26 Replies Latest reply: Feb 10, 2014 2:37 PM by Arnis Gubins RSS

    FrameMaker 10 Corrupts Searchable Text in SVG Files

    Jason L-S Community Member

      I notice that a font issue occurs with SVG files in FrameMaker 10.0.2.419 saved as PDF.

       

      1. Open a FrameMaker document that contains SVG images. The SVG images contain text. Text is in Arial font.

      2. Save FrameMaker file as PDF.

      3. Search PDF file for instances of words contained in SVG images. Text not found. Text is rendered correctly. But upon inspection, when copied and pasted into NotePad, text is rendered as squares:       .  Acrobat properties show that the original font is ArialMT.

       

      Things I already tried:

      Font embedding settings in Distiller. No effect.

      Using File > Print in FrameMaker instead of Save as PDF. No effect.

      Using a different version of Distiller. No effect.

      Using a printer instead of Distiller and printing a post-script file. No effect.

       

      It seems instead to be an issue with the FrameMaker patch that updated the version to 10.0.2.419. Prior to that patch, things seemed to work correctly. And things seem to work correctly in FrameMaker 8.

       

      Thanks,

       

      Jason

        • 1. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
          Arnis Gubins CommunityMVP

          Jason,

           

          Do you have the Adobe PDF printer instance as your default printer?

           

          Do you actually have the Arial font installed and told Distiller where it's located (Settings > Font Locations)?

           

          Exacly which joboptions have you used?

           

          Have you explicitly set the ArialMT font to the "Never Embed" options list?

           

          Did you actually multiple versions of Distiller on the same machine?? This can result in a totally hosed installation for creating PDFs.

          • 2. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
            Jason L-S Community Member

            Arnis,

             

            See my inline replies below:

             

            Do you have the Adobe PDF printer instance as your default printer?

            --Yes

             

            Do you actually have the Arial font installed and told Distiller where it's located (Settings > Font Locations)?

            --Yes I have Arial in C:\Windows\Fonts. But I can't find "ArialMT."

             

            Exacly which joboptions have you used?

            --Standard

             

            Have you explicitly set the ArialMT font to the "Never Embed" options list?

            --Yes. I've tried both. Never Embed and Always Embed. Neither works.

             

            Did you actually multiple versions of Distiller on the same machine?? This can result in a totally hosed installation for creating PDFs.

            --No. I used two different machines.

             

            Thanks for your help.

             

            Jason

            • 3. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
              Mike Wickham Community Member

              Have you run Windows update recently? A couple of months back, there was

              a Windows update that broke some PostScript fonts. Arial MT is the PS

              version of Arial. Microsoft has since released a fixed version of the

              update. Perhaps you didn't download it yet.

               

              You might also look to see if you've chosen the right font. On my

              Windows 7 computer, the TrueType version of the font shows as Arial and

              the PostScript version shows as Arial MT. So if you only see Arial

              installed in Windows\Fonts, you probably don't have Arial MT installed.

              So it can't embed in a PDF.

               

              Also, are you sure the text in your SWF is actually a font and not

              glyphs that have been converted to vector outlines? If they are

              outlines, they are graphics, not text, and cannot be searched.

              • 4. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                Jason L-S Community Member

                Mike,

                 

                Thanks for your consideration. See my comments below:

                 

                Have you run Windows update recently?

                • I searched for what you might have been talking about and found a discussion about KB2753842. I uninstalled it. No luck. I reinstalled it with the newest fix. No luck. This isn't the issue.

                 

                You might also look to see if you've chosen the right font.

                • I have looked in the Distiller Font settings. ArialMT is not in the C:\Windows\Fonts list as an installed font. However, I now think this might not be the issue. I have tried altering the source Visio file to use a font that I know is installed (e.g., Tahoma). I resaved the drawing as an SVG file, imported that drawing to FrameMaker, and then saved the FrameMaker file as a PDF file. The same thing happens. The font appears to be fine on screen. But if I search for the text, I can't find it. If I copy the text and paste it, the text is shown as squares. However this time I notice that if I use the Tools > Advanced Editing > Touch Up Text command in Acrobat Pro, I see that the font is embedded. The Original Font is the same as the Editing Font, and I can copy and paste the text sucessfully with that tool (the touch-up text tool). Is this some kind of security setting in Acrobat?

                 

                Also, are you sure the text in your SWF is actually a font and not

                glyphs

                • Yes. I'm sure of that. It's not glyphs.

