gladiaio / words2num2 / 36194400503
59%
main: 61%

LAST BUILD BRANCH: test/roundtrip-ratchet
DEFAULT BRANCH: main
Ran
Jobs 6
Files 6
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 59.459%. Remained the same
36194400503

Pull #25

github

jqueguiner 
fix(sentence): compounds the walker broke apart (deux cents, cinquante et une, a hundred, one and a half, pour cent)

Real-life finals from call-center / finance transcripts, run through
words2num_sentence, came back with the number split in two or a
connector swallowed:

  deux cents euros            -> 2 cents euros      (now 200 euros)
  quatre vingts               -> 4 vingts           (now 80)
  mille deux cent             -> 1002 cent          (now 1200)
  un million deux cent mille  -> 1000002 100000     (now 1200000)
  cinquante et une personnes  -> 50 une personnes   (now 51 personnes)
  vingt pour cent             -> 20 pour 100        (now 20 pour cent)
  diez por ciento             -> 10 por 100         (now 10 por ciento)
  veinte y uno                -> 20 1               (now 21)
  a hundred dollars           -> a 100 dollars      (now 100 dollars)
  one and a half hours        -> 1 half hours       (now 1 and a half hours)
  five point                  -> 5                  (now 5 point)
  two point five million      -> 2.5 1000000        (now 2.5 million)

Walker:
- a token that is not a number word on its own ("cents", "vingts") is
  still tried as the continuation of the run before the run is closed;
- a run never ends on a connector ("and", "a", "point", "et", "y"): the
  connector is emitted as a word instead of being eaten;
- en "a"/"an" opens a run in front of a scale word;
- a scale word right after a decimal stays a word;
- "pour cent" / "por ciento" / "per cent" / "por cento" / "per cento":
  the second word is the percent sign after the preposition, never 100.

Lookup:
- fr: num2words writes "deux cents", "quatre-vingts", "deux millions";
  speech drops or adds the plural -s freely, so a table miss retries the
  cent/cents, vingt/vingts, million(s), milliard(s) variants;
- fr "une" is the feminine "un" inside a number (apocope table), a lone
  "une" stays an article;
- es tens joined with the spoken connector ("veinte y uno"... (continued)
Pull Request #25: fix(sentence): compounds the walker broke apart (deux cents, cinquante et une, a hundred, one and a half, pour cent)

44 of 74 relevant lines covered (59.46%)

3.54 hits per line

Jobs
ID Job ID Ran Files Coverage
1 3.13 - 36194400503.1 6
59.46
GitHub Action Run
2 3.10 - 36194400503.2 6
59.46
GitHub Action Run
3 3.12 - 36194400503.3 6
59.46
GitHub Action Run
4 3.15 - 36194400503.4 6
58.9
GitHub Action Run
5 3.14 - 36194400503.5 6
58.9
GitHub Action Run
6 3.11 - 36194400503.6 6
59.46
GitHub Action Run
Source Files on build 36194400503