officeparser 7.0.1 → 7.0.3

package/dist/officeparser.browser.mjs

@@ -704,14 +704,18 @@ ${i}  </tr>
 `}return n+=`\\f0\\fs24
 `,n+=e,n+="}",{value:n,messages:this.messages}}async processNodeRecursive(e,n){let i=this.inTable;e.type==="table"&&(this.inTable=!0);let s=await super.processNodeRecursive(e,n);return this.inTable=i,s}async renderBody(e){let n="";this.inTable=!1;let i=async(s,a)=>{let l=this.getSemanticMapping(s);if(l){if(l.tag==="blockquote")return`${this.inTable?"\\pard\\intbl":"\\pard"}\\li720\\ri720\\sa120 ${a}\\par
 `;let c=l.tag.match(/^h([1-6])$/);if(c){let u=parseInt(c[1]),h=24+(6-u)*4;return`${this.inTable?"\\pard\\intbl":"\\pard"}\\s${u}\\sb240\\sa120{\\b\\fs${h} ${a}}\\par
-`}}switch(s.type){case"text":{let c=this.escapeRtf(s.text||""),u=s.formatting,h=s.metadata;if(this.config.includeFormatting&&u){let d="",f="";if(u.bold&&(d+="\\b ",f="\\b0 "+f),u.italic&&(d+="\\i ",f="\\i0 "+f),u.underline&&(d+="\\ul ",f="\\ul0 "+f),u.strikethrough&&(d+="\\strike ",f="\\strike0 "+f),u.color){let p=this.getColorIndex(u.color);d+=`\\cf${p+1} `}if(u.backgroundColor){let p=this.getColorIndex(u.backgroundColor);d+=`\\highlight${p+1} `}if(u.size){let p=parseInt(u.size);d+=`\\fs${p*2} `}c=`{\\rtlch\\f1 ${d}${c}${f}}`}if(h?.link){let d=h.linkType!=="external";if(!this.config.ignoreInternalLinks||!d)return`{\\field{\\*\\fldinst{HYPERLINK "${h.link}"}}{\\fldrslt ${c}}}`}return c}case"heading":{let u=s.metadata?.level||1,h=24+(6-u)*4;return`${this.inTable?"\\pard\\intbl":"\\pard"}\\s${u}\\sb240\\sa120{\\b\\fs${h} ${a}}\\par
-`}case"paragraph":{let c=this.inTable?"\\pard\\intbl":"\\pard";if(c+="\\sa120",this.config.includeFormatting&&s.metadata){let u=s.metadata;if(u.alignment&&(u.alignment==="center"?c+="\\qc":u.alignment==="right"?c+="\\qr":u.alignment==="justify"&&(c+="\\qj")),u.paragraphIndentation){let h=u.paragraphIndentation;h.left&&(c+=`\\li${h.left}`),h.right&&(c+=`\\ri${h.right}`),h.firstLine&&(c+=`\\fi${h.firstLine}`)}}return`${c} ${a}\\par
-`}case"list":{let c=s.metadata,u=((c?.indentation||0)+1)*360,h=c?.listType==="ordered",d=h?`${(c.itemIndex??0)+1}. `:"\\bullet ",f=h?"\\pndec":"\\pnbullet",p=c?.indentation||0;return`${this.inTable?"\\pard\\intbl":"\\pard"}\\li${u}\\fi-360\\ilvl${p}${f} ${d}${a}\\par
+`}}switch(s.type){case"text":{let c=this.escapeRtf(s.text||""),u=s.formatting,h=s.metadata;if(this.config.includeFormatting&&u){let d="",f="";if(u.bold&&(d+="\\b ",f="\\b0 "+f),u.italic&&(d+="\\i ",f="\\i0 "+f),u.underline&&(d+="\\ul ",f="\\ul0 "+f),u.strikethrough&&(d+="\\strike ",f="\\strike0 "+f),u.color){let p=this.getColorIndex(u.color);d+=`\\cf${p+1} `}if(u.backgroundColor){let p=this.getColorIndex(u.backgroundColor);d+=`\\highlight${p+1} `}if(u.size){let p=parseInt(u.size);d+=`\\fs${p*2} `}c=`{\\f0 ${d}${c}${f}}`}if(h?.link){let d=h.linkType!=="external";if(!this.config.ignoreInternalLinks||!d)return`{\\field{\\*\\fldinst{HYPERLINK "${h.link}"}}{\\fldrslt ${c}}}`}return c}case"heading":{let u=s.metadata?.level||1,h=24+(6-u)*4;return`${this.inTable?"\\pard\\intbl":"\\pard"}\\s${u}\\sb240\\sa120{\\b\\fs${h} ${a}}\\par
+`}case"paragraph":{let c=this.inTable?"\\pard\\intbl":"\\pard";if(c+="\\sa120",this.config.includeFormatting&&s.metadata){let u=s.metadata;u.alignment&&(u.alignment==="center"?c+="\\qc":u.alignment==="right"?c+="\\qr":u.alignment==="justify"&&(c+="\\qj"))}return`${c} ${a}\\par
+`}case"list":{let c=s.metadata,u=c?.indentation||0,h=(u+1)*360,d=c?.listType==="ordered",f=d?`${(c.itemIndex??0)+1}. `:"\\bullet ",p=d?"\\pndec":"\\pnbullet";return`${this.inTable?"\\pard\\intbl":"\\pard"}\\li${h}\\fi-360\\ilvl${u}${p} ${f}${a}\\par
 `}case"table":return`\\pard\\sa0
-${a}`;case"row":return`\\trowd\\trgaph108\\trleft-108
+${a}`;case"row":{let c=s.children||[],h=Math.floor(9e3/(c.length||1)),d="";for(let f=0;f<c.length;f++)d+=`\\clbrdrt\\brdrs\\brdrw10\\clbrdrl\\brdrs\\brdrw10\\clbrdrb\\brdrs\\brdrw10\\clbrdrr\\brdrs\\brdrw10\\cellx${(f+1)*h}`;return`\\trowd\\trgaph108\\trleft-108${d}
 ${a}\\row
-`;case"cell":return`${this.inTable?"\\pard\\intbl":"\\pard"}\\intbl\\sb60\\sa60 ${a}\\cell
-`;case"break":return s.metadata?.breakType==="page"?`\\page
+`}case"cell":return`\\pard\\intbl\\sb60\\sa60 ${a}\\cell
+`;case"image":{if(!this.config.includeImages)return"";let u=s.metadata?.attachmentName,h=this.ast.attachments.find(d=>d.name===u);if(h&&h.data){let d=h.extension==="png"?"pngblip":"jpegblip",f=atob(h.data),p="";for(let m=0;m<f.length;m++){let y=f.charCodeAt(m).toString(16);p+=y.length===1?"0"+y:y,m%64===63&&(p+=`
+`)}return`{\\pict\\${d}\\picwgoal4320\\pichgoal3240
+${p}
+}
+`}return""}case"break":return s.metadata?.breakType==="page"?`\\page
 `:`\\line
 `;default:return a}};for(let s of e.content)n+=await this.processNodeRecursive(s,i);return n}getColorIndex(e){let n=e.toUpperCase(),i=this.colorTable.indexOf(n);return i===-1&&(i=this.colorTable.length,this.colorTable.push(n)),i}escapeRtf(e){return e.replace(/\\/g,"\\\\").replace(/{/g,"\\{").replace(/}/g,"\\}").replace(/[^\x00-\x7F]/g,n=>`\\u${n.charCodeAt(0)}?`)}};B();var kg=class extends qn{constructor(t,e){super("text",t,e)}async generate(){let t="",e=this.config.textConfig.newlineDelimiter;this.config.renderMetadata&&this.ast.metadata&&(this.ast.metadata.title&&(t+=`Title: ${this.ast.metadata.title}${e}`),this.ast.metadata.author&&(t+=`Author: ${this.ast.metadata.author}${e}`),this.ast.metadata.created&&(t+=`Created: ${new Date(this.ast.metadata.created).toLocaleString()}${e}`),t+=`-------------------${e}${e}`);let n=async(i,s)=>{if(i.type==="text"||i.type==="code")return i.text||"";if(i.type==="break")return e;if(i.type==="image"){if(!this.config.includeImages)return"";let l=i.metadata;return`[Image: ${l?.altText||l?.attachmentName||"Untitled"}]${e}`}if(i.type==="table"&&this.config.textConfig.preserveLayout)return await this.renderTable(i,n,e);if(i.type==="list"&&this.config.textConfig.preserveLayout){let l=i.metadata,u=" ".repeat(4).repeat(l?.indentation||0),h=l?.listType==="ordered"?`${(l.itemIndex??0)+1}. `:"- ";return`${u}${h}${s.trimStart()}`+(s.endsWith(e)?"":e)}return["paragraph","heading","row","sheet","slide","note","list","table","code"].includes(i.type)?s.trim()===""?"":s+(s.endsWith(e)?"":e):s};for(let i of this.ast.content)t+=await this.processNodeRecursive(i,n);return{value:t.trim(),messages:this.messages}}async renderTable(t,e,n){if(!t.children||t.children.length===0)return"";let i=[],s=[];for(let l of t.children){let c=await this.handleOnNode(l);if(c===!1)continue;if(typeof c=="string"){i.push([c]);continue}let u=[];if(l.children)for(let h=0;h<l.children.length;h++){let d=l.children[h],f=(await this.processNodeRecursive(d,e)).trim().replace(/\r?\n/g," ");u.push(f),s[h]=Math.max(s[h]||0,f.length)}i.push(u)}let a=n;for(let l of i){a+="| ";for(let c=0;c<l.length;c++)a+=(l[c]||"").padEnd(s[c]||0)+" | ";a+=n}return a+n}};var Ka=class{static async generate(t,e,n){switch(e.toLowerCase()){case"text":return new kg(t,n).generate();case"md":return new Tg(t,n).generate();case"html":return new hu(t,n).generate();case"pdf":return new Ng(t,n).generate();case"csv":return new Eg(t,n).generate();case"rtf":return new Ig(t,n).generate();case"chunks":return new gg(t,n).generate();default:throw hn("EXTENSION_UNSUPPORTED",void 0,e)}}};function An(r,t,e,n,i,s){return{config:i,type:r,metadata:t,content:e,attachments:n,warnings:[],toText:s,async to(a,l){return Ka.generate(this,a,l)}}}var Gv=async(r,t)=>{let e=r.toString("utf-8"),n=t.csvDelimiter,i=[],s=[],a="",l=!1;for(let f=0;f<e.length;f++){let p=e[f],m=e[f+1];l?p==='"'?m==='"'?(a+='"',f++):l=!1:a+=p:p==='"'?l=!0:e.substring(f,f+n.length)===n?(s.push(a),a="",f+=n.length-1):p===`