                 

                Thanks!

                 

                Jason

                • 5. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                  Mike Wickham Community Member

                  It's probably late of me to say this, but I'm not even positive that text within an SVG would be searchable in a PDF. I don't use SVG graphics and haven't tried such a search. So it's possible you are trying to do something that can't be done. Someone else will have to give a definitive answer on that. There's a chance that text within a vector graphic isn't treated as searchable text.

                  Jason L-S wrote:

                   

                  If I copy the text and paste it, the text is shown as squares. However this time I notice that if I use the Tools > Advanced Editing > Touch Up Text command in Acrobat Pro, I see that the font is embedded. The Original Font is the same as the Editing Font, and I can copy and paste the text sucessfully with that tool (the touch-up text tool). Is this some kind of security setting in Acrobat?

                  I have no idea. Is it possible that your version of Acrobat got hosed, possibly by multiple installations? Adobe warns against having various combinations of Adobe Reader, Acrobat, or the PDF Creation Add-on that comes with FrameMaker on the same computer. For example, you shouldn't have Acrobat and the PDF Creation add-on on the same machine. Nor should you have Reader and Acrobat on the same computer. There might also be issues of using one version of Reader with another version of the PDF Creation add-on. Installing these combos can cause huge problems with the functionality of Acrobat-- especially, if differing versions were installed (such as, Adobe Reader 10, installed over Acrobat 9). So if you have, or have had this situation on your computer, you may need to uninstall Reader, Acrobat, and the PDF add-on (as many as are there), reboot, then reinstall only Acrobat to fix things up.

                  • 6. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                    Arnis Gubins CommunityMVP

                    Jason,

                     

                    If you're using VISIO to create your graphics, save them as PDFs from there and then import the PDF into FM. It will be much more relieable. Visio isn't very stellar at saving to other formats, but outputting to a PDF is quite solid.

                    • 7. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                      Jason L-S Community Member

                      Arnis,

                       

                      We cannot change our processes. SVG is the corporate standard. Thanks. Jason

                      • 8. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                        Error7103 Community Member

                        If you have a copy of Adobe Illustrator anywhere in the shop, you might try opening one of these SVGs, examining it (fontwise) and re-saving it (as SVG).

                         

                        If you don't have a copy, you could download a trial, and that would let you know if acquiring it is worthwhile.

                         

                        We have a vaguely similar problem with PDF imports, which are unstable in our production FM7.x environment, so we batch process PDF imports in Illustrator, re-saving as EPS.

                        • 9. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                          Jason L-S Community Member

                          Mike,

                           

                          What did you mean by:

                          Also, are you sure the text in your SWF is actually a font and not

                          glyphs

                           

                          I am using SVG files, not SWF files.

                           

                          What exactly is a Glyph? I am looking at the post-script code for this and I see some funky stuff in there about a Glyph library.

                           

                          Thanks,

                           

                          Jason

                          • 10. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                            Mike Wickham Community Member

                            Sorry about the SVG/SWF typo. And it looks like I slightly botched the

                            glyph term, too. I meant to say an "outlined" glyph. (A glyph is a

                            single character in a font, such as a letter, number, or punctuation

                            mark.) Programs that create vector graphics usually give you the option

                            of turning the text glyphs into vector shapes called outlines. (In

                            Illustrator you do this using Type> Create Outlines.)

                             

                            Why would a person convert text to outlines in their graphics? Doing so

                            keeps the file size small by eliminating the need to embed fonts in the

                            vector graphic-- or requiring the person viewing the file to have the

                            fonts installed on his computer.

                             

                            If the text glyphs in your vector graphics have been converted to

                            outlines, they will look exactly the same to you-- font characters are

                            vector drawings, too-- but they will no longer be part of a font and no

                            longer considered to be text. They are now graphics, so they will no

                            longer be searchable as text.

                             

                            If you load your graphic into the program that created it, you can click

                            on the text to see if each letter has a separate path (outlined), or if

                            the groups of characters act as normal lines of typed text.

                             

                            Of course, like I said before, I don't use SVG graphics, so don't know

                            if true text is searchable in them or not.

                            • 11. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                              Arnis Gubins CommunityMVP

                              I did a quick test with FM8 through FM11 with a small SVG and the results are identical in all cases, i.e. the text cannot be searched for and not copied.

