[gs-commits] ghostpdl branch, master, updated. jbig2dec-0.14-1860-g33f4f0c
[email protected] (Ken Sharp) Thu, 14 Nov 2019 15:25:15 +0000 (UTC)
| Newsgroups | gmane.comp.printing.ghostscript.cvs |
|---|---|
| Message-ID | <[email protected]> |
The ghostpdl branch, master has been updated
via 33f4f0c0ba5449fd535061e785090bc57849ae51 (commit)
from 98a8b291b135d49ffca23d9a289786971cf33a12 (commit)
----------------------------------------------------------------------
commit 33f4f0c0ba5449fd535061e785090bc57849ae51
Author: Ken Sharp <[email protected]>
Date: Thu Nov 14 11:20:40 2019 +0000
PDF interpreter - don't ignore SUBSTFONT
The PDF interpreter attempts to find a 'better' match for missing fonts
by using a list of TrueType mappings (eg ArialMT, CourierNew) and a
list of mappings (gs_font.ps, /.substitutefaces)
Failing to find a match there it attempts to use the FontDescriptor
Flags if present. As far as I can see that basically means we use
Times-Roman for serif fonts and Helvetica for sans serif.
This commit checks to see if SUBSTFONT is set. If it is we ignore all
this nonsense and simply use the substitute font the user told us to.
Otherwise we continue as before.
The diff is actually very simple, but the routine was very large so
I decided to move the actual substitution code out into a separate
function, to simplify the logic in pdffindfont which makes the
differences appear substantial. All that has really changed is that the
code to do the substitution, after checking for the existence of a
FontDescriptor, checks to see if SUBSTFONT is defined. If it is we
behave as if the font had no FontDescriptor and use the normal font
substitution machinery.
diff --git a/Resource/Init/pdf_font.ps b/Resource/Init/pdf_font.ps
index 8638904..60deb64 100644
--- a/Resource/Init/pdf_font.ps
+++ b/Resource/Init/pdf_font.ps
@@ -857,176 +857,222 @@ setglobal
/.pdfdfndict mark
/defaultfontname /Helvetica
.dicttomark readonly def
+
+% This code attempts to use the FontName and FontDescriptor to find a
+% 'better' match for a missing font than the default font. There
+% are a list of fonts and mappings here (TTfonts) and in gs_font.ps
+% (.substitutefaces) which we can use to map from a requested font
+% family name to a substitute family. If that fails we check the
+% FontDescriptor Flags to attempt a (very slightly) better match.
+% We use Times-Roman for serif fonts and Helvetica for non-serif.
+% This all seems pretty poor to me, but its long-standing, so we
+% won't attempt to meddle with it.
+/pdfsubstitutefont {
+ % Stack: font-res fontname fontdesc
+
+ dup /Flags oget
+ dup 16#40 and -6 bitshift % 1, oblique/italic
+ 1 index 16#40000 and -17 bitshift add % 2, bold
+ exch 16#2 and 2 bitshift add % 8, serif
+
+ % We should look at the fixed flag, too.
+ % Stack: font-res fontname fontdesc properties
+
+ % Even though /FontName is a required key in FontDescriptor dict
+ % (As of the PDF 1.4 Reference Manual), In the case of missing
+ % /FontName key, we substitue /BaseFont for the value of /FontName.
+ % Yet another case of broken PDF's that Adobe Reader accepts.
+ 1 index dup /FontName known
+ {
+ /FontName oget
+ dup type /nametype ne
+ {
+ ( **** Error: /FontName in FontDescriptor is not a name.\n)
+ pdfformaterror
+ ( Output may be incorrect.\n) pdfformaterror
+ cvn
+ } if
+ }
+ {
+ ( **** Error: FontDescriptor missing required /FontName key. BaseFont name used.\n)
+ pdfformaterror
+ ( Output may be incorrect.\n) pdfformaterror
+ pop 2 index % grab the BaseFont from the stack.
+ } ifelse
+ .remove_font_name_prefix
+ exch
+
+ % Analyzes font name and extract "Bold" and "Narrow" properties
+ % which are not described by the FontDescriptor Flags.