 `?(s.push(a),i.push(s),s=[],a=""):p==="\r"||(a+=p)}(a!==""||s.length>0)&&(s.push(a),i.push(s)),i.length>0&&i[i.length-1].length===1&&i[i.length-1][0]===""&&i.pop();let c=[];i.forEach((f,p)=>{if(f.length===1&&f[0].startsWith("#")){c.push({type:"comment",text:f[0]});return}let m=[];f.forEach((y,x)=>{if(y&&y.trim()!==""){let b={row:p,col:x};m.push({type:"cell",text:y,metadata:b,children:[{type:"text",text:y}]})}}),m.length>0&&c.push({type:"row",children:m})});let h={type:"sheet",metadata:{sheetName:"Sheet1"},children:c,rawContent:t.includeRawContent?e:void 0};return An("csv",{title:"Sheet1"},[h],[],t,()=>i.map(f=>f.filter(p=>p.trim()!=="").join(t.newlineDelimiter)).join(t.newlineDelimiter).replace(/\n{3,}/g,`
@@ -749,7 +753,7 @@ ${x}
 `})))}}}return W.filter(V=>V.text?.trim()||V.children&&V.children.length>0)};function q(T,I,_){let W=[];for(let z of Array.from(T?.childNodes||[])){if(!Oe(z))continue;let D=z,F=D.tagName;if(F==="p:sp")W.push(...$(D,I,_));else if(F==="p:pic"){let P=M(D,I,_);P&&W.push(P)}else if(F==="p:graphicFrame"){let P=j(D,I,_);P&&W.push(P)}else if(F==="p:grpSp"){let P=pt(D,"p:spTree");P&&W.push(...q(P,I,_))}}return W}for(let T of h)if(T.path.match(i)){let I=T.path.match(/slide(\d+)\.xml\.rels/);if(I){let _=parseInt(I[1]);x[_]={};let W=le(T.content.toString()),z=ut(W,"Relationship");for(let D=0;D<z.length;D++){let F=z[D].getAttribute("Id"),P=z[D].getAttribute("Type"),G=z[D].getAttribute("Target");if(F&&P&&G){let H="other";P.includes("relationships/image")?H="image":P.includes("relationships/hyperlink")?H="hyperlink":P.includes("relationships/chart")?H="chart":P.includes("relationships/slide")?H="slide":P.includes("relationships/notesSlide")&&(H="notes");let V=G;G.startsWith("http://")||G.startsWith("https://")||(V=V.split("/").pop()||""),x[_][F]={type:H,target:V}}}}}for(let T of h){if(T.path.match(a)||T.path.match(l)||T.path.match(i)||T.path.match(c))continue;let I=T.content.toString(),_=le(I,{locator:t.includeRawContent});t.includeRawContent&&y.push(I);let W=T.path.match(s),z=W?parseInt(W[1]):0,D=T.path.includes("notesSlide"),F={type:D?"note":"slide",children:[],metadata:{slideNumber:z,...D?{noteId:`slide-note-${z}`}:{}}};t.includeRawContent&&(F.rawContent=ee(_,I,t));let P=pt(_,"p:spTree");P&&F.children?.push(...q(P,z,I)),F.children&&F.children.length>0&&m.push(F)}let Q=[],Z=h.filter(T=>T.path.match(/ppt\/media\/.*/)),Y=h.filter(T=>T.path.match(/ppt\/charts\/chart\d+\.xml/));if(t.extractAttachments){for(let I of Z){let _=Mr(I.path.split("/").pop()||"image",I.content);if(Q.push(_),t.ocr&&_.mimeType.startsWith("image/"))try{_.ocrText=(await lr(I.content,{...t.ocrConfig})).trim()}catch(W){Ce("OCR_FAILED",t,_.name,W)}}for(let I of Y){let _={type:"chart",mimeType:"application/vnd.openxmlformats-officedocument.wordprocessingml.document",data:I.content.toString("base64"),name:I.path.split("/").pop()||"",extension:"xml"};Q.push(_);try{let W=await dl(I.content);_.chartData=W}catch(W){Ce("CHART_DATA_EXTRACTION_FAILED",t,I.path,W)}}let T=I=>{for(let _ of I){if("attachmentName"in(_.metadata||{})){let W=_.metadata,z=Q.find(D=>D.name===W.attachmentName);z&&(_.type==="image"&&(z.altText=W.altText,z.ocrText&&(_.text=z.ocrText)),_.type==="chart"&&(_.text=z.chartData?.rawTexts.join(t.newlineDelimiter)))}_.children&&T(_.children)}};T(m)}return!t.ignoreNotes&&t.putNotesAtLast&&m.sort((T,I)=>{let _=T.type==="note"?1:0,W=I.type==="note"?1:0;return _-W}),An("pptx",f,m,Q,t,()=>m.map(T=>{let I=_=>{let W="";return _.children?W+=_.children.map(I).filter(z=>z!="").join(_.children[0]?.children?t.newlineDelimiter:""):W+=_.text||"",W};return I(T)}).filter(T=>T!="").join(t.newlineDelimiter))};B();var lF={pngblip:"png",jpegblip:"jpeg",gifblip:"gif",tiffblip:"tiff",dibitmap:"bmp",wbitmap:"bmp"},cF={png:"image/png",jpeg:"image/jpeg",gif:"image/gif",tiff:"image/tiff",bmp:"image/bmp"},k2=class{constructor(t){U(this,"index",0);U(this,"buffer");U(this,"codePage",1252);U(this,"decoders",{});U(this,"pendingBytes",[]);U(this,"length");this.buffer=t,this.length=t.length}parse(){let t={type:"group",content:[]},e=[t];for(;this.index<this.length;){let n=this.buffer[this.index],i=e[e.length-1];if(n===123){this.index++,this.flushPendingText(i);let s={type:"group",content:[]};i.content.push(s),e.push(s)}else n===125?(this.index++,this.flushPendingText(i),e.length>1&&e.pop()):n===92?(this.index++,this.parseControl(i)):n===13||n===10?this.index++:this.parseText(i)}return this.flushPendingText(t),t}parseControl(t){if(this.index>=this.length)return;let e=this.buffer[this.index];if(e===123||e===125||e===92){this.pendingBytes.push(e),this.index++;return}if(e===39){if(this.index++,this.index+1<this.length){let l=String.fromCharCode(this.buffer[this.index],this.buffer[this.index+1]),c=parseInt(l,16);isNaN(c)||this.pendingBytes.push(c),this.index+=2}return}if(this.flushPendingText(t),e===42){t.content.push({type:"control",value:"*"}),this.index++;return}let n="";for(;this.index<this.length;){let l=this.buffer[this.index];if(l>=97&&l<=122||l>=65&&l<=90)n+=String.fromCharCode(l),this.index++;else break}let i,s=!1,a="";for(this.index<this.length&&this.buffer[this.index]===45&&(a+="-",this.index++);this.index<this.length;){let l=this.buffer[this.index];if(l>=48&&l<=57)a+=String.fromCharCode(l),this.index++,s=!0;else break}if(s&&(i=parseInt(a,10)),n!==""&&this.index<this.length&&this.buffer[this.index]===32&&this.index++,n==="bin"&&i!==void 0&&i>0){this.index+=i;return}n==="ansicpg"&&i!==void 0?this.codePage=i:n==="ansi"?this.codePage=1252:n==="mac"?this.codePage=1e4:n==="pc"?this.codePage=437:n==="pca"&&(this.codePage=850),t.content.push({type:"control",value:n,param:i}),(t.content.length===1&&t.type==="group"||t.content.length===2&&t.content[0].type==="control"&&t.content[0].value==="*")&&(t.destination=n)}parseText(t){for(;this.index<this.length;){let e=this.buffer[this.index];if(e===void 0||e===123||e===125||e===92||e===13||e===10)break;this.pendingBytes.push(e),this.index++}}flushPendingText(t){this.pendingBytes.length>0&&(t.content.push({type:"text",value:this.decodeBytes(this.pendingBytes,this.codePage)}),this.pendingBytes=[])}decodeBytes(t,e){let n=new Uint8Array(t);if(t.some(s=>s>127))try{return new TextDecoder("utf-8",{fatal:!0}).decode(n)}catch{}if(!this.decoders[e]){let s=`windows-${e}`;e===1e4?s="macintosh":e===437?s="ibm437":e===850&&(s="ibm850");try{this.decoders[e]=new TextDecoder(s)}catch{if(e!==1252)try{this.decoders[e]=new TextDecoder("windows-1252")}catch{return String.fromCharCode(...t)}else return String.fromCharCode(...t)}}let i=this.decoders[e].decode(n);if(e===1252&&/[\u0080-\u009F]/.test(i)){let s={"\x80":"\u20AC","\x82":"\u201A","\x83":"\u0192","\x84":"\u201E","\x85":"\u2026","\x86":"\u2020","\x87":"\u2021","\x88":"\u02C6","\x89":"\u2030","\x8A":"\u0160","\x8B":"\u2039","\x8C":"\u0152","\x8E":"\u017D","\x91":"\u2018","\x92":"\u2019","\x93":"\u201C","\x94":"\u201D","\x95":"\u2022","\x96":"\u2013","\x97":"\u2014","\x98":"\u02DC","\x99":"\u2122","\x9A":"\u0161","\x9B":"\u203A","\x9C":"\u0153","\x9E":"\u017E","\x9F":"\u0178"};return i.replace(/[\u0080-\u009F]/g,a=>s[a]||a)}return i}},qD=async(r,t)=>{let n=new k2(r).parse(),i=uF(n),s=hF(n),a=[],l=[],c=[],u=[],h=[],d=[],f=[],p={},m=a,y=0,x="left",b=!1,w,C,E,S=[],N,k,M=[],R=0,j=0,$=()=>M.length>0?M[M.length-1]:void 0,q=()=>{M.length===0&&M.push({rows:[],currentCells:[],currentCellContent:[],rowIndex:0})},Q=!1,Z=!1,Y=0,X=[],T={isMergedContinuation:!1},I=0,_=0,W={},z=!1,D=!1,F,P,G={},H=!1,V,L,tt={},rt={},at,ft=(O,K)=>O.bold!==K.bold||O.italic!==K.italic||O.underline!==K.underline||O.strikethrough!==K.strikethrough||O.size!==K.size||O.font!==K.font||O.color!==K.color||O.backgroundColor!==K.backgroundColor||O.subscript!==K.subscript||O.superscript!==K.superscript,J=()=>{if(f.length>0){let O=f.join(""),K={type:"text",text:O,formatting:{...p}};at&&(K.metadata={link:at,linkType:Ft(at)}),h.push(K),u.push(O),f=[]}},st=!1,lt=()=>{J();let O=u.length>0||h.length>0;if(Q&&!Z&&!st&&O){let K=[...u],Tt=[...h],xt=[...d];u=[],h=[],d=[],vt(),u=K,h=Tt,d=xt}if(O){let K=u.join(""),Tt="paragraph",xt;if(C===void 0&&h.length>0){let Ot=h[0];if(Ot.type==="text"&&Ot.formatting?.bold){let Yt=parseInt(Ot.formatting.size||"12");Yt>=14&&K.length<300&&(Yt>=22?C=1:Yt>=18?C=2:Yt>=16?C=3:C=4)}}if(!b&&y>300){let Ot=K.trim();(/^[\u2022\u00b7\-\u25cf\u25cb]/.test(Ot)||/^\d+[.\)]/.test(Ot))&&(b=!0,w=/^\d+[.\)]/.test(Ot)?"ordered":"unordered")}if(C!==void 0&&C>0)Tt="heading",xt={level:C},N=void 0,k=void 0;else if(b){Tt="list";let Ot=E||N,Yt=w||k||"unordered",se=0;if(Ot){rt[Ot]||(rt[Ot]={});let _n=Object.keys(rt[Ot]).map($s=>parseInt($s));for(let $s of _n)$s>y&&delete rt[Ot][$s];rt[Ot][y]===void 0?rt[Ot][y]=0:rt[Ot][y]++,se=rt[Ot][y]}xt={listType:Yt,indentation:y,listId:Ot||"",itemIndex:se,alignment:x},Ot&&(N=Ot),Yt&&(k=Yt)}let Mt={type:Tt,text:K,children:h,formatting:void 0,metadata:{...xt,anchorIds:S.length>0?[...S]:void 0}};t.includeRawContent&&d.length>0&&(Mt.rawContent=d.join("")),Q&&Z?(q(),$().currentCellContent.push(Mt)):m.push(Mt),u=[],h=[],d=[],S=[]}},dt=O=>{lt();let K=O||$();if(!K)return;let Tt={type:"cell",text:K.currentCellContent.map(xt=>xt.text).join(`