                               

                              Looking into this some more, your problem is actually more widespread than you think. The text in the SVG is getting CID Identiy-H encoded, which makes it virtually impossible to search for or extract from a PDF. The content is stored as octal  character codes that has to be deciphered through a toUnicode look-up table. If the table isn't passed into the PDF, the screen may show the correct letter (as the embedded font is subset to match the characters used in the text) , but Acrobat has no idea what the letter actually is - it's just drawing the "glyph" present in the subset font. Consequently, trying to copy & paste this results in a set of unknown markers (either question marks, squares or what have you in your text editor) depending upon the encoding that the editor is set for.

                               

                              Now the question is whether the problem is coming from FM or from the application (Visio in your case?) that is creating the SVG? Note: googling around indicates this is also happening with SVGs in InDesign (older versions, mind you, but it does use a completely different rendering engine for creating PDFs than FM) and  other apps, such as Xara & Quark.

                               

                              If you are using Visio to create your SVG files, which version of Visio? There are a lot of font-related "issues" in older versions of Visio. Do you have any options as to how the fonts are embedded and encoded into the SVG with the Visio export options? Has your Visio been upgraded lately?

                               

                              If, as you say this was working with older files in FM8, check out some of those same older "working" SVG files in FM10 and see if they still behave as they did before. If they do, then that kind of points the finger at the creating application for the SVGs. Something changed, but as far as I can tell when using SaveAsPDF (for RGB), or printing directly to PDF or printing to File and then manualy invoking PDF, then same process is still in use in FM (creating CMYK PDFs is a completely different process, but it did produce CID encoded fonts as well).

                               

                              Even though SVG may be your corporate standard, if the tools in the workflow have problems dealing with the format, you might need to re-evaluate what your using and why.

                              • 12. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                Jason L-S Community Member

                                Arnis,

                                 

                                You wrote, "I did a quick test with FM8 through FM11 with a small SVG and the results are identical in all cases, i.e. the text cannot be searched for and not copied."

                                 

                                This is not the same result I get. I did the same thing and found that the text from FM8 is searchable. I have some attachments I'd like to give you to see if you can duplicate my results. How can we do that?

                                 

                                Thanks,

                                 

                                Jason

                                • 13. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                  Arnis Gubins CommunityMVP

                                  I'll send you my contact info and we can take this offline for a bit.

                                  • 14. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                    Jason L-S Community Member

                                    Arnis,

                                     

                                    Thanks for your offline replies. To summarize, we found that it's either the CoolType engine or the SVG import filter. One of those is converting the fonts to the CID Identity-H format and rendering it unsearchable in the PDF file. In FM8, it depends upon the font specified as to whether or not it gets converted to the CID Identity-H format.

                                     

                                    Font TypeFM8FM10
                                    TTFOKCID
                                    OTFCIDCID
                                    Type 1 PSFailsCID

                                     

                                    So the newer versions of FM always convert SVG fonts to CID, while older version does not convert TrueType fonts to CID. That's why FM8 appeared to handle SVG-text conversions correctly.

                                     

                                    I took it upon myself to attempt swapping out the CoolType.dll file to see if an older version would work with FrameMaker 10. But this did not work, and depending on the .dll version, caused FrameMaker to crash.

                                     

                                    We talked about a potential solution with ExtendScripts where we might create a script that extracts the text strings from the SVG, which is a simple XML format with the text is clearly delineated-- so perhaps an XSLT to get the strings externally first-- puts the text strings in a look-up table associated with the svg filename and then in the documents adds those text strings (using the invisible colour) to the anchored frames containing the appropriate svg. But this seems like a big project to me. I don't think I could spare the time.

                                     

                                    We suggest (for now) to use another graphic file format OR re-enter the required text using either the Text tool or a text frame overtop of the graphic (use a custom colour so you can see it and then before generating the PDF, redefine the colour to render as white so it's invisible in the PDF). That way, a search in the PDF will stop at the graphic.

                                     

                                    We are reporting this as a bug.

                                     

                                    Jason

                                    • 15. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                      Error7103 Community Member

                                      > We cannot change our processes. SVG is the corporate standard.

                                       

                                      Your more flexible competitors probably appreciate that.

                                       

                                      > ... we might create a script ...

                                       

                                      Adobe Illustrator can be set to batch convert SVG to EPS or PDF, no extra charge. Font-based text in EPS survives as font-based text into PDF and is full-text searchable. Vectors in the SVG presumably remain vector as well.

                                       

                                      > We suggest (for now) to use another graphic file format ...

                                       

                                      I'd suggest EPS, as FM has to filter PDF imports into that anyway.

