@deepcitation/deepcitation-js 1.0.7 → 1.0.9

package/lib/client/DeepCitation.d.ts

@@ -1,5 +1,5 @@
-import type { Citation } from "../types/index";
-import type { CitationInput, ConvertFileInput, ConvertFileResponse, DeepCitationConfig, FileInput, PrepareConvertedFileOptions, PrepareFilesResult, UploadFileOptions, UploadFileResponse, VerifyCitationsFromLlmOutput, VerifyCitationsOptions, VerifyCitationsResponse } from "./types";
+import type { Citation } from "../types/index.js";
+import type { CitationInput, ConvertFileInput, ConvertFileResponse, DeepCitationConfig, FileInput, PrepareConvertedFileOptions, PrepareFilesResult, UploadFileOptions, UploadFileResponse, VerifyCitationsFromLlmOutput, VerifyCitationsOptions, VerifyCitationsResponse } from "./types.js";
 /**
  * DeepCitation client for file upload and citation verification.
  *

package/lib/client/DeepCitation.js

@@ -1,5 +1,5 @@
-import { getAllCitationsFromLlmOutput } from "../parsing/parseCitation";
-import { generateCitationKey } from "../react/utils";
+import { getAllCitationsFromLlmOutput } from "../parsing/parseCitation.js";
+import { generateCitationKey } from "../react/utils.js";
 const DEFAULT_API_URL = "https://api.deepcitation.com";
 /** Convert File/Blob/Buffer to a Blob suitable for FormData */
 function toBlob(file, filename) {

package/lib/client/index.d.ts

@@ -1,2 +1,2 @@
-export { DeepCitation } from "./DeepCitation";
-export type { DeepCitationConfig, UploadFileResponse, UploadFileOptions, VerifyCitationsResponse, VerifyCitationsOptions, CitationInput, FileInput, FileDataPart, PrepareFilesResult, VerifyCitationsFromLlmOutput, ConvertFileInput, ConvertFileResponse, PrepareConvertedFileOptions, } from "./types";
+export { DeepCitation } from "./DeepCitation.js";
+export type { DeepCitationConfig, UploadFileResponse, UploadFileOptions, VerifyCitationsResponse, VerifyCitationsOptions, CitationInput, FileInput, FileDataPart, PrepareFilesResult, VerifyCitationsFromLlmOutput, ConvertFileInput, ConvertFileResponse, PrepareConvertedFileOptions, } from "./types.js";

package/lib/client/index.js

@@ -1 +1 @@
-export { DeepCitation } from "./DeepCitation";
+export { DeepCitation } from "./DeepCitation.js";

package/lib/client/types.d.ts

@@ -1,4 +1,4 @@
-import type { Citation, FoundHighlightLocation } from "../types/index";
+import type { Citation, FoundHighlightLocation } from "../types/index.js";
 /**
  * Configuration options for the DeepCitation client
  */