 `),children:[...K.currentCellContent],metadata:{row:K.rowIndex,col:K.currentCells.length}};return K.currentCellContent=[],Tt},et=O=>{let K=O||$();if(!K)return;let Tt=dt(K);if(Tt&&(Tt.children&&Tt.children.length>0||Tt.text)&&K.currentCells.push(Tt),K.currentCells.length>0){let xt={type:"row",text:K.currentCells.map(Mt=>Mt.text).filter(Mt=>Mt!=="").join(t.newlineDelimiter),children:[...K.currentCells]};K.rows.push(xt),K.currentCells=[],K.rowIndex++}},vt=()=>{if(st)return;st=!0;let O=$();if(!O){st=!1;return}if(et(O),O.rows.length>0){Y++;let K={type:"table",text:O.rows.map(Tt=>Tt.text).join(`
 `),children:[...O.rows]};M.length>1?M[M.length-2].currentCellContent.push(K):m.push(K)}M.pop(),M.length===0&&(Q=!1,Z=!1),st=!1},bt=O=>{let K,Tt=xt=>{if(xt.type==="text"){let Mt=xt.value,Ot=Mt.match(/HYPERLINK\s+"([^"]+)"/i);if(Ot)return Ot[1];let Yt=Mt.match(/"(https?:\/\/[^"]+|mailto:[^"]+|#[^"]+)"/);if(Yt)return Yt[1]}else if(xt.type==="group"){let Mt=!1,Ot,Yt=!1;for(let se of xt.content)if(se.type==="text"){let _n=se.value.toUpperCase();_n.includes("HYPERLINK")&&(Mt=!0),_n.includes("\\L")&&(Yt=!0);let $s=se.value.match(/"([^"]+)"/);$s&&Mt&&(Ot=$s[1])}else if(se.type==="group"){let _n=Tt(se);_n&&(Ot=_n)}if(Ot)return Yt&&!Ot.startsWith("#")?"#"+Ot:Ot}};for(let xt of O.content)if(xt.type==="group"){let Mt=Tt(xt);if(Mt)return Mt}return K},Ft=O=>{let K=O.trim();return K.startsWith("#")||!/^[a-zA-Z][a-zA-Z0-9+.-]*:/.test(K)?"internal":"external"},Ct=O=>{let K="",Tt=xt=>{if(xt.type==="text")K+=xt.value;else if(xt.type==="group")for(let Mt of xt.content)Tt(Mt)};for(let xt of O.content)Tt(xt);return K},mt=O=>{let K,Tt=[];for(let Ot of O.content)if(Ot.type==="control"){let Yt=lF[Ot.value];Yt&&(K=Yt)}else Ot.type==="text"?Tt.push(Ot.value.replace(/[^0-9a-fA-F]/g,"")):Ot.type;let xt=Tt.join("");if(!K||xt.length===0)return;let Mt=cF[K];if(Mt)try{let Ot=ct.Buffer.from(xt,"hex"),Yt=K,se=`image_${c.length+1}.${Yt}`;return{type:"image",mimeType:Mt,data:Ot.toString("base64"),name:se,extension:Yt}}catch{return}},ot=O=>{if(!/^[a-zA-Z]/.test(O.value))return`\\${O.value}`;let K=`\\${O.value}`;return O.param!==void 0&&(K+=O.param),K+=" ",K},ht=O=>{let K="";for(let Tt of O.content)Tt.type==="text"?K+=Tt.value:Tt.type==="group"&&(K+=ht(Tt));return K.trim()},gt=O=>O.value.replace(/([\\{}])/g,"\\$1"),Et=(O,K,Tt=0)=>{if(O.type==="group"){let xt=["fonttbl","colortbl","stylesheet","info","macpict","pmmetafile","wmetafile","dibitmap","bitmap","object","nextGenerator","header","footer","nonshppict","xml","private","upnp","ud","filetbl","operator","author","creatim","revtim","printim","comment","fldinst","listtext","pntext"],Mt=!1,Ot=!1,Yt=!1,se=!1;if(O.destination){if(O.destination==="footnote")Ot=!0;else if(O.destination==="field"){let Ye=bt(O);Ye&&(J(),Yt=!0,at=Ye)}else if(O.destination==="listtable")z=!0;else if(z&&O.destination==="list")D=!0,F=void 0,P=void 0;else if(O.destination==="listoverridetable")H=!0;else if(H&&O.destination==="listoverride")V=void 0,L=void 0;else if(O.destination==="pict")t.extractAttachments?se=!0:Mt=!0;else if(xt.includes(O.destination))Mt=!0;else if(O.content.length>0&&O.content[0].type==="control"&&O.content[0].value==="*"){let Ye=["fldinst","nesttableprops"];if(t.extractAttachments&&Ye.push("shppict","listpicture"),!Ye.includes(O.destination||""))if(O.destination==="bkmkstart"){let Ke=ht(O);Ke&&!S.includes(Ke)&&S.push(Ke),Mt=!0}else O.destination,Mt=!0}}if(O.destination==="listtext"||O.destination==="pntext"){b||(y=0),b=!0;let Ye=Ct(O);if(Ye){let Ke=Ye.trim();["\xB7","\u2022","o","\xA7","\u25A0","\u25A1","\u25CF","\u25CB","\u25C6","\u25C7","\u25BA","\u25B8","\xB7","\u2022","\u25CF","\u25CB"].some(Va=>Ke.includes(Va))||Ke.length===1&&!/[0-9a-zA-Z]/.test(Ke)?w="unordered":(/^[0-9ivxlcdm]+[\.\)]/i.test(Ke)||/^[a-z][\.\)]/i.test(Ke))&&(w="ordered")}Mt=!0}if(Mt)return;if(d.push("{"),se){let Ye=mt(O);Ye&&(c.push(Ye),!z&&!D&&(lt(),m.push({type:"image",text:"",metadata:{attachmentName:Ye.name||`image_${c.length}`}})));for(let Ke of O.content)if(Ke.type==="control")d.push(ot(Ke));else if(Ke.type==="text")d.push(gt(Ke));else if(Ke.type==="group"){let nu=er=>{for(let Va of er.content)Va.type==="control"?d.push(ot(Va)):Va.type==="text"?d.push(gt(Va)):Va.type==="group"&&(d.push("{"),nu(Va),d.push("}"))};nu(O)}d.push("}");return}let _n=m;if(Ot){if(t.ignoreNotes)return;lt(),R++;let Ye="footnote";j===1?Ye="endnote":j===2&&(Ye=O.content.some(er=>er.type==="control"&&er.value==="ftnalt")?"endnote":"footnote");let Ke={type:"note",children:[],metadata:{noteId:R.toString(),noteType:Ye}};l.push(Ke),m=Ke.children}let $s={...K};for(let Ye of O.content){if(Ye.type==="group"&&Ye.destination==="fldinst"){d.push("{");let Ke=nu=>{for(let er of nu.content)er.type==="control"?d.push(ot(er)):er.type==="text"?d.push(gt(er)):er.type==="group"&&(d.push("{"),Ke(er),d.push("}"))};Ke(Ye),d.push("}");continue}Et(Ye,$s,Tt+1)}O.destination==="listtable"?z=!1:O.destination==="list"?(D=!1,F!==void 0&&P!==void 0&&(G[F]=P)):O.destination==="listoverridetable"?H=!1:H&&O.destination==="listoverride"&&L!==void 0&&V!==void 0&&(tt[L]=V),Ot&&(lt(),m=_n),Yt&&(J(),at=void 0),d.push("}")}else if(O.type==="text"){if(z||H){d.push(gt(O));return}ft(p,K)&&(J(),p={...K}),f.push(O.value),d.push(gt(O))}else if(O.type==="control"){if(d.push(ot(O)),z){if(O.value==="listid")F=O.param;else if(O.value==="levelnfc"||O.value==="levelnfcn"){let xt=O.param!==void 0&&(O.param===0||O.param>=0&&O.param<=4);(!P||P==="unordered"&&xt)&&(P=xt?"ordered":"unordered")}return}if(H){O.value==="listid"?V=O.param:O.value==="ls"&&(L=O.param);return}if(O.value==="par")lt(),p={...K};else if(O.value==="trowd"){let xt=$();Q&&xt&&xt.currentCellContent.length>0?(q(),M.push({rows:[],currentCells:[],currentCellContent:[],rowIndex:0})):Q||(Q=!0,q()),Z=!0,X=[],T={isMergedContinuation:!1},I=0}else if(O.value==="clvmrg")T.isMergedContinuation=!0;else if(O.value==="clmgf")T.isMergedContinuation=!1;else if(O.value==="cellx")X.push({...T}),T={isMergedContinuation:!1};else if(O.value==="cell"){Z=!0;let xt=!1;I<X.length&&(xt=X[I].isMergedContinuation),I++;let Mt=dt();if(Mt&&!xt){let Ot=$();Ot&&Ot.currentCells.push(Mt)}p={...K}}else if(O.value==="nestcell"){if(M.length>1){let Mt=$();Mt&&(Mt.rows.length>0||Mt.currentCells.length>0)&&vt()}Z=!0;let xt=dt();if(xt){let Mt=$();Mt&&Mt.currentCells.push(xt)}p={...K}}else if(O.value==="row")et(),p={...K},I=0,Z=!1;else if(O.value==="nestrow")M.length>1&&vt(),et(),p={...K},I=0;else if(O.value==="intbl")Q=!0,Z=!0,q();else if(O.value==="pard")Z=!1,y=0,x="left",b=!1,w=void 0,C=void 0,E=void 0,K.backgroundColor=void 0;else if(O.value==="tab")ft(p,K)&&(J(),p={...K}),f.push("	");else if(O.value==="line")ft(p,K)&&(J(),p={...K}),f.push(`
