@storyteller-platform/align 0.1.20 → 0.1.22

package/dist/align/align.cjs

@@ -91,6 +91,7 @@ var import_segmentation = require("../markup/segmentation.cjs");
 var import_getSentenceRanges = require("./getSentenceRanges.cjs");
 var import_search = require("./search.cjs");
 var import_slugify = require("./slugify.cjs");
+var import_textFragments = require("./textFragments.cjs");
 async function align(input, output, transcriptionsDir, audiobookDir, options) {
   var _stack = [];
   try {
@@ -126,6 +127,7 @@ async function align(input, output, transcriptionsDir, audiobookDir, options) {
       audiobookFiles,
       transcriptions,
       options.granularity,
+      options.textRef,
       options.primaryLocale,
       options.logger
     );
@@ -149,7 +151,7 @@ async function align(input, output, transcriptionsDir, audiobookDir, options) {
   }
 }
 class Aligner {
-  constructor(epub, audiofiles, transcriptions, granularity, languageOverride, logger) {
+  constructor(epub, audiofiles, transcriptions, granularity, textRef, languageOverride, logger) {
     this.epub = epub;
     this.audiofiles = audiofiles;
     this.languageOverride = languageOverride;
@@ -157,12 +159,14 @@ class Aligner {
     this.transcription = concatTranscriptions(transcriptions, audiofiles);
     this.getChapterSentences = (0, import_memoize.default)(this.getChapterSentences.bind(this));
     this.granularity = granularity ?? "sentence";
+    this.textRef = textRef ?? "id-fragment";
   }
   transcription;
   totalDuration = 0;
   alignedChapters = [];
   timing = (0, import_ghost_story.createAggregator)();
   granularity;
+  textRef;
   report = {
     chapters: []
   };
@@ -177,8 +181,59 @@ class Aligner {
     return segmentation.filter((s) => s.text.match(/\S/));
   }
   async writeAlignedChapter(alignedChapter) {
+    const locale = this.languageOverride ?? await this.epub.getLanguage() ?? new Intl.Locale("en-US");
     const { chapter, sentenceRanges, wordRanges, xml } = alignedChapter;
+    const sentences = await this.getChapterSentences(chapter.id);
+    const sentenceIdToFragment = new Map(
+      sentenceRanges.map((range) => [
+        range.id,
+        `${range.chapterId}-s${range.id}`
+      ])
+    );
+    const wordIdToFragment = new Map(
+      wordRanges.map((ranges) => [
+        // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+        ranges[0].sentenceId,
+        new Map(
+          ranges.map((range) => [
+            range.id,
+            `${range.chapterId}-s${range.sentenceId}-w${range.id}`
+          ])
+        )
+      ])
+    );
     const wordRangeMap = new Map(wordRanges.map((w) => [w[0].sentenceId, w]));
+    if (this.textRef === "text-fragment") {
+      const trie = new import_textFragments.TextFragmentTrie(
+        sentences.map((s) => s.text.replace("\n", " ")),
+        locale
+      );
+      for (const range of sentenceRanges) {
+        const sentence = sentences[range.id];
+        sentenceIdToFragment.set(
+          range.id,
+          trie.findMinimalFragment(
+            // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+            range.id
+          )
+        );
+        if (this.granularity === "word") {
+          const wordRanges2 = wordRangeMap.get(range.id);
+          const toFragment = wordIdToFragment.get(range.id);
+          const words = sentence.words.entries.filter((w) => w.text.match(/\S/));
+          const wordTrie = new import_textFragments.TextFragmentTrie(
+            words.map((w) => w.text.replace("\n", " ")),
+            locale
+          );
+          for (const wordRange of wordRanges2) {
+            toFragment.set(
+              wordRange.id,
+              wordTrie.findMinimalFragment(wordRange.id)
+            );
+          }
+        }
+      }
+    }
     const audiofiles = Array.from(
       new Set(sentenceRanges.map(({ audiofile }) => audiofile))
     );
@@ -215,7 +270,9 @@ class Aligner {
         chapter,
         this.granularity,
         sentenceRanges,
-        wordRangeMap
+        wordRangeMap,
+        sentenceIdToFragment,
+        wordIdToFragment
       ),
       "xml"
     );
@@ -348,16 +405,24 @@ class Aligner {
     };
   }
   narrowToAvailableBoundary(boundary) {
-    const narrowed = { ...boundary };
-    for (const chapter of this.alignedChapters) {
-      if (chapter.startOffset > narrowed.start && chapter.startOffset <= narrowed.end) {
-        narrowed.end = chapter.startOffset - 1;
-      }
-      if (chapter.endOffset < narrowed.end && chapter.endOffset >= narrowed.start) {
-        narrowed.start = chapter.endOffset + 1;
+    const available = [
+      -1,
+      ...this.alignedChapters.toSorted((a, b) => a.startOffset - b.startOffset).flatMap(({ startOffset, endOffset }) => [startOffset, endOffset]),
+      Infinity
+    ];
+    const withinBoundary = [];
+    for (let i = 0; i < available.length - 1; i += 2) {
+      const [start, end] = [available[i], available[i + 1]];
+      if (boundary.start <= start && boundary.end >= start || boundary.start <= end && boundary.end >= end) {
+        withinBoundary.push([
+          Math.max(boundary.start, start + 1),
+          Math.min(boundary.end, end - 1)
+        ]);
       }
     }
-    return narrowed;
+    const largestBoundary = (0, import_itertools.max)(withinBoundary, ([start, end]) => end - start);
+    if (!largestBoundary) return { start: boundary.start, end: boundary.end };
+    return { start: largestBoundary[0], end: largestBoundary[1] };
   }
   async alignBook(onProgress) {
     const locale = this.languageOverride ?? await this.epub.getLanguage() ?? new Intl.Locale("en-US");
@@ -467,7 +532,7 @@ class Aligner {
         alignedChapter.wordRanges[i] = (0, import_getSentenceRanges.expandEmptySentenceRanges)(wordRanges);
       }
       await this.writeAlignedChapter(alignedChapter);
-      collapsedStart += sentences.length - 1;
+      collapsedStart += sentences.length;
     }
     await this.epub.addMetadata({
       type: "meta",
@@ -495,7 +560,7 @@ class Aligner {
     return this.timing;
   }
 }
-function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
+function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges, sentenceIdToFragment, wordIdToFragment) {
   return [
     import_epub.Epub.createXmlElement(
       "smil",
@@ -522,7 +587,7 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                   },
                   [
                     import_epub.Epub.createXmlElement("text", {
-                      src: `../${chapter.href}#${chapter.id}-s${sentenceRange.id}`
+                      src: `../${chapter.href}#${sentenceIdToFragment.get(sentenceRange.id)}`
                     }),
                     import_epub.Epub.createXmlElement("audio", {
                       src: `../Audio/${(0, import_posix.basename)(sentenceRange.audiofile)}`,
@@ -533,12 +598,13 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                 );
               }
               const words = wordRanges.get(sentenceRange.id);
+              const wordToFragment = wordIdToFragment.get(sentenceRange.id);
               return import_epub.Epub.createXmlElement(
                 "seq",
                 {
                   id: `${chapter.id}-s${sentenceRange.id}`,
                   "epub:type": "text-range-small",
-                  "epub:textref": `../${chapter.href}#${chapter.id}-s${sentenceRange.id}`
+                  "epub:textref": `../${chapter.href}#${sentenceIdToFragment.get(sentenceRange.id)}`
                 },
                 words.map(
                   (word) => import_epub.Epub.createXmlElement(
@@ -548,7 +614,7 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                     },
                     [
                       import_epub.Epub.createXmlElement("text", {
-                        src: `../${chapter.href}#${chapter.id}-s${sentenceRange.id}-w${word.id}`
+                        src: `../${chapter.href}#${wordToFragment.get(word.id)}`
                       }),
                       import_epub.Epub.createXmlElement("audio", {
                         src: `../Audio/${(0, import_posix.basename)(word.audiofile)}`,

