> ## Documentation Index > Fetch the complete documentation index at: https://developers.phrase.com/llms.txt > Use this file to discover all available pages before exploring further. # Search translation memory (sync) > **Overview** - Searches a single translation memory using the primary TM scoring pipeline. The endpoint always returns up to 15 `SearchTMResponseDto` entries ordered by score. **Modes** - Standard segment search (default): normalize text, leverage penalties, thresholds, tag metadata, and context. This is the behavior when `phraseQuery=true`(default). The service calls the downstream `/search` endpoint with `wildcard=true` to allow combined exact/wildcard candidates in one pass. - Legacy phrase-only mode: set `phraseQuery=false` to disable the wildcard branch (`wildcard=false`, `exactIfQuotes=false`, `combinedExact=false`). Use when you need strict normalized matches without wildcard processing. **Request parameters** - `query` (required): search text. If `trimQuery=true` (default) leading/trailing whitespace is removed before sending to TM service. - `sourceLang` (required) and optional `targetLangs`: locales are normalized to BCP-47. When `targetLangs` is omitted the backend queries all TM targets. - `previousSegment` / `nextSegment`: optional context strings; when provided, the backend raises 101% matches that align with the surrounding segments. - `tagMetadata`: serialized inline tags; improves scoring and avoids dropping tag pairs. - `trimQuery`: default `true`. Set to `false` to preserve leading/trailing spaces (rare). - `phraseQuery`: default `true`. Controls whether wildcard/phrase helpers are enabled (see “Modes”). **Scoring & ordering** - Backend computes `grossScore` (raw similarity) and `score` (penalty-adjusted). 101% matches retain their priority even after penalties. - Ordering priority (per downstream service): `score DESC`, `grossScore DESC`, context match priority DESC, segment key ASC, TM priority ASC, timestamp DESC, TU id DESC. - Numeric replacements or uppercase normalization are not performed here (`replaceFigures=false`, `modifyTransText=false` in this controller). **Limits & behavior** - Returns maximum 15 matches; there is no `offset` or `totalFoundCount`. - Threshold is fixed to `0`, so all results above penalties are returned until limit. - Multiple target locales per TM are not supported, mirroring backend validation. - Logical operators (AND/OR/NOT) and wildcard expressions in `query` are interpreted only when `phraseQuery=true` enables the wildcard path. **Context & caching notes** - Downstream service may consult its 60-second cache for identical search inputs (including context and penalties). Freshly imported matches might appear with a short delay if cached results are reused. - If previous/next segments are omitted the search still runs, but 101% context boosts cannot trigger. **Error handling** - Invalid locales, blank `query`, or an inaccessible TM ID return `400`/`403`. - Backend may throw when multiple target locales are submitted for a single TM; the controller surfaces that error unchanged. ## OpenAPI ````yaml /openapi/phrase-tms-latest.json post /api2/v1/transMemories/{transMemoryUid}/search openapi: 3.0.0 info: description: |- Welcome to Phrase's TMS API documentation. Please visit our [help center](https://support.phrase.com/hc/en-us/sections/5709662083612) for more information about the APIs. If you have any questions, please contact [Support](https://support.phrase.com/hc/requests/new). Please, include the `User-Agent` header with the name of your application or project. It might be a good idea to include some sort of contact information as well, so that we can get in touch if necessary. Examples of excellent `User-Agent` headers: > User-Agent: Example mobile app (example@phrase.com)
User-Agent: ACME Inc Java 1.8 Client (http://acmeinc.com/contact) version: Latest title: Phrase TMS API servers: - url: https://cloud.memsource.com/web security: - ApiToken: [] - OAuth2: [] tags: - name: Additional Workflow Step - name: Analysis - name: Async Request - name: Authentication - name: Automations - name: Bilingual File - name: Business Unit - name: Buyer - name: Client - name: Connector - name: Conversations - name: Cost Center - name: Custom Fields - name: Custom File Type - name: Domain - name: Due Date Scheme - name: Email Template - name: File - name: Glossary - name: Import settings - name: Job - name: Language AI - name: Language Assets - name: Language Quality Assessment - name: Machine Translation - name: Machine Translation Settings - name: Mapping - name: Net Rate Scheme - name: Notifications - name: Price List - name: Project - name: Project Template - name: Provider - name: Quality Assurance - name: Quality Profile - name: Quote - name: Reference File - name: SCIM - name: Segment - name: Segmentation Rules - name: Service - name: Spell Check - name: SubDomain - name: Supported Languages - name: Term Base - name: Translation - name: Translation Memory - name: User - name: Vendor - name: Webhook - name: Workflow Step - name: Workflow changes - name: XML Assistant paths: /api2/v1/transMemories/{transMemoryUid}/search: post: tags: - Translation Memory summary: Search translation memory (sync) description: > **Overview** - Searches a single translation memory using the primary TM scoring pipeline. The endpoint always returns up to 15 `SearchTMResponseDto` entries ordered by score. **Modes** - Standard segment search (default): normalize text, leverage penalties, thresholds, tag metadata, and context. This is the behavior when `phraseQuery=true`(default). The service calls the downstream `/search` endpoint with `wildcard=true` to allow combined exact/wildcard candidates in one pass. - Legacy phrase-only mode: set `phraseQuery=false` to disable the wildcard branch (`wildcard=false`, `exactIfQuotes=false`, `combinedExact=false`). Use when you need strict normalized matches without wildcard processing. **Request parameters** - `query` (required): search text. If `trimQuery=true` (default) leading/trailing whitespace is removed before sending to TM service. - `sourceLang` (required) and optional `targetLangs`: locales are normalized to BCP-47. When `targetLangs` is omitted the backend queries all TM targets. - `previousSegment` / `nextSegment`: optional context strings; when provided, the backend raises 101% matches that align with the surrounding segments. - `tagMetadata`: serialized inline tags; improves scoring and avoids dropping tag pairs. - `trimQuery`: default `true`. Set to `false` to preserve leading/trailing spaces (rare). - `phraseQuery`: default `true`. Controls whether wildcard/phrase helpers are enabled (see “Modes”). **Scoring & ordering** - Backend computes `grossScore` (raw similarity) and `score` (penalty-adjusted). 101% matches retain their priority even after penalties. - Ordering priority (per downstream service): `score DESC`, `grossScore DESC`, context match priority DESC, segment key ASC, TM priority ASC, timestamp DESC, TU id DESC. - Numeric replacements or uppercase normalization are not performed here (`replaceFigures=false`, `modifyTransText=false` in this controller). **Limits & behavior** - Returns maximum 15 matches; there is no `offset` or `totalFoundCount`. - Threshold is fixed to `0`, so all results above penalties are returned until limit. - Multiple target locales per TM are not supported, mirroring backend validation. - Logical operators (AND/OR/NOT) and wildcard expressions in `query` are interpreted only when `phraseQuery=true` enables the wildcard path. **Context & caching notes** - Downstream service may consult its 60-second cache for identical search inputs (including context and penalties). Freshly imported matches might appear with a short delay if cached results are reused. - If previous/next segments are omitted the search still runs, but 101% context boosts cannot trigger. **Error handling** - Invalid locales, blank `query`, or an inaccessible TM ID return `400`/`403`. - Backend may throw when multiple target locales are submitted for a single TM; the controller surfaces that error unchanged. operationId: search parameters: - name: transMemoryUid in: path required: true schema: type: string requestBody: content: application/json: schema: $ref: '#/components/schemas/SearchRequestDtoV1' responses: '200': description: successful operation content: application/json: schema: $ref: '#/components/schemas/SearchResponseListTmDtoV1' '400': description: Bad request '401': description: Not authorized '403': description: Forbidden '404': description: Resource not found '405': description: Method not allowed '408': description: Timeout '410': description: Gone '415': description: Unsupported media type '429': description: Too many requests '500': description: Internal server error '501': description: Not implemented components: schemas: SearchRequestDtoV1: type: object required: - query - sourceLang properties: nextSegment: type: string phraseQuery: type: boolean description: 'Return both wildcard and exact search results. Default: true' previousSegment: type: string query: type: string sourceLang: type: string tagMetadata: type: array items: $ref: '#/components/schemas/TagMetadataDtoV1' targetLangs: type: array items: type: string trimQuery: type: boolean description: 'Remove leading and trailing whitespace from query. Default: true' SearchResponseListTmDtoV1: type: object properties: searchResults: type: array items: $ref: '#/components/schemas/SearchTMResponseDtoV1' TagMetadataDtoV1: type: object properties: content: type: string id: type: string transAttributes: type: string type: type: string SearchTMResponseDtoV1: type: object properties: grossScore: type: number format: double score: type: number format: double segmentId: type: string source: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMSegmentDtoV1' subSegment: type: boolean transMemory: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMTransMemoryDtoV1' translations: type: array items: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMSegmentDtoV1' SearchTMResponseDtoSearchTMSegmentDtoV1: type: object properties: client: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMClientDtoV1' createdAt: type: integer format: int64 createdBy: $ref: '#/components/schemas/UserReferenceV1' domain: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMDomainDtoV1' filename: type: string id: type: string key: type: string lang: type: string modifiedAt: type: integer format: int64 modifiedBy: $ref: '#/components/schemas/UserReferenceV1' nextSegment: type: string previousSegment: type: string project: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMProjectDtoV1' rtl: type: boolean subDomain: $ref: '#/components/schemas/SearchTMResponseDtoSearchTMSubDomainDtoV1' tagMetadata: type: array items: $ref: '#/components/schemas/TagMetadata' text: type: string SearchTMResponseDtoSearchTMTransMemoryDtoV1: type: object properties: id: type: string name: type: string reverse: type: boolean uid: type: string SearchTMResponseDtoSearchTMClientDtoV1: type: object properties: id: type: integer format: int64 name: type: string UserReferenceV1: type: object properties: email: type: string firstName: type: string id: type: string lastName: type: string role: type: string enum: - SYS_ADMIN - SYS_ADMIN_READ - ADMIN - PROJECT_MANAGER - LINGUIST - GUEST - SUBMITTER - PORTAL_MEMBER - BOT uid: type: string userName: type: string SearchTMResponseDtoSearchTMDomainDtoV1: type: object properties: id: type: integer format: int64 name: type: string SearchTMResponseDtoSearchTMProjectDtoV1: type: object properties: id: type: integer format: int64 name: type: string uid: type: string SearchTMResponseDtoSearchTMSubDomainDtoV1: type: object properties: id: type: integer format: int64 name: type: string TagMetadata: type: object properties: content: type: string id: type: string transAttributes: type: string type: type: string securitySchemes: ApiToken: description: >- Get a token from `auth/login` [endpoint](#operation/login) and then pass it in the `Authorization` HTTP header in every subsequent API call. For more information visit our [help center](https://support.phrase.com/hc/en-us/articles/5709662181404-API-Authentication-TMS-#token-0-0). type: apiKey name: Authorization in: header OAuth2: description: >- A standard OAuth 2.0 authorization code flow. For more information visit our [help center](https://support.phrase.com/hc/en-us/articles/5709662181404-API-Authentication-TMS-#oauth-2-0-0-1). type: oauth2 flows: authorizationCode: authorizationUrl: https://cloud.memsource.com/web/oauth/authorize tokenUrl: https://cloud.memsource.com/web/oauth/token scopes: {} ````