@alien_intelligence/data-api-client 1.4.14 → 1.11.0

package/src/src/models/Segment.ts

@@ -0,0 +1,148 @@
+/* tslint:disable */
+/* eslint-disable */
+/**
+ * Client Data API
+ * Client Cluster Data API - Manifest-based storage service
+ *
+ * The version of the OpenAPI document: 1.11.0
+ * 
+ *
+ * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
+ * https://openapi-generator.tech
+ * Do not edit the class manually.
+ */
+
+import { mapValues } from '../runtime';
+/**
+ * One citable unit of a processed document, as a span of the canonical text.
+ * @export
+ * @interface Segment
+ */
+export interface Segment {
+    /**
+     * Position in this document's segment list, dense from 0. This is the ordering key and half of the page document's primary key; it is not a page number.
+     * @type {number}
+     * @memberof Segment
+     */
+    seq: number;
+    /**
+     * What this unit is. 'page' covers folios, PDF pages and worksheets.
+     * @type {string}
+     * @memberof Segment
+     */
+    kind?: SegmentKindEnum;
+    /**
+     * What the document calls this unit - the string a citation shows. Sparse and not necessarily numeric: BnF folios skip, ocr-gateway reserves the index of a page it lost. Never derive ordering from it.
+     * @type {string}
+     * @memberof Segment
+     */
+    label: string;
+    /**
+     * 'source' if the document itself drew this boundary, 'derived' if it was invented to fit an indexing budget. Required rather than defaulted: a boundary whose provenance was never stated is not a boundary worth citing.
+     * @type {string}
+     * @memberof Segment
+     */
+    origin: SegmentOriginEnum;
+    /**
+     * 
+     * @type {string}
+     * @memberof Segment
+     */
+    splitRule?: SegmentSplitRuleEnum | null;
+    /**
+     * Offset of this segment's first character in the canonical text.
+     * @type {number}
+     * @memberof Segment
+     */
+    charStart: number;
+    /**
+     * Offset one past this segment's last character. Exclusive, so text[char_start:char_end] is the segment.
+     * @type {number}
+     * @memberof Segment
+     */
+    charEnd: number;
+}
+
+
+/**
+ * @export
+ */
+export const SegmentKindEnum = {
+    Page: 'page'
+} as const;
+export type SegmentKindEnum = typeof SegmentKindEnum[keyof typeof SegmentKindEnum];
+
+/**
+ * @export
+ */
+export const SegmentOriginEnum = {
+    Source: 'source',
+    Derived: 'derived'
+} as const;
+export type SegmentOriginEnum = typeof SegmentOriginEnum[keyof typeof SegmentOriginEnum];
+
+/**
+ * @export
+ */
+export const SegmentSplitRuleEnum = {
+    Paragraph: 'paragraph',
+    TableRow: 'table_row',
+    Line: 'line',
+    Fixed: 'fixed'
+} as const;
+export type SegmentSplitRuleEnum = typeof SegmentSplitRuleEnum[keyof typeof SegmentSplitRuleEnum];
+
+
+/**
+ * Check if a given object implements the Segment interface.
+ */
+export function instanceOfSegment(value: object): value is Segment {
+    if (!('seq' in value) || value['seq'] === undefined) return false;
+    if (!('label' in value) || value['label'] === undefined) return false;
+    if (!('origin' in value) || value['origin'] === undefined) return false;
+    if (!('charStart' in value) || value['charStart'] === undefined) return false;
+    if (!('charEnd' in value) || value['charEnd'] === undefined) return false;
+    return true;
+}
+
+export function SegmentFromJSON(json: any): Segment {
+    return SegmentFromJSONTyped(json, false);
+}
+
+export function SegmentFromJSONTyped(json: any, ignoreDiscriminator: boolean): Segment {
+    if (json == null) {
+        return json;
+    }
+    return {
+        
+        'seq': json['seq'],
+        'kind': json['kind'] == null ? undefined : json['kind'],
+        'label': json['label'],
+        'origin': json['origin'],
+        'splitRule': json['split_rule'] == null ? undefined : json['split_rule'],
+        'charStart': json['char_start'],
+        'charEnd': json['char_end'],
+    };
+}
+
+export function SegmentToJSON(json: any): Segment {
+    return SegmentToJSONTyped(json, false);
+}
+
+export function SegmentToJSONTyped(value?: Segment | null, ignoreDiscriminator: boolean = false): any {
+    if (value == null) {
+        return value;
+    }
+
+    return {
+        
+        'seq': value['seq'],
+        'kind': value['kind'],
+        'label': value['label'],
+        'origin': value['origin'],
+        'split_rule': value['splitRule'],
+        'char_start': value['charStart'],
+        'char_end': value['charEnd'],
+    };
+}
+

package/src/src/models/SnippetContext.ts

@@ -4,7 +4,7 @@
  * Client Data API
  * Client Cluster Data API - Manifest-based storage service
  *
- * The version of the OpenAPI document: 1.4.14
+ * The version of the OpenAPI document: 1.11.0
  * 
  *
  * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).