package/dist/align/align.d.cts

@@ -39,7 +39,8 @@ interface Report {
 }
 interface AlignOptions {
     reportsPath?: string | null | undefined;
-    granularity: "sentence" | "word" | null | undefined;
+    granularity?: "sentence" | "word" | null | undefined;
+    textRef?: "id-fragment" | "text-fragment" | null | undefined;
     primaryLocale?: Intl.Locale | null | undefined;
     logger?: Logger | null | undefined;
     onProgress?: ((progress: number) => void) | null | undefined;
@@ -55,8 +56,9 @@ declare class Aligner {
     private alignedChapters;
     private timing;
     private granularity;
+    private textRef;
     report: Report;
-    constructor(epub: Epub, audiofiles: string[], transcriptions: Pick<RecognitionResult, "transcript" | "timeline">[], granularity: "sentence" | "word" | null | undefined, languageOverride?: (Intl.Locale | null) | undefined, logger?: (Logger | null) | undefined);
+    constructor(epub: Epub, audiofiles: string[], transcriptions: Pick<RecognitionResult, "transcript" | "timeline">[], granularity: "sentence" | "word" | null | undefined, textRef: "id-fragment" | "text-fragment" | null | undefined, languageOverride?: (Intl.Locale | null) | undefined, logger?: (Logger | null) | undefined);
     private getChapterSentences;
     private writeAlignedChapter;
     private addChapterReport;

package/dist/align/align.d.ts

@@ -39,7 +39,8 @@ interface Report {
 }
 interface AlignOptions {
     reportsPath?: string | null | undefined;
-    granularity: "sentence" | "word" | null | undefined;
+    granularity?: "sentence" | "word" | null | undefined;
+    textRef?: "id-fragment" | "text-fragment" | null | undefined;
     primaryLocale?: Intl.Locale | null | undefined;
     logger?: Logger | null | undefined;
     onProgress?: ((progress: number) => void) | null | undefined;
@@ -55,8 +56,9 @@ declare class Aligner {
     private alignedChapters;
     private timing;
     private granularity;
+    private textRef;
     report: Report;
-    constructor(epub: Epub, audiofiles: string[], transcriptions: Pick<RecognitionResult, "transcript" | "timeline">[], granularity: "sentence" | "word" | null | undefined, languageOverride?: (Intl.Locale | null) | undefined, logger?: (Logger | null) | undefined);
+    constructor(epub: Epub, audiofiles: string[], transcriptions: Pick<RecognitionResult, "transcript" | "timeline">[], granularity: "sentence" | "word" | null | undefined, textRef: "id-fragment" | "text-fragment" | null | undefined, languageOverride?: (Intl.Locale | null) | undefined, logger?: (Logger | null) | undefined);
     private getChapterSentences;
     private writeAlignedChapter;
     private addChapterReport;

