fix: catch name variants when auto-blocking LF project orgs (CM-1199)

backend/src/services/organizationService.ts

@@ -6,7 +6,14 @@ import {
   organizationMergeAction,
   organizationUnmergeAction,
 } from '@crowd/audit-logs'
-import { Error400, Error404, Error409, mergeObjects, normalizeHostname } from '@crowd/common'
+import {
+  Error400,
+  Error404,
+  Error409,
+  generateOrganizationNameVariants,
+  mergeObjects,
+  normalizeHostname,
+} from '@crowd/common'
 import { unmergeRoles } from '@crowd/common_services'
 import {
   addMemberRole,
@@ -31,7 +38,7 @@ import {
 } from '@crowd/data-access-layer/src/organizations'
 import {
   decrementOrganizationMergeSuggestionCounts,
-  findLfSegmentByName,
+  findManyLfSegmentsByNames,
   getOrganizationsCommonProjectGroupSegmentIds,
 } from '@crowd/data-access-layer/src/segments'
 import { LoggerBase } from '@crowd/logging'
@@ -927,8 +934,11 @@ export default class OrganizationService extends LoggerBase {
         if (data.displayName) {
           // Block organization affiliation if a LF segment (project, subproject, or project group)
           // has the same name as the organization when creating one.
-          const lfSegment = await findLfSegmentByName(qx, data.displayName)
-          if (lfSegment) {
+          const lfSegments = await findManyLfSegmentsByNames(
+            qx,
+            generateOrganizationNameVariants(data.displayName),
+          )
+          if (lfSegments.length > 0) {
             this.log.info(
               { displayName: data.displayName },
               'Found segment with the same name as the organization, blocking affiliation!',

backend/src/services/segmentService.ts

@@ -1,9 +1,9 @@
 import { Transaction } from 'sequelize'
 
-import { Error400, validateNonLfSlug } from '@crowd/common'
+import { Error400, generateOrganizationNameVariants, validateNonLfSlug } from '@crowd/common'
 import {
   QueryExecutor,
-  findOrganizationsByName,
+  findManyOrganizationsByNames,
   updateOrganization,
 } from '@crowd/data-access-layer'
 import { ICreateInsightsProject, findBySlug } from '@crowd/data-access-layer/src/collections'
@@ -727,7 +727,10 @@ export default class SegmentService extends LoggerBase {
     })
 
     // Check if there is an existing organization with segment name
-    const organizations = await findOrganizationsByName(qx, segmentName)
+    const organizations = await findManyOrganizationsByNames(
+      qx,
+      generateOrganizationNameVariants(segmentName),
+    )
 
     if (organizations.length === 0) {
       return []

services/libs/common/src/index.ts

@@ -35,6 +35,7 @@ export * from './rawQueryParser'
 export * from './byteLength'
 export * from './domain'
 export * from './displayName'
+export * from './organization'
 export * from './country'
 export * from './jira'
 export * from './email'

services/libs/common/src/organization.ts

@@ -0,0 +1,59 @@
+export function generateOrganizationNameVariants(name: string): string[] {
+  const exact = name.trim().toLowerCase().replace(/\s+/g, ' ')
+  if (!exact) {
+    return []
+  }
+
+  const variants = new Set<string>([exact])
+  const add = (value: string) => {
+    const normalized = value.trim().toLowerCase().replace(/\s+/g, ' ')
+    if (normalized) {
+      variants.add(normalized)
+    }
+  }
+
+  let withoutParens = exact
+  if (exact.endsWith(')')) {
+    const open = exact.lastIndexOf('(')
+    if (open !== -1 && !exact.slice(open + 1, -1).includes(')')) {
+      withoutParens = exact.slice(0, open).trimEnd()
+    }
+  }
+  if (withoutParens !== exact && withoutParens.length >= 8) {
+    add(withoutParens)
+  }
+
+  for (const value of [...variants]) {
+    if (value.startsWith('the ') && value.slice(4).length >= 8) {
+      add(value.slice(4))
+    }
+  }
+
+  for (const value of [...variants]) {
+    for (const suffix of ['project', 'foundation', 'initiative']) {
+      const token = ` ${suffix}`
+      if (value.endsWith(token)) {
+        const base = value.slice(0, -token.length).trim()
+        if (base.length >= 6) {
+          add(base)
+        }
+      } else if (value.length >= 4 && !value.includes('(')) {
+        add(`${value}${token}`)
+      }
+    }
+  }
+
+  for (const value of [...variants]) {
+    if (value.includes('-')) {
+      add(value.replace(/-/g, ' '))
+    }
+    if (value.includes(' ')) {
+      add(value.replace(/ /g, '-'))
+    }
+    if (value.includes('.')) {
+      add(value.replace(/\./g, ''))
+    }
+  }
+
+  return [...variants]
+}

