@jbrowse/plugin-trix 4.2.1 → 5.0.0-beta.1

package/esm/TrixTextSearchAdapter/TrixTextSearchAdapter.d.ts

@@ -5,10 +5,14 @@ import type PluginManager from '@jbrowse/core/PluginManager';
 import type { AnyConfigurationModel } from '@jbrowse/core/configuration';
 import type { BaseTextSearchAdapter, BaseTextSearchArgs } from '@jbrowse/core/data_adapters/BaseAdapter';
 import type { getSubAdapterType } from '@jbrowse/core/data_adapters/dataAdapterCache';
+export declare function snippetAround(str: string, term: string, w?: number): string;
 export default class TrixTextSearchAdapter extends BaseAdapter implements BaseTextSearchAdapter {
-    indexingAttributes?: string[];
     trixJs: Trix;
-    tracksNames?: string[];
     constructor(config: AnyConfigurationModel, getSubAdapter?: getSubAdapterType, pluginManager?: PluginManager);
+    /**
+     * Returns list of results
+     * @param args - search options/arguments include: search query,
+     * searchType (prefix | full | exact), etc.
+     */
     searchIndex(args: BaseTextSearchArgs): Promise<BaseResult[]>;
 }

package/esm/TrixTextSearchAdapter/TrixTextSearchAdapter.js

@@ -3,68 +3,100 @@ import BaseResult from '@jbrowse/core/TextSearch/BaseResults';
 import { readConfObject } from '@jbrowse/core/configuration';
 import { BaseAdapter } from '@jbrowse/core/data_adapters/BaseAdapter';
 import { openLocation } from '@jbrowse/core/util/io';
+import { checkStopToken } from '@jbrowse/core/util/stopToken';
 function decodeURIComponentNoThrow(uri) {
     try {
         return decodeURIComponent(uri);
     }
-    catch (e) {
+    catch {
         return uri;
     }
 }
-function shorten(str, term, w = 15) {
-    const tidx = str.toLowerCase().indexOf(term);
-    return str.length < 40
-        ? str
-        : (Math.max(0, tidx - w) > 0 ? '...' : '') +
-            str.slice(Math.max(0, tidx - w), tidx + term.length + w).trim() +
-            (tidx + term.length < str.length ? '...' : '');
+// Build a search-result snippet: a window of `w` characters either side of
+// where `term` matched, ellipsized on whichever side was cut. Nothing like the
+// core `shorten`, which just elides a long name.
+export function snippetAround(str, term, w = 15) {
+    if (str.length < 40) {
+        return str;
+    }
+    const tidx = str.toLowerCase().indexOf(term.toLowerCase());
+    if (tidx === -1) {
+        return `${str.slice(0, 40).trim()}...`;
+    }
+    const start = Math.max(0, tidx - w);
+    const end = tidx + term.length + w;
+    return ((start > 0 ? '...' : '') +
+        str.slice(start, end).trim() +
+        (end < str.length ? '...' : ''));
 }
 export default class TrixTextSearchAdapter extends BaseAdapter {
-    indexingAttributes;
     trixJs;
-    tracksNames;
     constructor(config, getSubAdapter, pluginManager) {
         super(config, getSubAdapter, pluginManager);
         const ixFilePath = readConfObject(config, 'ixFilePath');
         const ixxFilePath = readConfObject(config, 'ixxFilePath');
         if (!ixFilePath) {
-            throw new Error('must provide out.ix');
+            throw new Error('TrixTextSearchAdapter requires an ixFilePath');
         }
         if (!ixxFilePath) {
-            throw new Error('must provide out.ixx');
+            throw new Error('TrixTextSearchAdapter requires an ixxFilePath');
         }
         this.trixJs = new Trix(openLocation(ixxFilePath, pluginManager), openLocation(ixFilePath, pluginManager), 1500);
     }
+    /**
+     * Returns list of results
+     * @param args - search options/arguments include: search query,
+     * searchType (prefix | full | exact), etc.
+     */
     async searchIndex(args) {
-        const query = args.queryString.toLowerCase();
-        const strs = query.split(' ');
+        const query = args.queryString.toLowerCase().trim();
+        // any run of whitespace separates words; splitting on a single ' ' let a
+        // double space or a tab produce an empty word that matches everything
+        const words = query.split(/\s+/);
         const results = await this.trixJs.search(query);
+        // the index read has landed but nothing below it has run yet; a broad
+        // prefix returns thousands of records to JSON-parse, decode and snippet,
+        // and the keystroke that asked for them is 50ms stale by now
+        checkStopToken(args.stopToken);
         const formatted = results
-            .filter(([, data]) => strs.every(r => decodeURIComponentNoThrow(data).toLowerCase().includes(r)))
             .map(([term, data]) => {
-            const result = JSON.parse(data.replaceAll('|', ','));
-            const [loc, trackId, ...rest] = result.map(record => decodeURIComponentNoThrow(record));
-            const labelFieldIdx = rest.findIndex(elt => !!elt);
-            const contextIdx = rest
-                .map(elt => elt.toLowerCase())
-                .findIndex(f => f.includes(term.toLowerCase()));
-            const labelField = rest[labelFieldIdx];
-            const contextField = rest[contextIdx];
-            const context = contextIdx !== -1 ? shorten(contextField, term) : undefined;
-            const label = shorten(labelField, term);
+            // record is ["loc"|"trackId"|"attr1"|...] with commas pre-encoded as
+            // pipes; swap them back to parse it as a JSON array, then URI-decode
+            const [loc, trackId, ...attrs] = JSON.parse(data.replaceAll('|', ',')).map(decodeURIComponentNoThrow);
+            return { term, loc, trackId, attrs };
+        })
+            // multi-word search: keep only entries whose attributes contain every word
+            .filter(({ attrs }) => {
+            const lower = attrs.join(' ').toLowerCase();
+            return words.every(w => lower.includes(w));
+        })
+            .map(({ term, loc, trackId, attrs }) => {
+            const termLower = term.toLowerCase();
+            const labelField = attrs.find(Boolean) ?? '';
+            const contextField = attrs.find(f => f.toLowerCase().includes(termLower));
+            const context = contextField
+                ? snippetAround(contextField, term)
+                : undefined;
+            const label = snippetAround(labelField, term);
             const displayString = !context || label.toLowerCase() === context.toLowerCase()
                 ? label
                 : `${label} (${context})`;
+            // exact match succeeds when any indexed attribute (name, id,
+            // description, ...) equals the query, so e.g. searching an ID works.
+            // Carried on the result as well as filtered on: only this adapter can
+            // judge it, and a caller that has the flag can ask one unrestricted
+            // question instead of an exact one and then a broad one
+            const exact = attrs.some(a => a.toLowerCase() === query);
             return new BaseResult({
                 locString: loc,
-                label: labelField,
+                label,
                 displayString,
-                matchedObject: result.map(record => decodeURIComponent(record)),
                 trackId,
+                exact,
             });
         });
         return args.searchType === 'exact'
-            ? formatted.filter(r => r.getLabel().toLowerCase() === args.queryString.toLowerCase())
+            ? formatted.filter(r => r.isExact())
             : formatted;
     }
 }

