viewing docx files

viewing docx files

[Devin Prater]

Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on MacOS 
Sierra. I run Emacs in the terminal, and use the Emacspeak package for 
access, since I am blind. I received an email (gnews), with an attachment, 
two docx files for reading. I was able to download the attachments to my ~/ 
directory. I opened the file (c-x c-f then tab completion), but it opened 
in the archive viewer. I've read online that there is a DocViewer, but I 
cannot get Emacs to open the files in that doc viewer.
Using PanDoc in a shell, I was able to convert the docx file to a usable 
HTML file, but if there's a way to natively view these types of files, I'd 
be saving a lot of time.
Thanks for any help.

Sent with AquaMail for Android
http://www.aqua-mail.com

Re: viewing docx files

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sat, Jan 28, 2017 at 09:36:44AM -0600, Devin Prater wrote:
> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on
> MacOS Sierra. I run Emacs in the terminal, and use the Emacspeak
> package for access, since I am blind. I received an email (gnews),
> with an attachment, two docx files for reading. I was able to
> download the attachments to my ~/ directory. I opened the file (c-x
> c-f then tab completion), but it opened in the archive viewer. I've
> read online that there is a DocViewer, but I cannot get Emacs to
> open the files in that doc viewer.
> Using PanDoc in a shell, I was able to convert the docx file to a
> usable HTML file, but if there's a way to natively view these types
> of files, I'd be saving a lot of time.
> Thanks for any help.

There is a DocView mode (cf. https://www.emacswiki.org/emacs/DocViewMode),
but it passes through PostScript (via GhostScript), in your case through
unoconv. The postscript seems to include the (plain) text, since it
supports text searching.

All in all, your approach with pandoc might prove superior for your
use case. Perhaps a little automation. There is an Emacs pandoc-mode,
but alas, I haven't any experience with that.

Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAliMwd0ACgkQBcgs9XrR2kasswCaA/I1zkn7LqGbdfo/k4LHw3EV
VP4AnRoha2WuRwokuN3F4bIXNaaBJ5jc
=mCW+
-----END PGP SIGNATURE-----

Re: viewing docx files

[Devin Prater]

Oh, I see. So DocViewer just converts everything to plaintext? Nah, I like 
the vibrancy of formatting and the way Emacspeak manifests that too much to 
deal with plain text where formatting was before. Thanks for the info, though.

Sent with AquaMail for Android
http://www.aqua-mail.com


On January 28, 2017 10:08:41 AM <tomas@tuxteam.de> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Sat, Jan 28, 2017 at 09:36:44AM -0600, Devin Prater wrote:
>> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on
>> MacOS Sierra. I run Emacs in the terminal, and use the Emacspeak
>> package for access, since I am blind. I received an email (gnews),
>> with an attachment, two docx files for reading. I was able to
>> download the attachments to my ~/ directory. I opened the file (c-x
>> c-f then tab completion), but it opened in the archive viewer. I've
>> read online that there is a DocViewer, but I cannot get Emacs to
>> open the files in that doc viewer.
>> Using PanDoc in a shell, I was able to convert the docx file to a
>> usable HTML file, but if there's a way to natively view these types
>> of files, I'd be saving a lot of time.
>> Thanks for any help.
>
> There is a DocView mode (cf. https://www.emacswiki.org/emacs/DocViewMode),
> but it passes through PostScript (via GhostScript), in your case through
> unoconv. The postscript seems to include the (plain) text, since it
> supports text searching.
>
> All in all, your approach with pandoc might prove superior for your
> use case. Perhaps a little automation. There is an Emacs pandoc-mode,
> but alas, I haven't any experience with that.
>
> Regards
> - -- tomás
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.12 (GNU/Linux)
>
> iEYEARECAAYFAliMwd0ACgkQBcgs9XrR2kasswCaA/I1zkn7LqGbdfo/k4LHw3EV
> VP4AnRoha2WuRwokuN3F4bIXNaaBJ5jc
> =mCW+
> -----END PGP SIGNATURE-----
>

Re: viewing docx files

[Eli Zaretskii]

> Date: Sat, 28 Jan 2017 17:07:57 +0100
> From: <tomas@tuxteam.de>
> 
> There is a DocView mode (cf. https://www.emacswiki.org/emacs/DocViewMode),
> but it passes through PostScript (via GhostScript), in your case through
> unoconv. The postscript seems to include the (plain) text, since it
> supports text searching.
> 
> All in all, your approach with pandoc might prove superior for your
> use case. Perhaps a little automation. There is an Emacs pandoc-mode,
> but alas, I haven't any experience with that.