package/dist/align/align.js

@@ -5,7 +5,7 @@ import {
 import { copyFile, mkdir, readFile, readdir, writeFile } from "node:fs/promises";
 import { dirname as autoDirname, join as autoJoin } from "node:path";
 import { basename, dirname, parse, relative } from "node:path/posix";
-import { enumerate } from "itertools";
+import { enumerate, max } from "itertools";
 import memoize from "memoize";
 import { isAudioFile, lookupAudioMime } from "@storyteller-platform/audiobook";
 import {
@@ -27,6 +27,7 @@ import {
 } from "./getSentenceRanges.js";
 import { findBoundaries } from "./search.js";
 import { slugify } from "./slugify.js";
+import { TextFragmentTrie } from "./textFragments.js";
 async function align(input, output, transcriptionsDir, audiobookDir, options) {
   var _stack = [];
   try {
@@ -62,6 +63,7 @@ async function align(input, output, transcriptionsDir, audiobookDir, options) {
       audiobookFiles,
       transcriptions,
       options.granularity,
+      options.textRef,
       options.primaryLocale,
       options.logger
     );
@@ -85,7 +87,7 @@ async function align(input, output, transcriptionsDir, audiobookDir, options) {
   }
 }
 class Aligner {
-  constructor(epub, audiofiles, transcriptions, granularity, languageOverride, logger) {
+  constructor(epub, audiofiles, transcriptions, granularity, textRef, languageOverride, logger) {
     this.epub = epub;
     this.audiofiles = audiofiles;
     this.languageOverride = languageOverride;
@@ -93,12 +95,14 @@ class Aligner {
     this.transcription = concatTranscriptions(transcriptions, audiofiles);
     this.getChapterSentences = memoize(this.getChapterSentences.bind(this));
     this.granularity = granularity ?? "sentence";
+    this.textRef = textRef ?? "id-fragment";
   }
   transcription;
   totalDuration = 0;
   alignedChapters = [];
   timing = createAggregator();
   granularity;
+  textRef;
   report = {
     chapters: []
   };
@@ -113,8 +117,59 @@ class Aligner {
     return segmentation.filter((s) => s.text.match(/\S/));
   }
   async writeAlignedChapter(alignedChapter) {
+    const locale = this.languageOverride ?? await this.epub.getLanguage() ?? new Intl.Locale("en-US");
     const { chapter, sentenceRanges, wordRanges, xml } = alignedChapter;
+    const sentences = await this.getChapterSentences(chapter.id);
+    const sentenceIdToFragment = new Map(
+      sentenceRanges.map((range) => [
+        range.id,
+        `${range.chapterId}-s${range.id}`
+      ])
+    );
+    const wordIdToFragment = new Map(
+      wordRanges.map((ranges) => [
+        // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+        ranges[0].sentenceId,
+        new Map(
+          ranges.map((range) => [
+            range.id,
+            `${range.chapterId}-s${range.sentenceId}-w${range.id}`
+          ])
+        )
+      ])
+    );
     const wordRangeMap = new Map(wordRanges.map((w) => [w[0].sentenceId, w]));
+    if (this.textRef === "text-fragment") {
+      const trie = new TextFragmentTrie(
+        sentences.map((s) => s.text.replace("\n", " ")),
+        locale
+      );
+      for (const range of sentenceRanges) {
+        const sentence = sentences[range.id];
+        sentenceIdToFragment.set(
+          range.id,
+          trie.findMinimalFragment(
+            // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+            range.id
+          )
+        );
+        if (this.granularity === "word") {
+          const wordRanges2 = wordRangeMap.get(range.id);
+          const toFragment = wordIdToFragment.get(range.id);
+          const words = sentence.words.entries.filter((w) => w.text.match(/\S/));
+          const wordTrie = new TextFragmentTrie(
+            words.map((w) => w.text.replace("\n", " ")),
+            locale
+          );
+          for (const wordRange of wordRanges2) {
+            toFragment.set(
+              wordRange.id,
+              wordTrie.findMinimalFragment(wordRange.id)
+            );
+          }
+        }
+      }
+    }
     const audiofiles = Array.from(
       new Set(sentenceRanges.map(({ audiofile }) => audiofile))
     );
