@storyteller-platform/align 0.1.27 → 0.1.29

package/dist/align/textFragments.cjs

@@ -18,130 +18,91 @@ var __copyProps = (to, from, except, desc) => {
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 var textFragments_exports = {};
 __export(textFragments_exports, {
-  TextFragmentTrie: () => TextFragmentTrie
+  TextFragmentFactory: () => TextFragmentFactory
 });
 module.exports = __toCommonJS(textFragments_exports);
 var import_itertools = require("itertools");
 var import_runes2 = require("runes2");
-class TextFragmentTrie {
-  root = new Node(null, "");
+class TextFragmentFactory {
+  runes;
   spans;
+  spanStarts;
+  charPositions;
   constructor(casedSpans, locale = new Intl.Locale("en-Latn-US")) {
     this.spans = casedSpans.map((span) => span.toLocaleLowerCase(locale));
-    for (const [i, span] of (0, import_itertools.enumerate)(this.spans)) {
-      const parents = [this.root];
-      for (const [j, char] of (0, import_itertools.enumerate)((0, import_runes2.runes)(span))) {
-        for (const [k, parent] of (0, import_itertools.enumerate)(parents)) {
-          const newNode = new Node(parent, char, { span: i, pos: j });
-          let node = parent.children.find((child) => child.eq(newNode));
-          if (!node) {
-            node = newNode;
-            parent.children.push(node);
-          } else {
-            node.indices.push({ span: i, pos: j });
-          }
-          parents[k] = node;
-        }
-        parents.push(this.root);
-      }
+    this.runes = (0, import_runes2.runes)(this.spans.join(""));
+    this.spanStarts = [];
+    let start = 0;
+    for (const span of this.spans) {
+      this.spanStarts.push(start);
+      start += span.length;
     }
-  }
-  findMinimalFragment(spanIndex) {
-    let node = this.root;
-    while (node.children.length) {
-      const candidates = node.children.filter(
-        (child2) => child2.indices.some(
-          ({ span: childSpanIndex }) => childSpanIndex === spanIndex
-        )
-      );
-      const child = (0, import_itertools.min)(
-        candidates,
-        // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-        (c) => c.indices.find((i) => i.span === spanIndex).pos
-      );
-      if (!child) {
-        return this.nodeToFragment(node, spanIndex, true);
-      }
-      if (child.indices.length === 1) {
-        return this.nodeToFragment(child, spanIndex);
-      }
-      node = child;
+    this.charPositions = /* @__PURE__ */ new Map();
+    for (const [i, char] of (0, import_itertools.enumerate)(this.runes)) {
+      const positions = this.charPositions.get(char) ?? [];
+      positions.push(i);
+      this.charPositions.set(char, positions);
     }
-    return this.nodeToFragment(node, spanIndex, true);
   }
-  nodeToFragment(node, spanIndex, findPrefix) {
+  findMinimalFragment(spanIndex) {
     const span = this.spans[spanIndex];
-    let fragment = ":~:text=";
-    let prefix = "";
-    if (findPrefix) {
-      const prev = this.spans[spanIndex - 1];
-      if (prev) {
-        const prefixes = node.indices.filter(({ span: s }) => s !== spanIndex).map(({ span: spanIndex2, pos }) => {
-          let startNode2 = node;
-          let startPos = pos;
-          while (startNode2.parent && startNode2.parent !== this.root) {
-            startPos -= startNode2.value.length;
-            startNode2 = startNode2.parent;
-          }
-          const prev2 = this.spans[spanIndex2 - 1];
-          const span2 = this.spans[spanIndex2];
