Installation problems on Darwin ARM

[MassiCav]

Hello Jess,
I encountered a problem with the installation on my laptop (Apple M1 Pro).

I was able to narrow down the problem to lines 50 and 57 of the installation script where the versions of the packages to download are asserted. Probably in the 'sed' part of the algorithm the processed token contains a couple of spaces at the beginning
(i.e. ' v0.21.1-19') that determine a failure with exit code 3 (and no error message).

I was able to overcome the problem by hardcoding the versions manually in the installation script.

BTW this is Massi from Zend, and since we both live in London (I am in Barnet), it would be lovely to catch up over a cup of coffee.

TIA and BR,
M.

[jessp01]

Hi Massi,

Nice to hear (or should I say read) from you.
I do not have a Darwin machine I can test this on and so, could you run the below in your shell:

$ curl -L https://api.github.com/repos/jessp01/zaje/releases/latest| grep tag_name|sed 's@\s*"tag_name": "\(.*\)".*@\1@'

and post the output?

Thanks,

[MassiCav]

Here it is:
' v0.21.1-19'

Mind the 2 spaces at the beginning.

BR,
M.

[jessp01]

I only see one space in your output but nonetheless, that's still a problem.
What happens when you run:

curl -L https://api.github.com/repos/jessp01/zaje/releases/latest| grep tag_name|sed 's@\s*"tag_name":\s*"\(.*\)".*@\1@'

?

What's baffling to me is that curl -L https://api.github.com/repos/jessp01/zaje/releases/latest| grep tag_name yields:

"tag_name": "v0.21.1-19",
So, not sure how sed 's@\s*"tag_name":\s*"\(.*\)".*@\1@' produces v0.21.1-19 on your machine...
To make things somewhat clearer, can you add set -x before line 50 and set+x after line 50, like so:

set -x
LATEST_VER=$(curl -sL "https://api.${GIT_BASE_DOMAIN}/repos/$GH_SPACE/$NAME/releases/latest"| grep tag_name|sed 's@\s*"tag_name":\s*"\(.*\)".*@\1@')
set +x

And post the output? On my machines (all Linux), this yields:

+ curl -sL https://api.github.com/repos/jessp01/zaje/releases/latest
+ grep tag_name
+ sed s@\s*"tag_name": "\(.*\)".*@\1@
+ LATEST_VER=v0.21.1-19

Cheers,

PS
Since curl -sL https://api.github.com/repos/jessp01/zaje/releases/latest returns a JSON, I could, of course, parse it with jq but that's quite the overkill for this simple task, plus, it would mean adding another dependency so, overall, does not make sense in this case.

[MassiCav]

Here is the output, down to line 64:

+ RED=
+ YELLOW=
+ BLUE=
+ NORMAL=
+ BOLD=
+ UNSET=
+ log_use_fancy_output
+ TPUT=/usr/bin/tput
+ EXPR=/usr/bin/expr
+ '[' -t 1 ']'
+ '[' xxterm-256color '!=' x ']'
+ '[' xxterm-256color '!=' xdumb ']'
+ '[' -x /usr/bin/tput ']'
+ '[' -x /usr/bin/expr ']'
+ FANCYTTY=0
+ case "$FANCYTTY" in
+ false
++ which curl
+ '[' '!' /usr/bin/curl ']'
+ GIT_BASE_DOMAIN=github.com
+ NAME=zaje
+ HIGHLIGHT_REPO_NAME=gohighlight
+ GH_SPACE=jessp01
++ curl -sL https://api.github.com/repos/jessp01/zaje/releases/latest
++ grep tag_name
++ sed 's@\s*"tag_name": "\(.*\)".*@\1@'
+ LATEST_VER='  v0.21.1-19'
++ uname
+ OS=Darwin
++ uname -m
+ ARCH=arm64
+ BIN_ARCHIVE=zaje_Darwin_arm64.tar.gz
++ curl -sL https://api.github.com/repos/jessp01/gohighlight/releases/latest
++ grep tag_name
++ sed 's@\s*"tag_name": "\(.*\)".*@\1@'
+ LATEST_HIGHLIGHT_VER='  v0.21.1-21'
+ HIGHLIGHT_SOURCE_ARCHIVE='  v0.21.1-21.tar.gz'
+ CONFIG_DIR=/Users/mcavicchioli/.config/zaje
+ LEXERS_DIR=/Users/mcavicchioli/.config/zaje/syntax_files
+ TMP_DIR=/tmp/zaje
+ FUNCTIONS_RC_FILE=/Users/mcavicchioli/.config/zaje/zaje_functions.rc

BR,
M.

