tinysoft
/
PdfConverter
5
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

1. 修复段落字体不一致问题 

2. 修复图片（印章为例）的位置不正确
3. 修复其他报错问题
4. 项目符号用可中文能识别的符号替代
This commit is contained in:
csh 2024-10-11 10:52:10 +08:00
parent 68ce759668
commit 2a7027476a
3 changed files with 108 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
TSDocxToPdf.tsf
View File

@ -20,6 +20,7 @@ public
function SetHeaderAndFooter();
function ProcessNumpages();
function CalculateTextCoordinates(): array of real;
function GetSymbol(symbol: string);
property Font read ReadFont;
function ReadFont();
@ -29,6 +30,7 @@ private
function InitCachePath(file: string);
function InitPdfEncoder();
function InitSectWare();
function InitSymbol();
function AllocateElementsToSectWare();
function SetHdr(type: string);
@ -57,6 +59,7 @@ private
ftr_point_: Point; // 页脚坐标
even_and_odd_flag_: boolean;
xml_file_: string;
symbol_: tableArray;
end;
type Point = class
@ -76,6 +79,7 @@ begin
{self.}InitDocxComponents(alias, file);
{self.}InitCachePath(file);
{self.}InitSectWare();
{self.}InitSymbol();
font_ware_ := new TSFontWare(pdf_);
current_page_ := nil;
page_array_ := array();
@ -118,8 +122,8 @@ begin
elements := sect_ware.Elements();
for _,element in elements do
begin
// if _ = 4 then break;
// if _ = 624 then
// if _ = 31 then break;
// if _ = 3 then
// println("_ = {}, xml_file_ = {}", _, xml_file_);
if element.LocalName = "p" then {self.}TransformP(text_point_, element, w, lb);
else if element.LocalName = "tbl" then {self.}TransformTbl(text_point_, element, w, lb);
@ -201,6 +205,19 @@ begin
{self.}AllocateElementsToSectWare();
end;
function TSDocxToPdf.InitSymbol();
begin
symbol_ := array(
"": "※",
"": "■",
"": "●",
"": "◆",
"": "●",
"": "■",
);
end;
function TSDocxToPdf.AllocateElementsToSectWare();
begin
elements := docx_components_ware_.Document.Body.Elements();
@ -440,3 +457,9 @@ begin
toc.LinkAnnot(dst);
toc.AddPageNumber(page);
end;
function TSDocxToPdf.GetSymbol(symbol: string);
begin
// println("symbol = {}, symbol_ = {}", symbol, symbol_);
return symbol_[symbol];
end;

