softwareheritage=> select lang, count, percent from swh_content_language_repartition();
| lang                |   count | percent |
|---------------------+---------+---------|
| python              | 1130544 | 16.1095 |
| javascript+lasso    | 1078207 | 15.3637 |
| c                   |  706431 | 10.0662 |
| unknown             |  665657 |  9.4852 |
| css+lasso           |  656440 |  9.3538 |
| xml                 |  524764 |  7.4775 |
| objective-c         |  340457 |  4.8513 |
| javascript+php      |  338407 |  4.8221 |
| html                |  224124 |  3.1936 |
| c++                 |  175327 |  2.4983 |
| perl6               |  165176 |  2.3536 |
| prolog              |  145120 |  2.0679 |
| rexx                |   81382 |  1.1596 |
| genshi              |   78082 |  1.1126 |
| velocity            |   57813 |   .8238 |
| swig                |   51430 |   .7328 |
| makefile            |   47717 |   .6799 |
| mason               |   45450 |   .6476 |
| bash                |   41872 |   .5966 |
| vb.net              |   36728 |   .5233 |
| perl                |   35840 |   .5107 |
| restructuredtext    |   35284 |   .5028 |
| xml+django/jinja    |   35145 |   .5008 |
| ini                 |   29649 |   .4225 |
| scalate-server-page |   28918 |   .4121 |
| numpy               |   23422 |   .3337 |
| actionscript-3      |   22484 |   .3204 |
| jasmin              |   21424 |   .3053 |
| xml+php             |   21051 |   .3000 |
| nix                 |   20153 |   .2872 |
| erb                 |   17455 |   .2487 |
| django/jinja        |   17346 |   .2472 |
| xml+ruby            |   13787 |   .1965 |
| objective-j         |   13172 |   .1877 |
| s                   |   11591 |   .1652 |
| smali               |    8764 |   .1249 |
| coq                 |    8260 |   .1177 |
| ruby                |    8012 |   .1142 |
| html+php            |    7323 |   .1043 |
| gas                 |    7061 |   .1006 |
| smarty              |    6129 |   .0873 |
| lasso               |    4798 |   .0684 |
| matlab              |    4737 |   .0675 |
| groff               |    4631 |   .0660 |
| hy                  |    3853 |   .0549 |
| html+django/jinja   |    3193 |   .0455 |
| diff                |    1722 |   .0245 |
| xml+smarty          |    1378 |   .0196 |
| rhtml               |    1202 |   .0171 |
| xml+lasso           |    1085 |   .0155 |
| cmake               |    1083 |   .0154 |
| php                 |    1072 |   .0153 |
| dtd                 |    1051 |   .0150 |
| javascript          |     737 |   .0105 |
| xslt                |     671 |   .0096 |
| logos               |     647 |   .0092 |
| html+genshi         |     542 |   .0077 |
| debian-sourcelist   |     476 |   .0068 |
| python-3            |     468 |   .0067 |
| tex                 |     236 |   .0034 |
| viml                |     228 |   .0032 |
| html+smarty         |     129 |   .0018 |
| http                |     120 |   .0017 |
| html+lasso          |     106 |   .0015 |
| qbasic              |      87 |   .0012 |
| xml+velocity        |      46 |   .0007 |
| aspx-vb             |      45 |   .0006 |
| groovy              |      39 |   .0006 |
| julia               |      38 |   .0005 |
| coffeescript        |      13 |   .0002 |
| logtalk             |      10 |   .0001 |
| text-only           |       8 |   .0001 |
| sql                 |       3 |   .0000 |
| rebol               |       3 |   .0000 |
| reg                 |       3 |   .0000 |
| common-lisp         |       2 |   .0000 |
| tcl                 |       1 |   .0000 |
| go                  |       1 |   .0000 |
| yaml                |       1 |   .0000 |
| apacheconf          |       1 |   .0000 |
| erlang              |       1 |   .0000 |
| css                 |       1 |   .0000 |

# sql

softwareheritage=> select count(*) from content_language;
  count
---------
 3410460
(1 row)


create type swh_content_language_repartition_signature as (
  lang    languages,
  count   bigint,
  percent text
);

create or replace function swh_content_language_repartition()
  returns setof swh_content_language_repartition_signature
  language plpgsql
as $$
declare
  total real;
begin
  select count(id) from content_language into total;

  return query
         select lang, count(id) as count, to_char((100.0 * count(id) / total), '999.99') as percent
         from content_language
         group by lang
         order by count desc;
end
$$;