PdfConverter/TSDocxToPdf.tsf

type TSDocxToPdf = class
public
    function Create(alias: string; file: string);
    function Destroy();
    function SaveToFile(alias: string; file: string): integer;
    function Transform();

    function GetCurrentPoint(): Point;
    function GetCachePath(image_path: string): string;
    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();

private
    function InitDocxComponents(alias: string; file: string);
    function InitCachePath(file: string);
    function InitPdfEncoder();
    function InitSectWare();
    function AllocateElementsToSectWare();

    function ResetCoordinates(sect_ware: TSSectWare);
    function TransformP(sect_ware: TSSectWare; paragraph: P);
    function TransformTbl(sect_ware: TSSectWare; table: Tbl);
    function TransformSdt(sect_ware: TSSectWare; sdt: Sdt);

    function PrintGrid(page: PdfPage; sect_ware: TSSectWare); // test

private
    pdf_: PdfFile;
    docx_components_: TSDocxComponentsWare; // TSDocxComponentsWare
    cache_path_: string; // 临时目录，用来存放临时文件
    sect_ware_array_: array of TSSectWare; // 页面布局组件数组
    font_ware_: TSFontWare;     // 字体部件
    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);
begin
    pdf_ := new PdfFile();
    {self.}InitPdfEncoder();
    {self.}InitDocxComponents(alias, file);
    {self.}InitCachePath(file);
    {self.}InitSectWare();
    font_ware_ := new TSFontWare(pdf_);
    current_page_ := nil;
    point_ := new TSPoint();
    page_array_ := array();
    toc_array_ := array();
end;

function TSDocxToPdf.Destroy();
begin
    removeDir("", cache_path_);
end;

function TSDocxToPdf.SaveToFile(alias: string; file: string): integer;
begin
    return pdf_.SaveToFile(alias, file);
end;

function TSDocxToPdf.Transform();
begin
    prev := nil;
    for _,sect_ware in sect_ware_array_ do
    begin
        if prev <> sect_ware then
        begin
            {self.}AddPage(sect_ware);
            prev := sect_ware;
        end
        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;

function TSDocxToPdf.GetCurrentPoint(): Point;
begin
    return point_;
end;

function TSDocxToPdf.GetCachePath(image_path: string): string;
begin
    return cache_path_ + extractFileName(image_path);
end;

function TSDocxToPdf.GetPdf(): PdfFile;
begin
    return pdf_;
end;

function TSDocxToPdf.ReadFont();
begin
    return font_ware_;
end;

function TSDocxToPdf.InitPdfEncoder();
begin
    pdf_.UseCNSFonts();
    pdf_.UseCNSEncodings();
    // pdf_.UseUTFEncodings();
end;

function TSDocxToPdf.InitDocxComponents(alias: string; file: string);
begin
    namespace "DOCX";
    docx_components_ := new TSDocxComponentsWare();
    [err, msg] := docx_components_.OpenFile(alias, file, nil);
    if err then raise "Open file error.";
end;

function TSDocxToPdf.InitCachePath(file: string);
begin
    path := format("%s_%s", extractFileName(file), formatDatetime("YYYYMMDDHHNNSSZZZ", now()));
    cache_dir := format("%s/funcext/PdfConverter/.cache", extractFileDir(sysExecName()));
    createDir("", cache_dir);
    cache_path_ := format("%s/%s/", cache_dir, path);
    createDir("", cache_path_);
end;

function TSDocxToPdf.InitSectWare();
begin
    document := docx_components_.Document;
    document.Deserialize();
    sect_ware_array_ := array();
    {self.}AllocateElementsToSectWare();
end;

