#58 - new behavior

Omnisearch no longer register extensions, PDFs are indexed after plaintext files, updated settings descriptions

src/main.ts

@@ -6,7 +6,7 @@ import { eventBus } from './globals'
 import { registerAPI } from '@vanakat/plugin-api'
 import api from './api'
 import { loadSearchHistory } from './search-history'
-import {isFileIndexable, showWelcomeNotice} from './utils'
+import {isFilePlaintext, showWelcomeNotice} from './utils'
 import { addNoteToReindex, addToIndex, removeFromIndex } from './notes-index'
 
 function _registerAPI(plugin: OmnisearchPlugin): void {
@@ -22,10 +22,7 @@ export default class OmnisearchPlugin extends Plugin {
     // additional files to index by Omnisearch
 
     await loadSettings(this)
-
-    this.registerExtensions(settings.indexedFileTypes, 'markdown')
     await loadSearchHistory()
-
     _registerAPI(this)
 
     if (settings.ribbonIcon) {
@@ -75,7 +72,7 @@ export default class OmnisearchPlugin extends Plugin {
       )
       this.registerEvent(
         this.app.vault.on('rename', async (file, oldPath) => {
-          if (file instanceof TFile && isFileIndexable(file.path)) {
+          if (file instanceof TFile && isFilePlaintext(file.path)) {
             removeFromIndex(oldPath)
             await addToIndex(file)
           }

src/notes-index.ts

@@ -2,8 +2,8 @@ import { Notice, TAbstractFile, TFile } from 'obsidian'
 import {
   extractHeadingsFromCache,
   getAliasesFromMetadata,
-  getTagsFromMetadata,
+  getTagsFromMetadata, isFileIndexable,
-  isFileIndexable,
+  isFilePlaintext,
   removeDiacritics,
   wait,
 } from './utils'
@@ -129,7 +129,7 @@ export function addNonExistingToIndex(name: string, parent: string): void {
  * @param path
  */
 export function removeFromIndex(path: string): void {
-  if (!isFileIndexable(path)) {
+  if (!isFilePlaintext(path)) {
     console.info(`"${path}" is not an indexable file`)
     return
   }

src/pdf-parser.ts

@@ -1,11 +1,11 @@
 import type { TFile } from 'obsidian'
-import {loadPdfJs} from "obsidian";
+import { loadPdfJs } from 'obsidian'
 
 let PDFJs: any = null
 
 // https://stackoverflow.com/a/59929946
 export async function getPdfText(file: TFile): Promise<string> {
-  PDFJs = PDFJs ?? await loadPdfJs()
+  PDFJs = PDFJs ?? (await loadPdfJs())
   const data = await app.vault.readBinary(file)
   const doc = await PDFJs.getDocument(data).promise
   const pageTexts = Array.from({ length: doc.numPages }, async (v, i) => {

src/search.ts

@@ -9,7 +9,7 @@ import {
   SPACE_OR_PUNCTUATION,
 } from './globals'
 import {
-  isFileIndexable,
+  isFilePlaintext,
   removeDiacritics,
   stringsToRegex,
   stripMarkdownCharacters,
@@ -42,7 +42,7 @@ const tokenize = (text: string): string[] => {
  * Initializes the MiniSearch instance,
  * and adds all the notes to the index
  */
-export async function initGlobalSearchIndex(force = false): Promise<void> {
+export async function initGlobalSearchIndex(): Promise<void> {
   const options: Options<IndexedNote> = {
     tokenize,
     processTerm: (term: string) =>
@@ -76,7 +76,7 @@ export async function initGlobalSearchIndex(force = false): Promise<void> {
     }
   }
 
-  if (!minisearchInstance || force) {
+  if (!minisearchInstance) {
     minisearchInstance = new MiniSearch(options)
     resetNotesCache()
   }
@@ -84,7 +84,7 @@ export async function initGlobalSearchIndex(force = false): Promise<void> {
   // Index files that are already present
   const start = new Date().getTime()
 
-  const allFiles = app.vault.getFiles().filter(f => isFileIndexable(f.path))
+  const allFiles = app.vault.getFiles().filter(f => isFilePlaintext(f.path))
 
   let files
   let notesSuffix
@@ -99,17 +99,15 @@ export async function initGlobalSearchIndex(force = false): Promise<void> {
   console.log(`Omnisearch - indexing ${files.length} ${notesSuffix}`)
 
   // This is basically the same behavior as MiniSearch's `addAllAsync()`.
-  // We index files by batches of 10
+  // We index markdown and plaintext files by batches of 10
   for (let i = 0; i < files.length; ++i) {
     if (i % 10 === 0) await wait(0)
     const file = files[i]
-    if (file) {
     if (getNoteFromCache(file.path)) {
       removeFromIndex(file.path)
     }
     await addToIndex(file)
   }
-  }
 
   if (files.length > 0) {
     const message = `Omnisearch - Indexed ${files.length} ${notesSuffix} in ${
@@ -123,6 +121,27 @@ export async function initGlobalSearchIndex(force = false): Promise<void> {
     }
 
     await saveIndexToFile()
+
+    // PDFs are indexed later, since they're heavier
+    await indexPDFs()
+  }
+}
+
+async function indexPDFs() {
+  if (settings.indexPDFs) {
+    console.warn("Omnisearch - Warnings on pdf.worker.min are due to some issues while reading PDFs file.")
+    const files = app.vault.getFiles().filter(f => f.path.endsWith('.pdf'))
+    for (const file of files) {
+      await wait(0)
+      if (getNoteFromCache(file.path)) {
+        removeFromIndex(file.path)
+      }
+      await addToIndex(file)
+      console.log(file.path)
+    }
+    if (settings.showIndexingNotices) {
+      new Notice(`Omnisearch - Indexed ${files.length} PDFs`)
+    }
   }
 }
 

