支持目录

TSDocxToPdf.tsf

@@ -10,6 +10,9 @@ public
     function GetPdf(): PdfFile;
     function GetNextPage(page: TSPage): TSPage;
     function AddPage(sect_ware: TSSectWare): TSPage;
+    function AdjustPageNumber(page: TSPage; num: integer);
+    function LinkToToc(anchor: string; page: TSPage);
+    function AddToc(anchor: string; toc: TSToc);
 
     property Font read ReadFont;
     function ReadFont();
@@ -37,6 +40,17 @@ private
     current_page_: TSPage;
     point_: TSPoint; // 定位坐标点
     page_array_: array of TSPage;
+    toc_array_: tableArray;
+end;
+
+type TSPoint = class
+    function Create();
+    begin
+        {self.}X := 0;
+        {self.}Y := 0;
+    end
+    X: real;
+    Y: real;
 end;
 
 function TSDocxToPdf.Create(alias: string; file: string);
@@ -50,7 +64,7 @@ begin
     current_page_ := nil;
     point_ := new TSPoint();
     page_array_ := array();
-    page_index_ := array();
+    toc_array_ := array();
 end;
 
 function TSDocxToPdf.Destroy();
@@ -76,8 +90,10 @@ begin
         elements := sect_ware.Elements;
         for _,element in elements do
         begin
+            // println("_ = {}", _);
             if element.LocalName = "p" then {self.}TransformP(sect_ware, element);
             else if element.LocalName = "tbl" then {self.}TransformTbl(sect_ware, element);
+            else if element.LocalName = "sdt" then {self.}TransformSdt(sect_ware, element);
         end
     end
 end;
@@ -114,7 +130,7 @@ begin
     namespace "DOCX";
     docx_components_ := new TSDocxComponentsWare();
     [err, msg] := docx_components_.OpenFile(alias, file, nil);
-    if err then raise "Create obejct 'TSDocxFile' failed.";
+    if err then raise "Open file error.";
 end;
 
 function TSDocxToPdf.InitCachePath(file: string);
@@ -172,9 +188,11 @@ begin
     {self.}ResetCoordinates(sect_ware);
 
     len := length(page_array_);
+    start := ifString(sect_ware.SectPr.PgNumType.Start) and trystrtoint(sect_ware.SectPr.PgNumType.Start, r) ? r : 0;
     current_page_ := new TSPage();
     current_page_.Index := len;
     current_page_.PdfPage := page;
+    current_page_.Number := len = 0 ? start : page_array_[len-1].Number + 1;
     page_array_[len] := current_page_;
 
 if sysparams["_PDF_PAGE_GRID_DEBUG_"] then
@@ -183,6 +201,12 @@ if sysparams["_PDF_PAGE_GRID_DEBUG_"] then
     return current_page_;
 end;
 
+function TSDocxToPdf.AdjustPageNumber(page: TSPage; num: integer);
+begin
+    for i:=page.Index to length(page_array_)-1 do
+        page_array_[i].Number := num++;
+end;
+
 function TSDocxToPdf.GetNextPage(page: TSPage);
 begin
     return page_array_[page.Index + 1];
@@ -259,5 +283,23 @@ end;
 
 function TSDocxToPdf.TransformSdt(sect_ware: TSSectWare; sdt: Sdt);
 begin
-    w := sect_ware.SectPr.PgSz.W - sect_ware.SectPr.PgMar.Right - sect_ware.SectPr.PgMar.Left;
+    ps := sdt.SdtContent.Ps();
+    for _,p in ps do
+        {self.}TransformP(sect_ware, p);
+end;
+
+function TSDocxToPdf.AddToc(anchor: string; toc: TSToc);
+begin
+    if ifarray(toc_array_[anchor]) then toc_array_[anchor] union= array(toc);
+    else toc_array_[anchor] := array(toc);
+end;
+
+function TSDocxToPdf.LinkToToc(anchor: string; page: TSPage);
+begin
+    arr := toc_array_[anchor];
+    if ifnil(arr) then return;
+    dst := page.PdfPage.CreateDestination();
+    for _,toc in arr do
+        toc.LinkAnnot(dst);
+    toc.AddPageNumber(page);
 end;