There's also antiword, and perhaps someone wrote a mode to run that
automatically on Word documents.

Re: viewing docx files

[Joost Kremers]

On Sat, Jan 28 2017, tomas@tuxteam.de wrote:
> All in all, your approach with pandoc might prove superior for 
> your
> use case. Perhaps a little automation. There is an Emacs 
> pandoc-mode,
> but alas, I haven't any experience with that.

pandoc-mode is intended to be used when you write the input file 
in Emacs, e.g., in Markdown or one of the other text-based input 
formats that Pandoc supports. It's not really helpful when you 
want to pipe a docx file through Pandoc and read the result.

-- 
Joost Kremers
Life has its moments

Re: viewing docx files

[Stefan Monnier]

> Oh, I see. So DocViewer just converts everything to plaintext?

Worse: it converts it to a set of PNG.  There is some support for
searching, but Emacs doesn't see the text really (other than as an
image), so Emacspeak won't b able to do anything useful with it.

So indeed, you'll be much better off going through something like HTML.

BTW, I think many people (including myself) would be interested in an
alternative to doc-view-mode which renders such documents as (decorated)
text rather than as images, e.g. going through HTML and then rendered
with shr.el.


        Stefan

Re: viewing docx files

[Jude DaShiell]

antiword can't handle .docx files last time I tried nor can catdoc.
On Sat, 
28 Jan 2017, Eli Zaretskii wrote:

> Date: Sat, 28 Jan 2017 11:14:57
> From: Eli Zaretskii <eliz@gnu.org>
> To: help-gnu-emacs@gnu.org
> Subject: Re: viewing docx files
> 
>> Date: Sat, 28 Jan 2017 17:07:57 +0100
>> From: <tomas@tuxteam.de>
>>
>> There is a DocView mode (cf. https://www.emacswiki.org/emacs/DocViewMode),
>> but it passes through PostScript (via GhostScript), in your case through
>> unoconv. The postscript seems to include the (plain) text, since it
>> supports text searching.
>>
>> All in all, your approach with pandoc might prove superior for your
>> use case. Perhaps a little automation. There is an Emacs pandoc-mode,
>> but alas, I haven't any experience with that.
>
> There's also antiword, and perhaps someone wrote a mode to run that
> automatically on Word documents.
>
>

-- 

Re: viewing docx files

[Tomas Nordin]

Devin Prater <r.d.t.prater@gmail.com> writes:

> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on MacOS 
> Sierra. I run Emacs in the terminal, and use the Emacspeak package for 
> access, since I am blind. I received an email (gnews), with an attachment, 
> two docx files for reading. I was able to download the attachments to my ~/ 

for converting docx files to text I use docx2txt, a tool made solely for
this purpose. 

> directory. I opened the file (c-x c-f then tab completion), but it

at this point (in dired with point on the docx file), I would invoke
dired-do-shell-command by hitting X or ! . By the prompt in the mini
buffer I say

    docx2txt <

and hit enter. The contents of the file as text is then available in a
new buffer "*Shell Command Output*"

This is Gnu/Linux Debian and I have got the tool via apt, I don't know
how it works on other systems. docx2txt is a command line tool and not
special to emacs.

-- 
Tomas Nordin | (The computing freedom explorer)
GPG Key: AB09AF78

Re: viewing docx files

[Narendra Joshi]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> BTW, I think many people (including myself) would be interested in an
> alternative to doc-view-mode which renders such documents as (decorated)
> text rather than as images, e.g. going through HTML and then rendered
> with shr.el.

Does `pdf-tools' do anything close?

Narendra

Re: viewing docx files

[Stefan Monnier]

>> BTW, I think many people (including myself) would be interested in an
>> alternative to doc-view-mode which renders such documents as (decorated)
>> text rather than as images, e.g. going through HTML and then rendered
>> with shr.el.
> Does `pdf-tools' do anything close?

Kind of, except that Emacs doesn't really see the text.  pdf-tools does
a great job at making the text accessible to Emacs on demand, but all
the features have to be supported "by hand".  If Emacs could really
directly see the text, then C-s would work without any need to write
specific code.  I.e. there would be no need for pdf-isearch.el,
pdf-occur.el, etc...

Clearly, it would suck at "rendering faithfully", since Emacs's
rendering engine is not sophisticated enough to handle the kind of
layouts seen in typical PDF files, so it wouldn't be a competitor to
pdf-tools.


