Add accept types

app/api/views/upload/catalog.py

@@ -6,12 +6,47 @@ from typing_extensions import Literal
 
 from ...models import DOCUMENT_CLASSIFICATION, SEQ2SEQ, SEQUENCE_LABELING
 
-CSV = 'CSV'
-FastText = 'fastText'
-JSON = 'JSON'
-JSONL = 'JSONL'
-EXCEL = 'Excel'
-CoNLL = 'CoNLL'
+
+class Format:
+    name = ''
+    accept_types = ''
+
+    @classmethod
+    def dict(cls):
+        return {
+            'name': cls.name,
+            'accept_types': cls.accept_types
+        }
+
+
+class CSV(Format):
+    name = 'CSV'
+    accept_types = 'text/csv'
+
+
+class FastText(Format):
+    name = 'fastText'
+    accept_types = 'text/plain'
+
+
+class JSON(Format):
+    name = 'JSON'
+    accept_types = 'application/json'
+
+
+class JSONL(Format):
+    name = 'JSONL'
+    accept_types = 'application/jsonl'
+
+
+class Excel(Format):
+    name = 'Excel'
+    accept_types = 'application/vnd.ms-excel, application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
+
+
+class CoNLL(Format):
+    name = 'CoNLL'
+    accept_types = 'text/*'
 
 
 class OptionColumn(BaseModel):
@@ -27,17 +62,21 @@ class OptionNone(BaseModel):
     pass
 
 
+class OptionCoNLL(BaseModel):
+    scheme: Literal['IOB2', 'IOE2', 'IOBES', 'BILOU'] = 'IOB2'
+
+
 class Options:
     options: Dict[str, List] = defaultdict(list)
 
     @classmethod
     def filter_by_task(cls, task_name: str):
         options = cls.options[task_name]
-        return [{'name': name, **option.schema()} for name, option in options]
+        return [{**format.dict(), **option.schema()} for format, option in options]
 
     @classmethod
-    def register(cls, task: str, name: str, option: Type[BaseModel]):
-        cls.options[task].append((name, option))
+    def register(cls, task: str, format: Type[Format], option: Type[BaseModel]):
+        cls.options[task].append((format, option))
 
 
 # Text Classification
@@ -45,7 +84,7 @@ Options.register(DOCUMENT_CLASSIFICATION, CSV, OptionDelimiter)
 Options.register(DOCUMENT_CLASSIFICATION, FastText, OptionNone)
 Options.register(DOCUMENT_CLASSIFICATION, JSON, OptionColumn)
 Options.register(DOCUMENT_CLASSIFICATION, JSONL, OptionColumn)
-Options.register(DOCUMENT_CLASSIFICATION, EXCEL, OptionColumn)
+Options.register(DOCUMENT_CLASSIFICATION, Excel, OptionColumn)
 
 # Sequence Labeling
 Options.register(SEQUENCE_LABELING, JSONL, OptionColumn)
@@ -55,4 +94,4 @@ Options.register(SEQUENCE_LABELING, CoNLL, OptionNone)
 Options.register(SEQ2SEQ, CSV, OptionDelimiter)
 Options.register(SEQ2SEQ, JSON, OptionColumn)
 Options.register(SEQ2SEQ, JSONL, OptionColumn)
-Options.register(SEQ2SEQ, EXCEL, OptionColumn)
+Options.register(SEQ2SEQ, Excel, OptionColumn)

app/api/views/upload/factory.py

@@ -11,26 +11,24 @@ def get_data_class(project_type: str):
 
 
 def get_dataset_class(format: str):
-    if format == catalog.CSV:
-        return dataset.CsvDataset
-    elif format == catalog.JSONL:
-        return dataset.JSONLDataset
-    elif format == catalog.JSONL:
-        return dataset.JSONDataset
-    elif format == catalog.FastText:
-        return dataset.FastTextDataset
-    elif format == catalog.EXCEL:
-        return dataset.ExcelDataset
-    else:
+    mapping = {
+        catalog.CSV.name: dataset.CsvDataset,
+        catalog.JSONL.name: dataset.JSONLDataset,
+        catalog.JSON.name: dataset.JSONDataset,
+        catalog.FastText.name: dataset.FastTextDataset,
+        catalog.Excel.name: dataset.ExcelDataset
+    }
+    if format not in mapping:
         ValueError(f'Invalid format: {format}')
+    return mapping[format]
 
 
 def get_label_class(project_type: str):
-    if project_type == DOCUMENT_CLASSIFICATION:
-        return label.CategoryLabel
-    elif project_type == SEQUENCE_LABELING:
-        return label.OffsetLabel
-    elif project_type == SEQ2SEQ:
-        return label.TextLabel
-    else:
+    mapping = {
+        DOCUMENT_CLASSIFICATION: label.CategoryLabel,
+        SEQUENCE_LABELING: label.OffsetLabel,
+        SEQ2SEQ: label.TextLabel
+    }
+    if project_type not in mapping:
         ValueError(f'Invalid project type: {project_type}')
+    return mapping[project_type]

frontend/domain/models/upload/catalog.ts

@@ -1,13 +1,14 @@
 export class Catalog {
   constructor(
     public name: string,
+    public accept_types: string,
     public properties: object
   ) {}
 
   static valueOf(
-    { name, properties }:
-    { name: string, properties: object }
+    { name, accept_types, properties }:
+    { name: string, accept_types: string, properties: object }
   ): Catalog {
-    return new Catalog(name, properties)
+    return new Catalog(name, accept_types, properties)
   }
 }

frontend/package.json

@@ -24,6 +24,7 @@
     "chart.js": "^2.9.3",
     "codemirror": "^5.55.0",
     "filepond": "^4.26.1",
+    "filepond-plugin-file-validate-type": "^1.2.6",
     "js-cookie": "^2.2.1",
     "lodash": "^4.17.21",
     "nuxt": "^2.11.0",

frontend/services/application/upload/catalogData.ts

@@ -3,10 +3,12 @@ import { Catalog } from '~/domain/models/upload/catalog'
 
 export class CatalogDTO {
   name: string
+  acceptTypes: string
   properties: object
 
   constructor(item: Catalog) {
     this.name = item.name
+    this.acceptTypes = item.accept_types
     this.properties = item.properties
   }
 }

frontend/yarn.lock

@@ -5731,6 +5731,11 @@ file-uri-to-path@1.0.0:
   resolved "https://registry.yarnpkg.com/file-uri-to-path/-/file-uri-to-path-1.0.0.tgz#553a7b8446ff6f684359c445f1e37a05dacc33dd"
   integrity sha512-0Zt+s3L7Vf1biwWZ29aARiVYLx7iMGnEUl9x33fbB/j3jR81u/O2LbqK+Bm1CDSNDKVtJ/YjwY7TUd5SkeLQLw==
 
+filepond-plugin-file-validate-type@^1.2.6:
+  version "1.2.6"
+  resolved "https://registry.yarnpkg.com/filepond-plugin-file-validate-type/-/filepond-plugin-file-validate-type-1.2.6.tgz#5a03e04fb8f70f0df37ff9a285d3d2343cdc55f2"
+  integrity sha512-3JFxpb4o5D4BssP/Drjn2GPpejg/QXzU4e3/Pq8IvIJuokHz95b6+RPYstQ+6ITKREYup0N8F/CO/OQfhVkUgg==
+
 filepond@^4.26.1:
   version "4.26.1"
   resolved "https://registry.yarnpkg.com/filepond/-/filepond-4.26.1.tgz#c0ab300fcf367d395129bf491290693d48142fe8"