package/lib/parsing/parseCitation.js

@@ -1,6 +1,50 @@
-import { sha1Hash } from "../utils/sha.js";
 import { normalizeCitations } from "./normalizeCitation.js";
 import { generateCitationKey } from "../react/utils.js";
+/**
+ * Parses a line_ids string that may contain individual numbers, ranges, or both.
+ * Examples: "1,2,3", "5-10", "1,5-7,10", "20-20"
+ *
+ * @param lineIdsString - The raw line_ids string (e.g., "1,5-7,10")
+ * @returns Sorted array of unique line IDs, or undefined if empty/invalid
+ */
+function parseLineIds(lineIdsString) {
+    if (!lineIdsString)
+        return undefined;
+    const lineIds = [];
+    const parts = lineIdsString.split(",");
+    for (const part of parts) {
+        const trimmed = part.trim();
+        if (!trimmed)
+            continue;
+        // Check if this part is a range (e.g., "5-10")
+        if (trimmed.includes("-")) {
+            const [startStr, endStr] = trimmed.split("-");
+            const start = parseInt(startStr, 10);
+            const end = parseInt(endStr, 10);
+            if (!isNaN(start) && !isNaN(end) && start <= end) {
+                // Expand the range
+                for (let i = start; i <= end; i++) {
+                    lineIds.push(i);
+                }
+            }
+            else if (!isNaN(start)) {
+                // If only start is valid, just use it
+                lineIds.push(start);
+            }
+        }
+        else {
+            // Single number
+            const num = parseInt(trimmed, 10);
+            if (!isNaN(num)) {
+                lineIds.push(num);
+            }
+        }
+    }
+    if (lineIds.length === 0)
+        return undefined;
+    // Sort and deduplicate
+    return [...new Set(lineIds)].sort((a, b) => a - b);
+}
 /**
  * Calculates the verification status of a citation based on the found highlight and search state.
  *
@@ -55,7 +99,6 @@ export const parseCitation = (fragment, mdAttachmentId, citationCounterRef, isVe
     const citationRegex = /<cite\s+file(?:_id|Id)='(\w{0,25})'\s+start_page[\_a-zA-Z]*='page[\_a-zA-Z]*(\d+)_index_(\d+)'\s+full_phrase='((?:[^'\\]|\\.)*)'\s+key_span='((?:[^'\\]|\\.)*)'\s+line(?:_ids|Ids)='([^']+)'(?:\s+(value|reasoning)='((?:[^'\\]|\\.)*)')?\s*\/>/g;
     const citationMatches = [...middleCite.matchAll(citationRegex)];
     const match = citationMatches?.[0];
-    const rawCitationMd = match?.[0];
     const pageNumber = match?.[2] ? parseInt(match?.[2]) : undefined;
     let fileId = match?.[1];
     let attachmentId = fileId?.length === 20 ? fileId : mdAttachmentId || match?.[1];
@@ -75,15 +118,9 @@ export const parseCitation = (fragment, mdAttachmentId, citationCounterRef, isVe
     }
     let lineIds;
     try {
-        // match[5] is line_ids
+        // match[6] is line_ids
         const lineIdsString = match?.[6]?.replace(/[A-Za-z_[\](){}:]/g, "");
-        lineIds = lineIdsString
-            ? lineIdsString
-                .split(",")
-                .map((id) => (isNaN(parseInt(id)) ? undefined : parseInt(id)))
-                .filter((id) => id !== undefined)
-                .sort((a, b) => a - b)
-            : undefined;
+        lineIds = lineIdsString ? parseLineIds(lineIdsString) : undefined;
     }
     catch (e) {
         if (isVerbose)
@@ -116,16 +153,13 @@ export const parseCitation = (fragment, mdAttachmentId, citationCounterRef, isVe
         }
         timestamps = { startTime, endTime };
     }
-    const fragmentContext = sha1Hash(fragment).toString().slice(0, 8);
     const citation = {
-        fragmentContext,
         fileId: attachmentId,
         pageNumber,
         fullPhrase,
         keySpan,
         citationNumber,
         lineIds,
-        rawCitationMd,
         beforeCite,
         value,
         timestamps,