64
range/Advanced/TSPdfParagraphRange.tsf
View File

@ -6,8 +6,6 @@ public
function SetTblStyleIdAndType(style_id: string; type: string);
function SetNumPages(num: integer);
function RangesToLines();
function FirstValidTSPage(): TSPage;
function IsSamePage(): boolean;
function GetLastPage(): TSPage;
function AdjustRangeOffset(page: TSPage; x_offset: real; y_offset: real);
function GetLineRangeArr(): array of TSPdfLineRange;
@ -72,17 +70,17 @@ end;
function TSPdfParagraphRange.Calc(): tableArray;
begin
// ppr.rpr是无效的，应该以ppr.pStyle为准
if ifnil(paragraph_.XmlChildPPr) then paragraph_.XmlChildPPr := new PPr();
{self.}SetPPr(paragraph_.PPr);
{self.}SetLvlText();
ppr_unit_decorator_ := new PPrUnitDecorator(paragraph_.PPr);
if not ppr_unit_decorator_.RPr.Sz.Val then ppr_unit_decorator_.RPr.Sz.Val := docx_to_pdf_.Font.GetDefaultSz();
{self.}ResetCoordinates();
{self.}EndX := {self.}StartX;
{self.}EndY := {self.}StartY;
{self.}CheckAndAddPage({self.}EndY, ppr_unit_decorator_.Spacing.Before);
{self.}TSPage := page_;
{self.}EndY -= ppr_unit_decorator_.Spacing.Before;
{self.}DynamicHeight += ppr_unit_decorator_.Spacing.Before;
{self.}SetLvlText();
elements := paragraph_.Elements();
empty_flag := true;
@ -94,6 +92,7 @@ begin
if element.LocalName = "r" then
begin
empty_flag := false;
if ifnil(element.XmlChildRPr) then element.XmlChildRPr := new RPr();
{self.}SetRPr(element.RPr, ppr_unit_decorator_);
if element.FldChar.FldCharType = "begin" then
continue;
@ -119,6 +118,7 @@ begin
numpages := false;
end
else if element.Br.Type = "page" then
// {self.}EndY := {self.}LowerBound;
{self.}CheckAndAddPage({self.}LowerBound, 1);
else if ifObj(element.Drawing.XmlNode) then {self.}RToDrawingRange(element);
else if ifObj(element.AlternateContent.XmlNode) then {self.}RAlternateContentToRange(element);
@ -228,7 +228,7 @@ begin
y := range.EndY;
end
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);
font_obj := docx_to_pdf_.Font.GetCNSFont(font_name, ppr.RPr.B, ppr.RPr.I);
line_space := {self.}GetParagraphLineSpace(max_size, ppr.Spacing.Line);
pg.PdfPage.SetFontAndSize(font_obj, max_size);
bottom -= line_space;
@ -268,7 +268,7 @@ begin
begin
if r.FldChar.FldCharType = "begin" then break;
// TODO：officexml项目是否应该保留赋值接口，如何统一
r.RPr.Lang.Val := "zh-CN";
if ifnil(r.XmlChildRPr) then r.XmlChildRPr := new RPr();
// r.RPr := new RPr();
{self.}SetRPr(r.RPr, ppr_unit_decorator_);
r.RPr.Color.Val := nil;
@ -339,6 +339,7 @@ begin
if {self.}CheckAndAddPage({self.}EndY, max(line_space, range.DynamicHeight)) then
{self.}DynamicHeight += diff;
if_newline := {self.}EndX + range.Width - {self.}StartX > {self.}Width + 1e-6;
if i = 0 then if_newline := 0;
if if_newline and range.Width < {self.}Width then
begin
offset := line_space > max_y ? (line_space - max_size) / 2 + max_size - max_size / 5 : max_y;
@ -356,7 +357,10 @@ begin
{self.}EndY -= max_value;
{self.}EndX := {self.}StartX;
// w:hanging
sz := range.RPr.SzCs.Val ? range.RPr.SzCs.Val : range.RPr.Sz.Val ? range.RPr.Sz.Val : docx_to_pdf_.Font.GetDefaultSz();
if range is class(TSPdfTextRange) then
sz := range.RPr.SzCs.Val ? range.RPr.SzCs.Val : range.RPr.Sz.Val ? range.RPr.Sz.Val : docx_to_pdf_.Font.GetDefaultSz();
else
sz := docx_to_pdf_.Font.GetDefaultSz();
{self.}EndX -= ppr.Ind.HangingChars ? ppr.Ind.HangingChars * sz : ppr.Ind.Hanging;
continue;
end
@ -430,11 +434,13 @@ begin
while pos <= length(text) do
begin
num := {self.}GetUtf8CharLength(text[pos]);
word := text[pos : pos+num-1];
word := utf8ToAnsi(word);
a_word := text[pos : pos+num-1];
word := utf8ToAnsi(a_word);
if num <> 1 and word = "?" then
word := utf8ToAnsi(docx_to_pdf_.GetSymbol(a_word));
pos += num;
font_name := rpr.RFonts.EastAsia ? rpr.RFonts.EastAsia : rpr.RFonts.Ascii;
font_obj := docx_to_pdf_.Font.GetFont(font_name, rpr.B, rpr.I);
font_obj := docx_to_pdf_.Font.GetCNSFont(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();
page_.PdfPage.SetFontAndSize(font_obj, rpr.Sz.Val);
word_width := page_.PdfPage.TextWidth(word);
@ -474,6 +480,10 @@ begin
xfrm := new XfrmUnitDecorator(anchor.Graphic.GraphicData.Pic.SpPr.Xfrm);
position_h := new PositionHUnitDecorator(anchor.PositionH);
position_v := new PositionVUnitDecorator(anchor.PositionV);
if position_h.RelativeFrom = "paragraph" then
x := {self.}StartY;
if position_v.RelativeFrom = "paragraph" then
y := {self.}StartY;
image_range := new TSPdfImageRange();
image_range.Image := image;
image_range.EndX := x + position_h.PosOffset.Text;
@ -493,6 +503,10 @@ begin
xfrm := new XfrmUnitDecorator(wsp.SpPr.Xfrm);
position_h := new PositionHUnitDecorator(anchor.PositionH);
position_v := new PositionVUnitDecorator(anchor.PositionV);
if position_h.RelativeFrom = "paragraph" then
x := {self.}StartY;
if position_v.RelativeFrom = "paragraph" then
y := {self.}StartY;
x += position_h.PosOffset.Text;
y -= position_v.PosOffset.Text;
w := xfrm.Ext.CX;
@ -576,20 +590,11 @@ begin
data := image.Data();
image_path := docx_to_pdf_.GetCachePath(image_path);
writeFile(rwBinary(), "", image_path, 0, length(data), data);
image_type := extractFileExt(image_path);
if image_type = ".emf" then
begin
image_old_path := image_path;
image_path := replaceStr(image_path, "emf", "png");
gdipconvimagetype("", image_old_path, "", image_path, "image/png");
image_type := ".png";
fileDelete("", image_old_path);
end
image := nil;
case image_type of
".png":
case GetImageFileType(data) of
"png":
image := docx_to_pdf_.GetPdf().LoadPngImageFromFile("", image_path);
".jpg", ".jpeg":
"jpg":
image := docx_to_pdf_.GetPdf().LoadJpegImageFromFile("", image_path);
end;
fileDelete("", image_path);
@ -604,6 +609,7 @@ begin
def := array(
('name': 'png', 'position': 0, 'value': array(0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A)),
('name': 'jpg', 'position': 0, 'value': array(0xFF, 0xD8)),
('name': 'emf', 'position': 40, 'value': array(0x20, 0x45, 0x4d, 0x46)),
);
for i:=0 to length(def)-1 do
begin
@ -681,7 +687,7 @@ begin
based_on := style.BasedOn.Val;
{self.}SetRPrByTblStyleId(rpr, based_on);
rpr.Copy(style.RPr);
end;
end
if table_style_type_ then
begin
tbl_style_pr := docx_components_ware_.GetTblStylePrByType(table_style_id_, table_style_type_);
@ -722,18 +728,6 @@ begin
{self.}Width -= ppr.Ind.RightChars ? ppr.Ind.RightChars * sz : ppr.Ind.Right;
end;
function TSPdfParagraphRange.FirstValidTSPage(): TSPage;
begin
range := range_array_[0];
return ifObj(range) ? range.TSPage : page_;
end;
function TSPdfParagraphRange.IsSamePage(): boolean;
begin
pg := range_array_[0] ? range.TSPage : {self.}TSPage;
return page_ = pg;
end;
function TSPdfParagraphRange.GetLastPage(): TSPage;
begin
return page_;