package/esm/TrixTextSearchAdapter/configSchema.d.ts

@@ -1,34 +1,76 @@
+export declare function normalizeSnapshot(snap: Record<string, unknown>): Record<string, unknown>;
+/**
+ * #config TrixTextSearchAdapter
+ * #trackType TextSearchAdapter
+ * #fileFormat textsearch | Trix index (.ix/.ixx) | Built by `jbrowse text-index`
+ *
+ * #example
+ * `jbrowse text-index` writes this entry into `aggregateTextSearchAdapters` for
+ * you. The `uri` shorthand points at the `.ix`; the sibling `.ixx` and
+ * `_meta.json` are derived from it, so all three only need spelling out when
+ * they are named against convention.
+ * ```js
+ * {
+ *   type: 'TrixTextSearchAdapter',
+ *   textSearchAdapterId: 'hg38-index',
+ *   uri: 'trix/hg38.ix',
+ *   assemblyNames: ['hg38'],
+ * }
+ * ```
+ */
 declare const TrixTextSearchAdapter: import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaType<{
-    ixFilePath: {
-        type: string;
-        defaultValue: {
-            uri: string;
-            locationType: string;
+    /**
+     * #slot
+     * location of the Trix `.ix` index written by `jbrowse text-index`: the
+     * sorted term-to-feature table the search box reads.
+     */
+    readonly ixFilePath: {
+        readonly type: "fileLocation";
+        readonly defaultValue: {
+            readonly uri: "out.ix";
+            readonly locationType: "UriLocation";
         };
     };
-    ixxFilePath: {
-        type: string;
-        defaultValue: {
-            uri: string;
-            locationType: string;
+    /**
+     * #slot
+     * location of the `.ixx` prefix index, which records where in the `.ix`
+     * each prefix begins. It is what makes a lookup a couple of range requests
+     * instead of a download of the whole index.
+     */
+    readonly ixxFilePath: {
+        readonly type: "fileLocation";
+        readonly defaultValue: {
+            readonly uri: "out.ixx";
+            readonly locationType: "UriLocation";
         };
     };
-    metaFilePath: {
-        type: string;
-        defaultValue: {
-            uri: string;
-            locationType: string;
+    /**
+     * #slot
+     * location of the `_meta.json` written beside the index, recording which
+     * tracks and assemblies it covers and the attributes it was built from.
+     */
+    readonly metaFilePath: {
+        readonly type: "fileLocation";
+        readonly defaultValue: {
+            readonly uri: "meta.json";
+            readonly locationType: "UriLocation";
         };
     };
-    tracks: {
-        type: string;
-        defaultValue: never[];
-        description: string;
+    /**
+     * #slot
+     */
+    readonly tracks: {
+        readonly type: "stringArray";
+        readonly defaultValue: readonly [];
+        readonly description: "List of tracks covered by text search adapter";
     };
-    assemblyNames: {
-        type: string;
-        defaultValue: never[];
-        description: string;
+    /**
+     * #slot
+     */
+    readonly assemblyNames: {
+        readonly type: "stringArray";
+        readonly defaultValue: readonly [];
+        readonly description: "List of assemblies covered by text search adapter";
     };
-}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, "textSearchAdapterId">>;
+}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined>>;
 export default TrixTextSearchAdapter;

