@coding01/docsjs 0.1.3 → 0.1.6

package/dist/index.d.cts

@@ -1,4 +1,4 @@
-export { D as DocsWordEditorChangeDetail, a as DocsWordEditorElementApi, b as DocsWordEditorErrorDetail, c as DocsWordEditorReadyDetail } from './types-DF14w1ol.cjs';
+export { D as DocsWordEditorChangeDetail, a as DocsWordEditorElementApi, b as DocsWordEditorErrorDetail, c as DocsWordEditorReadyDetail, d as DocxParseFeatureCounts, e as DocxParseReport, p as parseDocxToHtmlSnapshot, f as parseDocxToHtmlSnapshotWithReport } from './types-VvdwVF0_.cjs';
 
 declare class DocsWordElement extends HTMLElement {
     private rootRef;
@@ -48,6 +48,9 @@ interface SemanticStats {
     imageCount: number;
     anchorImageCount: number;
     wrappedImageCount: number;
+    ommlCount: number;
+    chartCount: number;
+    smartArtCount: number;
     listParagraphCount: number;
     commentRefCount: number;
     revisionInsCount: number;

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-export { D as DocsWordEditorChangeDetail, a as DocsWordEditorElementApi, b as DocsWordEditorErrorDetail, c as DocsWordEditorReadyDetail } from './types-DF14w1ol.js';
+export { D as DocsWordEditorChangeDetail, a as DocsWordEditorElementApi, b as DocsWordEditorErrorDetail, c as DocsWordEditorReadyDetail, d as DocxParseFeatureCounts, e as DocxParseReport, p as parseDocxToHtmlSnapshot, f as parseDocxToHtmlSnapshotWithReport } from './types-VvdwVF0_.js';
 
 declare class DocsWordElement extends HTMLElement {
     private rootRef;
@@ -48,6 +48,9 @@ interface SemanticStats {
     imageCount: number;
     anchorImageCount: number;
     wrappedImageCount: number;
+    ommlCount: number;
+    chartCount: number;
+    smartArtCount: number;
     listParagraphCount: number;
     commentRefCount: number;
     revisionInsCount: number;

package/dist/index.js

@@ -1,7 +1,9 @@
 import {
   DocsWordElement,
-  defineDocsWordElement
-} from "./chunk-PRPDJOB7.js";
+  defineDocsWordElement,
+  parseDocxToHtmlSnapshot,
+  parseDocxToHtmlSnapshotWithReport
+} from "./chunk-632UOG2B.js";
 
 // src/lib/semanticStats.ts
 function countElements(root, selector) {
@@ -25,6 +27,9 @@ function collectSemanticStatsFromDocument(doc) {
     imageCount: countElements(doc, "img"),
     anchorImageCount: countElements(doc, 'img[data-word-anchor="1"]'),
     wrappedImageCount: countElements(doc, "img[data-word-wrap]"),
+    ommlCount: countElements(doc, "[data-word-omml]"),
+    chartCount: countElements(doc, "[data-word-chart]"),
+    smartArtCount: countElements(doc, "[data-word-smartart]"),
     listParagraphCount,
     commentRefCount: countElements(doc, "[data-word-comment-ref]"),