-`);else if(O.value==="lquote")ft(p,K)&&(J(),p={...K}),f.push("\u2018");else if(O.value==="rquote")ft(p,K)&&(J(),p={...K}),f.push("\u2019");else if(O.value==="ldblquote")ft(p,K)&&(J(),p={...K}),f.push("\u201C");else if(O.value==="rdblquote")ft(p,K)&&(J(),p={...K}),f.push("\u201D");else if(O.value==="u"){if(O.param!==void 0){let xt=O.param;xt<0&&(xt+=65536),ft(p,K)&&(J(),p={...K}),f.push(String.fromCharCode(xt))}}else if(O.value==="b")K.bold=O.param!==0;else if(O.value==="i")K.italic=O.param!==0;else if(O.value==="ul")K.underline=O.param!==0;else if(O.value==="ulnone")K.underline=!1;else if(O.value==="strike")K.strikethrough=O.param!==0;else if(O.value==="plain")K.bold=!1,K.italic=!1,K.underline=!1,K.strikethrough=!1,K.subscript=!1,K.superscript=!1,K.size=void 0,K.font=void 0,K.color=void 0,K.backgroundColor=void 0;else if(O.value==="fs")O.param!==void 0&&(K.size=(O.param/2).toString()+"pt");else if(O.value==="f")O.param!==void 0&&i[O.param]&&(K.font=i[O.param]);else if(O.value==="cf")O.param!==void 0&&s[O.param]&&(K.color=s[O.param]);else if(O.value==="fet")O.param!==void 0&&(j=O.param);else if(O.value==="cb"||O.value==="highlight"||O.value==="chcbpat"||O.value==="cbpat")O.param!==void 0&&s[O.param]&&(K.backgroundColor=s[O.param]);else if(O.value==="sub")K.subscript=!0,K.superscript=!1;else if(O.value==="super")K.superscript=!0,K.subscript=!1;else if(O.value==="nosupersub")K.subscript=!1,K.superscript=!1;else if(O.value==="li"){if(O.param!==void 0){let xt=Math.round(O.param/720);b||(y=xt)}}else if(O.value==="ls"){if(O.param!==void 0){b||(y=0),b=!0,W[O.param]||(_++,W[O.param]=`rtf-list-${_}`),E=W[O.param];let xt=tt[O.param]!==void 0?tt[O.param]:O.param;G[xt]&&(w=G[xt])}}else O.value==="ilvl"?O.param!==void 0&&(b=!0,y=O.param):O.value==="pnlvl"?(b=!0,O.param!==void 0&&(y=O.param)):O.value==="levelnfc"||O.value==="pnf"?O.param!==void 0&&(b=!0,O.param===23?w="unordered":w="ordered"):O.value==="pndec"||O.value==="pnord"||O.value==="pnlcltr"||O.value==="pnucltr"?(b||(y=0),b=!0,w="ordered"):O.value==="pnbullet"||O.value==="pncard"?(b||(y=0),b=!0,w="unordered"):O.value==="s"?O.param!==void 0&&O.param>=1&&O.param<=9&&(C=O.param):O.value==="ql"?x="left":O.value==="qc"?x="center":O.value==="qr"?x="right":O.value==="qj"&&(x="justify")}};Et(n,{});let Pt=$();if((Q||Pt&&(Pt.rows.length>0||Pt.currentCells.length>0))&&vt(),lt(),!t.putNotesAtLast&&l.length>0&&(a.push(...l),l.length=0),t.ocr&&t.extractAttachments){for(let K of c)if(K.mimeType.startsWith("image/"))try{let Tt=ct.Buffer.from(K.data,"base64");K.ocrText=(await lr(Tt,{...t.ocrConfig})).trim()}catch(Tt){Ce("OCR_FAILED",t,K.name,Tt)}let O=K=>{for(let Tt of K){if(Tt.type==="image"&&Tt.metadata&&"attachmentName"in Tt.metadata){let xt=Tt.metadata,Mt=c.find(Ot=>Ot.name===xt.attachmentName);Mt&&(Mt.ocrText&&(Tt.text=Mt.ocrText),Mt.altText&&(xt.altText=Mt.altText))}Tt.children&&O(Tt.children)}};O(a)}let wt=O=>{for(let K of O){if(K.type==="note"&&K.children){let Tt=xt=>xt.children&&xt.children.length>0?xt.children.map(Tt).join(""):xt.text||"";K.text=K.children.map(Tt).join("").trim()}K.children&&wt(K.children)}};wt(a),wt(l);let St=An("rtf",{},a,c,t,()=>{let O=a.map(K=>K.text).join(t.newlineDelimiter);return t.putNotesAtLast&&l.length>0&&(O+=t.newlineDelimiter+l.map(K=>K.text).join(t.newlineDelimiter)),O});return t.putNotesAtLast&&l.length>0&&a.push(...l),St};function _2(r,t){for(let e of r.content)if(e.type==="group"){if(e.destination===t)return e;let n=_2(e,t);if(n)return n}return null}function uF(r){let t={},e=_2(r,"fonttbl");if(e){for(let n of e.content)if(n.type==="group"){let i,s="";for(let a of n.content)a.type==="control"&&a.value==="f"?i=a.param:a.type==="text"&&(s+=a.value);i!==void 0&&s&&(t[i]=s.replace(/;$/,"").trim())}}return t}function hF(r){let t={},e=_2(r,"colortbl");if(e){let n=0,i=0,s=0,a=0;for(let l of e.content)if(l.type==="control")l.value==="red"&&l.param!==void 0?i=l.param:l.value==="green"&&l.param!==void 0?s=l.param:l.value==="blue"&&l.param!==void 0&&(a=l.param);else if(l.type==="text"&&l.value===";"){let c=`#${i.toString(16).padStart(2,"0")}${s.toString(16).padStart(2,"0")}${a.toString(16).padStart(2,"0")}`;t[n]=c,n++,i=0,s=0,a=0}}return t}B();var VD=async(r,t)=>{let e=/word\/document[\d+]?.xml/,n=/word\/footnotes[\d+]?.xml/,i=/word\/endnotes[\d+]?.xml/,s=/word\/numbering[\d+]?.xml/,a=/(word\/)?media\/.*/,l=/docProps\/core[\d+]?.xml/,c=/docProps\/custom\.xml/,u=/word\/_rels\/document[\d+]?.xml\.rels/,h=/word\/styles[\d+]?.xml/,d=D=>{let F={},P=(dt,et)=>{let vt=pt(dt,et);if(vt){let bt=vt.getAttribute("w:val");return bt===null?!0:bt==="1"||bt==="true"||bt==="on"}return null},G=P(D,"w:b");G!==null&&(F.bold=G);let H=P(D,"w:i");H!==null&&(F.italic=H);let V=pt(D,"w:u");if(V){let dt=V.getAttribute("w:val");(!dt||dt!=="none")&&(F.underline=!0)}let L=P(D,"w:strike"),tt=P(D,"w:dstrike");L!==null?F.strikethrough=L:tt!==null&&(F.strikethrough=tt);let rt=pt(D,"w:sz");if(rt){let dt=rt.getAttribute("w:val");dt&&(F.size=(parseInt(dt,10)/2).toString()+"pt")}let at=pt(D,"w:color");if(at){let dt=at.getAttribute("w:val");dt&&dt!=="auto"&&(F.color="#"+dt)}let ft=pt(D,"w:shd");if(ft){let dt=ft.getAttribute("w:fill");dt&&dt!=="auto"&&(F.backgroundColor="#"+dt)}let J=pt(D,"w:highlight");if(J){let dt=J.getAttribute("w:val");if(dt&&dt!=="none"){let et={yellow:"#FFFF00",green:"#00FF00",cyan:"#00FFFF",magenta:"#FF00FF",blue:"#0000FF",red:"#FF0000",darkBlue:"#00008B",darkCyan:"#008B8B",darkGreen:"#006400",darkMagenta:"#8B008B",darkRed:"#8B0000",darkYellow:"#808000",darkGray:"#A9A9A9",lightGray:"#D3D3D3",black:"#000000"};F.backgroundColor=et[dt]||dt}}let st=pt(D,"w:rFonts");if(st){let dt=st.getAttribute("w:ascii")||st.getAttribute("w:hAnsi");dt&&(F.font=dt)}let lt=pt(D,"w:vertAlign");if(lt){let dt=lt.getAttribute("w:val");dt==="subscript"?F.subscript=!0:dt==="superscript"&&(F.superscript=!0)}return F},f=D=>{let F=pt(D,"w:ind");if(F){let P={},G=F.getAttribute("w:left")||F.getAttribute("w:start"),H=F.getAttribute("w:right")||F.getAttribute("w:end"),V=F.getAttribute("w:firstLine"),L=F.getAttribute("w:hanging");return G&&(P.left=parseInt(G,10)),H&&(P.right=parseInt(H,10)),V&&(P.firstLine=parseInt(V,10)),L&&(P.hanging=parseInt(L,10)),Object.keys(P).length>0?P:void 0}},p=D=>{let F=pt(D,"mc:Choice");if(F)return Array.from(F.childNodes);let P=pt(D,"mc:Fallback");return Array.from(P?P.childNodes:D.childNodes)},m=await ro(r,D=>!!D.match(e)||!!D.match(n)||!!D.match(i)||!!D.match(s)||!!D.match(l)||!!D.match(c)||!!D.match(u)||!!D.match(h)||!!t.extractAttachments&&!!D.match(a)),y=m.find(D=>D.path.match(l)),x=y?no(y.content.toString()):{},b=m.find(D=>D.path.match(c));if(b){let D=ku(b.content.toString());Object.keys(D).length>0&&(x.customProperties=D)}let w=new Map,C=new Map,E=[],S=[],N=m.filter(D=>D.path.match(a)),k=m.find(D=>D.path.match(u)),M={};if(k){let D=le(k.content.toString()),F=ut(D,"Relationship");for(let P of F){let G=P.getAttribute("Id"),H=P.getAttribute("Target");G&&H&&(M[G]=H)}}let R=m.find(D=>D.path.match(s)),j={};if(R){let D=le(R.content.toString()),F=ut(D,"w:num"),P=ut(D,"w:abstractNum"),G={};for(let H of P){let V=H.getAttribute("w:abstractNumId");V&&(G[V]=H)}for(let H of F){let V=H.getAttribute("w:numId"),tt=pt(H,"w:abstractNumId")?.getAttribute("w:val");if(V&&tt&&G[tt]){j[V]={};let rt=ut(G[tt],"w:lvl");for(let ft of rt){let J=ft.getAttribute("w:ilvl"),st=pt(ft,"w:numFmt"),lt=pt(ft,"w:lvlText"),dt=pt(ft,"w:start");J&&(j[V][J]={numFmt:st?.getAttribute("w:val")||"decimal",lvlText:lt?.getAttribute("w:val")||"",start:parseInt(dt?.getAttribute("w:val")||"1",10)})}let at=ut(H,"w:lvlOverride");for(let ft of at){let J=ft.getAttribute("w:ilvl");if(J&&j[V][J]){let st=pt(ft,"w:startOverride");st&&(j[V][J].start=parseInt(st.getAttribute("w:val")||"1",10))}}}}}let $=m.find(D=>D.path.match(h)),q={};if($){let D=le($.content.toString()),F=ut(D,"w:style");for(let P of F){let G=P.getAttribute("w:styleId");if(G){let H=pt(P,"w:rPr"),V=pt(P,"w:pPr"),L=H?d(H):{},tt,rt,at;if(V){let ft=pt(V,"w:jc");if(ft){let lt=ft.getAttribute("w:val");(lt==="left"||lt==="center"||lt==="right"||lt==="justify")&&(tt=lt)}let J=pt(V,"w:shd");if(J){let lt=J.getAttribute("w:fill");lt&&lt!