialization = ''
+        ir_datasets_id = component.get('ir_datasets_id')
+        if ir_datasets_id:
+            snippet = f'''
+            dataset = pt.get_dataset('irds:{ir_datasets_id}')
+
+            indexer = pt.IterDictIndexer('./index')
+            indexref = indexer.index(dataset.get_corpus_iter())
+            '''
+        else:
+            snippet = f'''
+            def get_corpus_iter():
+                # Iterate over the {component['display_name']} corpus
+                corpus = ...
+                for doc in corpus:
+                    yield {{'docno': doc.docno, 'text': doc.content}}
+
+            indexer = pt.IterDictIndexer('./index')
+            indexref = indexer.index(get_corpus_iter())
+            '''
+    elif component_type == 'document_processing':
+        tirex_submission_id = component.get('tirex_submission_id')
+        if tirex_submission_id:
+            snippet = f'''
+            transformed_docs = tira.pt.transform_documents('{tirex_submission_id}', dataset)
+            '''
+    elif component_type == 'query_processing':
+        tirex_submission_id = component.get('tirex_submission_id')
+        if tirex_submission_id:
+            snippet = f'''
+            topics = dataset.get_topics(variant='title')
+            transformed_queries = tira.pt.transform_queries('{tirex_submission_id}', topics)
+            '''
+    elif component_type in ('retrieval', 'reranking'):
+        tirex_submission_id = component.get('tirex_submission_id')
+        if tirex_submission_id:
+            snippet = f'''
+            run = tira.pt.from_retriever_submission('{tirex_submission_id}', dataset=dataset_id)
+            '''
+    elif component_type == 'dataset':
+        pass
+    else:
+        JsonResponse({'status': 1, 'message': f'Component type "{component_type}" does not exist...'})
 
-    component_definitions = "tira.pt.transform_queries('ir-benchmarks/ows/query-segmentation-hyb-i', dataset)\n"
+    if snippet:
+        snippet = textwrap.dedent(snippet).strip()
 
-    snippet = (dataset_initialization + additional_variables + component_definitions).strip()
+        if dataset_initialization:
+            snippet = dataset_initialization + '\n' + snippet
 
     return JsonResponse({'status': 0, 'context': {'snippet': snippet}})
 

application/src/tira/frontend-vuetify/src/IrComponents.vue

@@ -40,15 +40,16 @@
               <v-list-item v-for="link in vectorizedComponents[index][i-1].links">
                 <v-list-item-title><a :href="link.href" :target="link.target">{{ link.display_name }}</a></v-list-item-title>
               </v-list-item>
-              <v-list-item v-if="vectorizedComponents[index][i-1].code">
+              <v-list-item v-if="vectorizedComponents[index][i-1].tirex_submission_id">
                 <v-dialog>
                     <template v-slot:activator="{ props }">
-                      <v-list-item-title v-bind="props" class="show-code-button">Show code</v-list-item-title>
+                      <v-list-item-title v-bind="props" class="show-code-button" @click="fetch_code(index, i-1)">Show code</v-list-item-title>
                     </template>
                   <template v-slot:default="{ isActive }">
                     <v-card class="bg-grey-darken-3">
                       <v-card-text content="code">
-                        <code-snippet :title="'Example code snippet for ' + vectorizedComponents[index][i-1]?.display_name" :code="vectorizedComponents[index][i-1].code" expand_message=""/>
+                        <loading v-if="!code" loading="true"/>
+                        <code-snippet v-if="code" :title="'Example code snippet for ' + vectorizedComponents[index][i-1]?.display_name" :code="code" expand_message=""/>
                       </v-card-text>
 
                       <v-card-actions>
@@ -85,6 +86,7 @@ import {compareArrays, extractComponentTypesFromCurrentUrl, extractFocusTypesFro
 import CodeSnippet from "@/components/CodeSnippet.vue";
 
 interface Component {
+  identifier: string;
   display_name: string;
   components?: Component[];
   links?: { display_name: string; href: string; target: string }[];
@@ -103,8 +105,9 @@ export default {
       max_width: 1500,
       loading: true,
       tirex_components: [
-        {'display_name': 'loading', 'components': [{'display_name': 'loading'}], 'links': [{'display_name': '.', 'href': '.', 'target': '.'}]},
+        {'identifier': 'loading', 'display_name': 'loading', 'components': [{'identifier': 'loading', 'display_name': 'loading'}], 'links': [{'display_name': '.', 'href': '.', 'target': '.'}], 'tirex_submission_id': null},
       ],
+      code: '',
       colors: {
         'Dataset': 'green', 'Document Processing': 'yellow-lighten-1',
         'Query Processing': 'yellow-darken-4', 'Retrieval': 'cyan-lighten-1',
@@ -148,6 +151,11 @@ export default {
 