package/src/src/models/Transcription.ts

@@ -0,0 +1,133 @@
+/* tslint:disable */
+/* eslint-disable */
+/**
+ * Client Data API
+ * Client Cluster Data API - Manifest-based storage service
+ *
+ * The version of the OpenAPI document: 1.11.0
+ * 
+ *
+ * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
+ * https://openapi-generator.tech
+ * Do not edit the class manually.
+ */
+
+import { mapValues } from '../runtime';
+import type { TranscriptionWarnings } from './TranscriptionWarnings';
+import {
+    TranscriptionWarningsFromJSON,
+    TranscriptionWarningsFromJSONTyped,
+    TranscriptionWarningsToJSON,
+    TranscriptionWarningsToJSONTyped,
+} from './TranscriptionWarnings';
+import type { TranscriptionPage } from './TranscriptionPage';
+import {
+    TranscriptionPageFromJSON,
+    TranscriptionPageFromJSONTyped,
+    TranscriptionPageToJSON,
+    TranscriptionPageToJSONTyped,
+} from './TranscriptionPage';
+import type { TranscriptionUsage } from './TranscriptionUsage';
+import {
+    TranscriptionUsageFromJSON,
+    TranscriptionUsageFromJSONTyped,
+    TranscriptionUsageToJSON,
+    TranscriptionUsageToJSONTyped,
+} from './TranscriptionUsage';
+import type { TranscriptionDocumentInfo } from './TranscriptionDocumentInfo';
+import {
+    TranscriptionDocumentInfoFromJSON,
+    TranscriptionDocumentInfoFromJSONTyped,
+    TranscriptionDocumentInfoToJSON,
+    TranscriptionDocumentInfoToJSONTyped,
+} from './TranscriptionDocumentInfo';
+
+/**
+ * A transcribed document: its pages, what produced them, and what it cost.
+ * 
+ * Adopted verbatim from ocr-gateway's ``OcrResponse`` (D4). The *ingest* contract
+ * wraps this; it does not reshape it.
+ * @export
+ * @interface Transcription
+ */
+export interface Transcription {
+    /**
+     * The transcription, ordered by index. Read each page's 'origin' before citing it: 'derived' means the producer drew the boundary, not the document. Never empty - a document with zero pages is the absence of a result, not a degraded one, and must be reported as a failure rather than a 200.
+     * @type {Array<TranscriptionPage>}
+     * @memberof Transcription
+     */
+    pages: Array<TranscriptionPage>;
+    /**
+     * What produced the text: a model identifier, or the route name.
+     * @type {string}
+     * @memberof Transcription
+     */
+    model: string;
+    /**
+     * What the document turned out to be.
+     * @type {TranscriptionDocumentInfo}
+     * @memberof Transcription
+     */
+    document: TranscriptionDocumentInfo;
+    /**
+     * What the transcription cost, and whether a GPU was used.
+     * @type {TranscriptionUsage}
+     * @memberof Transcription
+     */
+    usageInfo: TranscriptionUsage;
+    /**
+     * 
+     * @type {TranscriptionWarnings}
+     * @memberof Transcription
+     */
+    warnings?: TranscriptionWarnings | null;
+}
+
+/**
+ * Check if a given object implements the Transcription interface.
+ */
+export function instanceOfTranscription(value: object): value is Transcription {
+    if (!('pages' in value) || value['pages'] === undefined) return false;
+    if (!('model' in value) || value['model'] === undefined) return false;
+    if (!('document' in value) || value['document'] === undefined) return false;
+    if (!('usageInfo' in value) || value['usageInfo'] === undefined) return false;
+    return true;
+}
+
+export function TranscriptionFromJSON(json: any): Transcription {
+    return TranscriptionFromJSONTyped(json, false);
+}
+
+export function TranscriptionFromJSONTyped(json: any, ignoreDiscriminator: boolean): Transcription {
+    if (json == null) {
+        return json;
+    }
+    return {
+        
+        'pages': ((json['pages'] as Array<any>).map(TranscriptionPageFromJSON)),
+        'model': json['model'],
+        'document': TranscriptionDocumentInfoFromJSON(json['document']),
+        'usageInfo': TranscriptionUsageFromJSON(json['usage_info']),
+        'warnings': json['warnings'] == null ? undefined : TranscriptionWarningsFromJSON(json['warnings']),
+    };
+}
+
+export function TranscriptionToJSON(json: any): Transcription {
+    return TranscriptionToJSONTyped(json, false);
+}
+
+export function TranscriptionToJSONTyped(value?: Transcription | null, ignoreDiscriminator: boolean = false): any {
+    if (value == null) {
+        return value;
+    }
+
+    return {
+        
+        'pages': ((value['pages'] as Array<any>).map(TranscriptionPageToJSON)),
+        'model': value['model'],
+        'document': TranscriptionDocumentInfoToJSON(value['document']),
+        'usage_info': TranscriptionUsageToJSON(value['usageInfo']),
+        'warnings': TranscriptionWarningsToJSON(value['warnings']),
+    };
+}
+