src/settings.ts

@@ -2,7 +2,6 @@ import { Plugin, PluginSettingTab, Setting, SliderComponent } from 'obsidian'
 import { writable } from 'svelte/store'
 import { notesCacheFilePath, searchIndexFilePath } from './globals'
 import type OmnisearchPlugin from './main'
-import { initGlobalSearchIndex } from './search'
 
 interface WeightingSettings {
   weightBasename: number
@@ -75,7 +74,7 @@ export class SettingsTab extends PluginSettingTab {
     const diacriticsDesc = new DocumentFragment()
     diacriticsDesc.createSpan({}, span => {
       span.innerHTML = `Normalize diacritics in search terms. Words like "brûlée" or "žluťoučký" will be indexed as "brulee" and "zlutoucky".<br/>
-        <strong>Changing this will trigger a full reindex.</strong>`
+        <strong>Needs a restart to fully take effect.</strong>`
     })
     new Setting(containerEl)
       .setName('Ignore diacritics')
@@ -84,7 +83,6 @@ export class SettingsTab extends PluginSettingTab {
         toggle.setValue(settings.ignoreDiacritics).onChange(async v => {
           settings.ignoreDiacritics = v
           await saveSettings(this.plugin)
-          await initGlobalSearchIndex(true)
         })
       )
 
@@ -93,7 +91,6 @@ export class SettingsTab extends PluginSettingTab {
     indexedFileTypesDesc.createSpan({}, span => {
       span.innerHTML = `In addition to standard <code>md</code> files, Omnisearch can also index other plain text files.<br/>
       Add extensions separated by a space. Example: <code>txt org</code>.<br />
-      This setting will also add these files in the navigation, and they will be treated as markdown.<br />
       <strong>Needs a restart to fully take effect.</strong>`
     })
     new Setting(containerEl)
@@ -114,7 +111,8 @@ export class SettingsTab extends PluginSettingTab {
     indexPDFsDesc.createSpan({}, span => {
       span.innerHTML = `Omnisearch will index your PDFs, and return them in search results.
         This feature is currently a work-in-progress, please report slowdowns or issues that you might experience.<br>
-        <strong>Changing this will trigger a full reindex.</strong>`
+        PDFs being quite slow to index, <strong style="color: var(--text-accent)">it is strongly recommended to also enable "Store index in file"</strong>.<br>
+        <strong>Needs a restart to fully take effect.</strong>`
     })
     new Setting(containerEl)
       .setName('BETA - Index PDFs')
@@ -123,9 +121,9 @@ export class SettingsTab extends PluginSettingTab {
         toggle.setValue(settings.indexPDFs).onChange(async v => {
           settings.indexPDFs = v
           await saveSettings(this.plugin)
-          await initGlobalSearchIndex(true)
         })
       )
+
     // Store index
     const serializedIndexDesc = new DocumentFragment()
     serializedIndexDesc.createSpan({}, span => {
@@ -133,7 +131,7 @@ export class SettingsTab extends PluginSettingTab {
         This results in faster loading times for bigger vaults and mobile devices.<br />
         <em>⚠️ Note: the index can become corrupted - if you notice any issue, disable and re-enable this option to clear the cache.</em><br/>
         <em>⚠️ Cache files in <code>.obsidian/plugins/omnisearch/</code> must not be synchronized.</em><br/>
-        <strong>Changing this will trigger a full reindex.</strong>
+        <strong>Needs a restart to fully take effect.</strong>
         `
     })
     new Setting(containerEl)
@@ -141,11 +139,18 @@ export class SettingsTab extends PluginSettingTab {
       .setDesc(serializedIndexDesc)
       .addToggle(toggle =>
         toggle.setValue(settings.storeIndexInFile).onChange(async v => {
+          try {
             await app.vault.adapter.remove(notesCacheFilePath)
+          } catch (e) {
+            console.warn(e)
+          }
+          try {
             await app.vault.adapter.remove(searchIndexFilePath)
+          } catch (e) {
+            console.warn(e)
+          }
           settings.storeIndexInFile = v
           await saveSettings(this.plugin)
-          await initGlobalSearchIndex(true)
         })
       )
 
@@ -311,7 +316,7 @@ export const DEFAULT_SETTINGS: OmnisearchSettings = {
 
   storeIndexInFile: false,
 
-  welcomeMessage: ''
+  welcomeMessage: '',
 } as const
 
 export let settings = Object.assign({}, DEFAULT_SETTINGS) as OmnisearchSettings

src/utils.ts

@@ -173,11 +173,15 @@ export function getCtrlKeyLabel(): 'ctrl' | '⌘' {
 }
 
 export function isFileIndexable(path: string): boolean {
-  return (
+  return (settings.indexPDFs && path.endsWith('.pdf')) || isFilePlaintext(path)
-    path.endsWith('.md') ||
+}
-    (settings.indexPDFs && path.endsWith('.pdf')) ||
+
-    settings.indexedFileTypes.some(t => path.endsWith(`.${t}`))
+export function isFilePlaintext(path: string): boolean {
-  )
+  return getPlaintextExtensions().some(t => path.endsWith(`.${t}`))
+}
+
+export function getPlaintextExtensions(): string[] {
+  return [...settings.indexedFileTypes, 'md']
 }
 
 export function getExtension(path: string): string {