        Stefan

Re: viewing docx files

[Tomas Nordin]

Devin Prater <r.d.t.prater@gmail.com> writes:

> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on MacOS 
> Sierra. I run Emacs in the terminal, and use the Emacspeak package for 
> access, since I am blind. I received an email (gnews), with an attachment, 
> two docx files for reading. I was able to download the attachments to my ~/ 
> directory. I opened the file (c-x c-f then tab completion), but it opened 

I wonder if you would like to eval and try this:

(defun docx2html (file)
  "Convert FILE to html in a buffer and display it."
  (interactive "f")
  (let ((html-buffer (format "*%s --> html*" file)))
    (call-process "pandoc" file html-buffer nil "--to=html")
    (switch-to-buffer html-buffer))
  )

After evaluation, say M-x docx2html and locate the docx file. See if it
works. It did not work for me but it seems to have to do with the
encoding of the characters in the test files I have. I mean, it works
such that I get the following message from pandoc in the new buffer:

    pandoc: Cannot decode byte '\xb1': Data.Text.Encoding.Fusion.streamUtf8: Invalid UTF-8 stream

Re: viewing docx files

[Joost Kremers]

On Sun, Jan 29 2017, Tomas Nordin wrote:
> Devin Prater <r.d.t.prater@gmail.com> writes:
>
>> Hi all. I'm running Gnu-Emacs (latest brew install emacs 
>> version) on MacOS 
>> Sierra. I run Emacs in the terminal, and use the Emacspeak 
>> package for 
>> access, since I am blind. I received an email (gnews), with an 
>> attachment, 
>> two docx files for reading. I was able to download the 
>> attachments to my ~/ 
>> directory. I opened the file (c-x c-f then tab completion), but 
>> it opened 
>
> I wonder if you would like to eval and try this:
>
> (defun docx2html (file)
>   "Convert FILE to html in a buffer and display it."
>   (interactive "f")
>   (let ((html-buffer (format "*%s --> html*" file)))
>     (call-process "pandoc" file html-buffer nil "--to=html")
>     (switch-to-buffer html-buffer))
>   )
>
> After evaluation, say M-x docx2html and locate the docx file. 
> See if it
> works. It did not work for me but it seems to have to do with 
> the
> encoding of the characters in the test files I have. I mean, it 
> works
> such that I get the following message from pandoc in the new 
> buffer:
>
>     pandoc: Cannot decode byte '\xb1': 
>     Data.Text.Encoding.Fusion.streamUtf8: Invalid UTF-8 stream

Pandoc only reads and writes UTF-8 and does no conversion. So if 
the files you want to convert & view are in another encoding, 
you'll need to reencode them first. Not sure if there's a tool to 
do that for docx files, though. iconv can convert text files from 
one encoding to another, but for that to work on docx files, 
you'll need to unzip them first (and zip them up again 
afterwards).

-- 
Joost Kremers
Life has its moments

Re: viewing docx files

[Jude DaShiell]

I wonder if the file utility can tell the difference between a docx-utf-8 
file and a docx-non-utf-8 file.  If that can work it may be possible to do 
a little docx inspection to find when to trigger the unzip->iconv->zip 
process and only trigger that process when necessary. On Sun, 29 Jan 2017, 
Joost Kremers wrote:

> Date: Sun, 29 Jan 2017 17:51:00
> From: Joost Kremers <joostkremers@fastmail.fm>
> To: Tomas Nordin <tomasn@posteo.net>
> Cc: Devin Prater <r.d.t.prater@gmail.com>, help-gnu-emacs@gnu.org
> Subject: Re: viewing docx files
> 
>
> On Sun, Jan 29 2017, Tomas Nordin wrote:
>> Devin Prater <r.d.t.prater@gmail.com> writes:
>> 
>>> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on MacOS 
>>> Sierra. I run Emacs in the terminal, and use the Emacspeak package for 
>>> access, since I am blind. I received an email (gnews), with an attachment, 
>>> two docx files for reading. I was able to download the attachments to my 
>>> ~/ directory. I opened the file (c-x c-f then tab completion), but it 
>>> opened 
>> 
>> I wonder if you would like to eval and try this:
>> 
>> (defun docx2html (file)
>>   "Convert FILE to html in a buffer and display it."
>>   (interactive "f")
>>   (let ((html-buffer (format "*%s --> html*" file)))
>>     (call-process "pandoc" file html-buffer nil "--to=html")
>>     (switch-to-buffer html-buffer))
>>   )
>> 
>> After evaluation, say M-x docx2html and locate the docx file. See if it
>> works. It did not work for me but it seems to have to do with the
>> encoding of the characters in the test files I have. I mean, it works
>> such that I get the following message from pandoc in the new buffer:
>>
>>     pandoc: Cannot decode byte '\xb1': 
>> Data.Text.Encoding.Fusion.streamUtf8: Invalid UTF-8 stream
>
> Pandoc only reads and writes UTF-8 and does no conversion. So if the files 
> you want to convert & view are in another encoding, you'll need to reencode 
> them first. Not sure if there's a tool to do that for docx files, though. 
> iconv can convert text files from one encoding to another, but for that to 
> work on docx files, you'll need to unzip them first (and zip them up again 
> afterwards).
>
>