package/esm/TrixTextSearchAdapter/configSchema.js

@@ -1,6 +1,50 @@
 import { ConfigurationSchema } from '@jbrowse/core/configuration';
-function x() { }
+export function normalizeSnapshot(snap) {
+    return typeof snap.uri === 'string'
+        ? {
+            ...snap,
+            // `uri` points at the `.ix` file; `.ixx` and `_meta.json` sit beside it
+            // (the `jbrowse text-index` naming convention), so derive all three
+            ixFilePath: {
+                uri: snap.uri,
+                baseUri: snap.baseUri,
+            },
+            ixxFilePath: {
+                uri: `${snap.uri}x`,
+                baseUri: snap.baseUri,
+            },
+            metaFilePath: {
+                uri: `${snap.uri.replace(/\.ix$/, '')}_meta.json`,
+                baseUri: snap.baseUri,
+            },
+        }
+        : snap;
+}
+/**
+ * #config TrixTextSearchAdapter
+ * #trackType TextSearchAdapter
+ * #fileFormat textsearch | Trix index (.ix/.ixx) | Built by `jbrowse text-index`
+ *
+ * #example
+ * `jbrowse text-index` writes this entry into `aggregateTextSearchAdapters` for
+ * you. The `uri` shorthand points at the `.ix`; the sibling `.ixx` and
+ * `_meta.json` are derived from it, so all three only need spelling out when
+ * they are named against convention.
+ * ```js
+ * {
+ *   type: 'TrixTextSearchAdapter',
+ *   textSearchAdapterId: 'hg38-index',
+ *   uri: 'trix/hg38.ix',
+ *   assemblyNames: ['hg38'],
+ * }
+ * ```
+ */
 const TrixTextSearchAdapter = ConfigurationSchema('TrixTextSearchAdapter', {
+    /**
+     * #slot
+     * location of the Trix `.ix` index written by `jbrowse text-index`: the
+     * sorted term-to-feature table the search box reads.
+     */
     ixFilePath: {
         type: 'fileLocation',
         defaultValue: {
@@ -8,6 +52,12 @@ const TrixTextSearchAdapter = ConfigurationSchema('TrixTextSearchAdapter', {
             locationType: 'UriLocation',
         },
     },
+    /**
+     * #slot
+     * location of the `.ixx` prefix index, which records where in the `.ix`
+     * each prefix begins. It is what makes a lookup a couple of range requests
+     * instead of a download of the whole index.
+     */
     ixxFilePath: {
         type: 'fileLocation',
         defaultValue: {
@@ -15,6 +65,11 @@ const TrixTextSearchAdapter = ConfigurationSchema('TrixTextSearchAdapter', {
             locationType: 'UriLocation',
         },
     },
+    /**
+     * #slot
+     * location of the `_meta.json` written beside the index, recording which
+     * tracks and assemblies it covers and the attributes it was built from.
+     */
     metaFilePath: {
         type: 'fileLocation',
         defaultValue: {
@@ -22,11 +77,17 @@ const TrixTextSearchAdapter = ConfigurationSchema('TrixTextSearchAdapter', {
             locationType: 'UriLocation',
         },
     },
+    /**
+     * #slot
+     */
     tracks: {
         type: 'stringArray',
         defaultValue: [],
         description: 'List of tracks covered by text search adapter',
     },
+    /**
+     * #slot
+     */
     assemblyNames: {
         type: 'stringArray',
         defaultValue: [],
@@ -34,21 +95,27 @@ const TrixTextSearchAdapter = ConfigurationSchema('TrixTextSearchAdapter', {
     },
 }, {
     explicitlyTyped: true,
-    explicitIdentifier: 'textSearchAdapterId',
-    preProcessSnapshot: snap => {
-        return snap.uri
-            ? {
-                ...snap,
-                ixFilePath: {
-                    uri: snap.uri,
-                    baseUri: snap.baseUri,
-                },
-                ixxFilePath: {
-                    uri: `${snap.uri}x`,
-                    baseUri: snap.baseUri,
-                },
-            }
-            : snap;
-    },
+    /**
+     * #identifier
+     * an explicit `textSearchAdapterId` is still honored when given
+     */
+    implicitIdentifier: 'textSearchAdapterId',
+    /**
+     * #preProcessSnapshot
+     *
+     *
+     * preprocessor to allow minimal config: `uri` points at the `.ix` file and
+     * the sibling `.ixx` and `_meta.json` are derived from it (the
+     * `jbrowse text-index` naming convention):
+     * ```json
+     * {
+     *   "type": "TrixTextSearchAdapter",
+     *   "uri": "file.ix",
+     *   "assemblyNames": ["hg19"],
+     *   "textSearchAdapterId": "hg19SearchIndex"
+     * }
+     * ```
+     */
+    preProcessSnapshot: normalizeSnapshot,
 });
 export default TrixTextSearchAdapter;