range/Advanced/TSPdfParagraphRange.tsf

@@ -13,15 +13,18 @@ private
     function SetLvlText();
     function GetImageFileType(data: binary): string;
     function GetParagraphLineSpace(size: real; line: integer): real;
-    function RangesToLines(ppr: PPr): tableArray;
+    function RangesToLines(ppr: PPrUnitDecorator): tableArray;
     function CheckAndAddPage(y: real; offset: real): boolean;
     function GetUtf8CharLength(byte: string): integer;
-    function RToTextRange(r: R; ppr: PPr);
+    function RToTextRange(r: R; ppr: PPrUnitDecorator; link: string);
-    function SplitTextToTextRange(text: string; rpr: RPr);
+    function SplitTextToTextRange(text: string; rpr: RPrUnitDecorator; link: string);
-    function RToDrawingRange(r: R; ppr: PPr);
+    function RToDrawingRange(r: R; ppr: PPrUnitDecorator);
-    function SetLinesAlignment(ppr: PPr);
+    function SetLinesAlignment(ppr: PPrUnitDecorator);
-    function ResetCoordinates(ppr: PPr);
+    function ResetCoordinates(ppr: PPrUnitDecorator);
     function NewLineRange(): TSPdfLineRange;
+    function BookMarkLinkToc();
+    function HyperlinkToToc(ppr: PPrUnitDecorator);
+    function HyperlinkToTextRange(hyperlink: Hyperlink; ppr: PPrUnitDecorator);
 
 private
     [weakref]docx_to_pdf_: TSDocxToPdf;
@@ -32,6 +35,8 @@ private
     extra_style_id_: string;
     range_array_: array of TSPdfBasicRange;
     line_range_array_: array of TSPdfLineRange;
+    hyperlink_array_: tableArray;
+    bookmark_array_: tableArray;
 end;
 
 function TSPdfParagraphRange.Create(docx_to_pdf: TSDocxToPdf; pg: TSPage; components: TSDocxComponentsWare; sect_ware: TSSectWare; paragraph: P);
@@ -44,11 +49,14 @@ begin
     extra_style_id_ := "";
     range_array_ := array();
     line_range_array_ := array();
+    hyperlink_array_ := array();
+    bookmark_array_ := array();
     {self.}TSPage := page_;
 end;
 
 function TSPdfParagraphRange.Calc(): tableArray;
 begin
+    // ppr.rpr是无效的，应该以ppr.pStyle为准
     {self.}SetPPr(paragraph_.PPr);
     {self.}SetLvlText();
     ppr := new PPrUnitDecorator(paragraph_.PPr);
@@ -58,37 +66,115 @@ begin
     {self.}CheckAndAddPage({self.}EndY, ppr.Spacing.Before);
     {self.}EndY -= ppr.Spacing.Before;
     {self.}DynamicHeight += ppr.Spacing.Before;
-    rs := paragraph_.Rs();
+
-    if length(rs) = 0 then
+    elements := paragraph_.Elements();
+    empty_flag := true;
+    bookmark_id := '';
+    bookmark_name := '';
+    bookmark_flag := false;
+    for _,element in elements do
+    begin
+        if not ifObj(element.XmlNode) then continue;
+        if element.LocalName = "r" then
+        begin
+            empty_flag := false;
+            {self.}SetRPr(element.RPr, ppr);
+            if element.Br.Type = "page" then
+                {self.}CheckAndAddPage(sect_ware_.SectPr.PgMar.Bottom, 1);
+            else if ifObj(element.Drawing.XmlNode) then {self.}RToDrawingRange(element, ppr);
+            else {self.}RToTextRange(element, ppr, bookmark_name);
+        end
+        else if element.LocalName = "hyperlink" then
+        begin
+            empty_flag := false;
+            {self.}HyperlinkToTextRange(element, ppr);
+        end
+        else if element.LocalName = "bookmarkStart" then
+        begin
+            bookmark_id := element.Id;
+            bookmark_name := element.Name;
+            bookmark_array_[bookmark_name] := array();
+        end
+        else if element.LocalName = "bookmarkEnd" then
+        begin
+            bookmark_id := '';
+            bookmark_name := '';
+        end
+    end
+    if empty_flag then
     begin
         line_space := {self.}GetParagraphLineSpace(ppr.RPr.Sz.Val, ppr.Spacing.Line);
         {self.}DynamicHeight += ppr.Spacing.After + line_space;
         {self.}EndY -= {self.}DynamicHeight;
     end
-    else begin
-        for _, r in rs do
-        begin
-            {self.}SetRPr(r.RPr, paragraph_.PPr);
-            if r.Br.Type = "page" then
-                {self.}CheckAndAddPage(sect_ware_.SectPr.PgMar.Bottom, 1);
-            else if ifObj(r.Drawing.XmlNode) then {self.}RToDrawingRange(r, ppr);
-            else {self.}RToTextRange(r, ppr);
-        end
-    end
     {self.}RangesToLines(ppr);
-    // hyperlinks := paragraph_.Hyperlinks(); // TOC
+    if hyperlink_array_ then {self.}HyperlinkToToc(ppr);
-    // for _,hyperlink in hyperlinks do
+    if bookmark_array_ then {self.}BookMarkLinkToc();
-    // begin
-    //     rs := hyperlink.Rs();
-    //     for _,r in rs do
-    //     begin
-    //         if r.FldChar.FldCharType = "begin" then break;
-    //         {self.}RToTextRange(r, ppr);
-    //     end
-    // end
 end;
 