-- 

Re: viewing docx files

[Yuri Khan]

On Mon, Jan 30, 2017 at 2:21 PM, Jude DaShiell <jdashiel@panix.com> wrote:
> I wonder if the file utility can tell the difference between a docx-utf-8
> file and a docx-non-utf-8 file.  If that can work it may be possible to do a
> little docx inspection to find when to trigger the unzip->iconv->zip process
> and only trigger that process when necessary.

Instead of iconv, use xmllint --encode utf-8. It will extract the
source encoding from the XML declaration at the top of the file, and
reencode from that to UTF-8. Trigger it unconditionally, for each
*.xml file in the archive.

Consider also trying to persuade Pandoc developers to support
non-UTF-8-encoded XML data.

Re: viewing docx files

[Tomas Nordin]

Tomas Nordin <tomasn@posteo.net> writes:

> Devin Prater <r.d.t.prater@gmail.com> writes:
>
>> Hi all. I'm running Gnu-Emacs (latest brew install emacs version) on MacOS 
>> Sierra. I run Emacs in the terminal, and use the Emacspeak package for 
>> access, since I am blind. I received an email (gnews), with an attachment, 
>> two docx files for reading. I was able to download the attachments to my ~/ 
>> directory. I opened the file (c-x c-f then tab completion), but it opened 
>
> I wonder if you would like to eval and try this:
>
> (defun docx2html (file)
>   "Convert FILE to html in a buffer and display it."
>   (interactive "f")
>   (let ((html-buffer (format "*%s --> html*" file)))
>     (call-process "pandoc" file html-buffer nil "--to=html")
>     (switch-to-buffer html-buffer))
>   )

Here is an updated version tested on my machine to work on all files
docx I can find. I have been assuming that you mean that it is a good
thing to have a buffer with hyper text markup. This version is ignoring
error output and provide the file as an argument to pandoc instead of
providing it as stdin. One should have pandoc version >= 1.13, I had an
older version before and docx reading was not supported.

(defun docx2html (file)
  "Convert FILE to html in a buffer and display it."
  (interactive "f")
  (let ((html-buffer (format "*%s --> html*" file)))
    (call-process "pandoc" nil `(,html-buffer nil) nil "--to=html" (expand-file-name file))
    (switch-to-buffer html-buffer))
  )

Those are the errors I have seen and we ignore:

pandoc: /usr/lib/x86_64-linux-gnu/liblua5.1.so.0: no version information
available (required by pandoc)

and

pandoc: unable to decommit memory: Invalid argument

It has been nice to experience this because with the html-buffer I
simply mark-whole-buffer and say shr-render-region, and have a nice
document in text in emacs.

--
Tomas

Re: viewing docx files

[Tomas Nordin]

Yuri Khan <yuri.v.khan@gmail.com> writes:

> Instead of iconv, use xmllint --encode utf-8. It will extract the
> source encoding from the XML declaration at the top of the file, and
> reencode from that to UTF-8. Trigger it unconditionally, for each
> *.xml file in the archive.
>
> Consider also trying to persuade Pandoc developers to support
> non-UTF-8-encoded XML data.

Seem like good ideas to keep in the pocket. I have been surfing around a
bit and it seems the docx xml files should be utf-8 encoded by default.

Re: viewing docx files

[Stefan Monnier]

> Consider also trying to persuade Pandoc developers to support
> non-UTF-8-encoded XML data.

I think the world will be a better place when all XML data is utf-8, so
better put pressure on the upstream tool generating the non-utf-8
XM data.


        Stefan

Re: viewing docx files

[Jude DaShiell]

Do any effective non-utf-8 to utf-8 translation tools exist?

On Tue, 31 Jan 2017, Stefan Monnier wrote:

> Date: Tue, 31 Jan 2017 09:56:09
> From: Stefan Monnier <monnier@iro.umontreal.ca>
> To: help-gnu-emacs@gnu.org
> Subject: Re: viewing docx files
> 
>> Consider also trying to persuade Pandoc developers to support
>> non-UTF-8-encoded XML data.
>
> I think the world will be a better place when all XML data is utf-8, so
> better put pressure on the upstream tool generating the non-utf-8
> XM data.
>
>
>        Stefan
>
>
>

--