✨ improve: sort search results by textual similarity; resolve #37

Squashed commit of the following:

commit b0afbb01381f42590b43e5095c4ef043803e57e9
Author: Huey <hello@huey.xyz>
Date:   Sat Jun 12 13:18:11 2021 +0800

    ⚡improve: sort search results using Ratcliff/Obershelp

commit 2ae77404f2296055059d90f2c1efdacb33a2e34d
Author: Huey <hello@huey.xyz>
Date:   Sat Jun 12 12:03:45 2021 +0800

    SG: return only matching citations

package.json

@@ -1,6 +1,6 @@
 {
   "name": "clerkent",
-  "version": "2.5.0",
+  "version": "2.6.0",
   "private": true,
   "description": "quick search for international caselaw and legislation",
   "repository": "https://github.com/lacuna-technologies/clerkent.git",
@@ -47,6 +47,7 @@
     "axios": "^0.21.1",
     "axios-cache-adapter": "^2.7.3",
     "cheerio": "^1.0.0-rc.9",
+    "gestalt-pattern-matcher": "^0.0.12",
     "memoizee": "^0.4.15",
     "qs": "^6.10.1",
     "react": "^17.0.2",

src/utils/Finder/CaseCitationFinder/utils.ts

@@ -23,4 +23,4 @@ export const sortCitationsByVolume = (abbrsList, citationsArray: any[], attribut
     abbrsList,
     citationsArray.map(c => c[attribute]),
   ).map(c => citationsArray.find(v => v[attribute] === c))
-}
+}

src/utils/scraper/AU/AU.ts

@@ -5,6 +5,7 @@ import Logger from '../../Logger'
 import Constants from '../../Constants'
 import { sortAUCitations } from '../../Finder/CaseCitationFinder/AU'
 import Helpers from '../../Helpers'
+import { sortByNameSimilarity } from '../utils'
 
 const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
   try {
@@ -16,10 +17,13 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.AU.id)
 
-    return sortAUCitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
-    )
+    return sortByNameSimilarity(
+      caseName,
+      sortAUCitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
+    ) 
   } catch (error) {
     Logger.error(error)
   }

src/utils/scraper/CA/CA.ts

@@ -5,6 +5,7 @@ import Logger from '../../Logger'
 import Constants from '../../Constants'
 import { sortCACitations } from '../../Finder/CaseCitationFinder/CA'
 import Helpers from '../../Helpers'
+import { sortByNameSimilarity } from '../utils'
 
 const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
   try {
@@ -16,10 +17,13 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.CA.id)
 
-    return sortCACitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
-    )
+    return sortByNameSimilarity(
+      caseName,
+      sortCACitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
+    ) 
   } catch (error) {
     Logger.error(error)
   }

src/utils/scraper/EU/EU.ts

@@ -5,6 +5,7 @@ import type Law from '../../../types/Law'
 import Constants from '../../Constants'
 import Helpers from '../../Helpers'
 import Logger from '../../Logger'
+import { sortByNameSimilarity } from '../utils'
 
 const getLegislation = EURLex.getLegislation
 
@@ -17,7 +18,10 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.EU.id)
 
-    return Helpers.uniqueBy(results, `citation`)
+    return sortByNameSimilarity(
+      caseName,
+      Helpers.uniqueBy(results, `citation`),
+    ) 
   } catch (error) {
     Logger.error(error)
   }
@@ -30,7 +34,7 @@ const getCaseByCitation = async (citation: string, court: string): Promise<Law.C
     try {
       return await option.getCaseByCitation(citation)
     } catch (error) {
-      console.error(error)
+      Logger.error(error)
     }
   }
   return []

src/utils/scraper/HK/HK.ts

@@ -7,6 +7,7 @@ import Logger from '../../Logger'
 import Constants from '../../Constants'
 import { sortHKCitations } from '../../Finder/CaseCitationFinder/HK'
 import Helpers from '../../Helpers'
+import { sortByNameSimilarity } from '../utils'
 
 const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
   try {
@@ -19,10 +20,13 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.HK.id)
 
