| 1 | #!/usr/bin/env bash
|
| 2 | #
|
| 3 | # Usage:
|
| 4 | # data_lang/htm8-test.sh
|
| 5 |
|
| 6 | : ${LIB_OSH=stdlib/osh}
|
| 7 | source $LIB_OSH/bash-strict.sh
|
| 8 | source $LIB_OSH/task-five.sh
|
| 9 |
|
| 10 | # parse with lazylex/html.py, or data_lang/htm8.py
|
| 11 |
|
| 12 | site-files() {
|
| 13 | find ../../oilshell/oilshell.org__deploy -name '*.html'
|
| 14 | }
|
| 15 |
|
| 16 | # Issues with lazylex/html.py
|
| 17 | #
|
| 18 | # - Token ID is annoying to express in Python
|
| 19 | # - re.DOTALL for newlines
|
| 20 | # - can we change that with [.\n]*?
|
| 21 | # - nongreedy match for --> and ?>
|
| 22 |
|
| 23 |
|
| 24 | test-site() {
|
| 25 | # 1.5 M lines of HTML - takes 3 xargs invocations!
|
| 26 | #
|
| 27 | # TODO:
|
| 28 | # - test that it lexes
|
| 29 | # - test that tags are balanced
|
| 30 |
|
| 31 | site-files | xargs wc -l
|
| 32 | }
|
| 33 |
|
| 34 | test-wwz() {
|
| 35 | echo 'TODO: download .wwz from CI'
|
| 36 | }
|
| 37 |
|
| 38 | task-five "$@"
|