+ % We also allow the font name analysis to override the Flags setting
+ % for Italic/Oblique as this gives us results more consistent with
+ % Acrobat.
+ 0 2 index //.fontnameproperties exec 7 and or
+
+ % Rebind the default font name to Helvetica so that
+ % fonts with no properties are handled correctly.
+ //.pdfdfndict begin .substitutefontname end
+
+ % Stack: font-res fontname fontdesc substname|null
+ Fontmap 1 index known
+ {//false}
+ { .buildnativefontmap pop NativeFontmap 1 index known not}
+ ifelse
+
+ {
+ % No available good substitution, use the standard one.
+ pop 1 index .substitutefont
+ } if
+
+ dup 3 index ne
+ {
+ QUIET not
+ {
+ (Substituting font ) print dup =only
+ ( for ) print 2 index =only (.) = flush
+ } if
+ % Send a warning to the device, in case its pdfwrite and we are in PDF/A mode
+ % In that case the substituted font may not (probably doesn't) have glyphs which
+ % match the /Widths of the original font, and that will cause a fault with PDF/A
+ % so we need to let the device know.
+ /SubstitutedFont /EventInfo .special_op
+ } if
+
+ pdffindcachedfont
+ % Stack: font-res fontname fontdesc font
+
+ % Some non-compliant files are missing FirstChar/LastChar,
+ % despite referencing a non-base14 font
+ 3 index /FirstChar knownoget
+ {
+ 0 % push an initial value for accumating the Widths
+ 0 string % and an empty string to interrogate the font
+ 7 index /Widths knownoget
+ {
+ 0 1 2 index length 1 sub
+ {
+ dup 2 index exch get
+ dup type /packedarraytype eq {exec} if % Handle entries which are indirect references (seriously ? !!)
+ dup 0 gt % We ignore entries of zero in the widths array
+ {
+ exch 5 index add % add FirstChar to the idx to get a char code
+ dup 64 gt 1 index 91 lt and % we only want to consider A~Z and....
+ 1 index 96 gt 2 index 123 lt and or % ... a~z
+ {
+ exch
+ 5 -1 roll add 4 1 roll % add the width to the accumulator on the stack
+ 1 string dup 0 4 -1 roll put
+ 3 -1 roll concatstrings exch % add the char code to the string
+ }
+ {
+ pop pop
+ } ifelse
+ }
+ {
+ pop pop
+ } ifelse
+ } for
+ pop % get rid of the Widths array
+ 3 -1 roll pop % get rid of the FirstChar value
+ dup length dup 0 gt
+ {
+ 3 -1 roll exch div % calculate an average width from the Widths array
+ gsave
+ 2 index 10 scalefont setfont
+ exch dup length exch stringwidth
+ grestore
+ pop 100 mul exch div div
+ % Only make the font smaller/narrower, not larger/wider
+ dup 1.0 lt
+ {
+ 0 0 2 index 0 0 6 array astore
+ exch //true .copyfontdict
+ dup /FontMatrix get 3 -1 roll matrix concatmatrix
+ exch dup 3 -1 roll /FontMatrix exch put
+ % we don't need to definfont here, we can leave that to .completefont below
+ }
+ {
+ pop
+ } ifelse
+ }
+ {
+ % no suitable Widths entries - carry on and hope for the best......
+ pop pop pop
+ } ifelse
+ }
+ {
+ % Broken file: no Widths array for non-base14 font - carry on and hope for the best......
+ ( **** Warning: Widths array missing in FontDescriptor for non-base14 font\n)
+ pdfformatwarning
+ pop pop pop
+ } ifelse
+ }
+ {
+ ( **** Warning: FirstChar value missing in FontDescriptor for non-base14 font\n)
+ pdfformatwarning
+ } ifelse
+ % Stack: font-res fontname fontdesc font
+
+ % If this is a small-caps font, replace the CharString
+ % entries for a..z.