                                      • 16. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                        dkcorreia4 Community Member

                                        I have just come across this issue today and we use Adobe FrameMaker 10.  We had 1 user recently report that ALL the SVG graphic text inside the PDF was a jumbled mess.  This is when we discovered we cannot search or copy/paste the text.  We had used Arial font and then did a test using Arial Unicode MS and now the user can see the text but the search and copy/paste issue remains.  This is important functionality that we really would like back.  We have used Frame for years and our company produces thousands of docs and searchability in graphics is important.  We also had problems with greek symbols such as (ohm) not showing up at all in the PDF.  Please advise.  Thanks

                                        • 17. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                          Arnis Gubins CommunityMVP

                                          As reported here, text (specifically it is a font subsetting issue) in SVGs is not handled properly in the current versions of FM.

                                           

                                          Your only options are either to possibly wait for the next release of the product (and hope that this has been addressed) or convert your current SVGs to a more FM ammenable format such as EPS.

                                          • 18. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                            jlangkamer Community Member

                                            You can report this as a bug. If you do, it may help elevate the status of this issue so developers actually pay attention.

                                             

                                            https://www.adobe.com/cfusion/mmform/index.cfm?name=wishform

                                            • 19. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                              Error7103 Community Member

                                              > We also had problems with greek symbols such as (ohm) not showing up at all in the PDF.

                                               

                                              How are you implementing ohm in the text?

                                               

                                              • As one of the (2) Unicode code points (\u03A9, \u2126) in a font that actually populates the chosen code point.
                                              • Or as the legacy \x57 with a codepage Symbol font applied via Character Format or override.
                                              • Or, now for something, completely different.
                                              • 20. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                dkcorreia4 Community Member

                                                These are symbols that were done in Visio and looked fine but once into Frame they no longer showed up correctly.  Thanks for any help.

                                                • 21. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                  dkcorreia4 Community Member

                                                  The jumbled font issue above was resolved by recreating the PDF (not the copy paste or search but that is lower priority).  SInce then we have had two customers report something similar.  It seems very random and It is nearly impossible to test because we cannot replicate the issue.  Does a bug exist in a certain version of Adobe Reader? I searched and could not find a definitive anwser.  As for the known bug in Frame.  Any progress in a fix?

                                                  • 22. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                    Arnis Gubins CommunityMVP

                                                    With the next FM release announced for next year (see: http://blogs.adobe.com/techcomm/2013/09/adobe-technical-communication-products-next-releas e-plan.html) , I highly doubt that Adobe will be issuing any further patches based upon past behaviours.

                                                    • 23. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                      ScriptoriumPub Community Member

                                                      Has this been fixed in version 12? Does anyone know?

                                                      • 24. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                        Arnis Gubins CommunityMVP

                                                        Hi Sarah,

                                                         

                                                        This ended up on the "Deferred" list in the final stages of the pre-release. I guess engineering considers that "seeing" the text in the PDF is good enough. [Why would anybody want to search for it...    ]

                                                        • 25. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                          dkcorreia4 Community Member

                                                          Thanks for the update.  This bug has far reaching effects and should would be addressed.  This glyph issue also corrupts the text when viewing in the FireFox PDF viewer.  This has been noticed by several customers.  When we did our testing of SVG graphics back in version 7 this was not an issue and went ahead with this format ever since.  Now thousands of graphics are in SVG and this is becoming increasingly problematic.  So not only did we lose the search functionality in PDF but actual viewability is is being compromised.  Our company has 100+ licenses of FrameMaker.  If you would like to see an example open this PDF if the Firefox default viewer.

                                                           

                                                          corrupt.PNG

                                                          • 26. Re: FrameMaker 10 Corrupts Searchable Text in SVG Files
                                                            Arnis Gubins CommunityMVP

                                                            The best way to get this issue addressed is to send emails to the product management team outlining how detrimental this bug is to your organization and how many licenses are at risk...

                                                             

                                                            TechComm Suite Sr. product manager: Kapil Verrma = kverma at adobe.com

                                                            FrameMaker product manager: Vikram Verma = vikverma at adobe.com

                                                            TechComm Sr. marketing manager: Saibal Bhattacharjee = saibal at adobe.com

                                                            TechComm evangelist: Maxwell Hoffmann = mhoffman at adobe.com

                                                             

                                                            FWIW, I looked at the PDF in Chrome and it looks fine. You might want to investigate if there are limitations/issues with the FireFox PDF viewer.