function TSDocxToPdf.AllocateElementsToSectWare();
begin
    elements := docx_components_.Document.Body.Elements();
    ware := new TSSectWare();
    fp := function(ware, sect);
    begin
        sect := new SectPrUnitDecorator(sect);
        sect.PgSz.Orient := sect.PgSz.Orient ? "portrait" : "landscape";
        ware.SectPr := sect;
        ware.Do();
    end
    for i:=0 to length(elements)-1 do
    begin
        element := elements[i];
        ware.AddElement(element);
        if element.LocalName = "p" and ifObj(element.PPr.SectPr.XmlNode) then
        begin
            ##fp(ware, element.PPr.SectPr);
            sect_ware_array_[length(sect_ware_array_)] := ware;
            ware := new TSSectWare();
        end
        else if element.LocalName = "sectPr" and i = length(elements)-1 then
        begin
            ##fp(ware, element);
            sect_ware_array_[length(sect_ware_array_)] := ware;
        end
    end
    // println("sect_ware_array_ = {}", sect_ware_array_);
end;

function TSDocxToPdf.AddPage(sect_ware: TSSectWare): TSPage;
begin
    page := pdf_.AddPage();
    page.SetWidth(sect_ware.SectPr.PgSz.W);
    page.SetHeight(sect_ware.SectPr.PgSz.H);
    {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
    {self.}PrintGrid(page, sect_ware);

    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];
end;

function TSDocxToPdf.ResetCoordinates(sect_ware: TSSectWare);
begin
    point_.X := sect_ware.SectPr.PgMar.Left;
    point_.Y := sect_ware.SectPr.PgSz.H - max(sect_ware.SectPr.PgMar.Top, sect_ware.SectPr.PgMar.Header);
end;

function TSDocxToPdf.PrintGrid(page: PdfPage; sect_ware: TSSectWare);
begin
    i := 0;
    while true do
    begin
        y := point_.Y - i * sect_ware.SectPr.DocGrid.LinePitch;
        if y <= sect_ware.SectPr.PgMar.Bottom then break;
        page.SetLineWidth(0.05);
        page.SetGrayStroke(0.75);
        page.MoveTo(sect_ware.SectPr.PgMar.Left, y);
        page.LineTo(sect_ware.SectPr.PgSz.W- sect_ware.SectPr.PgMar.Right, y);
        page.Stroke();
        i++;
    end

    x1 := sect_ware.SectPr.PgMar.Left;
    y1 := sect_ware.SectPr.PgSz.H - sect_ware.SectPr.PgMar.Top;
    x2 := sect_ware.SectPr.PgSz.W - sect_ware.SectPr.PgMar.Right;
    y2 := y1;
    x3 := x1;
    y3 := sect_ware.SectPr.PgMar.Bottom;
    x4 := x2;
    y4 := y3;
    page.SetLineWidth(0.05);
    page.SetGrayStroke(0.5);
    page.MoveTo(x1, y1);
    page.LineTo(x2, y2);
    page.Stroke();
    page.MoveTo(x1, y1);
    page.LineTo(x3, y3);
    page.Stroke();
    page.MoveTo(x2, y2);
    page.LineTo(x4, y4);
    page.Stroke();
    page.MoveTo(x3, y3);
    page.LineTo(x4, y4);
    page.Stroke();
end;

function TSDocxToPdf.TransformP(sect_ware: TSSectWare; paragraph: P);
begin
    w := sect_ware.SectPr.PgSz.W - sect_ware.SectPr.PgMar.Right - sect_ware.SectPr.PgMar.Left;
    range := new TSPdfParagraphRange(self, current_page_, docx_components_, sect_ware, paragraph);
    range.StartX := point_.X;
    range.StartY := point_.Y;
    range.Width := w;
    range.Calc();
    range.Do();
    point_.Y := range.EndY;
end;

function TSDocxToPdf.TransformTbl(sect_ware: TSSectWare; table: Tbl);
begin
    w := sect_ware.SectPr.PgSz.W - sect_ware.SectPr.PgMar.Right - sect_ware.SectPr.PgMar.Left;
    range := new TSPdfTableRange(self, current_page_, docx_components_, sect_ware, table);
    range.StartX := point_.X;
    range.StartY := point_.Y;
    range.Width := w;
    range.Calc();
    range.Do();
    point_.Y := range.EndY;
end;

function TSDocxToPdf.TransformSdt(sect_ware: TSSectWare; sdt: Sdt);
begin
    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;