Trying to translate a full Web page just hangs

[forthrin]

Is translating a full Web page supported? I can't really tell from the man pages or the help, but there was at least one bug report that referenced this. Anyway, trying to do this just hangs forever.

trans :fr http://www.w3.org/

[soimort]

Please follow https://github.com/soimort/translate-shell/blob/develop/CONTRIBUTING.md and report the information of trans -V.

[forthrin]

$ trans -V :fr http://www.w3.org/
gawk: /dev/fd/63:177: warning: regexp escape sequence `\"' is not a known regexp operator
Translate Shell       0.9.6.9

platform              Darwin
gawk (GNU Awk)        5.0.0
fribidi (GNU FriBidi) 1.0.5
audio player          mpg123
terminal pager        less
terminal type         xterm-256color
user locale           en_US.UTF-8 (English)
home language         en
source language       auto
target language       fr
translation engine    google
proxy                 [NONE]
user-agent            Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/602.1 (KHTML, like Gecko) Version/8.0 Safari/602.1 Epiphany/3.18.2
theme                 default
init file             [NONE]

[soimort]

On macOS, what trans actually does is simply invoking

$ open 'https://translate.google.com/translate?hl=en&sl=auto&tl=fr&u=http://www.w3.org/'

If the above doesn't work, try specifying an existing browser,

$ trans -browser firefox :fr http://www.w3.org/

[forthrin]

Aha! Well, it seems trans looks up the BROWSER environment variable, which is set to a special handler on my system. Unsetting this makes the browser launch as expected. So consider this "user error" :)

However, I was sort of wondering/expecting if trans would print a text-only version of the translated Web page to the terminal.

I see now that the whole thing is written in Bash, and creating a text-only version would at require at least a minimum of semi-intelligent HTML parsing and extraction, and quite often an imperfect result.

So unless some of the translation engines support text-only output themselves, or if there is a huge interest in this, this request is maybe out of scope?

[soimort]

Given that there is no universally accepted, easy way of transforming an HTML page into plain text, I'd say this feature is out of scope. Not to mention a lot of modern websites that make heavy use of JavaScript in their front-ends!

A reasonable solution would be using a terminal-based web browser such as w3m or lynx. Unfortunately, these browsers seem to be undermaintained these days (compared to the fast-developing Chrome, WebKit, and Firefox), and none of them supports Google Translate's frame embedding for website translation.

For me, I'm not a big command-line junkie trying to view everything text-only. So if it's a web page designed for displaying in a web browser, I see little reason to restrict oneself to a text-only terminal. I'm not going to have any time / interest to maintain such a nontrivial feature that I myself barely need to use.

[forthrin]

If I get the inspiration I may have a go. If so, I'll let you know.