feat: config options for fuzzy search

db-service/lib/cqn4sql.js

@@ -2203,7 +2203,7 @@ function cqn4sql(originalQuery, model) {
       const searchFunc = {
         func: 'search',
         args: [
-          searchIn.length > 1 ? { list: searchIn } : { ...searchIn[0] },
+          { list: searchIn },
           xpr.length === 1 && 'val' in xpr[0] ? xpr[0] : { xpr },
         ],
       }

hana/lib/cql-functions.js

@@ -24,10 +24,63 @@ const StandardFunctions = {
   contains: (...args) => args.length > 2 ? `CONTAINS(${args})` : `(CASE WHEN coalesce(locate(${args}),0)>0 THEN TRUE ELSE FALSE END)`,
   concat: (...args) => `(${args.map(a => (a.xpr ? `(${a})` : a)).join(' || ')})`,
   search: function (ref, arg) {
+    if (cds.env.hana.fuzzy === false) {
+      // REVISIT: remove once the protocol adapter only creates vals
+      arg = arg.xpr ? arg.xpr : arg
+      if (Array.isArray(arg)) arg = [{ val: arg.filter(a => a.val).map(a => a.val).join(' ') }]
+      else arg = [arg]
+      const searchTerms = arg[0].val
+          .match(/("")|("(?:[^"]|\\")*(?:[^\\]|\\\\)")|(\S*)/g)
+          .filter(el => el.length).map(el => `%${el.replace(/^\"|\"$/g, '').toLowerCase()}%`)
+
+      const columns = ref.list
+      const xpr = []
+      for (const s of searchTerms) {
+        const nestedXpr = []
+        for (const c of columns) {
+          if (nestedXpr.length) nestedXpr.push('or')
+          nestedXpr.push({ func: 'lower', args: [c]}, 'like', {val: s})
+        }
+        if (xpr.length) xpr.push('and')
+        xpr.push({xpr: nestedXpr})
+      }
+
+      const { toString } = ref
+      return `(CASE WHEN (${toString({ xpr })}) THEN TRUE ELSE FALSE END)`
+    }
+
+    // fuzziness config
+    const fuzzyIndex = cds.env.hana?.fuzzy || 0.7
+
+    const csnElements = ref.list
+    // if column specific value is provided, the configuration has to be defined on column level
+    if (csnElements.some(e => e.element?.['@Search.ranking'] || e.element?.['@Search.fuzzinessThreshold'])) {
+      csnElements.forEach(e => {
+        let fuzzy = `FUZZY`
+
+        // weighted search
+        const rank = e.element?.['@Search.ranking']?.['=']
+        if(rank === 'HIGH') fuzzy += ' WEIGHT 0.8'
+        else if(rank === 'LOW') fuzzy += ' WEIGHT 0.3'
+        else fuzzy += ' WEIGHT 0.5' // MEDIUM
+
+        // fuzziness
+        fuzzy+= ` MINIMAL TOKEN SCORE ${e.element?.['@Search.fuzzinessThreshold'] || fuzzyIndex}`
+        fuzzy+= " SIMILARITY CALCULATION MODE 'search'"
+
+        // rewrite ref to xpr to mix in search config
+        // ensure in place modification to reuse .toString method that ensures quoting
+        e.xpr = [{ ref: e.ref }, fuzzy]
+        delete e.ref
+      })
+    } else {
+      ref = `${ref} FUZZY MINIMAL TOKEN SCORE ${fuzzyIndex} SIMILARITY CALCULATION MODE 'search'`
+    }
+
     // REVISIT: remove once the protocol adapter only creates vals
     if (Array.isArray(arg.xpr)) arg = { val: arg.xpr.filter(a => a.val).map(a => a.val).join(' ') }
-    // REVISIT: make this more configurable
-    return (`(CASE WHEN SCORE(${arg} IN ${ref} FUZZY MINIMAL TOKEN SCORE 0.7 SIMILARITY CALCULATION MODE 'search') > 0 THEN TRUE ELSE FALSE END)`)
+
+    return (`(CASE WHEN SCORE(${arg} IN ${ref}) > 0 THEN TRUE ELSE FALSE END)`)
   },
 