package/src/src/models/TranscriptionDocumentInfo.ts

@@ -0,0 +1,115 @@
+/* tslint:disable */
+/* eslint-disable */
+/**
+ * Client Data API
+ * Client Cluster Data API - Manifest-based storage service
+ *
+ * The version of the OpenAPI document: 1.11.0
+ * 
+ *
+ * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
+ * https://openapi-generator.tech
+ * Do not edit the class manually.
+ */
+
+import { mapValues } from '../runtime';
+/**
+ * What the document turned out to be, and how it was handled.
+ * @export
+ * @interface TranscriptionDocumentInfo
+ */
+export interface TranscriptionDocumentInfo {
+    /**
+     * Canonical extension, e.g. 'xlsx'.
+     * @type {string}
+     * @memberof TranscriptionDocumentInfo
+     */
+    format: string;
+    /**
+     * Detected MIME type, from the signature rather than the name.
+     * @type {string}
+     * @memberof TranscriptionDocumentInfo
+     */
+    mime: string;
+    /**
+     * Size of the source document as received.
+     * @type {number}
+     * @memberof TranscriptionDocumentInfo
+     */
+    sizeBytes: number;
+    /**
+     * Pages the document is accounted for in. On the OCR route this is the source document's own page count, so a gap left by a failed chunk stays visible; on the extraction routes it is the emitted count, derived pages included.
+     * @type {number}
+     * @memberof TranscriptionDocumentInfo
+     */
+    totalPages: number;
+    /**
+     * Which path the document took. Anything but 'ocr' never reached a GPU.
+     * @type {string}
+     * @memberof TranscriptionDocumentInfo
+     */
+    route: TranscriptionDocumentInfoRouteEnum;
+}
+
+
+/**
+ * @export
+ */
+export const TranscriptionDocumentInfoRouteEnum = {
+    Text: 'text',
+    Tabular: 'tabular',
+    Html: 'html',
+    Ocr: 'ocr'
+} as const;
+export type TranscriptionDocumentInfoRouteEnum = typeof TranscriptionDocumentInfoRouteEnum[keyof typeof TranscriptionDocumentInfoRouteEnum];
+
+
+/**
+ * Check if a given object implements the TranscriptionDocumentInfo interface.
+ */
+export function instanceOfTranscriptionDocumentInfo(value: object): value is TranscriptionDocumentInfo {
+    if (!('format' in value) || value['format'] === undefined) return false;
+    if (!('mime' in value) || value['mime'] === undefined) return false;
+    if (!('sizeBytes' in value) || value['sizeBytes'] === undefined) return false;
+    if (!('totalPages' in value) || value['totalPages'] === undefined) return false;
+    if (!('route' in value) || value['route'] === undefined) return false;
+    return true;
+}
+
+export function TranscriptionDocumentInfoFromJSON(json: any): TranscriptionDocumentInfo {
+    return TranscriptionDocumentInfoFromJSONTyped(json, false);
+}
+
+export function TranscriptionDocumentInfoFromJSONTyped(json: any, ignoreDiscriminator: boolean): TranscriptionDocumentInfo {
+    if (json == null) {
+        return json;
+    }
+    return {
+        
+        'format': json['format'],
+        'mime': json['mime'],
+        'sizeBytes': json['size_bytes'],
+        'totalPages': json['total_pages'],
+        'route': json['route'],
+    };
+}
+
+export function TranscriptionDocumentInfoToJSON(json: any): TranscriptionDocumentInfo {
+    return TranscriptionDocumentInfoToJSONTyped(json, false);
+}
+
+export function TranscriptionDocumentInfoToJSONTyped(value?: TranscriptionDocumentInfo | null, ignoreDiscriminator: boolean = false): any {
+    if (value == null) {
+        return value;
+    }
+
+    return {
+        
+        'format': value['format'],
+        'mime': value['mime'],
+        'size_bytes': value['sizeBytes'],
+        'total_pages': value['totalPages'],
+        'route': value['route'],
+    };
+}
+