=="auto"&&(rt="#"+lt)}let st=f(V);st&&(at=st)}q[G]={formatting:L,alignment:tt,backgroundColor:rt,paragraphIndentation:at}}}}let Q={};if($){let D=le($.content.toString()),F=pt(D,"w:docDefaults");if(F){let P=pt(F,"w:rPrDefault");if(P){let G=pt(P,"w:rPr");G&&(Q=d(G))}}}let Z;if($){let D=le($.content.toString()),F=ut(D,"w:style");for(let P of F){let G=P.getAttribute("w:type"),H=P.getAttribute("w:default"),V=P.getAttribute("w:styleId");if(G==="paragraph"&&H==="1"&&V){Z=V;break}}!Z&&q.Normal&&(Z="Normal")}let Y=[],X=[],T={},I={},_=(D,F,P=[])=>{let G=pt(D,"w:numPr"),H=!!G,V=pt(D,"w:pPr"),tt=(V?pt(V,"w:pStyle"):null)?.getAttribute("w:val"),rt=tt?tt.startsWith("Heading")||tt==="Title":!1,at=tt&&q[tt]?q[tt]:{formatting:{}},ft=at.alignment;if(V){let mt=pt(V,"w:jc");if(mt){let ot=mt.getAttribute("w:val");(ot==="left"||ot==="center"||ot==="right"||ot==="justify")&&(ft=ot)}}let J=at.paragraphIndentation;if(V){let mt=f(V);mt&&(J={...J,...mt})}let st=at.backgroundColor;if(V){let mt=pt(V,"w:shd");if(mt){let ot=mt.getAttribute("w:fill");ot&&ot!=="auto"&&(st="#"+ot)}}let lt={...at.formatting};if(V){let mt=pt(V,"w:rPr");if(mt){let ot=d(mt);for(let ht in ot){let gt=ot[ht];gt===!1?delete lt[ht]:gt!==void 0&&(lt[ht]=gt)}}}let dt="",et=[],vt=mt=>{if(Oe(mt)&&mt.nodeName==="w:r"){let ot=mt,ht=pt(ot,"w:rPr"),gt={};for(let wt in lt)gt[wt]=lt[wt];let Et=ht?pt(ht,"w:rStyle"):null,Pt=Et?Et.getAttribute("w:val"):tt;if(Pt&&q[Pt])for(let wt in q[Pt].formatting)gt[wt]=q[Pt].formatting[wt];if(ht){let wt=d(ht);for(let kt in wt){let St=wt[kt];St===!1?delete gt[kt]:St!==void 0&&(gt[kt]=St)}}!gt.backgroundColor&&st&&(gt.backgroundColor=st);for(let wt of ot.childNodes)if(Oe(wt)){if(wt.tagName==="w:t"||wt.tagName==="t"){let kt=wt,St=kt.textContent||"";dt+=St;let O={type:"text",text:St,formatting:gt};t.includeRawContent&&(O.rawContent=ee(kt,F,t));let K=Pt||tt||Z;K&&(O.metadata={style:K}),et.push(O)}else if(t.includeBreakNodes&&(wt.tagName==="w:br"||wt.tagName==="br"||wt.tagName==="w:cr"||wt.tagName==="cr")){let kt=wt,St="textWrapping";if(wt.tagName==="w:cr"||wt.tagName==="cr")St="carriageReturn";else{let Tt=kt.getAttribute("w:type")||kt.getAttribute("type");Tt!==null&&(St=Tt)}let O;St==="textWrapping"&&kt.getAttribute("w:clear")!==null&&(O=kt.getAttribute("w:clear"));let K={type:"break",metadata:{breakType:St,clear:O}};t.includeRawContent&&(K.rawContent=ee(kt,F,t)),et.push(K)}else if(t.includeBreakNodes&&(wt.tagName==="w:lastRenderedPageBreak"||wt.tagName==="lastRenderedPageBreak")){let kt={type:"break",metadata:{breakType:"lastRenderedPage"}};t.includeRawContent&&(kt.rawContent=ee(wt,F,t)),et.push(kt)}}if(t.extractAttachments){let wt=ut(ot,"w:drawing"),kt=ut(ot,"w:pict"),St=[...wt,...kt];for(let O of St){let K=fx(O),Tt="",xt=pt(O,"wp:docPr");xt&&(Tt=xt.getAttribute("descr")||xt.getAttribute("title")||"");let Mt="",Ot=pt(O,"a:blip");if(Ot)Mt=Ot.getAttribute("r:embed")||"";else{let Yt=pt(O,"v:imagedata");Yt&&(Mt=Yt.getAttribute("r:id")||"")}if(Mt&&M[Mt]){let se=M[Mt].split("/").pop();if(se){let _n={type:"image",text:"",metadata:{attachmentName:se,altText:Tt}};t.includeRawContent&&(_n.rawContent=ee(O,F,t)),et.push(_n)}}else{let Yt={type:"image",text:""};t.includeRawContent&&(Yt.rawContent=ee(O,F,t)),et.push(Yt)}}}if(!t.ignoreNotes){let wt=pt(ot,"w:footnoteReference");if(wt){let St=wt.getAttribute("w:id");if(St&&w.has(St)){let O=w.get(St),K={type:"note",text:O.map(Tt=>Tt.text).join(" "),children:O,metadata:{noteType:"footnote",noteId:St}};t.putNotesAtLast?E.push(K):et.push(K)}}let kt=pt(ot,"w:endnoteReference");if(kt){let St=kt.getAttribute("w:id");if(St&&C.has(St)){let O=C.get(St),K={type:"note",text:O.map(Tt=>Tt.text).join(" "),children:O,metadata:{noteType:"endnote",noteId:St}};t.putNotesAtLast?E.push(K):et.push(K)}}}}else if(Oe(mt)&&mt.nodeName==="w:hyperlink"){let ot=mt,ht=ot.getAttribute("r:id"),gt=ot.getAttribute("w:anchor"),Et;gt&&!t.ignoreInternalLinks?Et={link:"#"+gt,linkType:"internal"}:ht&&M[ht]&&(Et={link:M[ht],linkType:"external"});let Pt=Array.from(ot.childNodes);for(let wt of Pt){let kt=et.length;if(vt(wt),Et)for(let St=kt;St<et.length;St++)et[St].type==="text"&&(et[St].metadata={...et[St].metadata??{},...Et})}}else if(Oe(mt)&&mt.nodeName==="w:bookmarkStart"){let ot=mt.getAttribute("w:name");ot&&!ot.startsWith("_GoBack")&&!t.ignoreInternalLinks&&bt.push(ot)}else if(Oe(mt)&&(mt.nodeName==="mc:AlternateContent"||mt.nodeName==="AlternateContent")){let ot=p(mt);for(let ht of ot)vt(ht)}else if(Oe(mt)&&(mt.nodeName==="w:pict"||mt.nodeName==="pict"||mt.nodeName==="w:drawing"||mt.nodeName==="drawing")){let ot=ut(mt,"w:txbxContent");for(let ht of ot){let gt=Array.from(ht.childNodes);for(let Et of gt)if(Oe(Et)&&Et.nodeName==="w:p"){let Pt=_(Et,F);et.push(...Pt.children||[]),dt+=Pt.text}}}else if(mt.childNodes.length>0)for(let ot of Array.from(mt.childNodes))vt(ot)},bt=[...P],Ft=Array.from(D.childNodes);for(let mt of Ft)vt(mt);let Ct=bt.length>0?{anchorIds:bt}:{};if(H){let mt=pt(G,"w:numId"),ot=pt(G,"w:ilvl"),ht=mt&&mt.getAttribute("w:val")||"0",gt=ot?parseInt(ot.getAttribute("w:val")||"0",10):0,Et="ordered",Pt=0;if(ht&&j[ht]){let kt=gt.toString();T[ht]||(T[ht]={}),T[ht][kt]||(T[ht][kt]=0),T[ht][kt]++;for(let O=gt+1;O<10;O++)T[ht][O.toString()]&&(T[ht][O.toString()]=0);Et=(j[ht][kt]?.numFmt||"decimal")==="bullet"?"unordered":"ordered",I[ht]||(I[ht]={}),I[ht][kt]===void 0?I[ht][kt]=(j[ht][kt]?.start??1)-1:I[ht][kt]++,Pt=I[ht][kt]}let wt={type:"list",text:dt,children:et,metadata:{listType:Et,indentation:gt,paragraphIndentation:J,alignment:ft||"left",listId:ht,itemIndex:Pt,style:tt,...Ct}};return t.includeRawContent&&(wt.rawContent=ee(D,F,t)),wt}else if(rt){let mt=tt&&parseInt(tt.replace("Heading",""),10)||1,ot={type:"heading",text:dt,children:et,metadata:{level:mt,alignment:ft,paragraphIndentation:J,style:tt??void 0,...Ct}};return t.includeRawContent&&(ot.rawContent=ee(D,F,t)),ot}else{let mt={type:"paragraph",text:dt,children:et,metadata:{alignment:ft,paragraphIndentation:J,style:tt??void 0,...Ct}};return t.includeRawContent&&(mt.rawContent=ee(D,F,t)),mt}},W=(D,F,P=[])=>{let G=[],H=Qe(D,"w:tr"),V=new Map;for(let L=0;L<H.length;L++){let tt=H[L],rt=[],at=Qe(tt,"w:tc"),ft=0;for(let st=0;st<at.length;st++){let lt=at[st],dt=pt(lt,"w:tcPr"),et=1;if(dt){let ht=pt(dt,"w:gridSpan");ht&&(et=parseInt(ht.getAttribute("w:val")||"1",10))}let vt=!1,bt=!1;if(dt){let ht=pt(dt,"w:vMerge");ht&&(bt=!0,(ht.getAttribute("w:val")==="restart"||!V.has(ft))&&(vt=!0))}let Ft=[],Ct="",mt=Array.from(lt.childNodes);for(let ht of mt)if(Oe(ht)&&ht.nodeName==="w:p"){let gt=_(ht,F);Ft.push(gt),Ct+=gt.text}else if(Oe(ht)&&ht.nodeName==="w:tbl"){let gt=W(ht,F);Ft.push(gt)}let ot={type:"cell",text:Ct,children:Ft,metadata:{row:L,col:ft}};if(et>1&&(ot.metadata.colSpan=et),bt)if(vt)V.set(ft,{node:ot,span:1}),rt.push(ot);else{let ht=V.get(ft);ht?