+ exch /Flags oget 16#20000 and 0 ne
+ {
+ //true .copyfontdict
+ dup /CharStrings 2 copy get dup length dict .copydict
+ % stack: font-res fontname font font /CharStrings CharStringsdict
+ 5 index /FirstChar get 97 .max
+ 6 index /LastChar get 122 .min 1 exch
+ {
+
+ % Stack: font-res fontname font' font' /CharStrings charstrings code
+ % Note that this only remaps a-z, not accented characters.
+ 6 index /Widths oget 1 index 8 index /FirstChar get sub oget
+ 1 string dup 0 5 -1 roll put
+
+ % Stack: font-res font' font' /CharStrings charstrings code
+ % width (x)
+ 2 index exch dup cvn exch
+ dup 0 2 copy get 32 sub put 4 -1 roll
+ {
+ % Stack: operand (X) width
+ 0 setcharwidth exch pop
+ currentfont /FontMatrix get matrix invertmatrix concat
+ 0.7 dup scale 0 0 moveto show
+ } /exec cvx 4 packedarray cvx put
+ } for put
+ } if
+
+ dup /FontName get 2 index ne
+ {
+ //true .copyfontdict
+ 2 copy exch /FontName exch put
+ } if
+ exch pop .completefont
+} bind executeonly def
+
/pdffindfont { % <font-resource> <fontname> pdffindfont <font>
- % If the font isn't available, synthesize one based on
- % its descriptor.
- dup /Font resourcestatus {
+ % If the font isn't available, synthesize one based on
+ % its descriptor.
+ dup /Font resourcestatus
+ {
pop pop pdffindcachedfont
- } {
+ }
+ {
1 index /FontDescriptor knownoget {
- % Stack: font-res fontname fontdesc
- dup /Flags oget
- dup 16#40 and -6 bitshift % 1, oblique/italic
- 1 index 16#40000 and -17 bitshift add % 2, bold
- exch 16#2 and 2 bitshift add % 8, serif
- % We should look at the fixed flag, too.
- % Stack: font-res fontname fontdesc properties
-
- % Even though /FontName is a required key in FontDescriptor dict
- % (As of the PDF 1.4 Reference Manual), In the case of missing
- % /FontName key, we substitue /BaseFont for the value of /FontName.
- % Yet another case of broken PDF's that Adobe Reader accepts.
- 1 index dup /FontName known {
- /FontName oget
- dup type /nametype ne {
- ( **** Error: /FontName in FontDescriptor is not a name.\n)
- pdfformaterror
- ( Output may be incorrect.\n) pdfformaterror
- cvn
- } if
- } {
- ( **** Error: FontDescriptor missing required /FontName key. BaseFont name used.\n)
- pdfformaterror
- ( Output may be incorrect.\n) pdfformaterror
- pop 2 index % grab the BaseFont from the stack.
- } ifelse
- .remove_font_name_prefix
- exch
- % Analyzes font name and extract "Bold" and "Narrow" properties
- % which are not described by the FontDescriptor Flags.
- % We also allow the font name analysis to override the Flags setting
- % for Italic/Oblique as this gives us results more consistent with
- % Acrobat.
- 0 2 index //.fontnameproperties exec 7 and or
- % Rebind the default font name to Helvetica so that
- % fonts with no properties are handled correctly.
- //.pdfdfndict begin .substitutefontname end
- % Stack: font-res fontname fontdesc substname|null
- Fontmap 1 index known
- {//false}
- { .buildnativefontmap pop NativeFontmap 1 index known not} ifelse
-
+ /SUBSTFONT where not
{
- % No available good substitution, use the standard one.
- pop 1 index .substitutefont
- } if
- dup 3 index ne {
- QUIET not {
- (Substituting font ) print dup =only
- ( for ) print 2 index =only (.) = flush
- } if
- % Send a warning to the device, in case its pdfwrite and we are in PDF/A mode
- % In that case the substituted font may not (probably doesn't) have glyphs which
- % match the /Widths of the original font, and that will cause a fault with PDF/A
- % so we need to let the device know.
