From 99b78a9ddf64dbb1e6dbf2e4251c5c19049fd141 Mon Sep 17 00:00:00 2001 From: Phuoc Cao Date: Sun, 3 Dec 2023 10:38:17 +0100 Subject: [PATCH] fixed cleanup function --- .../temp/clean_Nantes_anthologie_cinema_viet.htm | 11 ++++++++--- .../static/temp/clean_proposed_file_name.html | 5 ----- html_cleanup/static/temp/clean_www.icloud.com.html | 5 ----- html_cleanup/static/temp/errors_log | 12 +----------- html_cleanup/templates/view_log.jinja2 | 2 +- html_cleanup/views/default.py | 11 +++++++++-- 6 files changed, 19 insertions(+), 27 deletions(-) delete mode 100644 html_cleanup/static/temp/clean_proposed_file_name.html delete mode 100644 html_cleanup/static/temp/clean_www.icloud.com.html diff --git a/html_cleanup/static/temp/clean_Nantes_anthologie_cinema_viet.htm b/html_cleanup/static/temp/clean_Nantes_anthologie_cinema_viet.htm index 787a7d1..2e5f859 100644 --- a/html_cleanup/static/temp/clean_Nantes_anthologie_cinema_viet.htm +++ b/html_cleanup/static/temp/clean_Nantes_anthologie_cinema_viet.htm @@ -337,7 +337,11 @@ Mercredi 29.11, 14:45

-

+

https://www.3continents.com/fr/programme/2023/anthologie-du-cinema-vietnamien/

@@ -346,8 +350,9 @@ https://www.3continents.com/fr/programme/2023/anthologie-du-cinema-vietnamien/ -https://www.3continents.com/wp-content/uploads/f3c-prog-2023-40p-net-planche.pdf

+

https://www.3continents.com/wp-content/uploads/f3c-prog-2023-40p-net-planche.pdf

diff --git a/html_cleanup/static/temp/clean_proposed_file_name.html b/html_cleanup/static/temp/clean_proposed_file_name.html deleted file mode 100644 index 23b4f73..0000000 --- a/html_cleanup/static/temp/clean_proposed_file_name.html +++ /dev/null @@ -1,5 +0,0 @@ - - - -untitled - diff --git a/html_cleanup/static/temp/clean_www.icloud.com.html b/html_cleanup/static/temp/clean_www.icloud.com.html deleted file mode 100644 index 23b4f73..0000000 --- a/html_cleanup/static/temp/clean_www.icloud.com.html +++ /dev/null @@ -1,5 +0,0 @@ - - - -untitled - diff --git a/html_cleanup/static/temp/errors_log b/html_cleanup/static/temp/errors_log index 2f71fe6..a1174b9 100644 --- a/html_cleanup/static/temp/errors_log +++ b/html_cleanup/static/temp/errors_log @@ -1,11 +1 @@ -2023-11-29 17:22:16 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** -2023-11-29 17:23:56 [ Screenshot 2023-11-20 at 12.07.54.png ] Le format du fichier n'est pas valide. Téléchargement refusé. -30-11-2023 17:29 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** -30-11-2023 18:32 [ www.icloud.com.html ] *** Cleanup réussi. *** -02-12-2023 10:41 [ Nantes_anthologie_cinema_viet.htm ] Le format du fichier n'est pas valide (absence de tag ou ). Téléchargement refusé. -02-12-2023 10:44 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** -02-12-2023 10:45 [ cleanup_html.vbs.txt ] Le format du fichier n'est pas valide (absence de tag ou ). Téléchargement refusé. -02-12-2023 10:49 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** -02-12-2023 11:02 [ Nantes_anthologie_cinema_viet copy.htm ] *** Cleanup réussi. *** -02-12-2023 11:10 [ Nantes_anthologie_cinema_viet copy.htm ] *** Cleanup réussi. *** -02-12-2023 11:11 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** +03-12-2023 10:35 [ Nantes_anthologie_cinema_viet.htm ] *** Cleanup réussi. *** diff --git a/html_cleanup/templates/view_log.jinja2 b/html_cleanup/templates/view_log.jinja2 index f4cce28..7f73462 100644 --- a/html_cleanup/templates/view_log.jinja2 +++ b/html_cleanup/templates/view_log.jinja2 @@ -11,7 +11,7 @@ {% if not lines %}
- +
diff --git a/html_cleanup/views/default.py b/html_cleanup/views/default.py index ebd769b..9aff5b8 100644 --- a/html_cleanup/views/default.py +++ b/html_cleanup/views/default.py @@ -136,13 +136,19 @@ def clean_html(input_file, output_file): done = True skip_tag = False - tag = s.split(" ") + tag = s.split() if tag: tag[0] = tag[0].lower() if tag[0] in ["table", "div", "img", "a"]: - line_out += "<{}>".format(s) + line_out += "<{}".format(s) + if end_tag > 0: + # si end tag trouvé, fermer + line_out += ">" + else: + # sinon contine de recopier la ligne + skip_tag = False elif tag[0] in ["p"]: line_out += "

" elif tag[0] in ["b", "/b", "em", "/em", "i", "/i", "li", "li/", "/li", @@ -195,6 +201,7 @@ def view_log(request): lines = file.readlines() file.close() except Exception as e: + message = "Le fichier Journal est vide !" pass # lire tous les fichiers du dossier triés par date