Isefka-nneɣ ttwaxedmen ddaw n turagtin reative Commons.Ugar n telɣut
Ma tḥemmleḍ agbur-agi, mudd-d ma ulac aɣilif tawsa.
Umuɣ
Hello, how can I download sentences belonging to a particular user? There is "detailed sentences" option which let indicate user's name but doesn't let to choose it before downloading. Can I change it?
I don't think that's possible before downloading. Maybe the best thing to do is to download the sentences_detailed file of the desired language and then filter the user's sentences with a spreadsheet.
Accepting this approach, I must download more than one million English sentences, when I really need about 30,000 right now. I guess the file would be too big to process it.
Plain scrapping from Tatoeba.org using Beautiful Soup looks better than that, but maybe it's possible to scrap the data from its source database without html tags?
> maybe it's possible to scrap the data from its source database without html tags?
if you have coding skills, maybe you can use the API (in beta).
https://en.wiki.tatoeba.org/articles/show/api#
That's exactly what I need, thanks a lot!
How can I get more than 1000 sentences unlike using the Tatoeba search? Is there any variable for that in the API? I tried "count=all" or "count=5000", but it didn't help.
FI
Jos tiedoston lataaminen ei ole ongelma itsessään, niin senhän voisi ladata ja sitten muokata tekstitiedostona. Poistaa vaan suurimman osan datasta, niin että jäljelle jää vain haluttu määrä lauseita.
EN
If downloading the file is not an issue as such, you could do that and edit the file (as a text file) to remove most of the data, leaving behind only a usable amount.
you'd be surprised, it's only 27mb compressed and 156mb uncompressed
using the fantastic `tatoebatools` python library (made by lbdx, funnily enough) i was able to filter for my own sentences in 5 seconds
granted, your computer might be weaker than my 4-year-old core i5 laptop, but it should still take less time than figuring out the api *or* beautifulsoup, let alone running them
Where can I learn more about the 'tatoebatools' library? (I can assume it permits you having and processing all the Tatoeba database on your own computer, right?)
there's some documentation on the pypi page https://pypi.org/project/tatoebatools/
and yeah that's what it is, it handles all the downloading and csv quirks for you
Thanks! Btw, do I need to download the database just once or every time using the library? Unfortunately, I didn't have enough of patience to wait for downloading the database but if it's just one time I can wait.
it saves it after downloading but it does redownload it if there's an update
updates are released every saturday at 6:30am utc
Thanks for the answer, I'll try it, too.
🍎 Here are files with only 4 fields, omitting the dates.
sentence_ID + tab + language_code + tab + text + tab + sentence_owner
🥝 All the exported sentences
http://study.aitech.ac.jp/4flds...2022-12-17.zip
228 MB
🥝 Just the English sentences
http://study.aitech.ac.jp/4flds...2022-12-17.zip
27 MB
🥝 Just the 891,125 sentences on List 907
http://study.aitech.ac.jp/4flds...2022-12-17.zip
12 MB
Perhaps one of these will help you.
Thanks a lot!
Where can I get updates with new sentences?
At the bottom of every page on Tatoeba there’s a link to the Downloads page, which is updated weekly.
https://tatoeba.org/downloads
If it is the English sentences on List 907 that you are interested in, you can download them at any time from this URL.
https://tatoeba.org/en/sentence...s/download/907
You will also have the option to include any linked Russian sentence.
Note that this export doesn't include the names of the sentence owners, though.