package/lib/prompts/citationPrompts.d.ts

@@ -1,4 +1,4 @@
-export declare const CITATION_MARKDOWN_SYNTAX_PROMPT = "\nCitation syntax to use within Markdown:\n\u2022 To support any ideas or information that requires a citation from the provided content, use the following citation syntax:\n<cite file_id='file_id' start_page_key='page_number_PAGE_index_INDEX' full_phrase='the verbatim text of the terse phrase inside <file_text />; remember to escape quotes and newlines inside the full_phrase to remain as valid JSON' key_span='the verbatim value or words within full_phrase that best support the citation' line_ids='2-6' reasoning='the terse logic used to conclude the citation' />\n\n\u2022 Very important: for page numbers, only use the page number and page index info from the page_number_PAGE_index_INDEX format (e.g. <page_number_1_index_0>) and never from the contents inside the page.\n\u2022 start_page_key, full_phrase, and line_ids are required for each citation.\n\u2022 Infer line_ids, as we only provide the first, last, and every 5th line. When copying a previous <cite />, use the full info from the previous citation without changing the start_page_key, line_ids, or any other <cite /> attributes.\n\u2022 Use refer to line_ids inclusively, and use a range (or single) for each citation, split multiple sequential line_ids into multiple citations.\n\u2022 These citations will be replaced and displayed in-line as a numeric element (e.g. [1]), the markdown preceding <cite /> should read naturally with only one <cite /> per sentence with rare exceptions for two <cite /> in a sentence. <cite /> often present best at the end of the sentence, and are not grouped at the end of the document.\n\u2022 The full_phrase should be the exact verbatim text of the phrase or paragraph from the source document to support the insight or idea.\n\u2022 We do NOT put the full_phrase inside <cite ...></cite>; we only use full_phrase inside the full_phrase attribute.\n";
+export declare const CITATION_MARKDOWN_SYNTAX_PROMPT = "\nCitation syntax to use within Markdown:\n\u2022 To support any ideas or information that requires a citation from the provided content, use the following citation syntax:\n<cite file_id='file_id' start_page_key='page_number_PAGE_index_INDEX' full_phrase='the verbatim text of the terse phrase inside <file_text />; remember to escape quotes and newlines inside the full_phrase to remain as valid JSON' key_span='the verbatim value or 1-3 words within full_phrase that best support the citation' line_ids='2-6' reasoning='the terse logic used to conclude the citation' />\n\n\u2022 Very important: for page numbers, only use the page number and page index info from the page_number_PAGE_index_INDEX format (e.g. <page_number_1_index_0>) and never from the contents inside the page.\n\u2022 start_page_key, full_phrase, and line_ids are required for each citation.\n\u2022 Infer line_ids, as we only provide the first, last, and every 5th line. When copying a previous <cite />, use the full info from the previous citation without changing the start_page_key, line_ids, or any other <cite /> attributes.\n\u2022 Use refer to line_ids inclusively, and use a range (or single) for each citation, split multiple sequential line_ids into multiple citations.\n\u2022 These citations will be replaced and displayed in-line as a numeric element (e.g. [1]), the markdown preceding <cite /> should read naturally with only one <cite /> per sentence with rare exceptions for two <cite /> in a sentence. <cite /> often present best at the end of the sentence, and are not grouped at the end of the document.\n\u2022 The full_phrase should be the exact verbatim text of the phrase or paragraph from the source document to support the insight or idea.\n\u2022 We do NOT put the full_phrase inside <cite ...></cite>; we only use full_phrase inside the full_phrase attribute.\n";
 export declare const AV_CITATION_MARKDOWN_SYNTAX_PROMPT = "\n\u2022 To support any ideas or information that requires a citation from the provided content, use the following citation syntax:\n<cite file_id='file_id' full_phrase='the verbatim text of the phrase; remember to escape quotes and newlines inside the full_phrase to remain as valid JSON' timestamps='HH:MM:SS.SSS-HH:MM:SS.SSS' reasoning='the logic connecting the form section requirements to the supporting source citation' />\n\u2022 These citations are displayed in-line or in the relevant list item, and are not grouped at the end of the document.\n";
 export interface WrapSystemPromptOptions {
     /** The original system prompt to wrap with citation instructions */

package/lib/prompts/citationPrompts.js

@@ -1,7 +1,7 @@
 export const CITATION_MARKDOWN_SYNTAX_PROMPT = `
 Citation syntax to use within Markdown:
 • To support any ideas or information that requires a citation from the provided content, use the following citation syntax:
-<cite file_id='file_id' start_page_key='page_number_PAGE_index_INDEX' full_phrase='the verbatim text of the terse phrase inside <file_text />; remember to escape quotes and newlines inside the full_phrase to remain as valid JSON' key_span='the verbatim value or words within full_phrase that best support the citation' line_ids='2-6' reasoning='the terse logic used to conclude the citation' />
+<cite file_id='file_id' start_page_key='page_number_PAGE_index_INDEX' full_phrase='the verbatim text of the terse phrase inside <file_text />; remember to escape quotes and newlines inside the full_phrase to remain as valid JSON' key_span='the verbatim value or 1-3 words within full_phrase that best support the citation' line_ids='2-6' reasoning='the terse logic used to conclude the citation' />
 
 • Very important: for page numbers, only use the page number and page index info from the page_number_PAGE_index_INDEX format (e.g. <page_number_1_index_0>) and never from the contents inside the page.
 • start_page_key, full_phrase, and line_ids are required for each citation.
@@ -125,7 +125,7 @@ export const CITATION_JSON_OUTPUT_FORMAT = {
         },
         keySpan: {
             type: "string",
-            description: "the verbatim value or words within fullPhrase that best support the citation",
+            description: "the verbatim value or 1-3 words within fullPhrase that best support the citation",
         },
         lineIds: {
             type: "array",

package/lib/prompts/index.d.ts

@@ -1,3 +1,3 @@
-export * from "./promptCompression";
-export * from "./citationPrompts";
-export * from "./types";
+export * from "./promptCompression.js";
+export * from "./citationPrompts.js";
+export * from "./types.js";

package/lib/prompts/index.js

@@ -1,3 +1,3 @@
-export * from "./promptCompression";
-export * from "./citationPrompts";
-export * from "./types";
+export * from "./promptCompression.js";
+export * from "./citationPrompts.js";
+export * from "./types.js";

package/lib/prompts/promptCompression.d.ts

@@ -1,4 +1,4 @@
-import { CompressedResult } from "./types";
+import { CompressedResult } from "./types.js";
 /**
  * Compress all occurrences of `ids` inside `obj`, returning a new object
  * plus the `prefixMap` needed to decompress.

package/lib/react/utils.js

@@ -13,7 +13,6 @@ export function generateCitationKey(citation) {
         citation.lineIds?.join(",") || "",
         citation.timestamps?.startTime || "",
         citation.timestamps?.endTime || "",
-        citation.fragmentContext || "",
     ];
     return sha1Hash(keyParts.join("|")).slice(0, 16);
 }
@@ -31,7 +30,10 @@ export function generateCitationInstanceId(citationKey) {
 export function getCitationDisplayText(citation, options = {}) {
     const { displayCitationValue = false, fallbackDisplay } = options;
     if (displayCitationValue) {
-        return citation.value || citation.citationNumber?.toString() || fallbackDisplay || "";
+        return (citation.value ||
+            citation.citationNumber?.toString() ||
+            fallbackDisplay ||
+            "");
     }
     return citation.citationNumber?.toString() || "";
 }

package/lib/types/citation.d.ts

@@ -1,5 +1,5 @@
-import { type ScreenBox } from "./boxes";
-import { type FoundHighlightLocation } from "./foundHighlight";
+import { type ScreenBox } from "./boxes.js";
+import { type FoundHighlightLocation } from "./foundHighlight.js";
 export type OutputImageFormat = "jpeg" | "png" | "avif" | undefined | null;
 export declare const DEFAULT_OUTPUT_IMAGE_FORMAT: "avif";
 export interface VerifyCitationResponse {
@@ -30,8 +30,6 @@ export interface Citation {
         endTime?: string;
         startTime?: string;
     };
-    fragmentContext?: string | null;
-    rawCitationMd?: string;
     beforeCite?: string;
 }
 export interface CitationStatus {

package/lib/types/foundHighlight.d.ts

@@ -1,6 +1,6 @@
-import { type Citation } from "./citation";
-import { type SearchState } from "./search";
-import { type PdfSpaceItem } from "./boxes";
+import { type Citation } from "./citation.js";
+import { type SearchState } from "./search.js";
+import { type PdfSpaceItem } from "./boxes.js";
 export declare const NOT_FOUND_HIGHLIGHT_INDEX = -1;
 export declare const PENDING_HIGHLIGHT_INDEX = -2;
 export declare const BLANK_HIGHLIGHT_LOCATION: FoundHighlightLocation;

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@deepcitation/deepcitation-js",
-    "version": "1.0.7",
+    "version": "1.0.9",
     "description": "DeepCitation JavaScript SDK for deterministic AI citation verification",
     "type": "module",
     "private": false,