#184 - Verbose logging

This commit is contained in:
Simon Cambier
2023-03-11 14:43:29 +01:00
parent 58db940b68
commit 24bfb94b71
6 changed files with 61 additions and 28 deletions

View File

@@ -9,6 +9,7 @@ import { chsRegex, getChsSegmenter, SPACE_OR_PUNCTUATION } from '../globals'
import { settings } from '../settings'
import {
chunkArray,
logDebug,
removeDiacritics,
splitCamelCase,
stringsToRegex,
@@ -122,11 +123,13 @@ export class Omnisearch {
* @param paths
*/
public async addFromPaths(paths: string[]): Promise<void> {
logDebug('Adding files', paths)
let documents = (
await Promise.all(
paths.map(async path => await cacheManager.getDocument(path))
)
).filter(d => !!d?.path)
logDebug('Sorting documents to first index markdown')
// Index markdown files first
documents = sortBy(documents, d => (d.path.endsWith('.md') ? 0 : 1))
@@ -138,13 +141,14 @@ export class Omnisearch {
// Split the documents in smaller chunks to add them to minisearch
const chunkedDocs = chunkArray(documents, 500)
for (const docs of chunkedDocs) {
logDebug('Indexing into search engine', docs)
// Update the list of indexed docs
docs.forEach(doc => this.indexedDocuments.set(doc.path, doc.mtime))
// Discard files that may have been already added (though it shouldn't happen)
const alreadyAdded = docs.filter(doc => this.minisearch.has(doc.path))
this.removeFromPaths(alreadyAdded.map(o => o.path))
// Add docs to minisearch
await this.minisearch.addAllAsync(docs)
}
@@ -175,6 +179,8 @@ export class Omnisearch {
return []
}
logDebug('Starting search for', query)
let results = this.minisearch.search(query.segmentsToStr(), {
prefix: term => term.length >= options.prefixLength,
// length <= 3: no fuzziness
@@ -192,6 +198,8 @@ export class Omnisearch {
},
})
logDebug('Found', results.length, 'results')
// Filter query results to only keep files that match query.extensions (if any)
if (query.extensions.length) {
results = results.filter(r => {
@@ -247,6 +255,8 @@ export class Omnisearch {
}
}
logDebug('Sorting and limiting results')
// Sort results and keep the 50 best
results = results.sort((a, b) => b.score - a.score).slice(0, 50)
@@ -257,6 +267,7 @@ export class Omnisearch {
// If the search query contains quotes, filter out results that don't have the exact match
const exactTerms = query.getExactTerms()
if (exactTerms.length) {
logDebug('Filtering with quoted terms')
results = results.filter(r => {
const document = documents.find(d => d.path === r.id)
const title = document?.path.toLowerCase() ?? ''
@@ -270,6 +281,7 @@ export class Omnisearch {
// If the search query contains exclude terms, filter out results that have them
const exclusions = query.exclusions
if (exclusions.length) {
logDebug('Filtering with exclusions')
results = results.filter(r => {
const content = stripMarkdownCharacters(
documents.find(d => d.path === r.id)?.content ?? ''
@@ -277,6 +289,8 @@ export class Omnisearch {
return exclusions.every(q => !content.includes(q.value))
})
}
logDebug('Deduping')
// FIXME:
// Dedupe results - clutch for https://github.com/scambier/obsidian-omnisearch/issues/129
results = results.filter(
@@ -337,17 +351,13 @@ export class Omnisearch {
})
}
// Extract tags from the query
const tags = query.segments
.filter(s => s.value.startsWith('#'))
.map(s => s.value)
const documents = await Promise.all(
results.map(async result => await cacheManager.getDocument(result.id))
)
// Map the raw results to get usable suggestions
const resultNotes = results.map(result => {
logDebug('Locating matches for', result.id)
let note = documents.find(d => d.path === result.id)
if (!note) {
// throw new Error(`Omnisearch - Note "${result.id}" not indexed`)
@@ -363,6 +373,12 @@ export class Omnisearch {
query.segments.forEach(s => {
s.value = s.value.replace(/^#/, '')
})
// Extract tags from the query
const tags = query.segments
.filter(s => s.value.startsWith('#'))
.map(s => s.value)
// Clean search matches that match quoted expressions,
// and inject those expressions instead
const foundWords = [
@@ -376,13 +392,15 @@ export class Omnisearch {
// Tags, starting with #
...tags,
].filter(w => w.length > 1 || /\p{Emoji}/u.test(w))
logDebug('Matching tokens:', foundWords)
// console.log(foundWords)
logDebug('Getting matches locations...')
const matches = this.getMatches(
note.content,
stringsToRegex(foundWords),
query
)
logDebug('Matches:', matches)
const resultNote: ResultNote = {
score: result.score,
foundWords,