(ht.span++,ht.node.metadata.rowSpan=ht.span,Ft.length>0&&(ht.node.children||(ht.node.children=[]),ht.node.children.push(...Ft),ht.node.text+=" "+Ct)):rt.push(ot)}else V.delete(ft),rt.push(ot);ft+=et}let J={type:"row",children:rt};G.push(J)}return{type:"table",children:G}};if(!t.ignoreNotes){let D=m.find(P=>P.path.match(n));if(D){let P=le(D.content.toString()),G=D.content.toString(),H=ut(P,"w:footnote");for(let V of H){let L=V.getAttribute("w:id");if(!L||L==="-1"||L==="0")continue;let tt=ut(V,"w:p");w.set(L,tt.map(rt=>_(rt,G)))}}let F=m.find(P=>P.path.match(i));if(F){let P=le(F.content.toString()),G=F.content.toString(),H=ut(P,"w:endnote");for(let V of H){let L=V.getAttribute("w:id");if(!L||L==="-1"||L==="0")continue;let tt=ut(V,"w:p");C.set(L,tt.map(rt=>_(rt,G)))}}}for(let D of m){if(D.path.match(a)||D.path.match(s)||D.path.match(u)||D.path.match(h)||D.path.match(n)||D.path.match(i))continue;let F=D.content.toString();t.includeRawContent&&X.push(F);let P=le(F,{locator:t.includeRawContent}),G=pt(P,"w:body");if(G){let H=Array.from(G.childNodes),V=[];for(let L of H)if(Oe(L)){if(L.nodeName==="w:p")Y.push(_(L,F,V)),V=[];else if(L.nodeName==="w:tbl")Y.push(W(L,F,V)),V=[];else if(L.nodeName==="w:bookmarkStart"){let tt=L.getAttribute("w:name");tt&&!tt.startsWith("_GoBack")&&!t.ignoreInternalLinks&&V.push(tt)}}}}if(t.extractAttachments){for(let D of N){let F=Mr(D.path.split("/").pop()||"image",D.content);if(S.push(F),t.ocr&&F.mimeType.startsWith("image/"))try{F.ocrText=(await lr(D.content,{...t.ocrConfig})).trim()}catch(P){Ce("OCR_FAILED",t,F.name,P)}}if(t.ocr){let D=F=>{for(let P of F){if(P.type==="image"&&"attachmentName"in(P.metadata||{})){let G=P.metadata,H=S.find(V=>V.name===G.attachmentName);H&&H.ocrText&&(P.text=H.ocrText,H.altText=G.altText)}P.children&&D(P.children)}};D(Y)}}t.putNotesAtLast&&E.length>0&&Y.push(...E);let z=()=>Y.map(D=>{let F=P=>{let G="";return P.children?G+=P.children.map(F).filter(H=>H!="").join(P.children[0]?.children?t.newlineDelimiter:""):P.type==="break"?G+=t.newlineDelimiter:G+=P.text||"",G};return F(D)}).filter(D=>D!="").join(t.newlineDelimiter);return An("docx",{...x,formatting:Q,styleMap:q},Y,S,t,z)};var rl=class{static async parseOffice(t,e,n){let i,s={};typeof e=="function"?(i=e,s=n||{}):s=e||{};let a=fv(s),l=[],c=a.onWarning;a.onWarning=f=>{l.push(f),c&&c(f)};let u=ct.Buffer.alloc(0),h=a.fileType??"",d;try{if(!t)throw hn("IMPROPER_ARGUMENTS",a);if(t instanceof ArrayBuffer)u=ct.Buffer.from(t);else if(ct.Buffer.isBuffer(t))u=t;else if(typeof t=="string"){d=t,Dg("path-parsing");let p=await Promise.resolve().then(()=>(iN(),nN));if(!p.existsSync(t))throw hn("FILE_DOES_NOT_EXIST",a,t);if(p.lstatSync(t).isDirectory())throw hn("LOCATION_NOT_FOUND",a,t);u=p.readFileSync(t),h=h||t.split(".").pop()||""}else throw hn("INVALID_INPUT",a);if(u.length>0&&!h)try{let{fileTypeFromBuffer:p}=await I2(),m=await p(u);m&&(h=m.ext)}catch(p){Ce("FILE_TYPE_DETECTION_FAILED",a,{error:p})}else if(u.length>0&&h)try{let{fileTypeFromBuffer:p}=await I2(),m=await p(u);m&&m.ext.toLowerCase()!==h.toLowerCase()&&Ce("BUFFER_TYPE_MISMATCH",a,{detected:m.ext,expected:h})}catch(p){Ce("FILE_TYPE_DETECTION_FAILED",a,{error:p})}if(!h)throw hn("IMPROPER_BUFFERS",a);let f;switch(h.toLowerCase()){case"docx":f=await VD(u,a);break;case"pptx":f=await WD(u,a);break;case"xlsx":f=await GE(u,a);break;case"odt":case"odp":case"ods":f=await qE(u,a);break;case"pdf":f=await HD(u,a);break;case"rtf":f=await qD(u,a);break;case"csv":f=await Gv(u,a);break;case"html":f=await HE(u,a);break;case"md":f=await WE(u,a);break;default:throw hn("EXTENSION_UNSUPPORTED",a,h)}return f.warnings=l,i&&i(f),f}catch(f){let p=dv(f,a,d);throw i&&i(void 0,p),p}}static async terminateOcr(){await jE()}};B();var yb=class{static async convert(t,e,n){let i={...n?.parseConfig,onWarning:n?.onWarning||n?.parseConfig?.onWarning};i.ocr=!1,Object.keys(i).forEach(c=>i[c]===void 0&&delete i[c]),i.extractAttachments=n?.generatorConfig?.includeImages!==!1||n?.generatorConfig?.includeCharts!==!1;let s=await rl.parseOffice(t,i),a={...n?.generatorConfig,onWarning:n?.onWarning||n?.generatorConfig?.onWarning},l=await Ka.generate(s,e,a);return l.messages=[...s.warnings||[],...l.messages],l}};var M4=rl.parseOffice,O4=rl.terminateOcr,F4=yb.convert,P4=Ka.generate;var R4=rl;export{yb as OfficeConverter,ov as OfficeErrorType,Ka as OfficeGenerator,rl as OfficeParser,lv as OfficeWarningType,F4 as convert,R4 as default,P4 as generate,M4 as parseOffice,O4 as terminateOcr};
+`);else if(O.value==="lquote")ft(p,K)&&(J(),p={...K}),f.push("\u2018");else if(O.value==="rquote")ft(p,K)&&(J(),p={...K}),f.push("\u2019");else if(O.value==="ldblquote")ft(p,K)&&(J(),p={...K}),f.push("\u201C");else if(O.value==="rdblquote")ft(p,K)&&(J(),p={...K}),f.push("\u201D");else if(O.value==="u"){if(O.param!==void 0){let xt=O.param;xt<0&&(xt+=65536),ft(p,K)&&(J(),p={...K}),f.push(String.fromCharCode(xt))}}else if(O.value==="b")K.bold=O.param!==0;else if(O.value==="i")K.italic=O.param!==0;else if(O.value==="ul")K.underline=O.param!==0;else if(O.value==="ulnone")K.underline=!1;else if(O.value==="strike")K.strikethrough=O.param!==0;else if(O.value==="plain")K.bold=!1,K.italic=!1,K.underline=!1,K.strikethrough=!1,K.subscript=!1,K.superscript=!1,K.size=void 0,K.font=void 0,K.color=void 0,K.backgroundColor=void 0;else if(O.value==="fs")O.param!==void 0&&(K.size=(O.param/2).toString()+"pt");else if(O.value==="f")O.param!==void 0&&i[O.param]&&(K.font=i[O.param]);else if(O.value==="cf")O.param!==void 0&&s[O.param]&&(K.color=s[O.param]);else if(O.value==="fet")O.param!==void 0&&(j=O.param);else if(O.value==="cb"||O.value==="highlight"||O.value==="chcbpat"||O.value==="cbpat")O.param!==void 0&&s[O.param]&&(K.backgroundColor=s[O.param]);else if(O.value==="sub")K.subscript=!0,K.superscript=!1;else if(O.value==="super")K.superscript=!0,K.subscript=!1;else if(O.value==="nosupersub")K.subscript=!1,K.superscript=!1;else if(O.value==="li"){if(O.param!==void 0){let xt=Math.round(O.param/720);b||(y=xt)}}else if(O.value==="ls"){if(O.param!==void 0){b||(y=0),b=!0,W[O.param]||(_++,W[O.param]=`rtf-list-${_}`),E=W[O.param];let xt=tt[O.param]!==void 0?tt[O.param]:O.param;G[xt]&&(w=G[xt])}}else O.value==="ilvl"?O.param!==void 0&&(b=!0,y=O.param):O.value==="pnlvl"?(b=!0,O.param!==void 0&&(y=O.param)):O.value==="levelnfc"||O.value==="pnf"?O.param!==void 0&&(b=!0,O.param===23?w="unordered":w="ordered"):O.value==="pndec"||O.value==="pnord"||O.value==="pnlcltr"||O.value==="pnucltr"?(b||(y=0),b=!0,w="ordered"):O.value==="pnbullet"||O.value==="pncard"?(b||(y=0),b=!0,w="unordered"):O.value==="s"?O.param!==void 0&&O.param>=1&&O.param<=9&&(C=O.param):O.value==="ql"?x="left":O.value==="qc"?x="center":O.value==="qr"?x="right":O.value==="qj"&&(x="justify")}};Et(n,{});let Pt=$();if((Q||Pt&&(Pt.rows.length>0||Pt.currentCells.length>0))&&vt(),lt(),!t.putNotesAtLast&&l.length>0&&(a.push(...l),l.length=0),t.ocr&&t.extractAttachments){for(let K of c)if(K.mimeType.startsWith("image/"))try{let Tt=ct.Buffer.from(K.data,"base64");K.ocrText=(await lr(Tt,{...t.ocrConfig})).trim()}catch(Tt){Ce("OCR_FAILED",t,K.name,Tt)}let O=K=>{for(let Tt of K){if(Tt.type==="image"&&Tt.metadata&&"attachmentName"in Tt.metadata){let xt=Tt.metadata,Mt=c.find(Ot=>Ot.name===xt.attachmentName);Mt&&(Mt.ocrText&&(Tt.text=Mt.ocrText),Mt.altText&&(xt.altText=Mt.altText))}Tt.children&&O(Tt.children)}};O(a)}let wt=O=>{for(let K of O){if(K.type==="note"&&K.children){let Tt=xt=>xt.children&&xt.children.length>0?xt.children.map(Tt).join(""):xt.text||"";K.text=K.children.map(Tt).join("").trim()}K.children&&wt(K.children)}};wt(a),wt(l);let St=An("rtf",{},a,c,t,()=>{let O=a.map(K=>K.text).join(t.newlineDelimiter);return t.putNotesAtLast&&l.length>0&&(O+=t.newlineDelimiter+l.map(K=>K.text).join(t.newlineDelimiter)),O});return t.putNotesAtLast&&l.length>0&&a.push(...l),St};function _2(r,t){for(let e of r.content)if(e.type==="group"){if(e.destination===t)return e;let n=_2(e,t);if(n)return n}return null}function uF(r){let t={},e=_2(r,"fonttbl");if(e){for(let n of e.content)if(n.type==="group"){let i,s="";for(let a of n.content)a.type==="control"&&a.value==="f"?i=a.param:a.type==="text"&&(s+=a.value);i!==void 0&&s&&(t[i]=s.replace(/;$/,"").trim())}}return t}function hF(r){let t={},e=_2(r,"colortbl");if(e){let n=0,i=0,s=0,a=0;for(let l of e.content)if(l.type==="control")l.value==="red"&&l.param!==void 0?i=l.param:l.value==="green"&&l.param!