@@ -151,7 +206,9 @@ class Aligner {
         chapter,
         this.granularity,
         sentenceRanges,
-        wordRangeMap
+        wordRangeMap,
+        sentenceIdToFragment,
+        wordIdToFragment
       ),
       "xml"
     );
@@ -284,16 +341,24 @@ class Aligner {
     };
   }
   narrowToAvailableBoundary(boundary) {
-    const narrowed = { ...boundary };
-    for (const chapter of this.alignedChapters) {
-      if (chapter.startOffset > narrowed.start && chapter.startOffset <= narrowed.end) {
-        narrowed.end = chapter.startOffset - 1;
-      }
-      if (chapter.endOffset < narrowed.end && chapter.endOffset >= narrowed.start) {
-        narrowed.start = chapter.endOffset + 1;
+    const available = [
+      -1,
+      ...this.alignedChapters.toSorted((a, b) => a.startOffset - b.startOffset).flatMap(({ startOffset, endOffset }) => [startOffset, endOffset]),
+      Infinity
+    ];
+    const withinBoundary = [];
+    for (let i = 0; i < available.length - 1; i += 2) {
+      const [start, end] = [available[i], available[i + 1]];
+      if (boundary.start <= start && boundary.end >= start || boundary.start <= end && boundary.end >= end) {
+        withinBoundary.push([
+          Math.max(boundary.start, start + 1),
+          Math.min(boundary.end, end - 1)
+        ]);
       }
     }
-    return narrowed;
+    const largestBoundary = max(withinBoundary, ([start, end]) => end - start);
+    if (!largestBoundary) return { start: boundary.start, end: boundary.end };
+    return { start: largestBoundary[0], end: largestBoundary[1] };
   }
   async alignBook(onProgress) {
     const locale = this.languageOverride ?? await this.epub.getLanguage() ?? new Intl.Locale("en-US");
@@ -403,7 +468,7 @@ class Aligner {
         alignedChapter.wordRanges[i] = expandEmptySentenceRanges(wordRanges);
       }
       await this.writeAlignedChapter(alignedChapter);
-      collapsedStart += sentences.length - 1;
+      collapsedStart += sentences.length;
     }
     await this.epub.addMetadata({
       type: "meta",
@@ -431,7 +496,7 @@ class Aligner {
     return this.timing;
   }
 }
-function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
+function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges, sentenceIdToFragment, wordIdToFragment) {
   return [
     Epub.createXmlElement(
       "smil",
@@ -458,7 +523,7 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                   },
                   [
                     Epub.createXmlElement("text", {
-                      src: `../${chapter.href}#${chapter.id}-s${sentenceRange.id}`
+                      src: `../${chapter.href}#${sentenceIdToFragment.get(sentenceRange.id)}`
                     }),
                     Epub.createXmlElement("audio", {
                       src: `../Audio/${basename(sentenceRange.audiofile)}`,
@@ -469,12 +534,13 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                 );
               }
               const words = wordRanges.get(sentenceRange.id);
+              const wordToFragment = wordIdToFragment.get(sentenceRange.id);
               return Epub.createXmlElement(
                 "seq",
                 {
                   id: `${chapter.id}-s${sentenceRange.id}`,
                   "epub:type": "text-range-small",
-                  "epub:textref": `../${chapter.href}#${chapter.id}-s${sentenceRange.id}`
+                  "epub:textref": `../${chapter.href}#${sentenceIdToFragment.get(sentenceRange.id)}`
                 },
                 words.map(
                   (word) => Epub.createXmlElement(
@@ -484,7 +550,7 @@ function createMediaOverlay(chapter, granularity, sentenceRanges, wordRanges) {
                     },
                     [
                       Epub.createXmlElement("text", {
-                        src: `../${chapter.href}#${chapter.id}-s${sentenceRange.id}-w${word.id}`
+                        src: `../${chapter.href}#${wordToFragment.get(word.id)}`
                       }),
                       Epub.createXmlElement("audio", {
                         src: `../Audio/${basename(word.audiofile)}`,