60
ware/TSFontWare.tsf
View File

@ -1,7 +1,9 @@
type TSFontWare = class
public
function Create(pdf: PdfFile);
function GetFont(name: string; bold: boolean; italic: boolean);
function GetFontByText(text: string; name: string; bold: boolean; italic: boolean);
function GetAsciiFont(name: string; bold: boolean; italic: boolean);
function GetCNSFont(name: string; bold: boolean; italic: boolean);
function UseExternalFont();
function SetSubstitutionRules(source: string; target: string);
function SetDefaultSz(value: real);
@ -9,7 +11,8 @@ public
private
function GetExternalFont(name:string; bold: boolean; italic: boolean);
function GetBuiltInFont(name:string; bold: boolean; italic: boolean);
function GetBuiltInCNSFont(name:string; bold: boolean; italic: boolean);
function GetBuiltInAsciiFont(name:string; bold: boolean; italic: boolean);
private
[weakref]pdf_: PdfFile;
@ -25,7 +28,12 @@ function TSFontWare.Create(pdf: PdfFile);
begin
pdf_ := pdf;
use_built_in_font_ := true;
substitution_rules_ := array("宋体": "SimSun", "黑体": "SimHei");
substitution_rules_ := array("宋体": "SimSun",
"黑体": "SimHei",
"Courier New": "Courier",
"Helvetica": "Helvetica",
"Times New Roman": "Times-Roman",
);
external_reference_ := array();
default_sz_ := 10.5;
end;
@ -63,12 +71,21 @@ begin
return font;
end;
function TSFontWare.GetFont(name: string; bold: boolean; italic: boolean);
function TSFontWare.GetFontByText(text: string; name: string; bold: boolean; italic: boolean);
begin
return use_built_in_font_ ? {self.}GetBuiltInFont(name, bold, italic) : {self.}GetExternalFont(name, bold, italic);
len := length(text);
if len > 1 then
return {self.}GetCNSFont(name, bold, italic);
else
return {self.}GetAsciiFont(name, bold, italic);
end;
function TSFontWare.GetBuiltInFont(name: string; bold: boolean; italic: boolean);
function TSFontWare.GetCNSFont(name: string; bold: boolean; italic: boolean);
begin
return use_built_in_font_ ? {self.}GetBuiltInCNSFont(name, bold, italic) : {self.}GetExternalFont(name, bold, italic);
end;
function TSFontWare.GetBuiltInCNSFont(name: string; bold: boolean; italic: boolean);
begin
font_name := substitution_rules_[name];
if ifnil(font_name) then font_name := "SimSun";
@ -82,6 +99,37 @@ begin
return font;
end;
function TSFontWare.GetAsciiFont(name: string; bold: boolean; italic: boolean);
begin
return use_built_in_font_ ? {self.}GetBuiltInAsciiFont(name, bold, italic) : {self.}GetExternalFont(name, bold, italic);
end;
function TSFontWare.GetBuiltInAsciiFont(name:string; bold: boolean; italic: boolean);
begin
font_name := substitution_rules_[name];
if ifnil(font_name) then font_name := "Times-Roman";
if font_name = "Courier" or font_name = "Helvetica" then
begin
if bold and italic then
font_name += "-BoldOblique";
else if bold then
font_name += "-Bold";
else if italic then
font_name += "-Oblique";
end
else if font_name = "Times-Roman" then
begin
if bold and italic then
font_name := "Times-BoldItalic";
else if bold then
font_name += "Times-Bold";
else if italic then
font_name += "Times-Italic";
end
font := pdf_.GetFont(font_name, "");
return font;
end;
function TSFontWare.SetSubstitutionRules(source: string; target: string);
begin
substitution_rules_[source] := target;