korpRequest for corpus_config

CHANGELOG.md

@@ -4,6 +4,7 @@
 
 ### Changed
 
+- The `corpus_config_url` setting is replaced by `get_corpus_ids`, see [doc/frontend_devel.md](./doc/frontend_devel.md)
 - The `httpConfAddMethod*` util functions were refactored:
   - The `$.ajax` case of `httpConfAddMethod` was extracted into `ajaxConfAddMethod`
   - `httpConfAddMethodFetch` was renamed to `fetchConfAddMethod`

app/scripts/backend/common.ts

@@ -3,11 +3,13 @@ import { fetchConfAddMethod } from "@/util"
 import { getAuthorizationHeader } from "@/components/auth/auth"
 import settings from "@/settings"
 import { API, Response } from "./types"
+import { omitBy } from "lodash"
 
 export async function korpRequest<K extends keyof API>(
     endpoint: K,
     params: API[K]["params"]
 ): Promise<Response<API[K]["response"]>> {
+    params = omitBy(params, (value) => value == null) as API[K]["params"]
     const { url, request } = fetchConfAddMethod(settings.korp_backend_url + "/" + endpoint, params)
     request.headers = { ...request.headers, ...getAuthorizationHeader() }
     const response = await fetch(url, request)

app/scripts/backend/types/corpus-config.ts

@@ -0,0 +1,11 @@
+/** @format */
+/** @see https://ws.spraakbanken.gu.se/docs/korp#tag/Information/paths/~1corpus_config/get */
+import { Config } from "@/settings/config.types"
+
+export type CorpusConfigParams = {
+    mode: string
+    corpus?: string
+    include_lab?: string
+}
+
+export type CorpusConfigResponse = Config

app/scripts/backend/types/index.ts

@@ -1,5 +1,6 @@
 /** @format */
 
+import { CorpusConfigParams, CorpusConfigResponse } from "./corpus-config"
 import { CorpusInfoParams, CorpusInfoResponse } from "./corpus-info"
 import { CountParams, CountResponse } from "./count"
 import { LoglikeParams, LoglikeResponse } from "./loglike"
@@ -9,6 +10,10 @@ export * from "./common"
 
 /** Maps a Korp backend endpoint name to the expected parameters and response */
 export type API = {
+    corpus_config: {
+        params: CorpusConfigParams
+        response: CorpusConfigResponse
+    }
     corpus_info: {
         params: CorpusInfoParams
         response: CorpusInfoResponse

app/scripts/data_init.ts

@@ -46,6 +46,8 @@ type InfoData = Record<string, Pick<CorpusTransformed, "info" | "private_struct_
  * Fetch CWB corpus info (Size, Updated etc).
  */
 async function getInfoData(corpusIds: string[]): Promise<InfoData> {
+    if (!corpusIds.length) return {}
+
     const params = { corpus: corpusIds.map((id) => id.toUpperCase()).join(",") }
     const data = await korpRequest("corpus_info", params)
     if ("ERROR" in data) {
@@ -98,27 +100,19 @@ async function getConfig(): Promise<Config> {
         return corpusConfig
     } catch {}
 
-    let configUrl: string
-    // The corpora to include can be defined elsewhere can in a mode
-    if (settings.corpus_config_url) {
-        configUrl = await settings.corpus_config_url()
-    } else {
-        const labParam = process.env.ENVIRONMENT == "staging" ? "&include_lab" : ""
-        configUrl = `${settings.korp_backend_url}/corpus_config?mode=${currentMode}${labParam}`
-    }
-    let response: Response
-    try {
-        response = await fetch(configUrl)
-    } catch (error) {
-        throw Error("Config request failed")
-    }
+    // The corpora to include are normally given by the mode config, but allow defining it elsewhere (used by Mink)
+    const corpusIds = settings.get_corpus_ids ? await settings.get_corpus_ids() : undefined
+
+    const config = await korpRequest("corpus_config", {
+        mode: currentMode,
+        corpus: corpusIds?.join(",") || undefined,
+    })
 
