def download_srt_from_3rd_party(lang_codes=None, **kwargs):
"""Download subtitles specified by command line args"""
lang_codes = lang_codes or get_all_prepped_lang_codes()
bad_languages = {}
for lang_code in lang_codes:
lang_code = lcode_to_ietf(lang_code)
try:
lang_map_filepath = get_lang_map_filepath(lang_code)
if not os.path.exists(lang_map_filepath):
videos = {} # happens if an unknown set for subtitles.
else:
with open(lang_map_filepath, "r") as fp:
videos = json.load(fp)
except Exception as e:
error_msg = "Error in subtitles metadata file for %s: %s" % (lang_code, e)
logging.error(error_msg)
bad_languages[lang_code] = error_msg
continue
try:
download_if_criteria_met(videos, lang_code=lang_code, **kwargs)
except Exception as e:
error_msg = "Error downloading subtitles for %s: %s" % (lang_code, e)
logging.error(error_msg)
bad_languages[lang_code] = error_msg
continue
# now report final results
if bad_languages:
outstr = "Failed to download subtitles for the following languages: %s" % (bad_languages.keys())
outstr += "\n" + str(bad_languages)
logging.error(outstr)
def update_json(youtube_id, lang_code, downloaded, api_response, time_of_attempt):
"""Update language_srt_map to reflect download status
lang_code in IETF format
"""
# Open JSON file
filepath = get_lang_map_filepath(lang_code)
try:
with open(filepath, "r") as fp:
language_srt_map = json.load(fp)
except Exception as e:
logging.error("Something went wrong while trying to open the json file (%s): %s" % (filepath, e))
return False
# create updated entry
entry = language_srt_map[youtube_id]
entry["downloaded"] = downloaded
entry["api_response"] = api_response
entry["last_attempt"] = time_of_attempt
if api_response == "success":
entry["last_success"] = time_of_attempt
# update full-size JSON with new information
language_srt_map[youtube_id].update(entry)
# write it to file
json_file = open(filepath, "wb")
json_file.write(json.dumps(language_srt_map))
json_file.close()
logging.debug("File updated.")
return True
def clear_subtitles_cache(lang_codes=None, locale_root=LOCALE_ROOT):
"""
Language codes will be converted to django format (e.g. en_US)
"""
lang_codes = lang_codes or os.listdir(locale_root)
for lang_code in lang_codes:
lang_code = lcode_to_ietf(lang_code)
# Clear the status file
lm_file = get_lang_map_filepath(lang_code)
with open(lm_file, "r") as fp:
download_status = json.load(fp)
for key in download_status:
download_status[key] = {
u"downloaded": False,
u"last_success": u"",
u"last_attempt": u"",
u"api_response": u"",
}
with open(lm_file, "w") as fp:
json.dump(download_status, fp)
# Delete all srt files
srt_path = get_srt_path(lang_code)
if os.path.exists(srt_path):
shutil.rmtree(srt_path)
def update_language_srt_map(map_file=SRTS_JSON_FILEPATH):
"""
Translate the srts_remote_availability dictionary into language specific files
that can be used by the cache_subtitles command.
Note: srt map deals with amara, so uses ietf codes (e.g. en-us)
"""
# Load the current download status
try:
with open(map_file) as fp:
api_info_map = json.load(fp)
except Exception as e:
# Must be corrupted; start from scratch!
logging.warn("Could not open %s for updates; starting from scratch. Error=%s" % (map_file, e))
api_info_map = {}
# Next we want to iterate through those and create a big srt dictionary organized by language code
remote_availability_map = {}
for youtube_id, data in api_info_map.items():
languages = data.get("language_codes", [])
for lang_code in languages:
lang_code = lcode_to_ietf(lang_code)
if not lang_code in remote_availability_map:
#logging.info("Creating language section '%s'" % lang_code)
remote_availability_map[lang_code] = {}
# This entry will be valid if it's new, otherwise it will be overwitten later
remote_availability_map[lang_code][youtube_id] = {
"downloaded": False,
"api_response": "",
"last_attempt": "",
"last_success": "",
}
# Finally we need to iterate through that dictionary and create individual files for each language code
for lang_code, new_data in remote_availability_map.items():
# Try to open previous language file
lang_map_filepath = get_lang_map_filepath(lang_code)
if not os.path.exists(lang_map_filepath):
lang_map = {}
else:
try:
with open(lang_map_filepath, "r") as fp:
lang_map = json.load(fp)
except Exception as e:
logging.error("Language download status mapping for (%s) is corrupted (%s), rewriting it." % (lang_code, e))
lang_map = {}
# First, check to see if it's empty (e.g. no subtitles available for any videos)
if not new_data:
logging.info("Subtitle support for %s has been terminated; removing." % lang_code)
if os.path.exists(lang_map_filepath):
os.remove(lang_map_filepath)
continue
# Compare how many empty entries you are adding and add them to master map
old_yt_ids = set(new_data.keys())
new_yt_ids = set(lang_map.keys())
yt_ids_to_add = set(new_data.keys()) - set(lang_map.keys())
yt_ids_to_delete = set(lang_map.keys()) - set(new_data.keys())
if yt_ids_to_add:
logging.info("Adding %d new YouTube IDs to language (%s)" % (len(yt_ids_to_add), lang_code))
for yt_id in yt_ids_to_add:
lang_map[yt_id] = new_data.get(yt_id)
if yt_ids_to_delete:
logging.info("Deleting %d old YouTube IDs from language (%s) because they are no longer supported." % (len(yt_ids_to_delete), lang_code))
for yt_id in yt_ids_to_delete:
lang_map.pop(yt_id, None)
# Write the new file to the correct location
logging.debug("Writing %s" % lang_map_filepath)
ensure_dir(os.path.dirname(lang_map_filepath))
with open(lang_map_filepath, 'w') as outfile:
json.dump(lang_map, outfile)
# Update the big mapping with the most accurate numbers
remote_availability_map[lang_code].update(lang_map)
# Finally, remove any files not found in the current map at all.
if lang_map_filepath:
for filename in os.listdir(os.path.dirname(lang_map_filepath)):
lang_code = lang_code = filename.split("_")[0]
if not lang_code in remote_availability_map:
file_to_remove = get_lang_map_filepath(lang_code)
logging.info("Subtitle support for %s has been terminated; removing." % lang_code)
if os.path.exists(file_to_remove):
os.remove(file_to_remove)
else:
logging.warn("Subtitles metadata for %s not found; skipping deletion of non-existent file %s." % (lang_code, file_to_remove))
return remote_availability_map