-function TSPdfParagraphRange.Do();
+function TSPdfParagraphRange.BookMarkLinkToc();
+begin
+    for name,arr in bookmark_array_ do
+        if arr[0] then docx_to_pdf_.LinkToToc(name, arr[0].TSPage);
+end;
+
+function TSPdfParagraphRange.HyperlinkToToc(ppr: PPrUnitDecorator);
+begin
+    for anchor,arr in hyperlink_array_ do  // 整理hyperlink发送到docx_to_pdf_
+    begin
+        pg := arr[0].TSPage;
+        left := {self.}StartX;
+        bottom := arr[0].EndY;
+        right := left + {self.}Width;
+        top := bottom + arr[0].RPr.Sz.Val;
+        x := arr[0].EndX + arr[0].Width;
+        y := arr[0].EndY;
+        for _,range in arr do
+        begin
+            if range.TSPage <> pg then // 说明目录文字换页
+            begin
+                rect := array(left, bottom, right, top);
+                toc := new TSToc(ppr, rect, pg, x, y);
+                docx_to_pdf_.AddToc(anchor, toc);
+                pg := range.TSPage;
+                left := range.EndX;
+                top := range.EndY + range.RPr.Sz.Val;
+                right := left;
+                bottom := range.EndY;
+                continue;
+            end
+            top := range.EndY + arr[0].RPr.Sz.Val;
+            x := range.EndX + range.Width;
+        end
+        rect := array(left, bottom, right, top);
+        font_name := ppr.RPr.RFonts.EastAsia ? ppr.RPr.RFonts.EastAsia : ppr.RPr.RFonts.Ascii;
+        font_obj := docx_to_pdf_.Font.GetFont(font_name, ppr.RPr.B, ppr.RPr.I);
+        toc := new TSToc(ppr, rect, pg, x, y, font_obj);
+        docx_to_pdf_.AddToc(anchor, toc);
+    end
+end;
+
+function TSPdfParagraphRange.HyperlinkToTextRange(hyperlink: Hyperlink; ppr: PPrUnitDecorator);
+begin
+    i := length(range_array_);
+    rs := hyperlink.Rs();
+    for _,r in rs do
+    begin
+        if r.FldChar.FldCharType = "begin" then break;
+        r.RPr := new RPr();
+        {self.}SetRPr(r.RPr, ppr);
+        {self.}RToTextRange(r, ppr, hyperlink.Anchor);
+    end
+    arr := array();
+    while i < length(range_array_) do
+    begin
+        arr[length(arr)] := range_array_[i];
+        i++;
+    end
+    hyperlink_array_[hyperlink.Anchor] := arr;
+end;
+
+function TSPdfParagraphRange.Do();override;
 begin
     for _,line_range in line_range_array_ do
         line_range.Do();
@@ -108,7 +194,7 @@ begin
     return line_range;
 end;
 
-function TSPdfParagraphRange.RangesToLines(ppr: PPr);
+function TSPdfParagraphRange.RangesToLines(ppr: PPrUnitDecorator);
 begin
     line_range := {self.}NewLineRange();
     i := 0;
@@ -167,7 +253,7 @@ begin
     {self.}SetLinesAlignment(ppr);
 end;
 
-function TSPdfParagraphRange.SetLinesAlignment(ppr: PPr);
+function TSPdfParagraphRange.SetLinesAlignment(ppr: PPrUnitDecorator);
 begin
     if length(line_range_array_) = 0 then return;
     idx := length(line_range_array_)-1;
@@ -200,14 +286,20 @@ begin
         return 4;
 end;
 
