Set format from file extension when creating corpus

CHANGELOG.md

@@ -15,6 +15,7 @@ As this project is a user-facing application, the places in the semantic version
 
 ### Changed
 
+- A corpus created by upload now gets its format set to match the uploaded files
 - A corpus created by upload (not through the creation form) now gets its name set to the id, instead of remaining empty
 
 ### Fixed

src/api/corpusConfig.js

@@ -68,13 +68,13 @@ export async function makeConfig(id, options) {
     ],
   };
 
-  if (format === "xml") {
-    if (!textAnnotation) {
-      throw new TypeError("Text annotation setting is required for XML");
+  if (format == "xml") {
+    // The text annotation setting is required if XML, but it may be set later
+    if (textAnnotation) {
+      config.import.text_annotation = textAnnotation;
+      config.export.source_annotations = [`${textAnnotation} as text`, "..."];
     }
-    config.import.text_annotation = textAnnotation;
-    config.export.source_annotations = [`${textAnnotation} as text`, "..."];
-  } else if (format === "pdf") {
+  } else if (format == "pdf") {
     config.export.source_annotations = ["text", "page:number"];
   }
 
@@ -141,3 +141,14 @@ export async function parseConfig(configYaml) {
     ).params.value,
   };
 }
+
+/** Check if the config looks ready to run. May throw anything. */
+export function validateConfig(config) {
+  if (!config.format) {
+    throw new TypeError("Format missing");
+  }
+
+  if (config.format == "xml" && !config.textAnnotation) {
+    throw new TypeError("Text annotation setting is required for XML");
+  }
+}

src/corpus/corpusState.composable.js

@@ -1,9 +1,10 @@
 import { computed } from "vue";
 import { useI18n } from "vue-i18n";
-import { FORMATS_EXT } from "@/api/corpusConfig";
+import { validateConfig } from "@/api/corpusConfig";
 import useConfig from "./config/config.composable";
 import useJob from "./job/job.composable";
 import useSources from "./sources/sources.composable";
+import { getException } from "@/util";
 
 /** The "corpus state" is related to the job status, but is more about predicting what action the user needs to take. */
 export function useCorpusState(corpusId) {
@@ -38,7 +39,7 @@ export function useCorpusState(corpusId) {
   });
 
   const isConfigValid = computed(
-    () => config.value && FORMATS_EXT.includes(config.value.format)
+    () => !getException(() => validateConfig(config.value))
   );
 
   const hasMetadata = computed(

src/corpus/createCorpus.composable.js

@@ -30,9 +30,13 @@ export default function useCreateCorpus() {
     const corpusId = await createCorpus().catch(alertError);
     if (!corpusId) return;
 
+    // Get file extension of first file, assuming all are using the same extension.
+    const format = files[0]?.name.split(".").pop();
+
     // Create a minimal config.
     const config = {
       name: { swe: corpusId, eng: corpusId },
+      format,
     };
 
     const results = await Promise.allSettled([

src/i18n/locales/en.yaml

@@ -270,7 +270,7 @@ api.code.failed_uploading_sources: Failed to upload source files
 api.code.failed_uploading_sources_file_size: Some files could not be uploaded, because they exceed the maximum file size of {max_file_size} MB
 api.code.failed_uploading_sources_invalid_file_extension: 'File "{file}" could not be uploaded, because it has an invalid filename extension. Please use any of the supported file formats: xml, txt, odt, docx, pdf.'
 api.code.failed_uploading_sources_invalid_xml: File "{file}" could not be uploaded, because it contains invalid XML
-api.code.failed_uploading_sources_incompatible_file_extension: File "{file}" could not be uploaded, because it has an different filename extension. Please check that the file format of this corpus is configured to agree with your source files.
+api.code.failed_uploading_sources_incompatible_file_extension: File "{file}" could not be uploaded, because it has a different filename extension. Please check that the file format of this corpus is configured to agree with your source files.
 api.code.uploaded_sources: The source files were successfully added
 api.code.data_too_large: Request data too large (max {max_content_length} MB per upload)
 api.code.listing_sources: Listing current source files for the corpus "{corpus_id}"

src/util.js

@@ -67,3 +67,13 @@ export function setKeys(obj, keys, defaultValue = null) {
 
 /** Create a random string of around 11 chars in the [0-9a-z] range. */
 export const randomString = () => Math.random().toString(36).slice(2);
+
+/** Execute callback, catch and return any exception, otherwise return undefined. */
+export const getException = (f) => {
+  try {
+    f();
+  } catch (e) {
+    return e;
+  }
+  return undefined;
+};

src/util.test.js

@@ -3,6 +3,7 @@ import {
   ensureExtension,
   formatDate,
   formatSeconds,
+  getException,
   pathJoin,
   randomString,
   setKeys,
@@ -64,3 +65,19 @@ describe("randomString", () => {
     expect(fails).toEqual([]);
   });
 });
+
+describe("getException", () => {
+  test("translate success to undefined", () => {
+    const f = () => "foobar";
+    const exception = getException(f);
+    expect(exception).toBeUndefined();
+  });
+  test("reflect exception", () => {
+    const f = () => {
+      throw new EvalError("Leverpastej");
+    };
+    const exception = getException(f);
+    expect(exception.name).toBe("EvalError");
+    expect(exception.message).toBe("Leverpastej");
+  });
+});