package/dist/align/getSentenceRanges.cjs

@@ -275,6 +275,7 @@ async function getSentenceRanges(transcriptionText, mappedTimeline, sentences, c
             if (start2 && end2) {
               perSentenceWordRanges.push({
                 id: k,
+                chapterId,
                 sentenceId: j + chapterSentenceIndex + slice[0],
                 start: end2.audiofile === start2.audiofile ? start2.start : 0,
                 audiofile: end2.audiofile,

package/dist/align/getSentenceRanges.d.cts

@@ -19,6 +19,7 @@ type SentenceRange = {
 };
 type WordRange = {
     id: number;
+    chapterId: string;
     sentenceId: number;
     start: number;
     end: number;

package/dist/align/getSentenceRanges.d.ts

@@ -19,6 +19,7 @@ type SentenceRange = {
 };
 type WordRange = {
     id: number;
+    chapterId: string;
     sentenceId: number;
     start: number;
     end: number;

package/dist/align/getSentenceRanges.js

@@ -247,6 +247,7 @@ async function getSentenceRanges(transcriptionText, mappedTimeline, sentences, c
             if (start2 && end2) {
               perSentenceWordRanges.push({
                 id: k,
+                chapterId,
                 sentenceId: j + chapterSentenceIndex + slice[0],
                 start: end2.audiofile === start2.audiofile ? start2.start : 0,
                 audiofile: end2.audiofile,