     revisionInsCount: countElements(doc, '[data-word-revision="ins"]'),
@@ -55,7 +60,7 @@ function clamp01(v) {
 }
 function calculateFidelityScore(expected, actual) {
   const structure = clamp01(
-    (ratioScore(actual.paragraphCount, expected.paragraphCount) + ratioScore(actual.headingCount, expected.headingCount) + ratioScore(actual.tableCount, expected.tableCount) + ratioScore(actual.tableCellCount, expected.tableCellCount) + ratioScore(actual.imageCount, expected.imageCount) + ratioScore(actual.listParagraphCount, expected.listParagraphCount)) / 6
+    (ratioScore(actual.paragraphCount, expected.paragraphCount) + ratioScore(actual.headingCount, expected.headingCount) + ratioScore(actual.tableCount, expected.tableCount) + ratioScore(actual.tableCellCount, expected.tableCellCount) + ratioScore(actual.imageCount, expected.imageCount) + ratioScore(actual.ommlCount, expected.ommlCount) + ratioScore(actual.chartCount, expected.chartCount) + ratioScore(actual.smartArtCount, expected.smartArtCount) + ratioScore(actual.listParagraphCount, expected.listParagraphCount)) / 9
   );
   const styleProxy = clamp01(ratioScore(actual.textCharCount, expected.textCharCount));
   const pagination = clamp01(ratioScore(actual.pageSpacerCount, expected.pageSpacerCount));
@@ -67,6 +72,8 @@ export {
   calculateFidelityScore,
   collectSemanticStatsFromDocument,
   collectSemanticStatsFromHtml,
-  defineDocsWordElement
+  defineDocsWordElement,
+  parseDocxToHtmlSnapshot,
+  parseDocxToHtmlSnapshotWithReport
 };
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/lib/semanticStats.ts","../src/lib/fidelityScore.ts"],"sourcesContent":["export interface SemanticStats {\n  paragraphCount: number;\n  headingCount: number;\n  tableCount: number;\n  tableCellCount: number;\n  imageCount: number;\n  anchorImageCount: number;\n  wrappedImageCount: number;\n  listParagraphCount: number;\n  commentRefCount: number;\n  revisionInsCount: number;\n  revisionDelCount: number;\n  pageBreakCount: number;\n  pageSpacerCount: number;\n  textCharCount: number;\n}\n\nfunction countElements(root: ParentNode, selector: string): number {\n  return root.querySelectorAll(selector).length;\n}\n\nfunction isListLikeParagraph(p: Element): boolean {\n  if (p.hasAttribute(\"data-word-list\")) return true;\n  if (p.querySelector(\"span.