       return componentSet;
     },
+    fetch_code(index: number, i: number) {
+      this.code = ''
+      get('/api/tirex-snippet?component='+ this.vectorizedComponents[index][i].tirex_submission_id)
+        .then((message) => {this.code = message['context']['snippet']})
+    },
     colorOfComponent(c:string) : string {
       return this.colors[c] ?? "grey"
     },
@@ -172,8 +180,8 @@ export default {
     is_collapsed(component:any) {
       return !this.computed_expanded_entries.includes(component.display_name)
     },
-    filtered_sub_components(component:any) : {display_name: string, subItems: number, pos: number, links: any[], focus_type: string|undefined|null, component_type: string|undefined|null}[] {
-      let ret: {display_name: string, subItems: number, pos: number, links: any[], focus_type: string|undefined|null, component_type: string|undefined|null}[] = []
+    filtered_sub_components(component:any) : {display_name: string, subItems: number, pos: number, links: any[], focus_type: string|undefined|null, component_type: string|undefined|null, tirex_submission_id: string|undefined|null}[] {
+      let ret: {display_name: string, subItems: number, pos: number, links: any[], focus_type: string|undefined|null, component_type: string|undefined|null, tirex_submission_id: string|undefined|null}[] = []
 
       if (this.is_collapsed(component) || !component['components']) {
         return ret
@@ -188,6 +196,7 @@ export default {
             'links': c.hasOwnProperty('links') ? c['links'] : null,
             'focus_type': c.hasOwnProperty('focus_type') ? c['focus_type'] : null,
             'component_type': c.hasOwnProperty('component_type') ? c['component_type'] : null,
+            'tirex_submission_id': c['tirex_submission_id']
           })
 
           for (let sub_c of this.filtered_sub_components(c)) {
@@ -198,6 +207,7 @@ export default {
               'links': sub_c['links'],
               'focus_type': sub_c.hasOwnProperty('focus_type') ? sub_c['focus_type'] : null,
               'component_type': sub_c.hasOwnProperty('component_type') ? sub_c['component_type'] : null,
+              'tirex_submission_id': sub_c['tirex_submission_id']
             })
           }
         }
@@ -284,8 +294,7 @@ export default {
             ret = ret.concat(terms[i]);
           }
         }
-
-        }
+      }
 
       return ret
     },
@@ -306,7 +315,7 @@ export default {
         let c = this.tirex_components[i]
 
         // we set row 0, aka the headers
-        ret[0][i] = {'display_name': c.display_name, 'links': c.links, 'collapsed': this.is_collapsed(c), 'subItems':this.countSubItems(c), 'hide': false}
+        ret[0][i] = {'display_name': c.display_name, 'links': c.links, 'collapsed': this.is_collapsed(c), 'subItems':this.countSubItems(c), 'hide': false, 'tirex_submission_id': ''}
 
         // we loop through each categories subcomponents and enrich them with information needed for the grid display
         for (let subcomponent of this.filtered_sub_components(c)) {
@@ -320,7 +329,7 @@ export default {
             'links': subcomponent.links,
             'collapsed': this.is_collapsed(subcomponent),
             'hide': this.hide_component(subcomponent),
-            'code': "this will show example code for executing this components method"
+            'tirex_submission_id': subcomponent['tirex_submission_id'] || null
           }
         }
       }

application/src/tira/urls.py

@@ -107,6 +107,7 @@
     path('api/registration/add_registration/<str:vm_id>/<str:task_id>', data_api.add_registration, name='add_registration'),
     path('api/submissions-for-task/<str:task_id>/<str:user_id>/<str:submission_type>', data_api.submissions_for_task, name="submissions_for_task"),
     path('api/tirex-components', data_api.tirex_components, name='tirex_components'),
+    path('api/tirex-snippet', data_api.get_snippet_to_run_components, name='get_snippet_to_run_components'),
     path('api/snippets-for-tirex-components', data_api.get_snippet_to_run_components, name='get_snippet_to_run_components'),
     path('api/re-ranking-datasets/<str:task_id>', data_api.reranking_datasets, name='reranking_datasets'),
     path('api/submissions-of-user/<str:vm_id>', data_api.submissions_of_user, name='submissions_of_user'),