-          return (prev2 ?? "") + span2.slice(0, startPos);
-        });
-        const reversedPrefixes = prefixes.map((p) => (0, import_runes2.runes)(p).toReversed());
-        for (const [i2, char] of (0, import_itertools.enumerate)((0, import_runes2.runes)(prev).toReversed())) {
-          prefix = char + prefix;
-          for (const [j, p] of (0, import_itertools.enumerate)([...reversedPrefixes.toReversed()])) {
-            if (p[i2] !== char) {
-              reversedPrefixes.splice(reversedPrefixes.length - 1 - j, 1);
-            }
-          }
-          if (reversedPrefixes.length === 0) {
-            break;
-          }
+    if (!span) throw new RangeError(`Span index ${spanIndex} out of bounds`);
+    const startPos = this.spanStarts[spanIndex];
+    const chars = (0, import_runes2.runes)(span);
+    const firstChar = chars[0];
+    const candidates = this.charPositions.get(firstChar)?.filter((pos) => pos < startPos) ?? [];
+    let i = 1;
+    while (i < chars.length && candidates.length) {
+      const toRemove = [];
+      const char = chars[i];
+      for (let j = 0; j < candidates.length; j++) {
+        const candidate = candidates[j];
+        if (this.runes[candidate + i] !== char) {
+          toRemove.push(j);
         }
       }
+      toRemove.toReversed().map((r) => candidates.splice(r, 1));
+      i++;
     }
-    if (prefix) {
-      fragment += `${encodeTextFragmentPart(prefix)}-,`;
-    }
-    let startNode = node;
-    let start = "";
-    while (startNode) {
-      start = startNode.value + start;
-      startNode = startNode.parent;
-    }
+    let fragment = "";
+    const start = chars.slice(0, i).join("");
     fragment += encodeTextFragmentPart(start);
-    const remainingSentence = span.slice(start.length + node.value.length);
+    const remainingSpan = span.slice(i);
     let end = "";
-    let i = remainingSentence.length - 1;
-    while (remainingSentence.indexOf(end) !== i + 1 && i >= node.value.length) {
-      end = remainingSentence.slice(i);
-      i--;
+    let e = remainingSpan.length - 1;
+    if (remainingSpan.at(-1) === "\n") e--;
+    while (remainingSpan.indexOf(end) !== e + 1 && e >= 0) {
+      end = remainingSpan.slice(e);
+      e--;
     }
     if (end) {
       fragment += `,${encodeTextFragmentPart(end)}`;
     }
-    return fragment;
+    if (candidates.length) {
+      let p = 1;
+      while (p < startPos) {
+        const toRemove = [];
+        const char = this.runes[startPos - p];
+        for (let j = 0; j < candidates.length; j++) {
+          const candidate = candidates[j];
+          if (this.runes[candidate - p] !== char) {
+            toRemove.push(j);
+          }
+        }
+        toRemove.toReversed().map((r) => candidates.splice(r, 1));
+        p++;
+        if (!candidates.length) break;
+      }
+      const prefix = this.runes.slice(startPos - p + 1, startPos).join("");
+      fragment = `${encodeTextFragmentPart(prefix)}-,${fragment}`;
+    }
+    return `:~:text=${fragment}`;
   }
 }
 function encodeTextFragmentPart(part) {
   return encodeURIComponent(part).replaceAll(/-/g, "%2d").replaceAll(/,/g, "%2c");
 }
-class Node {
-  constructor(parent, value, firstIndex) {
-    this.parent = parent;
-    this.value = value;
-    if (firstIndex !== void 0) {
-      this.indices.push(firstIndex);
-    }
-  }
-  children = [];
-  indices = [];
-  eq(other) {
-    return this.value === other.value;
-  }
-}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
-  TextFragmentTrie
+  TextFragmentFactory
 });