   // Date and Time Functions

hana/test/fuzzy.cds

@@ -1 +1,5 @@
-using {sap.capire.bookshop.Books as Books} from '../../test/bookshop/db/schema.cds';
+using {sap.capire.bookshop.BooksAnnotated as BooksAnnotated} from '../../test/bookshop/db/schema.cds';
+
+annotate BooksAnnotated with @cds.search: {title, descr, currency.code};
+annotate BooksAnnotated:title with @(Search.ranking: HIGH, Search.fuzzinessThreshold: 0.9);
+annotate BooksAnnotated:descr with @(Search.ranking: LOW, Search.fuzzinessThreshold: 0.6);

hana/test/fuzzy.test.js

@@ -1,19 +1,78 @@
 const cds = require('../../test/cds')
 
-describe('Fuzzy search', () => {
+describe('search', () => {
   const { expect } = cds.test(__dirname, 'fuzzy.cds')
 
-  test('select', async () => {
-    const { Books } = cds.entities('sap.capire.bookshop')
-    const res = await SELECT.from(Books).where({
-      func: 'contains',
-      args: [
-        { list: [{ ref: ['title'] }, { ref: ['descr'] }] },
-        { val: 'poem' },
-        { func: 'FUZZY', args: [{ val: 0.8 }, { val: 'similarCalculationMode=searchCompare' }] }
-      ]
+  beforeEach (() => {
+    delete cds.env.hana.fuzzy
+  })
+
+  describe('fuzzy', () => {
+    test('default', async () => {
+      const { Books } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(Books).search('"autobio"').columns('1')
+      const {sql} = cqn.toSQL()
+      expect(sql).to.include('FUZZY MINIMAL TOKEN SCORE 0.7')
+      const res = await cqn
+      expect(res.length).to.be(2) // Eleonora and Jane Eyre
+    })
+
+    //HCE returns different result than HXE
+    test.skip('multiple search terms', async () => {
+      const { Books } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(Books).search('"autobio" "jane"').columns('1')
+      const {sql, values} = cqn.toSQL()
+      expect(sql).to.include('FUZZY MINIMAL TOKEN SCORE 0.7')
+      expect(values[0]).to.eq('"autobio" "jane"') // taken as is
+      const res = await cqn
+      expect(res.length).to.be(2) // Eleonora and Jane Eyre
+    })
+
+    test('global config', async () => {
+      cds.env.hana.fuzzy = 1
+      const { Books } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(Books).search('"autobio"').columns('1')
+      const {sql} = cqn.toSQL()
+      expect(sql).to.include('FUZZY MINIMAL TOKEN SCORE 1')
+      const res = await cqn
+      expect(res.length).to.be(2) // Eleonora and Jane Eyre
     })
 
-    expect(res).to.have.property('length').to.be.eq(1)
+    test('annotations', async () => {
+      const { BooksAnnotated } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(BooksAnnotated).search('"heights"').columns('1')
+      const {sql} = cqn.toSQL()
+      expect(sql).to.include('title FUZZY WEIGHT 0.8 MINIMAL TOKEN SCORE 0.9')
+      expect(sql).to.include('code FUZZY WEIGHT 0.5 MINIMAL TOKEN SCORE 0.7')
+      expect(sql).to.include('descr FUZZY WEIGHT 0.3 MINIMAL TOKEN SCORE 0.6')
+
+      const res = await SELECT.from(BooksAnnotated).search('"heights"')
+      expect(res[0].title).to.eq('Wuthering Heights')
+    })
+  })
+
+  describe('like', () => {
+    beforeEach (() => cds.env.hana.fuzzy = false)
+    test('fallback - 1 search term', async () => {
+      const { Books } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(Books).search('"autobio"').columns('1')
+      const {sql} = cqn.toSQL()
+      // 5 columns to be searched createdBy, modifiedBy, title, descr, currency_code
+      expect(sql.match(/(like)/g).length).to.be(5)
+      const res = await cqn
+      expect(res.length).to.be(2) // Eleonora and Jane Eyre
+    })
+
+    test('fallback - 2 search terms', async () => {
+      const { Books } = cds.entities('sap.capire.bookshop')
+      const cqn = SELECT.from(Books).search('"autobio"', '"Jane"').columns('1')
+      const {sql, values} = cqn.toSQL()
+      // 5 columns to be searched createdBy, modifiedBy, title, descr, currency_code
+      expect(sql.match(/(like)/g).length).to.be(10)
+      expect(values).to.include('%autobio%')
+      expect(values).to.include('%jane%')
+      const res = await cqn
+      expect(res.length).to.be(1) // Jane Eyre
+    })
   })
 })

test/bookshop/db/schema.cds

@@ -67,4 +67,5 @@ entity C : managed {
       B   : Integer;
       toB : Composition of many B
               on toB.ID = $self.B;
-}
+};
+entity BooksAnnotated as projection on Books;