services/libs/data-access-layer/src/organizations/base.ts

@@ -1,6 +1,7 @@
 import {
   DEFAULT_TENANT_ID,
   UnrepeatableError,
+  generateOrganizationNameVariants,
   generateUUIDv1,
   normalizeHostname,
 } from '@crowd/common'
@@ -18,7 +19,7 @@ import {
 } from '@crowd/types'
 
 import { QueryExecutor } from '../queryExecutor'
-import { findLfSegmentByName } from '../segments'
+import { findManyLfSegmentsByNames } from '../segments'
 import { QueryOptions, QueryResult, prepareBulkInsert, queryTable, queryTableById } from '../utils'
 import { prepareSelectColumns } from '../utils'
 
@@ -131,24 +132,23 @@ export async function findOrgsByIds(
   return results
 }
 
-export async function findOrganizationsByName(
+export async function findManyOrganizationsByNames(
   qx: QueryExecutor,
-  name: string,
-  options: { limit?: number } = {},
+  names: string[],
 ): Promise<IDbOrganization[]> {
-  const { limit } = options
+  const normalized = names.map((name) => name.trim().toLowerCase()).filter(Boolean)
+  if (normalized.length === 0) {
+    return []
+  }
 
   return qx.select(
     `
       select ${prepareSelectColumns(ORG_SELECT_COLUMNS, 'o')}
       from organizations o
-      where lower(trim(o."displayName")) = lower(trim($(name)))
-      ${limit !== undefined ? 'limit $(limit)' : ''}
+      where o."deletedAt" is null
+        and trim(lower(o."displayName")) in ($(names:csv))
     `,
-    {
-      name,
-      limit,
-    },
+    { names: normalized },
   )
 }
 
@@ -584,9 +584,9 @@ export async function findOrCreateOrganization(
 
     if (!existing) {
       const organizations = await logExecutionTimeV2(
-        async () => findOrganizationsByName(qe, data.displayName, { limit: 1 }),
+        async () => findManyOrganizationsByNames(qe, [data.displayName]),
         log,
-        'organizationService -> findOrCreateOrganization -> findOrganizationsByName',
+        'organizationService -> findOrCreateOrganization -> findManyOrganizationsByNames',
       )
 
       if (organizations.length > 0) {
@@ -673,8 +673,11 @@ export async function findOrCreateOrganization(
 
       // Block organization affiliation if a segment (project, subproject, or project group)
       // has the same name as the organization when creating one.
-      const lfSegment = await findLfSegmentByName(qe, displayName)
-      if (lfSegment) {
+      const lfSegments = await findManyLfSegmentsByNames(
+        qe,
+        generateOrganizationNameVariants(displayName),
+      )
+      if (lfSegments.length > 0) {
         payload.isAffiliationBlocked = true
       }
 

services/libs/data-access-layer/src/segments/index.ts

@@ -34,19 +34,27 @@ export async function findProjectGroupByName(
   )
 }
 
-export async function findLfSegmentByName(
+export async function findManyLfSegmentsByNames(
   qx: QueryExecutor,
-  name: string,
-): Promise<SegmentData | null> {
-  return qx.selectOneOrNone(
+  names: string[],
+): Promise<SegmentData[]> {
+  const normalized = names.map((name) => name.trim().toLowerCase()).filter(Boolean)
+  if (normalized.length === 0) {
+    return []
+  }
+
+  return qx.select(
     `
       SELECT *
       FROM segments
       WHERE "isLF" = true
-        AND trim(lower(name)) = trim(lower($(name)))
-      LIMIT 1;
+        AND (
+          trim(lower(name)) IN ($(names:csv))
+          OR trim(both FROM regexp_replace(trim(lower(name)), '\\s*\\([^)]*\\)\\s*$', ''))
+            IN ($(names:csv))
+        )
     `,
-    { name },
+    { names: normalized },
   )
 }