-    return sortHKCitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
-    )
+    return sortByNameSimilarity(
+      caseName,
+      sortHKCitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
+    ) 
   } catch (error) {
     Logger.error(error)
   }

src/utils/scraper/NZ/NZ.ts

@@ -5,6 +5,7 @@ import Logger from '../../Logger'
 import Helpers from '../../Helpers'
 import Constants from '../../Constants'
 import { sortNZCitations } from '../../Finder/CaseCitationFinder/NZ'
+import { sortByNameSimilarity } from '../utils'
 
 const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
   try {
@@ -16,9 +17,12 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.NZ.id)
 
-    return sortNZCitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
+    return sortByNameSimilarity(
+      caseName,
+      sortNZCitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
     )
   } catch (error) {
     Logger.error(error)

src/utils/scraper/SG/SG.ts

@@ -7,6 +7,7 @@ import Helpers from '../../Helpers'
 import Logger from '../../Logger'
 import Constants from '../../Constants'
 import { sortSGCitations } from '../../Finder/CaseCitationFinder/SG'
+import { sortByNameSimilarity } from '../utils'
 
 const getLegislation = SSO.getLegislation
 
@@ -21,9 +22,12 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.SG.id)
 
-    return sortSGCitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
+    return sortByNameSimilarity(
+      caseName,
+      sortSGCitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
     )
   } catch (error) {
     Logger.error(error)

src/utils/scraper/SG/SGSC.ts

@@ -29,14 +29,18 @@ const parseCase = ($: cheerio.Root, cheerioElement: cheerio.Element): Law.Case =
   }
 }
 
+const trimLeadingPageZeros = (citation: string) => citation.replace(/ 0+([1-9]+)$/, ` $1`)
+
 const getCaseByCitation = async (citation: string): Promise<Law.Case[]> => {
   const { data } = await Request.get(getSearchResults(citation))
   const $ = cheerio.load(data)
 
   const results = $(`.judgmentpage`)
     .map((_, element) => parseCase($, element))
     .get()
-    .filter((match: Law.Case)=> match.citation.toLowerCase() === citation.toLowerCase())
+    .filter(({ citation: scrapedCitation })=> (
+      trimLeadingPageZeros(scrapedCitation).toLowerCase() === citation.toLowerCase()
+    ))
   Logger.log(`SGSC scrape results`, results)
   return results
 }

src/utils/scraper/SG/SLW.ts

@@ -29,10 +29,7 @@ const getCaseByCitation = async (citation: string): Promise<Law.Case[]> => {
 
   const results = data
     .map(([name, link]) => parseCase(name, link))
-    .filter(({ citation: scrapedCitation }) => (
-      Helpers.isCitationValid(scrapedCitation) &&
-      citation === scrapedCitation
-    ))
+    .filter(({ citation: scrapedCitation }) => citation.toLowerCase() === scrapedCitation.toLowerCase())
   Logger.log(`SLW scrape results`, results)
   return results
 }

src/utils/scraper/UK/UK.ts

@@ -6,6 +6,7 @@ import Logger from '../../Logger'
 import Helpers from '../../Helpers'
 import { sortUKCitations } from '../../Finder/CaseCitationFinder/UK'
 import Constants from '../../Constants'
+import { sortByNameSimilarity } from '../utils'
 
 const getLegislation = LegislationGovUk.getLegislation
 const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
@@ -18,9 +19,12 @@ const getCaseByName = async (caseName: string): Promise<Law.Case[]> => {
     .flatMap(({ value }: PromiseFulfilledResult<Law.Case[]>) => value)
     .filter(({ jurisdiction }) => jurisdiction === Constants.JURISDICTIONS.UK.id)
 
-    return sortUKCitations(
-      Helpers.uniqueBy(results, `citation`),
-      `citation`,
+    return sortByNameSimilarity(
+      caseName,
+      sortUKCitations(
+        Helpers.uniqueBy(results, `citation`),
+        `citation`,
+      ),
     )
   } catch (error) {
     Logger.error(error)

src/utils/scraper/utils.ts

@@ -0,0 +1,8 @@
+import GestaltSimilarity from 'gestalt-pattern-matcher'
+import Law from '../../types/Law'
+
+export const sortByNameSimilarity = (query: string, cases: Law.Case[]) => cases.sort((a, b) => {
+  const similarityA = GestaltSimilarity(query, a.name)
+  const similarityB = GestaltSimilarity(query, b.name)
+  return similarityB - similarityA
+})