[jessp01]

Parsing with JQ would add a dependency, but definitely makes things more portable I guess.
As I wrote, it's a serious overkill for something as simple as this:)

I'm still unclear as to how we get to:

LATEST_VER=' v0.21.1-19'
from the line in question but changing it thusly:

LATEST_VER=$(curl -L https://api.github.com/repos/jessp01/zaje/releases/latest| grep tag_name|sed 's@\s*"tag_name":\s*"\(.*\)".*@\1@')| xargs echo -n

Should fix this. If so, you of course also need to add the | xargs echo -n in line 56.

Please confirm that this works and I'll push the change.

Cheers,

[MassiCav]

Adding the 'xargs' pipe on lines 50 and 57 fixes the 'zaje' installation, TYVM.

Now my problem is installing 'super-zaje'.
After installing 'tesseract' with 'brew', having both Tesseract and Leptonica code installed I experience this:

go install github.com/jessp01/zaje/cmd/super-zaje@latest

# github.com/otiai10/gosseract/v2
tessbridge.cpp:5:10: fatal error: 'leptonica/allheaders.h' file not found

Here are the headers:

eza --long --git /opt/homebrew/include/tesseract/

.rw-r--r--@  32k mcavicchioli 10 Nov  2024 -I baseapi.h
.rw-r--r--@  21k mcavicchioli 10 Nov  2024 -I capi.h
.rw-r--r--@ 1.2k mcavicchioli 10 Nov  2024 -I export.h
.rw-r--r--@  10k mcavicchioli 10 Nov  2024 -I ltrresultiterator.h
.rw-r--r--@ 7.2k mcavicchioli 10 Nov  2024 -I ocrclass.h
.rw-r--r--@ 4.5k mcavicchioli 10 Nov  2024 -I osdetect.h
.rw-r--r--@  15k mcavicchioli 10 Nov  2024 -I pageiterator.h
.rw-r--r--@  12k mcavicchioli 10 Nov  2024 -I publictypes.h
.rw-r--r--@  10k mcavicchioli 10 Nov  2024 -I renderer.h
.rw-r--r--@ 9.5k mcavicchioli 10 Nov  2024 -I resultiterator.h
.rw-r--r--@ 6.3k mcavicchioli 10 Nov  2024 -I unichar.h
.rw-r--r--@ 1.1k mcavicchioli 10 Nov  2024 -I version.h

eza --long --git /opt/homebrew/include/leptonica/

.rw-r--r--@ 262k mcavicchioli 16 Oct  2024 -I allheaders.h
.rw-r--r--@ 2.3k mcavicchioli 16 Oct  2024 -I alltypes.h
.rw-r--r--@ 4.2k mcavicchioli 16 Oct  2024 -I array.h
.rw-r--r--@ 5.6k mcavicchioli 16 Oct  2024 -I array_internal.h
.rw-r--r--@  11k mcavicchioli 16 Oct  2024 -I arrayaccess.h
.rw-r--r--@ 2.7k mcavicchioli 16 Oct  2024 -I bbuffer.h
.rw-r--r--@ 6.6k mcavicchioli 16 Oct  2024 -I bilateral.h
.rw-r--r--@ 3.1k mcavicchioli 16 Oct  2024 -I bmf.h
.rw-r--r--@  47k mcavicchioli 16 Oct  2024 -I bmfdata.h
.rw-r--r--@ 5.3k mcavicchioli 16 Oct  2024 -I bmp.h
.rw-r--r--@ 2.3k mcavicchioli 16 Oct  2024 -I ccbord.h
.rw-r--r--@ 5.3k mcavicchioli 16 Oct  2024 -I ccbord_internal.h
.rw-r--r--@ 3.3k mcavicchioli 16 Oct  2024 -I colorfill.h
.rw-r--r--@  11k mcavicchioli 16 Oct  2024 -I dewarp.h
.rw-r--r--@  205 mcavicchioli 16 Oct  2024 -I endianness.h
.rw-r--r--@  27k mcavicchioli 16 Oct  2024 -I environ.h
.rw-r--r--@ 3.8k mcavicchioli 16 Oct  2024 -I gplot.h
.rw-r--r--@ 7.1k mcavicchioli 16 Oct  2024 -I hashmap.h
.rw-r--r--@ 4.3k mcavicchioli 16 Oct  2024 -I heap.h
.rw-r--r--@  12k mcavicchioli 16 Oct  2024 -I imageio.h
.rw-r--r--@ 7.2k mcavicchioli 16 Oct  2024 -I jbclass.h
.rw-r--r--@ 1.9k mcavicchioli 16 Oct  2024 -I leptwin.h
.rw-r--r--@ 3.5k mcavicchioli 16 Oct  2024 -I list.h
.rw-r--r--@ 9.1k mcavicchioli 16 Oct  2024 -I morph.h
.rw-r--r--@  54k mcavicchioli 16 Oct  2024 -I pix.h
.rw-r--r--@  20k mcavicchioli 16 Oct  2024 -I pix_internal.h
.rw-r--r--@ 3.7k mcavicchioli 16 Oct  2024 -I ptra.h
.rw-r--r--@ 3.4k mcavicchioli 16 Oct  2024 -I queue.h
.rw-r--r--@ 3.7k mcavicchioli 16 Oct  2024 -I rbtree.h
.rw-r--r--@  12k mcavicchioli 16 Oct  2024 -I readbarcode.h
.rw-r--r--@  15k mcavicchioli 16 Oct  2024 -I recog.h
.rw-r--r--@ 6.3k mcavicchioli 16 Oct  2024 -I regutils.h
.rw-r--r--@ 3.1k mcavicchioli 16 Oct  2024 -I stack.h
.rw-r--r--@ 2.9k mcavicchioli 16 Oct  2024 -I stringcode.h
.rw-r--r--@ 3.4k mcavicchioli 16 Oct  2024 -I sudoku.h
.rw-r--r--@ 3.5k mcavicchioli 16 Oct  2024 -I watershed.h