package/dist/align/textFragments.d.cts

@@ -1,23 +1,10 @@
-declare class TextFragmentTrie {
-    private root;
+declare class TextFragmentFactory {
+    private runes;
     private spans;
+    private spanStarts;
+    private charPositions;
     constructor(casedSpans: string[], locale?: Intl.Locale);
     findMinimalFragment(spanIndex: number): string;
-    nodeToFragment(node: Node, spanIndex: number, findPrefix?: boolean): string;
-}
-declare class Node {
-    parent: Node | null;
-    value: string;
-    children: Node[];
-    indices: {
-        span: number;
-        pos: number;
-    }[];
-    constructor(parent: Node | null, value: string, firstIndex?: {
-        span: number;
-        pos: number;
-    });
-    eq(other: Node): boolean;
 }
 
-export { TextFragmentTrie };
+export { TextFragmentFactory };

package/dist/align/textFragments.d.ts

@@ -1,23 +1,10 @@
-declare class TextFragmentTrie {
-    private root;
+declare class TextFragmentFactory {
+    private runes;
     private spans;
+    private spanStarts;
+    private charPositions;
     constructor(casedSpans: string[], locale?: Intl.Locale);
     findMinimalFragment(spanIndex: number): string;
-    nodeToFragment(node: Node, spanIndex: number, findPrefix?: boolean): string;
-}
-declare class Node {
-    parent: Node | null;
-    value: string;
-    children: Node[];
-    indices: {
-        span: number;
-        pos: number;
-    }[];
-    constructor(parent: Node | null, value: string, firstIndex?: {
-        span: number;
-        pos: number;
-    });
-    eq(other: Node): boolean;
 }
 
-export { TextFragmentTrie };
+export { TextFragmentFactory };

