How to manually install PDF tools correctly on Windows?

Ulrich Windl

Having followed https://www.zotero.org/support/pdf_fulltext_indexing I installed pdftotext-Win32.exe and pdfinfo-Win32.exe successfully in addition to pdftotext-Win32.version and pdfinfo-Win32.version.
Unfortunately When starting PDF indexing the status changes to "Unknown" shortly after I see a "DOS Box" for an instant. Also the Version is displayed as "Unknown" in the settings dialog.
Maybe the documentation should be more explicit what is supposed to be in the versions files. And if something goes wrong with indexing, a more descriptive message should appear. Help, please!
Using latest Zotero Standalone (as of yesterday) and xpdf tools (3.03 from today)...

aurimas

So you compiled the binaries yourself? Since the instruction page does not mention specific modifications to the source that are needed to work with Zotero and the link to modified source is dead (I agree that we need to fix the instructions), I assume that you compiled unmodified xpdf tools, which is why this is not working (the Command Prompt box flashing is a strong indication of this happening).

You can either download the binaries from zotero.org or compile from modified source code (I'll have to look for links for both). In either case, May I first ask why you are going for the manual installation steps? Is something not working with the automatic method? Do you get an error?

Ulrich Windl

I downloaded xpdf tools from the xpds site as binaries. Otherwiese I would have said that I compiled them myself.
Yes I'd prefer to download the Windows binaries from Zotero if I were able to locate them. The reference page isn't really helpful locating the Windows binaries; that's why I'm asking for help.
On manual installation: Have you ever thought of using Zotero in an Intranet? Maybe people have to get the clouds out of their heads first... ;-)

dstillman

On manual installation: Have you ever thought of using Zotero in an Intranet? Maybe people have to get the clouds out of their heads first... ;-)

But you already need a computer with an internet connection in order to download them manually... So just install them through Zotero on the same computer with the internet connection where you downloaded Zotero itself and then copy the binaries along with the Zotero binary. The manual instructions are not for situations like yours.

dstillman

If you insist, though:

https://www.zotero.org/download/xpdf/pdfinfo-Win32.exe-3.02
https://www.zotero.org/download/xpdf/pdftotext-Win32.exe-3.02

The files have to be renamed to pdfinfo-Win32.exe and pdftotext-Win32.exe, and there should be corresponding pdf*-Win32.exe.version text files containing "3.02". Or something like that. It's not meant to be done this way. If that doesn't work, do what I said above.

(We're also planning to switch to a non-binary solution (Mozilla's pdf.js) at some point. And Standalone should be made to bundle these itself in the meantime, since it's already platform specific.)

Ulrich Windl

(Maybe I'm in computer business for too long: I grew up in times where you got software on a physical media to install; even updates. That method still works in Intranets. OK now?)
I downloaded the binaries sucessfully, and indexing worked. However the installed version is still displayed as "UNKNOWN". What should be in the "*.version" text files exactly to make it work?

dstillman

Like I said, if it's not working, copy the files over from a working installation. This is a waste of all of our time.

Ulrich Windl

Updating incomplete documentation is a waste of your time?

adamsmith

documentation is a wiki. We try to make sure that the parts needed by most regular users are up-to-date and accurate and even that can be a challenge at times.

Dan presents you a solution that seems perfectly usable. Why not use it instead of being snarky? If it doesn't work for you, then tell us why.

dstillman

Updating incomplete documentation is a waste of your time?

No, it's waste of time to help you debug problems with an installation method we don't recommend when there's a much easier solution that you could have done days ago.

Anyhow, the version file doesn't really matter, but you're likely missing an ".exe" in the filename. I gave you the correct filename above. (I also clarified the Windows filename in the documentation, but there's no expectation that people on Windows would be compiling the tools by hand, which is all those instructions are for.)

Ulrich Windl

Let me add: First, there's a button in Zotero that refers to the Wiki page for manual installation. I followed that path to find out that the documentation does not cover Windows. Second I was specifically asking what the content of a ".version" file should look like. The wiki and this thread leaves that question still unanswered. I wonder: Would it be more work to add two lines to the wiki page than complaining about my use case?

dstillman

Me, above:

there should be corresponding pdf*-Win32.exe.version text files containing "3.02".

And:

If that doesn't work, do what I said above.

Ulrich Windl

I managed to get the version thing running: I had been confused by "in the format pdftotext-{platform}.version". For Windows "{platform}" would include ".exe". So I had files pdftotext-Win32.version" and "pdfinfo-Win32.version", while I needed "pdftotext-Win32.exe.version" and "pdfinfo-Win32.exe.version".
I was playing with different end-of-line conventions as the documentation said nothing on that (no EOL, UNIX EOL, DOS EOL, Mac EOL), ...
The wiki page is a little bit better now.
Maybe explain that "{platform}" was added to the "version" files to allow sharing one profile between different platforms (I guess this is the rationale behind). The ".exe" in ".version" does not make much sense IMHO, unless you state that versions files are named by simply adding ".version" to the executable name.

akhalifa

This is what i did to solve the win32 error of indexing unindexed items.

1. Download the binary files from ftp://ftp.foolabs.com/pub/xpdf/xpdfbin-win-3.04.zip
2. Extract them.
3. BASED ON YOUR SYSTEM 32/64bit. Rename the extracted binaries pdfinfo to pdfinfo-Win32 and pdftotext to pdftotext-Win32.
4. copy and replace those two files to the C:\Users\XXXX\AppData\Roaming\Zotero\Zotero\Profiles\XXXX\zotero folder.
IT WORKED!!

dstillman

akhalifa: Note that, if you do it that way, Zotero won't have number-of-pages information for PDF files, because that requires a custom build of pdfinfo, and you might see a black command prompt window pop up briefly every time Zotero indexes something. (Not sure if the latter is still the case.)

akhalifa

TRUE! CMD runs and closes quickly.

How may I fix that and get the number-of-pages information?

dstillman

See the page linked at the top of this thread.