__word-list-marker\")) return true;\n  const style = (p.getAttribute(\"style\") ?? \"\").toLowerCase();\n  return style.includes(\"mso-list\");\n}\n\nexport function collectSemanticStatsFromDocument(doc: Document): SemanticStats {\n  const paragraphs = Array.from(doc.querySelectorAll(\"p\"));\n  const listParagraphCount = paragraphs.filter((p) => isListLikeParagraph(p)).length;\n  const textCharCount = (doc.body.textContent ?? \"\").replace(/\\s+/g, \"\").length;\n\n  return {\n    paragraphCount: paragraphs.length,\n    headingCount: countElements(doc, \"h1,h2,h3,h4,h5,h6\"),\n    tableCount: countElements(doc, \"table\"),\n    tableCellCount: countElements(doc, \"td,th\"),\n    imageCount: countElements(doc, \"img\"),\n    anchorImageCount: countElements(doc, 'img[data-word-anchor=\"1\"]'),\n    wrappedImageCount: countElements(doc, \"img[data-word-wrap]\"),\n    listParagraphCount,\n    commentRefCount: countElements(doc, \"[data-word-comment-ref]\"),\n    revisionInsCount: countElements(doc, '[data-word-revision=\"ins\"]'),\n    revisionDelCount: countElements(doc, '[data-word-revision=\"del\"]'),\n    pageBreakCount: countElements(doc, \"[data-word-page-break='1']\"),\n    pageSpacerCount: countElements(doc, \"[data-word-page-spacer='1']\"),\n    textCharCount\n  };\n}\n\nexport function collectSemanticStatsFromHtml(rawHtml: string): SemanticStats {\n  const parser = new DOMParser();\n  const doc = parser.parseFromString(rawHtml, \"text/html\");\n  return collectSemanticStatsFromDocument(doc);\n}\n","import type { SemanticStats } from \"./semanticStats\";\n\nexport interface FidelityScore {\n  structure: number;\n  styleProxy: number;\n  pagination: number;\n  overall: number;\n}\n\nfunction ratioScore(actual: number, expected: number): number {\n  if (expected <= 0 && actual <= 0) return 1;\n  if (expected <= 0 || actual < 0) return 0;\n  const delta = Math.abs(actual - expected);\n  const penalty = delta / expected;\n  return Math.max(0, 1 - penalty);\n}\n\nfunction clamp01(v: number): number {\n  if (v < 0) return 0;\n  if (v > 1) return 1;\n  return v;\n}\n\nexport function calculateFidelityScore(expected: SemanticStats, actual: SemanticStats): FidelityScore {\n  const structure = clamp01(\n    (\n      ratioScore(actual.paragraphCount, expected.paragraphCount) +\n      ratioScore(actual.headingCount, expected.headingCount) +\n      ratioScore(actual.tableCount, expected.tableCount) +\n      ratioScore(actual.tableCellCount, expected.tableCellCount) +\n      ratioScore(actual.imageCount, expected.imageCount) +\n      ratioScore(actual.listParagraphCount, expected.listParagraphCount)\n    ) / 6\n  );\n\n  const styleProxy = clamp01(ratioScore(actual.textCharCount, expected.textCharCount));\n  const pagination = clamp01(ratioScore(actual.pageSpacerCount, expected.pageSpacerCount));\n  const overall = clamp01(structure * 0.6 + styleProxy * 0.25 + pagination * 0.15);\n\n  return { structure, styleProxy, pagination, overall };\n}\n"],"mappings":";;;;;;AAiBA,SAAS,cAAc,MAAkB,UAA0B;AACjE,SAAO,KAAK,iBAAiB,QAAQ,EAAE;AACzC;AAEA,SAAS,oBAAoB,GAAqB;AAChD,MAAI,EAAE,aAAa,gBAAgB,EAAG,QAAO;AAC7C,MAAI,EAAE,cAAc,yBAAyB,EAAG,QAAO;AACvD,QAAM,SAAS,EAAE,aAAa,OAAO,KAAK,IAAI,YAAY;AAC1D,SAAO,MAAM,SAAS,UAAU;AAClC;AAEO,SAAS,iCAAiC,KAA8B;AAC7E,QAAM,aAAa,MAAM,KAAK,IAAI,iBAAiB,GAAG,CAAC;AACvD,QAAM,qBAAqB,WAAW,OAAO,CAAC,MAAM,oBAAoB,CAAC,CAAC,EAAE;AAC5E,QAAM,iBAAiB,IAAI,KAAK,eAAe,IAAI,QAAQ,QAAQ,EAAE,EAAE;AAEvE,SAAO;AAAA,IACL,gBAAgB,WAAW;AAAA,IAC3B,cAAc,cAAc,KAAK,mBAAmB;AAAA,IACpD,YAAY,cAAc,KAAK,OAAO;AAAA,IACtC,gBAAgB,cAAc,KAAK,OAAO;AAAA,IAC1C,YAAY,cAAc,KAAK,KAAK;AAAA,IACpC,kBAAkB,cAAc,KAAK,2BAA2B;AAAA,IAChE,mBAAmB,cAAc,KAAK,qBAAqB;AAAA,IAC3D;AAAA,IACA,iBAAiB,cAAc,KAAK,yBAAyB;AAAA,IAC7D,kBAAkB,cAAc,KAAK,4BAA4B;AAAA,IACjE,kBAAkB,cAAc,KAAK,4BAA4B;AAAA,IACjE,gBAAgB,cAAc,KAAK,4BAA4B;AAAA,IAC/D,iBAAiB,cAAc,KAAK,6BAA6B;AAAA,IACjE;AAAA,EACF;AACF;AAEO,SAAS,6BAA6B,SAAgC;AAC3E,QAAM,SAAS,IAAI,UAAU;AAC7B,QAAM,MAAM,OAAO,gBAAgB,SAAS,WAAW;AACvD,SAAO,iCAAiC,GAAG;AAC7C;;;AC9CA,SAAS,WAAW,QAAgB,UAA0B;AAC5D,MAAI,YAAY,KAAK,UAAU,EAAG,QAAO;AACzC,MAAI,YAAY,KAAK,SAAS,EAAG,QAAO;AACxC,QAAM,QAAQ,KAAK,IAAI,SAAS,QAAQ;AACxC,QAAM,UAAU,QAAQ;AACxB,SAAO,KAAK,IAAI,GAAG,IAAI,OAAO;AAChC;AAEA,SAAS,QAAQ,GAAmB;AAClC,MAAI,IAAI,EAAG,QAAO;AAClB,MAAI,IAAI,EAAG,QAAO;AAClB,SAAO;AACT;AAEO,SAAS,uBAAuB,UAAyB,QAAsC;AACpG,QAAM,YAAY;AAAA,KAEd,WAAW,OAAO,gBAAgB,SAAS,cAAc,IACzD,WAAW,OAAO,cAAc,SAAS,YAAY,IACrD,WAAW,OAAO,YAAY,SAAS,UAAU,IACjD,WAAW,OAAO,gBAAgB,SAAS,cAAc,IACzD,WAAW,OAAO,YAAY,SAAS,UAAU,IACjD,WAAW,OAAO,oBAAoB,SAAS,kBAAkB,KAC/D;AAAA,EACN;AAEA,QAAM,aAAa,QAAQ,WAAW,OAAO,eAAe,SAAS,aAAa,CAAC;AACnF,QAAM,aAAa,QAAQ,WAAW,OAAO,iBAAiB,SAAS,eAAe,CAAC;AACvF,QAAM,UAAU,QAAQ,YAAY,MAAM,aAAa,OAAO,aAAa,IAAI;AAE/E,SAAO,EAAE,WAAW,YAAY,YAAY,QAAQ;AACtD;","names":[]}
+{"version":3,"sources":["../src/lib/semanticStats.ts","../src/lib/fidelityScore.ts"],"sourcesContent":["export interface SemanticStats {\n  paragraphCount: number;\n  headingCount: number;\n  tableCount: number;\n  tableCellCount: number;\n  imageCount: number;\n  anchorImageCount: number;\n  wrappedImageCount: number;\n  ommlCount: number;\n  chartCount: number;\n  smartArtCount: number;\n  listParagraphCount: number;\n  commentRefCount: number;\n  revisionInsCount: number;\n  revisionDelCount: number;\n  pageBreakCount: number;\n  pageSpacerCount: number;\n  textCharCount: number;\n}\n\nfunction countElements(root: ParentNode, selector: string): number {\n  return root.querySelectorAll(selector).length;\n}\n\nfunction isListLikeParagraph(p: Element): boolean {\n  if (p.hasAttribute(\"data-word-list\")) return true;\n  if (p.querySelector(\"span.__word-list-marker\")) return true;\n  const style = (p.getAttribute(\"style\") ?? \"\").toLowerCase();\n  return style.includes(\"mso-list\");\n}\n\nexport function collectSemanticStatsFromDocument(doc: Document): SemanticStats {\n  const paragraphs = Array.from(doc.querySelectorAll(\"p\"));\n  const listParagraphCount = paragraphs.filter((p) => isListLikeParagraph(p)).length;\n  const textCharCount = (doc.body.textContent ?? \"\").replace(/\\s+/g, \"\").length;\n\n  return {\n    paragraphCount: paragraphs.length,\n    headingCount: countElements(doc, \"h1,h2,h3,h4,h5,h6\"),\n    tableCount: countElements(doc, \"table\"),\n    tableCellCount: countElements(doc, \"td,th\"),\n    imageCount: countElements(doc, \"img\"),\n    anchorImageCount: countElements(doc, 'img[data-word-anchor=\"1\"]'),\n    wrappedImageCount: countElements(doc, \"img[data-word-wrap]\"),\n    ommlCount: countElements(doc, \"[data-word-omml]\"),\n    chartCount: countElements(doc, \"[data-word-chart]\"),\n    smartArtCount: countElements(doc, \"[data-word-smartart]\"),\n    listParagraphCount,\n    commentRefCount: countElements(doc, \"[data-word-comment-ref]\"),\n    revisionInsCount: countElements(doc, '[data-word-revision=\"ins\"]'),\n    revisionDelCount: countElements(doc, '[data-word-revision=\"del\"]'),\n    pageBreakCount: countElements(doc, \"[data-word-page-break='1']\"),\n    pageSpacerCount: countElements(doc, \"[data-word-page-spacer='1']\"),\n    textCharCount\n  };\n}\n\nexport function collectSemanticStatsFromHtml(rawHtml: string): SemanticStats {\n  const parser = new DOMParser();\n  const doc = parser.parseFromString(rawHtml, \"text/html\");\n  return collectSemanticStatsFromDocument(doc);\n}\n","import type { SemanticStats } from \"./semanticStats\";\n\nexport interface FidelityScore {\n  structure: number;\n  styleProxy: number;\n  pagination: number;\n  overall: number;\n}\n\nfunction ratioScore(actual: number, expected: number): number {\n  if (expected <= 0 && actual <= 0) return 1;\n  if (expected <= 0 || actual < 0) return 0;\n  const delta = Math.abs(actual - expected);\n  const penalty = delta / expected;\n  return Math.max(0, 1 - penalty);\n}\n\nfunction clamp01(v: number): number {\n  if (v < 0) return 0;\n  if (v > 1) return 1;\n  return v;\n}\n\nexport function calculateFidelityScore(expected: SemanticStats, actual: SemanticStats): FidelityScore {\n  const structure = clamp01(\n    (\n      ratioScore(actual.paragraphCount, expected.paragraphCount) +\n      ratioScore(actual.headingCount, expected.headingCount) +\n      ratioScore(actual.tableCount, expected.tableCount) +\n      ratioScore(actual.tableCellCount, expected.tableCellCount) +\n      