- /SubstitutedFont /EventInfo .special_op
- } if
- pdffindcachedfont
- % Stack: font-res fontname fontdesc font
-
- % Some non-compliant files are missing FirstChar/LastChar,
- % despite referencing a non-base14 font
- 3 index /FirstChar knownoget
- {
- 0 % push an initial value for accumating the Widths
- 0 string % and an empty string to interrogate the font
- 7 index /Widths knownoget
- {
- 0 1 2 index length 1 sub
- {
- dup 2 index exch get
- dup type /packedarraytype eq {exec} if % Handle entries which are indirect references (seriously ? !!)
- dup 0 gt % We ignore entries of zero in the widths array
- {
- exch 5 index add % add FirstChar to the idx to get a char code
- dup 64 gt 1 index 91 lt and % we only want to consider A~Z and....
- 1 index 96 gt 2 index 123 lt and or % ... a~z
- {
- exch
- 5 -1 roll add 4 1 roll % add the width to the accumulator on the stack
- 1 string dup 0 4 -1 roll put
- 3 -1 roll concatstrings exch % add the char code to the string
- }
- {
- pop pop
- } ifelse
- }
- {
- pop pop
- } ifelse
- } for
- pop % get rid of the Widths array
- 3 -1 roll pop % get rid of the FirstChar value
- dup length dup 0 gt
- {
- 3 -1 roll exch div % calculate an average width from the Widths array
-
- gsave
- 2 index 10 scalefont setfont
- exch dup length exch stringwidth
- grestore
- pop 100 mul exch div div
- % Only make the font smaller/narrower, not larger/wider
- dup 1.0 lt
- {
- 0 0 2 index 0 0 6 array astore
- exch //true .copyfontdict
- dup /FontMatrix get 3 -1 roll matrix concatmatrix
- exch dup 3 -1 roll /FontMatrix exch put
- % we don't need to definfont here, we can leave that to .completefont below
- }
- {
- pop
- } ifelse
- }
- {
- % no suitable Widths entries - carry on and hope for the best......
- pop pop pop
- } ifelse
- }
- {
- % Broken file: no Widths array for non-base14 font - carry on and hope for the best......
- ( **** Warning: Widths array missing in FontDescriptor for non-base14 font\n)
- pdfformatwarning
- pop pop pop
- } ifelse
+ pdfsubstitutefont
}
{
- ( **** Warning: FirstChar value missing in FontDescriptor for non-base14 font\n)
- pdfformatwarning
- } ifelse
- % Stack: font-res fontname fontdesc font
- % If this is a small-caps font, replace the CharString
- % entries for a..z.
-
- exch /Flags oget 16#20000 and 0 ne {
- //true .copyfontdict
- dup /CharStrings 2 copy get dup length dict .copydict
- % stack: font-res fontname font font /CharStrings CharStringsdict
- 5 index /FirstChar get 97 .max
- 6 index /LastChar get 122 .min 1 exch {
- % Stack: font-res fontname font' font' /CharStrings charstrings code
- % Note that this only remaps a-z, not accented characters.
- 6 index /Widths oget 1 index 8 index /FirstChar get sub oget
- 1 string dup 0 5 -1 roll put
- % Stack: font-res font' font' /CharStrings charstrings code
- % width (x)
- 2 index exch dup cvn exch
- dup 0 2 copy get 32 sub put 4 -1 roll {
- % Stack: operand (X) width
- 0 setcharwidth exch pop
- currentfont /FontMatrix get matrix invertmatrix concat
- 0.7 dup scale 0 0 moveto show
- } /exec cvx 4 packedarray cvx put
- } for put
- } if
- dup /FontName get 2 index ne {
- //true .copyfontdict
- 2 copy exch /FontName exch put
- } if
- exch pop .completefont
- } {
- % No descriptor available, use the default algorithm.
+ % User has defined SUBSTFONT, use the defined substitute font, do not
+ % try to be clever
+ pop % the dictionary containing SUBSTFONT
+ pop % the FontDescriptor
+ pdffindcachedfont
+ }ifelse
+ }
+ {
+ % No descriptor available, use the default algorithm.
pdffindcachedfont
} ifelse
} ifelse
Summary of changes:
Resource/Init/pdf_font.ps | 358 ++++++++++++++++++++++++++--------------------
1 file changed, 202 insertions(+), 156 deletions(-)