-    if (!response.ok) {
-        console.error("Something wrong with corpus config", response.statusText)
-        throw Error("Something wrong with corpus config")
+    if ("ERROR" in config) {
+        throw Error(config.ERROR.value)
     }
 
-    return (await response.json()) as Config
+    return config
 }
 
 /**

app/scripts/settings/app-settings.types.ts

@@ -15,7 +15,7 @@ export type AppSettings = {
     backendURLMaxLength: number
     common_struct_types?: Record<string, Attribute>
     config_dependent_on_authentication?: boolean
-    corpus_config_url?: () => Promise<string>
+    get_corpus_ids?: () => Promise<string[]>
     corpus_info_link?: {
         url_template: string
         label: LangString

doc/frontend_devel.md

@@ -109,7 +109,6 @@ settings that affect the frontend.
     that may be added automatically to a corpus. See [backend documentation](https://github.com/spraakbanken/korp-backend)
     for more information about how to define attributes.
 - __config_dependent_on_authentication__ - Boolean. If true, backend config will not be fetched until login check has finished.
-- __corpus_config_url__ - Async function returning a url string. Configuration for the selected mode is fetched from here at app initialization. If not given, the default is `<korp_backend_url>/corpus_config?mode=<mode>`, see the [`corpus_config`](https://ws.spraakbanken.gu.se/docs/korp#tag/Information/paths/~1corpus_config/get) API.
 - __corpus_info_link__ - Object. Use this to render a link for each corpus in the corpus chooser.
   - __url_template__ - String or translation object. A URL containing a token "%s", which will be replaced with the corpus id.
   - __label__ - String or translation object. The label is the the same for all corpora.
@@ -134,6 +133,7 @@ settings that affect the frontend.
     - __label__: String or translation object.
     - __params__: Object. This is translated to URL search params when the link is clicked.
     - __hint__: String or translation object. Can contain HTML.
+- __get_corpus_ids__ - Async function returning a list of strings. The corpus ids are passed as the `corpus=` param to the `<korp_backend_url>/corpus_config?mode=<mode>` call, see the [`corpus_config`](https://ws.spraakbanken.gu.se/docs/korp#tag/Information/paths/~1corpus_config/get) API.
 - __group_statistics__ - List of attribute names. Attributes that either have a rank or a numbering used for multi-word units. For example, removing `:2` from `ta_bort..vbm.1:2`, to get the lemgram of this word: `ta_bort..vbm.1`.
 - __has_timespan__ - Boolean. If the backend supports the `timespan` call, used in corpus chooser for example. Default: `true`
 - __hits_per_page_values__ - Array of integer. The available page sizes. Default: `[25, 50, 75, 100]`
@@ -228,12 +228,14 @@ If no mode is given, mode is `default`.
 
 It then looks for mode-specific code in `<configDir>/modes/<mode>_mode.js`. Mode code may overwrite values from `config.yml` by altering the `settings` object imported from `@/settings`.
 
-It then looks for settings for this specific mode, the **corpus config**. If it exists at `<configDir>/modes/<mode>_corpus_config.json`, it will be loaded from there. Otherwise, it retrieves it from the url given by the `corpus_config_url` option, which defaults to:
+It then looks for settings for this specific mode, the **corpus config**. If it exists at `<configDir>/modes/<mode>_corpus_config.json`, it will be loaded from there. Otherwise, it retrieves it from the backend:
 
 ```
 https://<korp_backend_url>/corpus_config?mode=<mode>
 ```
 
+Normally, the mode param is enough for the backend to know what corpora to include. Alternatively, it is possible to specify corpus ids in the `corpus=` param, by assigning a function to the `get_corpus_ids` setting (in `<mode>_mode.js`). The function can be async and should return a list of corpus ids.
+
 See the [`corpus_config`](https://ws.spraakbanken.gu.se/docs/korp#tag/Information/paths/~1corpus_config/get) API for more information.
 
 ## Parallel corpora