-function TSPdfParagraphRange.RToTextRange(r: R; ppr: PPr);
+function TSPdfParagraphRange.RToTextRange(r: R; ppr: PPrUnitDecorator; link: string);
 begin
-    rpr := new RPrUnitDecorator(r.RPr);
+    if r.Anchor then
-    text := r.T.Text;
+    begin
-    if ifString(text) then {self.}SplitTextToTextRange(text, rpr);
+        {self.}HyperlinkToTextRange(r, ppr);
+    end
+    else begin
+        rpr := new RPrUnitDecorator(r.RPr);
+        text := r.T.Text;
+        if ifString(text) then {self.}SplitTextToTextRange(text, rpr, link);
+    end
 end;
 
-function TSPdfParagraphRange.SplitTextToTextRange(text: string; rpr: RPr);
+function TSPdfParagraphRange.SplitTextToTextRange(text: string; rpr: RPrUnitDecorator; link: string);
 begin
     pos := 1;
     while pos <= length(text) do
@@ -219,7 +311,6 @@ begin
         font_name := rpr.RFonts.EastAsia ? rpr.RFonts.EastAsia : rpr.RFonts.Ascii;
         font_obj := docx_to_pdf_.Font.GetFont(font_name, rpr.B, rpr.I);
         if not rpr.Sz.Val then rpr.Sz.Val := rpr.SzCs.Val ? rpr.SzCs.Val : docx_to_pdf_.Font.GetDefaultSz();
-        // println("page_.PdfPage = {}, font_obj = {}, sz = {}", page_.PdfPage, font_obj, rpr.Sz.Val);
         page_.PdfPage.SetFontAndSize(font_obj, rpr.Sz.Val);
         word_width := page_.PdfPage.TextWidth(word);
         text_range := new TSPdfTextRange();
@@ -228,10 +319,11 @@ begin
         text_range.Font := font_obj;
         text_range.Width := word_width;
         range_array_[length(range_array_)] := text_range;
+        if ifarray(bookmark_array_[link]) then bookmark_array_[link] union= array(text_range);
     end
 end;
 
-function TSPdfParagraphRange.RToDrawingRange(r: R; ppr: PPr);
+function TSPdfParagraphRange.RToDrawingRange(r: R; ppr: PPrUnitDecorator);
 begin
     xfrm := new XfrmUnitDecorator(r.Drawing._Inline.Graphic.GraphicData.Pic.SpPr.Xfrm);
     rels_adapter := docx_components_ware_.GetDocumentRelsAdapter();
@@ -318,11 +410,11 @@ end;
 
 function TSPdfParagraphRange.SetRPr(var rpr; ppr: PPr);
 begin
-    // rpr，先继承ppr，再继承样式，最后继承自己
     new_rpr := new RPr();
-    style_id := rpr.RStyle.Val;
+    styles := docx_components_ware_.GetStyles();
-    new_rpr.Copy(ppr.RPr);
+    new_rpr.Copy(styles.DocDefaults.RPrDefault.RPr);
-    {self.}SetRPrByStyleId(new_rpr, style_id);
+    {self.}SetRPrByStyleId(new_rpr, ppr.PStyle.Val);
+    {self.}SetRPrByStyleId(new_rpr, rpr.RStyle.Val);
     new_rpr.Copy(rpr);
     rpr.Copy(new_rpr);
 end;
@@ -350,7 +442,7 @@ begin
     {self.}SplitTextToTextRange(lvl_text, rpr);
 end;
 
-function TSPdfParagraphRange.ResetCoordinates(ppr: PPr);
+function TSPdfParagraphRange.ResetCoordinates(ppr: PPrUnitDecorator);
 begin
     // 根据段落的间距确定新的坐标
     sz := ppr.RPr.SzCs.Val ? ppr.RPr.SzCs.Val : ppr.RPr.Sz.Val ? ppr.RPr.Sz.Val : docx_to_pdf_.Font.GetDefaultSz();