package/src/src/models/TranscriptionPage.ts

@@ -0,0 +1,144 @@
+/* tslint:disable */
+/* eslint-disable */
+/**
+ * Client Data API
+ * Client Cluster Data API - Manifest-based storage service
+ *
+ * The version of the OpenAPI document: 1.11.0
+ * 
+ *
+ * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
+ * https://openapi-generator.tech
+ * Do not edit the class manually.
+ */
+
+import { mapValues } from '../runtime';
+import type { PageDimensions } from './PageDimensions';
+import {
+    PageDimensionsFromJSON,
+    PageDimensionsFromJSONTyped,
+    PageDimensionsToJSON,
+    PageDimensionsToJSONTyped,
+} from './PageDimensions';
+import type { FigureImage } from './FigureImage';
+import {
+    FigureImageFromJSON,
+    FigureImageFromJSONTyped,
+    FigureImageToJSON,
+    FigureImageToJSONTyped,
+} from './FigureImage';
+
+/**
+ * One page of transcribed content, and whether the document drew its edges.
+ * @export
+ * @interface TranscriptionPage
+ */
+export interface TranscriptionPage {
+    /**
+     * Zero-indexed page number. Pages keep their source index, so a page lost to a failed chunk leaves a visible gap rather than a silently shorter document.
+     * @type {number}
+     * @memberof TranscriptionPage
+     */
+    index: number;
+    /**
+     * The page's content. Consumers rebuild the document by joining these with a blank line; the join rule is part of the contract (see PAGE_JOIN).
+     * @type {string}
+     * @memberof TranscriptionPage
+     */
+    markdown: string;
+    /**
+     * 'source' if the document itself drew this boundary (a PDF page, a worksheet, a folio); 'derived' if the producer split a unit to keep it inside an indexing position budget.
+     * @type {string}
+     * @memberof TranscriptionPage
+     */
+    origin?: TranscriptionPageOriginEnum;
+    /**
+     * 
+     * @type {string}
+     * @memberof TranscriptionPage
+     */
+    splitRule?: TranscriptionPageSplitRuleEnum | null;
+    /**
+     * Figure crops with bounding boxes. Always empty on extraction routes.
+     * @type {Array<FigureImage>}
+     * @memberof TranscriptionPage
+     */
+    images?: Array<FigureImage>;
+    /**
+     * 
+     * @type {PageDimensions}
+     * @memberof TranscriptionPage
+     */
+    dimensions?: PageDimensions | null;
+}
+
+
+/**
+ * @export
+ */
+export const TranscriptionPageOriginEnum = {
+    Source: 'source',
+    Derived: 'derived'
+} as const;
+export type TranscriptionPageOriginEnum = typeof TranscriptionPageOriginEnum[keyof typeof TranscriptionPageOriginEnum];
+
+/**
+ * @export
+ */
+export const TranscriptionPageSplitRuleEnum = {
+    Paragraph: 'paragraph',
+    TableRow: 'table_row',
+    Line: 'line',
+    Fixed: 'fixed'
+} as const;
+export type TranscriptionPageSplitRuleEnum = typeof TranscriptionPageSplitRuleEnum[keyof typeof TranscriptionPageSplitRuleEnum];
+
+
+/**
+ * Check if a given object implements the TranscriptionPage interface.
+ */
+export function instanceOfTranscriptionPage(value: object): value is TranscriptionPage {
+    if (!('index' in value) || value['index'] === undefined) return false;
+    if (!('markdown' in value) || value['markdown'] === undefined) return false;
+    return true;
+}
+
+export function TranscriptionPageFromJSON(json: any): TranscriptionPage {
+    return TranscriptionPageFromJSONTyped(json, false);
+}
+
+export function TranscriptionPageFromJSONTyped(json: any, ignoreDiscriminator: boolean): TranscriptionPage {
+    if (json == null) {
+        return json;
+    }
+    return {
+        
+        'index': json['index'],
+        'markdown': json['markdown'],
+        'origin': json['origin'] == null ? undefined : json['origin'],
+        'splitRule': json['split_rule'] == null ? undefined : json['split_rule'],
+        'images': json['images'] == null ? undefined : ((json['images'] as Array<any>).map(FigureImageFromJSON)),
+        'dimensions': json['dimensions'] == null ? undefined : PageDimensionsFromJSON(json['dimensions']),
+    };
+}
+
+export function TranscriptionPageToJSON(json: any): TranscriptionPage {
+    return TranscriptionPageToJSONTyped(json, false);
+}
+
+export function TranscriptionPageToJSONTyped(value?: TranscriptionPage | null, ignoreDiscriminator: boolean = false): any {
+    if (value == null) {
+        return value;
+    }
+
+    return {
+        
+        'index': value['index'],
+        'markdown': value['markdown'],
+        'origin': value['origin'],
+        'split_rule': value['splitRule'],
+        'images': value['images'] == null ? undefined : ((value['images'] as Array<any>).map(FigureImageToJSON)),
+        'dimensions': PageDimensionsToJSON(value['dimensions']),
+    };
+}
+