==void 0?s=l.param:l.value==="blue"&&l.param!==void 0&&(a=l.param);else if(l.type==="text"&&l.value===";"){let c=`#${i.toString(16).padStart(2,"0")}${s.toString(16).padStart(2,"0")}${a.toString(16).padStart(2,"0")}`;t[n]=c,n++,i=0,s=0,a=0}}return t}B();var VD=async(r,t)=>{let e=/word\/document[\d+]?.xml/,n=/word\/footnotes[\d+]?.xml/,i=/word\/endnotes[\d+]?.xml/,s=/word\/numbering[\d+]?.xml/,a=/(word\/)?media\/.*/,l=/docProps\/core[\d+]?.xml/,c=/docProps\/custom\.xml/,u=/word\/_rels\/document[\d+]?.xml\.rels/,h=/word\/styles[\d+]?.xml/,d=D=>{let F={},P=(dt,et)=>{let vt=pt(dt,et);if(vt){let bt=vt.getAttribute("w:val");return bt===null?!0:bt==="1"||bt==="true"||bt==="on"}return null},G=P(D,"w:b");G!==null&&(F.bold=G);let H=P(D,"w:i");H!==null&&(F.italic=H);let V=pt(D,"w:u");if(V){let dt=V.getAttribute("w:val");(!dt||dt!=="none")&&(F.underline=!0)}let L=P(D,"w:strike"),tt=P(D,"w:dstrike");L!==null?F.strikethrough=L:tt!==null&&(F.strikethrough=tt);let rt=pt(D,"w:sz");if(rt){let dt=rt.getAttribute("w:val");dt&&(F.size=(parseInt(dt,10)/2).toString()+"pt")}let at=pt(D,"w:color");if(at){let dt=at.getAttribute("w:val");dt&&dt!=="auto"&&(F.color="#"+dt)}let ft=pt(D,"w:shd");if(ft){let dt=ft.getAttribute("w:fill");dt&&dt!=="auto"&&(F.backgroundColor="#"+dt)}let J=pt(D,"w:highlight");if(J){let dt=J.getAttribute("w:val");if(dt&&dt!=="none"){let et={yellow:"#FFFF00",green:"#00FF00",cyan:"#00FFFF",magenta:"#FF00FF",blue:"#0000FF",red:"#FF0000",darkBlue:"#00008B",darkCyan:"#008B8B",darkGreen:"#006400",darkMagenta:"#8B008B",darkRed:"#8B0000",darkYellow:"#808000",darkGray:"#A9A9A9",lightGray:"#D3D3D3",black:"#000000"};F.backgroundColor=et[dt]||dt}}let st=pt(D,"w:rFonts");if(st){let dt=st.getAttribute("w:ascii")||st.getAttribute("w:hAnsi");dt&&(F.font=dt)}let lt=pt(D,"w:vertAlign");if(lt){let dt=lt.getAttribute("w:val");dt==="subscript"?F.subscript=!0:dt==="superscript"&&(F.superscript=!0)}return F},f=D=>{let F=pt(D,"w:ind");if(F){let P={},G=F.getAttribute("w:left")||F.getAttribute("w:start"),H=F.getAttribute("w:right")||F.getAttribute("w:end"),V=F.getAttribute("w:firstLine"),L=F.getAttribute("w:hanging");return G&&(P.left=parseInt(G,10)),H&&(P.right=parseInt(H,10)),V&&(P.firstLine=parseInt(V,10)),L&&(P.hanging=parseInt(L,10)),Object.keys(P).length>0?P:void 0}},p=D=>{let F=pt(D,"mc:Choice");if(F)return Array.from(F.childNodes);let P=pt(D,"mc:Fallback");return Array.from(P?P.childNodes:D.childNodes)},m=await ro(r,D=>!!D.match(e)||!!D.match(n)||!!D.match(i)||!!D.match(s)||!!D.match(l)||!!D.match(c)||!!D.match(u)||!!D.match(h)||!!t.extractAttachments&&!!D.match(a)),y=m.find(D=>D.path.match(l)),x=y?no(y.content.toString()):{},b=m.find(D=>D.path.match(c));if(b){let D=ku(b.content.toString());Object.keys(D).length>0&&(x.customProperties=D)}let w=new Map,C=new Map,E=[],S=[],N=m.filter(D=>D.path.match(a)),k=m.find(D=>D.path.match(u)),M={};if(k){let D=le(k.content.toString()),F=ut(D,"Relationship");for(let P of F){let G=P.getAttribute("Id"),H=P.getAttribute("Target");G&&H&&(M[G]=H)}}let R=m.find(D=>D.path.match(s)),j={};if(R){let D=le(R.content.toString()),F=ut(D,"w:num"),P=ut(D,"w:abstractNum"),G={};for(let H of P){let V=H.getAttribute("w:abstractNumId");V&&(G[V]=H)}for(let H of F){let V=H.getAttribute("w:numId"),tt=pt(H,"w:abstractNumId")?.getAttribute("w:val");if(V&&tt&&G[tt]){j[V]={};let rt=ut(G[tt],"w:lvl");for(let ft of rt){let J=ft.getAttribute("w:ilvl"),st=pt(ft,"w:numFmt"),lt=pt(ft,"w:lvlText"),dt=pt(ft,"w:start");J&&(j[V][J]={numFmt:st?.getAttribute("w:val")||"decimal",lvlText:lt?.getAttribute("w:val")||"",start:parseInt(dt?.getAttribute("w:val")||"1",10)})}let at=ut(H,"w:lvlOverride");for(let ft of at){let J=ft.getAttribute("w:ilvl");if(J&&j[V][J]){let st=pt(ft,"w:startOverride");st&&(j[V][J].start=parseInt(st.getAttribute("w:val")||"1",10))}}}}}let $=m.find(D=>D.path.match(h)),q={};if($){let D=le($.content.toString()),F=ut(D,"w:style");for(let P of F){let G=P.getAttribute("w:styleId");if(G){let H=pt(P,"w:rPr"),V=pt(P,"w:pPr"),L=H?d(H):{},tt,rt,at;if(V){let ft=pt(V,"w:jc");if(ft){let lt=ft.getAttribute("w:val");(lt==="left"||lt==="center"||lt==="right"||lt==="justify")&&(tt=lt)}let J=pt(V,"w:shd");if(J){let lt=J.getAttribute("w:fill");lt&&lt!=="auto"&&(rt="#"+lt)}let st=f(V);st&&(at=st)}q[G]={formatting:L,alignment:tt,backgroundColor:rt,paragraphIndentation:at}}}}let Q={};if($){let D=le($.content.toString()),F=pt(D,"w:docDefaults");if(F){let P=pt(F,"w:rPrDefault");if(P){let G=pt(P,"w:rPr");G&&(Q=d(G))}}}let Z;if($){let D=le($.content.toString()),F=ut(D,"w:style");for(let P of F){let G=P.getAttribute("w:type"),H=P.getAttribute("w:default"),V=P.getAttribute("w:styleId");if(G==="paragraph"&&H==="1"&&V){Z=V;break}}!Z&&q.Normal&&(Z="Normal")}let Y=[],X=[],T={},I={},_=(D,F,P=[])=>{let G=pt(D,"w:numPr"),H=!!G,V=pt(D,"w:pPr"),tt=(V?pt(V,"w:pStyle"):null)?.getAttribute("w:val"),rt=tt?tt.startsWith("Heading")||tt==="Title":!1,at=tt&&q[tt]?q[tt]:{formatting:{}},ft=at.alignment;if(V){let mt=pt(V,"w:jc");if(mt){let ot=mt.getAttribute("w:val");(ot==="left"||ot==="center"||ot==="right"||ot==="justify")&&(ft=ot)}}let J=at.paragraphIndentation;if(V){let mt=f(V);mt&&(J={...J,...mt})}let st=at.backgroundColor;if(V){let mt=pt(V,"w:shd");if(mt){let ot=mt.getAttribute("w:fill");ot&&ot!=="auto"&&(st="#"+ot)}}let lt={...at.formatting};if(V){let mt=pt(V,"w:rPr");if(mt){let ot=d(mt);for(let ht in ot){let gt=ot[ht];gt===!1?delete lt[ht]:gt!==void 0&&(lt[ht]=gt)}}}let dt="",et=[],vt=mt=>{if(Oe(mt)&&mt.nodeName==="w:r"){let ot=mt,ht=pt(ot,"w:rPr"),gt={};for(let wt in lt)gt[wt]=lt[wt];let Et=ht?pt(ht,"w:rStyle"):null,Pt=Et?Et.getAttribute("w:val"):tt;if(Pt&&q[Pt])for(let wt in q[Pt].formatting)gt[wt]=q[Pt].formatting[wt];if(ht){let wt=d(ht);for(let kt in wt){let St=wt[kt];St===!1?delete gt[kt]:St!==void 0&&(gt[kt]=St)}}!gt.backgroundColor&&st&&(gt.backgroundColor=st);for(let wt of ot.childNodes)if(Oe(wt)){if(wt.tagName==="w:t"||wt.tagName==="t"){let kt=wt,St=kt.textContent||"";dt+=St;let O={type:"text",text:St,formatting:gt};t.includeRawContent&&(O.rawContent=ee(kt,F,t));let K=Pt||tt||Z;K&&(O.metadata={style:K}),et.push(O)}else if(t.includeBreakNodes&&(wt.tagName==="w:br"||wt.tagName==="br"||wt.tagName==="w:cr"||wt.tagName==="cr")){let kt=wt,St="textWrapping";if(wt.tagName==="w:cr"||wt.tagName==="cr")St="carriageReturn";else{let Tt=kt.getAttribute("w:type")||kt.getAttribute("type");Tt!==null&&(St=Tt)}let O;St==="textWrapping"&&kt.getAttribute("w:clear")!==null&&(O=kt.getAttribute("w:clear"));let K={type:"break",metadata:{breakType:St,clear:O}};t.includeRawContent&&(K.rawContent=ee(kt,F,t)),et.push(K)}else if(t.includeBreakNodes&&(wt.tagName==="w:lastRenderedPageBreak"||wt.tagName==="lastRenderedPageBreak")){let kt={type:"break",metadata:{breakType:"lastRenderedPage"}};t.includeRawContent&&(kt.rawContent=ee(wt,F,t)),et.push(kt)}}if(t.extractAttachments){let wt=ut(ot,"w:drawing"),kt=ut(ot,"w:pict"),St=[...wt,...kt];for(let O of St){let K=fx(O),Tt="",xt=pt(O,"wp:docPr");xt&&(Tt=xt.getAttribute("descr")||xt.getAttribute("title")||"");let Mt="",Ot=pt(O,"a:blip");if(Ot)Mt=Ot.getAttribute("r:embed")||"";else{let Yt=pt(O,"v:imagedata");Yt&&(Mt=Yt.getAttribute("r:id")||"")}if(Mt&&M[Mt]){let se=M[Mt].split("/").pop();if(se){let _n={type:"image",text:"",metadata:{attachmentName:se,altText:Tt}};t.includeRawContent&&(_n.rawContent=ee(O,F,t)),et.push(_n)}}else{let Yt={type:"image",text:""};t.includeRawContent&&(Yt.rawContent=ee(O,F,t)),et.push(Yt)}}}if(!t.ignoreNotes){let wt=pt(ot,"w:footnoteReference");if(wt){let St=wt.getAttribute("w:id");if(St&&w.has(St)){let O=w.get(St),K={type:"note",text:O.map(Tt=>Tt.text).join(" "),children:O,metadata:{noteType:"footnote",noteId:St}};t.putNotesAtLast?E.push(K):et.push(K)}}let kt=pt(ot,"w:endnoteReference");if(kt){let St=kt.getAttribute("w:id");if(St&&C.has(St)){let O=C.get(St),K={type:"note",text:O.map(Tt=>Tt.text).join(" "),children:O,metadata:{noteType:"endnote",noteId:St}};t.putNotesAtLast?E.push(K):et.push(K)}}}}else if(Oe(mt)&&mt.nodeName==="w:hyperlink"){let ot=mt,ht=ot.getAttribute("r:id"),gt=ot.getAttribute("w:anchor"),Et;gt&&!t.ignoreInternalLinks?Et={link:"#"+gt,linkType:"internal"}:ht&&M[ht]&&(Et={link:M[ht],linkType:"external"});let Pt=Array.from(ot.childNodes);for(let wt of Pt){let kt=et.length;if(vt(wt),Et)for(let St=kt;St<et.length;St++)et[St].type==="text"&&(et[St].metadata={...et[St].metadata??