ratioScore(actual.imageCount, expected.imageCount) +\n      ratioScore(actual.ommlCount, expected.ommlCount) +\n      ratioScore(actual.chartCount, expected.chartCount) +\n      ratioScore(actual.smartArtCount, expected.smartArtCount) +\n      ratioScore(actual.listParagraphCount, expected.listParagraphCount)\n    ) / 9\n  );\n\n  const styleProxy = clamp01(ratioScore(actual.textCharCount, expected.textCharCount));\n  const pagination = clamp01(ratioScore(actual.pageSpacerCount, expected.pageSpacerCount));\n  const overall = clamp01(structure * 0.6 + styleProxy * 0.25 + pagination * 0.15);\n\n  return { structure, styleProxy, pagination, overall };\n}\n"],"mappings":";;;;;;;;AAoBA,SAAS,cAAc,MAAkB,UAA0B;AACjE,SAAO,KAAK,iBAAiB,QAAQ,EAAE;AACzC;AAEA,SAAS,oBAAoB,GAAqB;AAChD,MAAI,EAAE,aAAa,gBAAgB,EAAG,QAAO;AAC7C,MAAI,EAAE,cAAc,yBAAyB,EAAG,QAAO;AACvD,QAAM,SAAS,EAAE,aAAa,OAAO,KAAK,IAAI,YAAY;AAC1D,SAAO,MAAM,SAAS,UAAU;AAClC;AAEO,SAAS,iCAAiC,KAA8B;AAC7E,QAAM,aAAa,MAAM,KAAK,IAAI,iBAAiB,GAAG,CAAC;AACvD,QAAM,qBAAqB,WAAW,OAAO,CAAC,MAAM,oBAAoB,CAAC,CAAC,EAAE;AAC5E,QAAM,iBAAiB,IAAI,KAAK,eAAe,IAAI,QAAQ,QAAQ,EAAE,EAAE;AAEvE,SAAO;AAAA,IACL,gBAAgB,WAAW;AAAA,IAC3B,cAAc,cAAc,KAAK,mBAAmB;AAAA,IACpD,YAAY,cAAc,KAAK,OAAO;AAAA,IACtC,gBAAgB,cAAc,KAAK,OAAO;AAAA,IAC1C,YAAY,cAAc,KAAK,KAAK;AAAA,IACpC,kBAAkB,cAAc,KAAK,2BAA2B;AAAA,IAChE,mBAAmB,cAAc,KAAK,qBAAqB;AAAA,IAC3D,WAAW,cAAc,KAAK,kBAAkB;AAAA,IAChD,YAAY,cAAc,KAAK,mBAAmB;AAAA,IAClD,eAAe,cAAc,KAAK,sBAAsB;AAAA,IACxD;AAAA,IACA,iBAAiB,cAAc,KAAK,yBAAyB;AAAA,IAC7D,kBAAkB,cAAc,KAAK,4BAA4B;AAAA,IACjE,kBAAkB,cAAc,KAAK,4BAA4B;AAAA,IACjE,gBAAgB,cAAc,KAAK,4BAA4B;AAAA,IAC/D,iBAAiB,cAAc,KAAK,6BAA6B;AAAA,IACjE;AAAA,EACF;AACF;AAEO,SAAS,6BAA6B,SAAgC;AAC3E,QAAM,SAAS,IAAI,UAAU;AAC7B,QAAM,MAAM,OAAO,gBAAgB,SAAS,WAAW;AACvD,SAAO,iCAAiC,GAAG;AAC7C;;;ACpDA,SAAS,WAAW,QAAgB,UAA0B;AAC5D,MAAI,YAAY,KAAK,UAAU,EAAG,QAAO;AACzC,MAAI,YAAY,KAAK,SAAS,EAAG,QAAO;AACxC,QAAM,QAAQ,KAAK,IAAI,SAAS,QAAQ;AACxC,QAAM,UAAU,QAAQ;AACxB,SAAO,KAAK,IAAI,GAAG,IAAI,OAAO;AAChC;AAEA,SAAS,QAAQ,GAAmB;AAClC,MAAI,IAAI,EAAG,QAAO;AAClB,MAAI,IAAI,EAAG,QAAO;AAClB,SAAO;AACT;AAEO,SAAS,uBAAuB,UAAyB,QAAsC;AACpG,QAAM,YAAY;AAAA,KAEd,WAAW,OAAO,gBAAgB,SAAS,cAAc,IACzD,WAAW,OAAO,cAAc,SAAS,YAAY,IACrD,WAAW,OAAO,YAAY,SAAS,UAAU,IACjD,WAAW,OAAO,gBAAgB,SAAS,cAAc,IACzD,WAAW,OAAO,YAAY,SAAS,UAAU,IACjD,WAAW,OAAO,WAAW,SAAS,SAAS,IAC/C,WAAW,OAAO,YAAY,SAAS,UAAU,IACjD,WAAW,OAAO,eAAe,SAAS,aAAa,IACvD,WAAW,OAAO,oBAAoB,SAAS,kBAAkB,KAC/D;AAAA,EACN;AAEA,QAAM,aAAa,QAAQ,WAAW,OAAO,eAAe,SAAS,aAAa,CAAC;AACnF,QAAM,aAAa,QAAQ,WAAW,OAAO,iBAAiB,SAAS,eAAe,CAAC;AACvF,QAAM,UAAU,QAAQ,YAAY,MAAM,aAAa,OAAO,aAAa,IAAI;AAE/E,SAAO,EAAE,WAAW,YAAY,YAAY,QAAQ;AACtD;","names":[]}