package/src/src/models/TranscriptionUsage.ts

@@ -0,0 +1,100 @@
+/* tslint:disable */
+/* eslint-disable */
+/**
+ * Client Data API
+ * Client Cluster Data API - Manifest-based storage service
+ *
+ * The version of the OpenAPI document: 1.11.0
+ * 
+ *
+ * NOTE: This class is auto generated by OpenAPI Generator (https://openapi-generator.tech).
+ * https://openapi-generator.tech
+ * Do not edit the class manually.
+ */
+
+import { mapValues } from '../runtime';
+/**
+ * What the transcription cost. ``gpu_used`` is the field that matters.
+ * @export
+ * @interface TranscriptionUsage
+ */
+export interface TranscriptionUsage {
+    /**
+     * Pages transcribed and returned.
+     * @type {number}
+     * @memberof TranscriptionUsage
+     */
+    pagesProcessed: number;
+    /**
+     * Pages that did not survive. Every one of them is named in warnings.missing_page_ranges, and processed + failed == document.total_pages.
+     * @type {number}
+     * @memberof TranscriptionUsage
+     */
+    pagesFailed?: number;
+    /**
+     * Figures actually delivered in pages[].images, not a producer's own tally.
+     * @type {number}
+     * @memberof TranscriptionUsage
+     */
+    figuresExtracted?: number;
+    /**
+     * The caller's wall clock. Chunks run concurrently, so this is not their sum.
+     * @type {number}
+     * @memberof TranscriptionUsage
+     */
+    processingTimeS: number;
+    /**
+     * False means this document never reached a GPU fleet.
+     * @type {boolean}
+     * @memberof TranscriptionUsage
+     */
+    gpuUsed: boolean;
+}
+
+/**
+ * Check if a given object implements the TranscriptionUsage interface.
+ */
+export function instanceOfTranscriptionUsage(value: object): value is TranscriptionUsage {
+    if (!('pagesProcessed' in value) || value['pagesProcessed'] === undefined) return false;
+    if (!('processingTimeS' in value) || value['processingTimeS'] === undefined) return false;
+    if (!('gpuUsed' in value) || value['gpuUsed'] === undefined) return false;
+    return true;
+}
+
+export function TranscriptionUsageFromJSON(json: any): TranscriptionUsage {
+    return TranscriptionUsageFromJSONTyped(json, false);
+}
+
+export function TranscriptionUsageFromJSONTyped(json: any, ignoreDiscriminator: boolean): TranscriptionUsage {
+    if (json == null) {
+        return json;
+    }
+    return {
+        
+        'pagesProcessed': json['pages_processed'],
+        'pagesFailed': json['pages_failed'] == null ? undefined : json['pages_failed'],
+        'figuresExtracted': json['figures_extracted'] == null ? undefined : json['figures_extracted'],
+        'processingTimeS': json['processing_time_s'],
+        'gpuUsed': json['gpu_used'],
+    };
+}
+
+export function TranscriptionUsageToJSON(json: any): TranscriptionUsage {
+    return TranscriptionUsageToJSONTyped(json, false);
+}
+
+export function TranscriptionUsageToJSONTyped(value?: TranscriptionUsage | null, ignoreDiscriminator: boolean = false): any {
+    if (value == null) {
+        return value;
+    }
+
+    return {
+        
+        'pages_processed': value['pagesProcessed'],
+        'pages_failed': value['pagesFailed'],
+        'figures_extracted': value['figuresExtracted'],
+        'processing_time_s': value['processingTimeS'],
+        'gpu_used': value['gpuUsed'],
+    };
+}
+