It seems a problem of paths recognition I guess so that the headers can be picked up.

I tried adding this SHELL variables:

export LIBRARY_PATH="/opt/homebrew/lib"
export CPATH="/opt/homebrew/include" 

After that the error changes to the following:

go install github.com/jessp01/zaje/cmd/super-zaje@latest

# github.com/otiai10/gosseract/v2
In file included from tessbridge.cpp:6:
In file included from /opt/homebrew/include/tesseract/baseapi.h:27:
In file included from /opt/homebrew/include/tesseract/resultiterator.h:23:
In file included from /opt/homebrew/include/tesseract/ltrresultiterator.h:24:
/opt/homebrew/include/tesseract/unichar.h:23:10: fatal error: 'cstring' file not found

A sign that now the basic system libraries cannot be picked up.

BR,
M.

[jessp01]

Hi @MassiCav ,

Try:

$ CGO_CFLAGS="-I/opt/homebrew/include" go install github.com/jessp01/zaje/cmd/super-zaje@latest

[MassiCav]

Errors stay exactly the same.
When only with 'CGO_CFLAGS' the error is:

# github.com/otiai10/gosseract/v2
tessbridge.cpp:5:10: fatal error: 'leptonica/allheaders.h' file not found

Whe the 'CGO_CFLAGS' is set together with 'LIBRARY_PATH' and 'CPATH' the error is:

# github.com/otiai10/gosseract/v2
In file included from tessbridge.cpp:6:
In file included from /opt/homebrew/include/tesseract/baseapi.h:27:
In file included from /opt/homebrew/include/tesseract/resultiterator.h:23:
In file included from /opt/homebrew/include/tesseract/ltrresultiterator.h:24:
/opt/homebrew/include/tesseract/unichar.h:23:10: fatal error: 'cstring' file not found

BR,
M.

[jessp01]

Hi @MassiCav ,

This really isn't an issue with ZAJE's code anymore but still happy to help..

Try adding -stdlib=libc++ (see https://stackoverflow.com/questions/19774778/when-is-it-necessary-to-use-the-flag-stdlib-libstdc and sirfz/tesserocr#162)

Like I wrote, I don't have a Darwin system to test on. If adding this does not solve the problem, you can also submit an issue here: https://github.com/otiai10/gosseract, it's a well established project, I imagine one of the contributors will have access to a Darwin machine.

[MassiCav]

It did not work.

I saw you referenced this issue on the Tesserocr project.
I will be waiting for them replying to it.

TYVM and all the best to you,
M.

[jessp01]

My recommendation would be to submit an issue to gosseract.
Closing this for now but it would be good if you posted an update and referenced the issue here for the benefit of others.

Best,