{},...Et})}}else if(Oe(mt)&&mt.nodeName==="w:bookmarkStart"){let ot=mt.getAttribute("w:name");ot&&!ot.startsWith("_GoBack")&&!t.ignoreInternalLinks&&bt.push(ot)}else if(Oe(mt)&&(mt.nodeName==="mc:AlternateContent"||mt.nodeName==="AlternateContent")){let ot=p(mt);for(let ht of ot)vt(ht)}else if(Oe(mt)&&(mt.nodeName==="w:pict"||mt.nodeName==="pict"||mt.nodeName==="w:drawing"||mt.nodeName==="drawing")){let ot=ut(mt,"w:txbxContent");for(let ht of ot){let gt=Array.from(ht.childNodes);for(let Et of gt)if(Oe(Et)&&Et.nodeName==="w:p"){let Pt=_(Et,F);et.push(...Pt.children||[]),dt+=Pt.text}}}else if(mt.childNodes.length>0)for(let ot of Array.from(mt.childNodes))vt(ot)},bt=[...P],Ft=Array.from(D.childNodes);for(let mt of Ft)vt(mt);let Ct=bt.length>0?{anchorIds:bt}:{};if(H){let mt=pt(G,"w:numId"),ot=pt(G,"w:ilvl"),ht=mt&&mt.getAttribute("w:val")||"0",gt=ot?parseInt(ot.getAttribute("w:val")||"0",10):0,Et="ordered",Pt=0;if(ht&&j[ht]){let kt=gt.toString();T[ht]||(T[ht]={}),T[ht][kt]||(T[ht][kt]=0),T[ht][kt]++;for(let O=gt+1;O<10;O++)T[ht][O.toString()]&&(T[ht][O.toString()]=0);Et=(j[ht][kt]?.numFmt||"decimal")==="bullet"?"unordered":"ordered",I[ht]||(I[ht]={}),I[ht][kt]===void 0?I[ht][kt]=(j[ht][kt]?.start??1)-1:I[ht][kt]++,Pt=I[ht][kt]}let wt={type:"list",text:dt,children:et,metadata:{listType:Et,indentation:gt,paragraphIndentation:J,alignment:ft||"left",listId:ht,itemIndex:Pt,style:tt,...Ct}};return t.includeRawContent&&(wt.rawContent=ee(D,F,t)),wt}else if(rt){let mt=tt&&parseInt(tt.replace("Heading",""),10)||1,ot={type:"heading",text:dt,children:et,metadata:{level:mt,alignment:ft,paragraphIndentation:J,style:tt??void 0,...Ct}};return t.includeRawContent&&(ot.rawContent=ee(D,F,t)),ot}else{let mt={type:"paragraph",text:dt,children:et,metadata:{alignment:ft,paragraphIndentation:J,style:tt??void 0,...Ct}};return t.includeRawContent&&(mt.rawContent=ee(D,F,t)),mt}},W=(D,F,P=[])=>{let G=[],H=Qe(D,"w:tr"),V=new Map;for(let L=0;L<H.length;L++){let tt=H[L],rt=[],at=Qe(tt,"w:tc"),ft=0;for(let st=0;st<at.length;st++){let lt=at[st],dt=pt(lt,"w:tcPr"),et=1;if(dt){let ht=pt(dt,"w:gridSpan");ht&&(et=parseInt(ht.getAttribute("w:val")||"1",10))}let vt=!1,bt=!1;if(dt){let ht=pt(dt,"w:vMerge");ht&&(bt=!0,(ht.getAttribute("w:val")==="restart"||!V.has(ft))&&(vt=!0))}let Ft=[],Ct="",mt=Array.from(lt.childNodes);for(let ht of mt)if(Oe(ht)&&ht.nodeName==="w:p"){let gt=_(ht,F);Ft.push(gt),Ct+=gt.text}else if(Oe(ht)&&ht.nodeName==="w:tbl"){let gt=W(ht,F);Ft.push(gt)}let ot={type:"cell",text:Ct,children:Ft,metadata:{row:L,col:ft}};if(et>1&&(ot.metadata.colSpan=et),bt)if(vt)V.set(ft,{node:ot,span:1}),rt.push(ot);else{let ht=V.get(ft);ht?(ht.span++,ht.node.metadata.rowSpan=ht.span,Ft.length>0&&(ht.node.children||(ht.node.children=[]),ht.node.children.push(...Ft),ht.node.text+=" "+Ct)):rt.push(ot)}else V.delete(ft),rt.push(ot);ft+=et}let J={type:"row",children:rt};G.push(J)}return{type:"table",children:G}};if(!t.ignoreNotes){let D=m.find(P=>P.path.match(n));if(D){let P=le(D.content.toString()),G=D.content.toString(),H=ut(P,"w:footnote");for(let V of H){let L=V.getAttribute("w:id");if(!L||L==="-1"||L==="0")continue;let tt=ut(V,"w:p");w.set(L,tt.map(rt=>_(rt,G)))}}let F=m.find(P=>P.path.match(i));if(F){let P=le(F.content.toString()),G=F.content.toString(),H=ut(P,"w:endnote");for(let V of H){let L=V.getAttribute("w:id");if(!L||L==="-1"||L==="0")continue;let tt=ut(V,"w:p");C.set(L,tt.map(rt=>_(rt,G)))}}}for(let D of m){if(D.path.match(a)||D.path.match(s)||D.path.match(u)||D.path.match(h)||D.path.match(n)||D.path.match(i))continue;let F=D.content.toString();t.includeRawContent&&X.push(F);let P=le(F,{locator:t.includeRawContent}),G=pt(P,"w:body");if(G){let H=Array.from(G.childNodes),V=[];for(let L of H)if(Oe(L)){if(L.nodeName==="w:p")Y.push(_(L,F,V)),V=[];else if(L.nodeName==="w:tbl")Y.push(W(L,F,V)),V=[];else if(L.nodeName==="w:bookmarkStart"){let tt=L.getAttribute("w:name");tt&&!tt.startsWith("_GoBack")&&!t.ignoreInternalLinks&&V.push(tt)}}}}if(t.extractAttachments){for(let D of N){let F=Mr(D.path.split("/").pop()||"image",D.content);if(S.push(F),t.ocr&&F.mimeType.startsWith("image/"))try{F.ocrText=(await lr(D.content,{...t.ocrConfig})).trim()}catch(P){Ce("OCR_FAILED",t,F.name,P)}}if(t.ocr){let D=F=>{for(let P of F){if(P.type==="image"&&"attachmentName"in(P.metadata||{})){let G=P.metadata,H=S.find(V=>V.name===G.attachmentName);H&&H.ocrText&&(P.text=H.ocrText,H.altText=G.altText)}P.children&&D(P.children)}};D(Y)}}t.putNotesAtLast&&E.length>0&&Y.push(...E);let z=()=>Y.map(D=>{let F=P=>{let G="";return P.children?G+=P.children.map(F).filter(H=>H!="").join(P.children[0]?.children?t.newlineDelimiter:""):P.type==="break"?G+=t.newlineDelimiter:G+=P.text||"",G};return F(D)}).filter(D=>D!="").join(t.newlineDelimiter);return An("docx",{...x,formatting:Q,styleMap:q},Y,S,t,z)};var rl=class{static async parseOffice(t,e,n){let i,s={};typeof e=="function"?(i=e,s=n||{}):s=e||{};let a=fv(s),l=[],c=a.onWarning;a.onWarning=f=>{l.push(f),c&&c(f)};let u=ct.Buffer.alloc(0),h=a.fileType??"",d;try{if(!t)throw hn("IMPROPER_ARGUMENTS",a);if(t instanceof ArrayBuffer)u=ct.Buffer.from(t);else if(t instanceof Uint8Array)u=ct.Buffer.from(t.buffer,t.byteOffset,t.byteLength);else if(ct.Buffer.isBuffer(t))u=t;else if(typeof t=="string"){d=t,Dg("path-parsing");let p=await Promise.resolve().then(()=>(iN(),nN));if(!p.existsSync(t))throw hn("FILE_DOES_NOT_EXIST",a,t);if(p.lstatSync(t).isDirectory())throw hn("LOCATION_NOT_FOUND",a,t);u=p.readFileSync(t),h=h||t.split(".").pop()||""}else throw hn("INVALID_INPUT",a);if(u.length>0&&!h)try{let{fileTypeFromBuffer:p}=await I2(),m=await p(u);m&&(h=m.ext)}catch(p){Ce("FILE_TYPE_DETECTION_FAILED",a,{error:p})}else if(u.length>0&&h)try{let{fileTypeFromBuffer:p}=await I2(),m=await p(u);m&&m.ext.toLowerCase()!==h.toLowerCase()&&Ce("BUFFER_TYPE_MISMATCH",a,{detected:m.ext,expected:h})}catch(p){Ce("FILE_TYPE_DETECTION_FAILED",a,{error:p})}if(!h)throw hn("IMPROPER_BUFFERS",a);let f;switch(h.toLowerCase()){case"docx":f=await VD(u,a);break;case"pptx":f=await WD(u,a);break;case"xlsx":f=await GE(u,a);break;case"odt":case"odp":case"ods":f=await qE(u,a);break;case"pdf":f=await HD(u,a);break;case"rtf":f=await qD(u,a);break;case"csv":f=await Gv(u,a);break;case"html":f=await HE(u,a);break;case"md":f=await WE(u,a);break;default:throw hn("EXTENSION_UNSUPPORTED",a,h)}return f.warnings=l,i&&i(f),f}catch(f){let p=dv(f,a,d);throw i&&i(void 0,p),p}}static async terminateOcr(){await jE()}};B();var yb=class{static async convert(t,e,n){let i={...n?.parseConfig,onWarning:n?.onWarning||n?.parseConfig?.onWarning};i.ocr=!1,Object.keys(i).forEach(c=>i[c]===void 0&&delete i[c]),i.extractAttachments=n?.generatorConfig?.includeImages!==!1||n?.generatorConfig?.includeCharts!==!1;let s=await rl.parseOffice(t,i),a={...n?.generatorConfig,onWarning:n?.onWarning||n?.generatorConfig?.onWarning},l=await Ka.generate(s,e,a);return l.messages=[...s.warnings||[],...l.messages],l}};var M4=rl.parseOffice,O4=rl.terminateOcr,F4=yb.convert,P4=Ka.generate;var R4=rl;export{yb as OfficeConverter,ov as OfficeErrorType,Ka as OfficeGenerator,rl as OfficeParser,lv as OfficeWarningType,F4 as convert,R4 as default,P4 as generate,M4 as parseOffice,O4 as terminateOcr};
 /*! Bundled license information:
 
 ieee754/index.js: