--- title: "Retrieve a record" method: GET path: "/v1beta/versions/{version_id}/record" tags: ["Versions"] --- # Retrieve a record `GET /v1beta/versions/{version_id}/record` Retrieves the processed content record for a given version. The content oneof is populated based on source_format: document formats (pdf, docx, markdown) populate the document field; spreadsheet formats (xlsx, csv) populate the spreadsheet field. ## Path parameters - `version_id` string, required — Version ID to retrieve the record for. Pattern: ver_[0-9a-hjkmnp-tv-z]{26} ## Response `200` Success - union - object — GetRecordResponse contains the processed content and layout for a version. The content oneof is populated based on source_format: - document formats (pdf, docx, markdown): document field - spreadsheet formats (xlsx, csv): spreadsheet field - `id` string — Unique record ID (UUID format). - `source_format` string — Source format of the original upload (e.g., pdf, docx, xlsx, csv). - `document` DocumentBundle, required — DocumentBundle groups a document record with its optional spatial layout. - `layout` QcrV1alphaDocumentLayout — Top-level layout structure, joined to a DocumentRecord by record_id. Used with the Block-based record representation (DocumentRecord.blocks). - `bindings` QcrV1alphaLayoutBinding[] — sparse element → layout entries - union - object — LayoutBinding attaches layout data to a specific element in the record. One binding per addressable element. Elements without layout data have no binding entry (sparse). - … - object — LayoutBinding attaches layout data to a specific element in the record. One binding per addressable element. Elements without layout data have no binding entry (sparse). - … - object — LayoutBinding attaches layout data to a specific element in the record. One binding per addressable element. Elements without layout data have no binding entry (sparse). - … - object — LayoutBinding attaches layout data to a specific element in the record. One binding per addressable element. Elements without layout data have no binding entry (sparse). - … - `defaults` QcrV1alphaLayoutDefaults — Document-wide layout defaults. - `dimensions` QcrV1alphaPhysicalDimensions — Physical dimensions in millimeters. - `height_mm` number, float - `width_mm` number, float - `direction` 'DIRECTION_UNSPECIFIED' | 'DIRECTION_LTR' | 'DIRECTION_RTL' — Text direction / reading order. - `margins` QcrV1alphaMargins — Page/section margins in millimeters. - `bottom_mm` number, float - `left_mm` number, float - `right_mm` number, float - `top_mm` number, float - `orientation` 'ORIENTATION_UNSPECIFIED' | 'ORIENTATION_PORTRAIT' | 'ORIENTATION_LANDSCAPE' - `page_size` 'PAGE_SIZE_UNSPECIFIED' | 'PAGE_SIZE_A4' | 'PAGE_SIZE_LETTER' | 'PAGE_SIZE_LEGAL' | 'PAGE_SIZE_A3' | 'PAGE_SIZE_TABLOID' | 'PAGE_SIZE_CUSTOM' - `extracted_at` string, date-time - `pages` QcrV1alphaPageGeometry[] — page-level geometry (all pages) - `dimensions` QcrV1alphaPhysicalDimensions — Physical dimensions in millimeters. - `height_mm` number, float - `width_mm` number, float - `orientation` 'ORIENTATION_UNSPECIFIED' | 'ORIENTATION_PORTRAIT' | 'ORIENTATION_LANDSCAPE' - `page_number` integer — 0-indexed - `record_id` string — references DocumentRecord.id - `source` QcrV1alphaSourceReference — SourceReference links a record back to the source it was extracted from. Three typed fields identify WHAT, WHERE, and WHEN. Everything else — format, content hash, storage location, checksums, external URLs — goes in the attributes bag, because the relevant metadata varies by source type. Examples by source type: File upload: source_id = "intg_01j8k..." (upload channel instance) remote_id = "req_01h2xcejqtf2nbrexx3vqjhp41" (upload request ID) attributes = {format: "pdf", location: "uploads/.../original.pdf", content_hash: "a1b2c3...", storage_etag: "\"abc\""} Gmail integration: source_id = "intg_01k9m..." (Gmail connector instance) remote_id = "18abc123def" (Gmail message ID) attributes = {format: "eml", thread_id: "...", external_url: "https://mail.google.com/..."} Confluence sync: source_id = "intg_01p3q..." (Confluence connector instance) remote_id = "12345678" (Confluence page ID) attributes = {format: "html", space_key: "ENG", external_url: "https://acme.atlassian.net/wiki/..."} - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `ingested_at` string, date-time — When this content was ingested into the system. - `remote_id` string — The content's coordinates in the external system — whatever identifier the source system uses for this specific piece of content. Examples: upload request ID, Gmail message ID, Confluence page ID, S3 object key, Merge remote_id. - `source_id` string — Identifies the integration or connector instance that produced this record. Stable across re-syncs of the same source. Example: "intg_01j8k..." (opaque ID from integration registry). - `record` QcrV1alphaDocumentRecord — DocumentRecord is the QCR representation of an ingested document. Immutable once extracted. Source-format-agnostic. Structure is a flat sequence of blocks. Sections provide optional grouping and can nest, but block-level elements (paragraphs, tables, etc.) are leaves — they do not contain other blocks. This constrains nesting to only what is semantically valid and gives every element a natural positional address. Addressing convention: every element is addressable by its index path through the typed containment hierarchy. Examples: blocks[2] → third block blocks[2].section.blocks[0] → first block inside a section blocks[4].table.rows[1].cells[0] → second row, first cell of a table blocks[3].list.items[2] → third item of a list These addresses are deterministic for a given record and are used by DocumentLayout to join layout data to semantic content. - `blocks` QcrV1alphaBlock[] - union - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - object — Block is the unit of document content. Each block carries exactly one type-specific payload via oneof. Blocks do not nest generically — only SectionData can contain child blocks, enforcing valid containment at the schema level. - … - `id` string - `metadata` QcrV1alphaDocumentMetadata - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `extracted_at` string, date-time - `language` string - `title` string - `source` QcrV1alphaSourceReference — SourceReference links a record back to the source it was extracted from. Three typed fields identify WHAT, WHERE, and WHEN. Everything else — format, content hash, storage location, checksums, external URLs — goes in the attributes bag, because the relevant metadata varies by source type. Examples by source type: File upload: source_id = "intg_01j8k..." (upload channel instance) remote_id = "req_01h2xcejqtf2nbrexx3vqjhp41" (upload request ID) attributes = {format: "pdf", location: "uploads/.../original.pdf", content_hash: "a1b2c3...", storage_etag: "\"abc\""} Gmail integration: source_id = "intg_01k9m..." (Gmail connector instance) remote_id = "18abc123def" (Gmail message ID) attributes = {format: "eml", thread_id: "...", external_url: "https://mail.google.com/..."} Confluence sync: source_id = "intg_01p3q..." (Confluence connector instance) remote_id = "12345678" (Confluence page ID) attributes = {format: "html", space_key: "ENG", external_url: "https://acme.atlassian.net/wiki/..."} - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `ingested_at` string, date-time — When this content was ingested into the system. - `remote_id` string — The content's coordinates in the external system — whatever identifier the source system uses for this specific piece of content. Examples: upload request ID, Gmail message ID, Confluence page ID, S3 object key, Merge remote_id. - `source_id` string — Identifies the integration or connector instance that produced this record. Stable across re-syncs of the same source. Example: "intg_01j8k..." (opaque ID from integration registry). - object — GetRecordResponse contains the processed content and layout for a version. The content oneof is populated based on source_format: - document formats (pdf, docx, markdown): document field - spreadsheet formats (xlsx, csv): spreadsheet field - `id` string — Unique record ID (UUID format). - `source_format` string — Source format of the original upload (e.g., pdf, docx, xlsx, csv). - `spreadsheet` SpreadsheetBundle, required — SpreadsheetBundle groups a spreadsheet record with its optional grid layout. - `layout` QcrV1alphaSpreadsheetLayout - `defaults` QcrV1alphaSpreadsheetLayoutDefaults - `column_width_mm` number, float - `row_height_mm` number, float - `extracted_at` string, date-time - `record_id` string - `sheet_layouts` QcrV1alphaSheetLayout[] - `cell_bindings` QcrV1alphaCellLayoutBinding[] - `alignment` union - … - `col` integer - `row` integer - `text_wrap` boolean - `column_formats` QcrV1alphaColumnFormatBinding[] - `column_index` integer — index into DataTable.columns[] - `number_format_category` 'NUMBER_FORMAT_CATEGORY_UNSPECIFIED' | 'NUMBER_FORMAT_CATEGORY_GENERAL' | 'NUMBER_FORMAT_CATEGORY_NUMBER' | 'NUMBER_FORMAT_CATEGORY_CURRENCY' | 'NUMBER_FORMAT_CATEGORY_ACCOUNTING' | 'NUMBER_FORMAT_CATEGORY_PERCENTAGE' | 'NUMBER_FORMAT_CATEGORY_FRACTION' | 'NUMBER_FORMAT_CATEGORY_SCIENTIFIC' | 'NUMBER_FORMAT_CATEGORY_TEXT' | 'NUMBER_FORMAT_CATEGORY_DATE' | 'NUMBER_FORMAT_CATEGORY_TIME' | 'NUMBER_FORMAT_CATEGORY_DATE_TIME' | 'NUMBER_FORMAT_CATEGORY_DURATION' - `number_format_pattern` string — e.g. "#,##0.00", "yyyy-mm-dd", "0.00%" - `table_name` string — references DataTable.name - `columns` QcrV1alphaColumnLayout[] - `col` integer - `is_hidden` boolean - `width_mm` number, float - `direction` 'DIRECTION_UNSPECIFIED' | 'DIRECTION_LTR' | 'DIRECTION_RTL' — Text direction / reading order. - `frozen_pane` QcrV1alphaFrozenPane - `frozen_col` integer — 0 = no col freeze; N = cols 0..N-1 frozen - `frozen_row` integer — 0 = no row freeze; N = rows 0..N-1 frozen - `is_hidden` boolean - `merged_regions` QcrV1alphaMergedRegion[] - `end_col` integer - `end_row` integer - `start_col` integer - `start_row` integer - `rows` QcrV1alphaRowLayout[] - `height_mm` number, float - `is_hidden` boolean - `row` integer - `sheet_name` string — references Sheet.name - `source` QcrV1alphaSourceReference — SourceReference links a record back to the source it was extracted from. Three typed fields identify WHAT, WHERE, and WHEN. Everything else — format, content hash, storage location, checksums, external URLs — goes in the attributes bag, because the relevant metadata varies by source type. Examples by source type: File upload: source_id = "intg_01j8k..." (upload channel instance) remote_id = "req_01h2xcejqtf2nbrexx3vqjhp41" (upload request ID) attributes = {format: "pdf", location: "uploads/.../original.pdf", content_hash: "a1b2c3...", storage_etag: "\"abc\""} Gmail integration: source_id = "intg_01k9m..." (Gmail connector instance) remote_id = "18abc123def" (Gmail message ID) attributes = {format: "eml", thread_id: "...", external_url: "https://mail.google.com/..."} Confluence sync: source_id = "intg_01p3q..." (Confluence connector instance) remote_id = "12345678" (Confluence page ID) attributes = {format: "html", space_key: "ENG", external_url: "https://acme.atlassian.net/wiki/..."} - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `ingested_at` string, date-time — When this content was ingested into the system. - `remote_id` string — The content's coordinates in the external system — whatever identifier the source system uses for this specific piece of content. Examples: upload request ID, Gmail message ID, Confluence page ID, S3 object key, Merge remote_id. - `source_id` string — Identifies the integration or connector instance that produced this record. Stable across re-syncs of the same source. Example: "intg_01j8k..." (opaque ID from integration registry). - `record` QcrV1alphaSpreadsheetRecord — SpreadsheetRecord is the QCR representation of an ingested spreadsheet workbook (XLSX, CSV, Parquet). Immutable once extracted. Source-format-agnostic. Hierarchy: SpreadsheetRecord 1:M Sheet 1:M DataTable/SheetImage/SheetChart. The semantic core is DataTable: named, typed columns applied to ordered rows. The extractor identifies headers and schemas so consumers don't have to. - `id` string - `metadata` QcrV1alphaSpreadsheetMetadata - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `extracted_at` string, date-time - `title` string - `named_ranges` QcrV1alphaNamedRange[] - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `name` string - `range_ref` string — A1-notation - `sheet_name` string — empty = workbook-scoped - `sheets` QcrV1alphaSheet[] - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `charts` QcrV1alphaSheetChart[] - `anchor_col` integer - `anchor_row` integer - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `chart_type` string — e.g., "bar", "line", "pie" - `rendered` QcrV1alphaVectorImageData — VectorImageData represents a vector graphic (SVG, diagram, chart). - … - `source_ranges` string[] — A1-notation data ranges - `title` string - `images` QcrV1alphaSheetImage[] - `anchor_col` integer - `anchor_row` integer - `height_mm` number, float - `image` QcrV1alphaRasterImageData — RasterImageData represents a raster image (PNG, JPEG, WebP, etc.). - … - `offset_x_mm` number, float — horizontal offset within anchor cell - `offset_y_mm` number, float — vertical offset within anchor cell - `width_mm` number, float - `name` string — unique within workbook, used as join key by SheetLayout - `tables` QcrV1alphaDataTable[] - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `columns` QcrV1alphaColumnDef[] — the schema - … - `has_total_row` boolean - `name` string — from ListObject or extractor-assigned - `range_ref` string — A1-notation origin, e.g. "A1:D100" - `rows` QcrV1alphaDataRow[] — ordered data rows (excludes header) - … - `source` QcrV1alphaSourceReference — SourceReference links a record back to the source it was extracted from. Three typed fields identify WHAT, WHERE, and WHEN. Everything else — format, content hash, storage location, checksums, external URLs — goes in the attributes bag, because the relevant metadata varies by source type. Examples by source type: File upload: source_id = "intg_01j8k..." (upload channel instance) remote_id = "req_01h2xcejqtf2nbrexx3vqjhp41" (upload request ID) attributes = {format: "pdf", location: "uploads/.../original.pdf", content_hash: "a1b2c3...", storage_etag: "\"abc\""} Gmail integration: source_id = "intg_01k9m..." (Gmail connector instance) remote_id = "18abc123def" (Gmail message ID) attributes = {format: "eml", thread_id: "...", external_url: "https://mail.google.com/..."} Confluence sync: source_id = "intg_01p3q..." (Confluence connector instance) remote_id = "12345678" (Confluence page ID) attributes = {format: "html", space_key: "ENG", external_url: "https://acme.atlassian.net/wiki/..."} - `attributes` Struct — `Struct` represents a structured data value, consisting of fields which map to dynamically typed values. In some languages, `Struct` might be supported by a native representation. For example, in scripting languages like JS a struct is represented as an object. The details of that representation are described together with the proto support for the language. The JSON representation for `Struct` is JSON object. - `ingested_at` string, date-time — When this content was ingested into the system. - `remote_id` string — The content's coordinates in the external system — whatever identifier the source system uses for this specific piece of content. Examples: upload request ID, Gmail message ID, Confluence page ID, S3 object key, Merge remote_id. - `source_id` string — Identifies the integration or connector instance that produced this record. Stable across re-syncs of the same source. Example: "intg_01j8k..." (opaque ID from integration registry). ## Other responses - `400` — Invalid request parameters - `401` — Authentication failed - missing or invalid API key - `403` — Authorization failed - valid key but insufficient permissions - `404` — Resource not found - `429` — Too many requests - `500` — Internal server error ## Changes - **2026-04-13** `96e3f520e606` — 1 info - endpoint added [Change history](https://skmtc.dev/factify-inc/apis/factify-api/changes/v1beta/versions/:version_id/record/get.md) --- [API](https://skmtc.dev/factify-inc/apis/factify-api.md) · [All operations](https://skmtc.dev/factify-inc/apis/factify-api/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/factify-inc/factify-api/revisions/8baabdfb82bf/schema)