package/dist/align/textFragments.js

@@ -1,124 +1,85 @@
 import "../chunk-BIEQXUOY.js";
-import { enumerate, min } from "itertools";
+import { enumerate } from "itertools";
 import { runes } from "runes2";
-class TextFragmentTrie {
-  root = new Node(null, "");
+class TextFragmentFactory {
+  runes;
   spans;
+  spanStarts;
+  charPositions;
   constructor(casedSpans, locale = new Intl.Locale("en-Latn-US")) {
     this.spans = casedSpans.map((span) => span.toLocaleLowerCase(locale));
-    for (const [i, span] of enumerate(this.spans)) {
-      const parents = [this.root];
-      for (const [j, char] of enumerate(runes(span))) {
-        for (const [k, parent] of enumerate(parents)) {
-          const newNode = new Node(parent, char, { span: i, pos: j });
-          let node = parent.children.find((child) => child.eq(newNode));
-          if (!node) {
-            node = newNode;
-            parent.children.push(node);
-          } else {
-            node.indices.push({ span: i, pos: j });
-          }
-          parents[k] = node;
-        }
-        parents.push(this.root);
-      }
+    this.runes = runes(this.spans.join(""));
+    this.spanStarts = [];
+    let start = 0;
+    for (const span of this.spans) {
+      this.spanStarts.push(start);
+      start += span.length;
     }
-  }
-  findMinimalFragment(spanIndex) {
-    let node = this.root;
-    while (node.children.length) {
-      const candidates = node.children.filter(
-        (child2) => child2.indices.some(
-          ({ span: childSpanIndex }) => childSpanIndex === spanIndex
-        )
-      );
-      const child = min(
-        candidates,
-        // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-        (c) => c.indices.find((i) => i.span === spanIndex).pos
-      );
-      if (!child) {
-        return this.nodeToFragment(node, spanIndex, true);
-      }
-      if (child.indices.length === 1) {
-        return this.nodeToFragment(child, spanIndex);
-      }
-      node = child;
+    this.charPositions = /* @__PURE__ */ new Map();
+    for (const [i, char] of enumerate(this.runes)) {
+      const positions = this.charPositions.get(char) ?? [];
+      positions.push(i);
+      this.charPositions.set(char, positions);
     }
-    return this.nodeToFragment(node, spanIndex, true);
   }
-  nodeToFragment(node, spanIndex, findPrefix) {
+  findMinimalFragment(spanIndex) {
     const span = this.spans[spanIndex];
-    let fragment = ":~:text=";
-    let prefix = "";
-    if (findPrefix) {
-      const prev = this.spans[spanIndex - 1];
-      if (prev) {
-        const prefixes = node.indices.filter(({ span: s }) => s !== spanIndex).map(({ span: spanIndex2, pos }) => {
-          let startNode2 = node;
-          let startPos = pos;
-          while (startNode2.parent && startNode2.parent !== this.root) {
-            startPos -= startNode2.value.length;
-            startNode2 = startNode2.parent;
-          }
-          const prev2 = this.spans[spanIndex2 - 1];
-          const span2 = this.spans[spanIndex2];
-          return (prev2 ?? "") + span2.slice(0, startPos);
-        });
-        const reversedPrefixes = prefixes.map((p) => runes(p).toReversed());
-        for (const [i2, char] of enumerate(runes(prev).toReversed())) {
-          prefix = char + prefix;
-          for (const [j, p] of enumerate([...reversedPrefixes.toReversed()])) {
-            if (p[i2] !== char) {
-              reversedPrefixes.splice(reversedPrefixes.length - 1 - j, 1);
-            }
-          }
-          if (reversedPrefixes.length === 0) {
-            break;
-          }
+    if (!span) throw new RangeError(`Span index ${spanIndex} out of bounds`);
+    const startPos = this.spanStarts[spanIndex];
+    const chars = runes(span);
+    const firstChar = chars[0];
+    const candidates = this.charPositions.get(firstChar)?.filter((pos) => pos < startPos) ?? [];
+    let i = 1;
+    while (i < chars.length && candidates.length) {
+      const toRemove = [];
+      const char = chars[i];
+      for (let j = 0; j < candidates.length; j++) {
+        const candidate = candidates[j];
+        if (this.runes[candidate + i] !== char) {
+          toRemove.push(j);
         }
       }
+      toRemove.toReversed().map((r) => candidates.splice(r, 1));
+      i++;
     }
-    if (prefix) {
-      fragment += `${encodeTextFragmentPart(prefix)}-,`;
-    }
-    let startNode = node;
-    let start = "";
-    while (startNode) {
-      start = startNode.value + start;
-      startNode = startNode.parent;
-    }
+    let fragment = "";
+    const start = chars.slice(0, i).join("");
     fragment += encodeTextFragmentPart(start);
-    const remainingSentence = span.slice(start.length + node.value.length);
+    const remainingSpan = span.slice(i);
     let end = "";
-    let i = remainingSentence.length - 1;
-    while (remainingSentence.indexOf(end) !== i + 1 && i >= node.value.length) {
-      end = remainingSentence.slice(i);
-      i--;
+    let e = remainingSpan.length - 1;
+    if (remainingSpan.at(-1) === "\n") e--;
+    while (remainingSpan.indexOf(end) !== e + 1 && e >= 0) {
+      end = remainingSpan.slice(e);
+      e--;
     }
     if (end) {
       fragment += `,${encodeTextFragmentPart(end)}`;
     }
-    return fragment;
+    if (candidates.length) {
+      let p = 1;
+      while (p < startPos) {
+        const toRemove = [];
+        const char = this.runes[startPos - p];
+        for (let j = 0; j < candidates.length; j++) {
+          const candidate = candidates[j];
+          if (this.runes[candidate - p] !== char) {
+            toRemove.push(j);
+          }
+        }
+        toRemove.toReversed().map((r) => candidates.splice(r, 1));
+        p++;
+        if (!candidates.length) break;
+      }
+      const prefix = this.runes.slice(startPos - p + 1, startPos).join("");
+      fragment = `${encodeTextFragmentPart(prefix)}-,${fragment}`;
+    }
+    return `:~:text=${fragment}`;
   }
 }
 function encodeTextFragmentPart(part) {
   return encodeURIComponent(part).replaceAll(/-/g, "%2d").replaceAll(/,/g, "%2c");
 }
-class Node {
-  constructor(parent, value, firstIndex) {
-    this.parent = parent;
-    this.value = value;
-    if (firstIndex !== void 0) {
-      this.indices.push(firstIndex);
-    }
-  }
-  children = [];
-  indices = [];
-  eq(other) {
-    return this.value === other.value;
-  }
-}
 export {
-  TextFragmentTrie
+  TextFragmentFactory
 };