range/Advanced/TSPdfTableRange.tsf

@@ -7,7 +7,6 @@ public
 private
     function ProcessTrData(grid_cols: array of GridCol; tr: Tr; tbl_pr: TblPr): array of TSPdfAbstractRange;
     function ResetCoordinates(tbl_pr: TblPr; grid_cols: array of GridCol);
-    function CheckAndAddPage(y: real; offset: real): boolean;
     function SetTblPr(tbl_pr: TblPr);
     function SetTblPrByStyleId(var tbl_pr: TblPr; style_id: string);
     function SetTrPr(var tr_pr: TrPr);
@@ -22,7 +21,6 @@ private
     [weakref]table_: Tbl;
     [weakref]page_: TSPage;
     [weakref]range_array_: tableArray;
-    point_: TSPoint;
 end;
 
 function TSPdfTableRange.Create(docx_to_pdf: TSDocxToPdf; pg: PdfPage; components: Components; sect_ware: TSSectWare; table: Tbl);
@@ -60,19 +58,6 @@ begin
             range.Do();
 end;
 
-function TSPdfTableRange.CheckAndAddPage(y: real; offset: real): boolean;
-begin
-    if y - offset < sect_ware_.SectPr.PgMar.Bottom then
-    begin
-        page_ := docx_to_pdf_.GetNextPage(page_);
-        if ifnil(page_) then page_ := docx_to_pdf_.AddPage(sect_ware_);
-        point := docx_to_pdf_.GetCurrentPoint();
-        {self.}EndY := point.Y;
-        return true;
-    end
-    return false;
-end;
-
 function TSPdfTableRange.ProcessTrData(grid_cols: array of GridCol; tr: Tr; tbl_pr: TblPr): array of TSPdfAbstractRange;
 begin
     {self.}SetTrPr(tr.TrPr);

range/Advanced/TSPdfTocRange.tsf

@@ -1,6 +1,9 @@
 type TSPdfTocRange = class(TSPdfBasicRange)
 public
     function Create();
+
+public
+    TSPage: TSPage;
 end;
 
 function TSPdfTocRange.Create();

utils/TSPage.tsf

@@ -1,5 +1,6 @@
 type TSPage = class
 public
-    PdfPage;
+    PdfPage: PdfPage;
-    Index;
+    Index: integer;
+    Number: integer; // 页码
 end;

utils/TSPoint.tsf

@@ -1,10 +0,0 @@
-type TSPoint = class
-    function Create();
-    begin
-        {self.}X := 0;
-        {self.}Y := 0;
-    end
-    X: real;
-    Y: real;
-end;
-

utils/TSToc.tsf

@@ -0,0 +1,51 @@
+type TSToc = class
+uses TSPdfEnumerations;
+public
+    function Create(ppr: PPrUnitDecorator; rect: array of real; page: TSPage; x: real; y: real; font: PdfFont);
+    function LinkAnnot(dst: PdfDestination);
+    function AddPageNumber(page: TSPage);
+
+private
+    ppr_: PPrUnitDecorator;
+    rect_: array of real;
+    font_: PdfFont;
+    [weakref]tspage_: TSPage;
+    x_;
+    y_;
+end;
+
+function TSToc.Create(ppr: PPrUnitDecorator; rect: array of real; page: TSPage; x: real; y: real; font: PdfFont);
+begin
+    ppr_ := ppr;
+    rect_ := rect;
+    tspage_ := page;
+    x_ := x;
+    y_ := y;
+    font_ := font;
+end;
+
+function TSToc.LinkAnnot(dst: PdfDestination);
+begin
+    annot := tspage_.PdfPage.CreateLinkAnnot(rect_, dst);
+    annot.LinkAnnotSetHighlightMode(TSPdfEnumerations.ANNOT_NO_HIGHTLIGHT);
+end;
+
+function TSToc.AddPageNumber(page: TSPage);
+begin
+    number := tostring(page.Number);
+    symbol := ".";
+    tspage_.PdfPage.SetFontAndSize(font_, ppr_.RPr.Sz.Val);
+    w := tspage_.PdfPage.TextWidth(symbol);
+    number_sz := tspage_.PdfPage.TextWidth(number);
+    x := rect_[2] - number_sz;
+    tspage_.PdfPage.BeginText();
+    tspage_.PdfPage.TextOut(x, y_, number);
+    tspage_.PdfPage.EndText();
+    while x_ < x - w do
+    begin
+        tspage_.PdfPage.BeginText();
+        tspage_.PdfPage.TextOut(x_, y_, symbol);
+        tspage_.PdfPage.EndText();
+        x_ += w;
+    end
+end;