package/dist/align/parse.cjs

@@ -34,6 +34,12 @@ const alignParser = (0, import_core.object)("Alignment", {
     "--epub",
     (0, import_valueparser.path)({ mustExist: true, type: "file", extensions: [".epub"] })
   ),
+  textRef: (0, import_core.withDefault)(
+    (0, import_core.option)("--text-ref", (0, import_core.choice)(["id-fragment", "text-fragment"]), {
+      description: import_core.message`Whether to use text fragments rather than element id fragments to identify text ranges in generated media overlays.`
+    }),
+    "id-fragment"
+  ),
   reports: (0, import_core.optional)((0, import_core.option)("--reports", (0, import_valueparser.path)({ type: "directory" })))
 });
 const alignCommand = (0, import_core.command)(

package/dist/align/parse.d.cts

@@ -3,10 +3,12 @@ import * as _optique_core from '@optique/core';
 declare const alignParser: _optique_core.Parser<"sync", {
     readonly audiobook: string;
     readonly epub: string;
+    readonly textRef: "id-fragment" | "text-fragment";
     readonly reports: string | undefined;
 }, {
     readonly audiobook: _optique_core.ValueParserResult<string> | undefined;
     readonly epub: _optique_core.ValueParserResult<string> | undefined;
+    readonly textRef: [_optique_core.ValueParserResult<"id-fragment" | "text-fragment"> | undefined] | undefined;
     readonly reports: [_optique_core.ValueParserResult<string> | undefined] | undefined;
 }>;
 declare const alignCommand: _optique_core.Parser<"sync", {
@@ -16,6 +18,7 @@ declare const alignCommand: _optique_core.Parser<"sync", {
 } & {
     readonly audiobook: string;
     readonly epub: string;
+    readonly textRef: "id-fragment" | "text-fragment";
     readonly reports: string | undefined;
 } & {
     readonly noProgress: boolean;

package/dist/align/parse.d.ts

@@ -3,10 +3,12 @@ import * as _optique_core from '@optique/core';
 declare const alignParser: _optique_core.Parser<"sync", {
     readonly audiobook: string;
     readonly epub: string;
+    readonly textRef: "id-fragment" | "text-fragment";
     readonly reports: string | undefined;
 }, {
     readonly audiobook: _optique_core.ValueParserResult<string> | undefined;
     readonly epub: _optique_core.ValueParserResult<string> | undefined;
+    readonly textRef: [_optique_core.ValueParserResult<"id-fragment" | "text-fragment"> | undefined] | undefined;
     readonly reports: [_optique_core.ValueParserResult<string> | undefined] | undefined;
 }>;
 declare const alignCommand: _optique_core.Parser<"sync", {
@@ -16,6 +18,7 @@ declare const alignCommand: _optique_core.Parser<"sync", {
 } & {
     readonly audiobook: string;
     readonly epub: string;
+    readonly textRef: "id-fragment" | "text-fragment";
     readonly reports: string | undefined;
 } & {
     readonly noProgress: boolean;

package/dist/align/parse.js

@@ -1,12 +1,14 @@
 import "../chunk-BIEQXUOY.js";
 import {
+  choice,
   command,
   constant,
   merge,
   message,
   object,
   option,
-  optional
+  optional,
+  withDefault
 } from "@optique/core";
 import { path } from "@optique/run/valueparser";
 import {
@@ -23,6 +25,12 @@ const alignParser = object("Alignment", {
     "--epub",
     path({ mustExist: true, type: "file", extensions: [".epub"] })
   ),
+  textRef: withDefault(
+    option("--text-ref", choice(["id-fragment", "text-fragment"]), {
+      description: message`Whether to use text fragments rather than element id fragments to identify text ranges in generated media overlays.`
+    }),
+    "id-fragment"
+  ),
   reports: optional(option("--reports", path({ type: "directory" })))
 });
 const alignCommand = command(