package/esm/index.js

@@ -1,6 +1,6 @@
 import Plugin from '@jbrowse/core/Plugin';
 import TextSearchAdapterType from '@jbrowse/core/pluggableElementTypes/TextSearchAdapterType';
-import configSchema from "./TrixTextSearchAdapter/configSchema.js";
+import configSchema from './TrixTextSearchAdapter/configSchema.js';
 export default class TrixPlugin extends Plugin {
     name = 'TrixPlugin';
     install(pluginManager) {
@@ -10,7 +10,7 @@ export default class TrixPlugin extends Plugin {
                 displayName: 'Trix text search adapter',
                 configSchema,
                 description: 'Trix text search adapter',
-                getAdapterClass: () => import("./TrixTextSearchAdapter/TrixTextSearchAdapter.js").then(d => d.default),
+                getAdapterClass: () => import('./TrixTextSearchAdapter/TrixTextSearchAdapter.js').then(d => d.default),
             });
         });
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jbrowse/plugin-trix",
-  "version": "4.2.1",
+  "version": "5.0.0-beta.1",
   "type": "module",
   "description": "JBrowse 2 trix text search adapter",
   "keywords": [
@@ -21,11 +21,11 @@
     "esm"
   ],
   "peerDependencies": {
-    "react": ">=18.0.0"
+    "react": ">=19.0.0"
   },
   "dependencies": {
-    "@gmod/trix": "^3.0.10",
-    "@jbrowse/core": "^4.2.1"
+    "@gmod/trix": "^3.0.22",
+    "@jbrowse/core": "^5.0.0-beta.1"
   },
   "publishConfig": {
     "access": "public"
@@ -33,16 +33,13 @@
   "sideEffects": false,
   "scripts": {
     "build": "pnpm run /^build:/",
-    "test": "cd ../..; jest --passWithNoTests plugins/trix",
+    "test": "jest --config ../../jest.config.js --passWithNoTests --testPathPatterns plugins/trix",
     "prebuild": "pnpm clean",
-    "build:esm": "tsc -p tsconfig.build.esm.json",
+    "build:esm": "node ../../node_modules/typescript7/bin/tsc --build tsconfig.build.esm.json",
     "clean": "rimraf esm *.tsbuildinfo"
   },
   "types": "esm/index.d.ts",
   "exports": {
-    ".": {
-      "types": "./esm/index.d.ts",
-      "import": "./esm/index.js"
-    }
+    ".": "./esm/index.js"
   }
 }