package/dist/cli/bin.cjs

@@ -227,6 +227,7 @@ async function main() {
           {
             granularity: parsed.granularity,
             textRef: parsed.textRef,
+            outFormat: parsed.outFormat,
             primaryLocale: parsed.language,
             logger,
             ...!parsed.noProgress && parsed.logLevel === "silent" && {
@@ -346,6 +347,7 @@ async function main() {
             {
               granularity: parsed.granularity,
               textRef: parsed.textRef,
+              outFormat: parsed.outFormat,
               primaryLocale,
               logger,
               ...!parsed.noProgress && parsed.logLevel === "silent" && {

package/dist/cli/bin.js

@@ -178,6 +178,7 @@ async function main() {
           {
             granularity: parsed.granularity,
             textRef: parsed.textRef,
+            outFormat: parsed.outFormat,
             primaryLocale: parsed.language,
             logger,
             ...!parsed.noProgress && parsed.logLevel === "silent" && {
@@ -297,6 +298,7 @@ async function main() {
             {
               granularity: parsed.granularity,
               textRef: parsed.textRef,
+              outFormat: parsed.outFormat,
               primaryLocale,
               logger,
               ...!parsed.noProgress && parsed.logLevel === "silent" && {

package/dist/errorAlign/graphMetadata.cjs

@@ -34,7 +34,7 @@ module.exports = __toCommonJS(graphMetadata_exports);
 var import_itertools = require("itertools");
 var import_memoize = __toESM(require("memoize"), 1);
 var import_backtraceGraph = require("./backtraceGraph.cjs");
-var import_editDistance = require("./editDistance.cjs");
+var import_native = require("./native.cjs");
 var import_utils = require("./utils.cjs");
 class SubgraphMetadata {
   constructor(refRaw, hypRaw, refTokenMatches, hypTokenMatches, refNorm, hypNorm) {
@@ -52,7 +52,7 @@ class SubgraphMetadata {
     this.hypCharTypes = getCharTypes(this.hyp);
     this.refIndexMap = createIndexMap(refTokenMatches);
     this.hypIndexMap = createIndexMap(hypTokenMatches);
-    const { backtraceMatrix } = (0, import_editDistance.computeErrorAlignDistanceMatrix)(
+    const { backtraceMatrix } = (0, import_native.computeErrorAlignDistanceMatrix)(
       this.ref,
       this.hyp,
       true

package/dist/errorAlign/graphMetadata.js

@@ -2,7 +2,7 @@ import "../chunk-BIEQXUOY.js";
 import { range } from "itertools";
 import memoize from "memoize";
 import { BacktraceGraph } from "./backtraceGraph.js";
-import { computeErrorAlignDistanceMatrix } from "./editDistance.js";
+import { computeErrorAlignDistanceMatrix } from "./native.js";
 import { END_DELIMITER, START_DELIMITER, categorizeChar } from "./utils.js";
 class SubgraphMetadata {
   constructor(refRaw, hypRaw, refTokenMatches, hypTokenMatches, refNorm, hypNorm) {

package/dist/markup/parseDom.cjs

@@ -87,7 +87,7 @@ function findFootnotePairs(root) {
   const pairs = /* @__PURE__ */ new Map();
   (0, import_model.descendants)(root, (node, pos) => {
     if (node instanceof import_model.NoterefNode) {
-      const id = node.attrs["href"]?.slice(1);
+      const id = node.attrs["href"]?.split("#")[1];
       if (id) {
         noterefs.set(id, pos);
       }

package/dist/markup/parseDom.js

@@ -72,7 +72,7 @@ function findFootnotePairs(root) {
   const pairs = /* @__PURE__ */ new Map();
   descendants(root, (node, pos) => {
     if (node instanceof NoterefNode) {
-      const id = node.attrs["href"]?.slice(1);
+      const id = node.attrs["href"]?.split("#")[1];
       if (id) {
         noterefs.set(id, pos);
       }