def refine_poetry(crawled_poetry, url, language):
'''
'''
# Get image of the poetry
f_image = download_image(get_poetry_img_url(url, language))
if f_image is False:
print 'ERR:: failed to download image'
return False
# Get text from poetry image : ocr_text
ret = image_to_text_for_reindeer(f_image, language)
if ret is False:
print 'ERR: falied to convert image to text'
return False
ocr_text = ret.strip()
# Delete the `f_image` file from the storage
try:
os.remove(f_image)
except:
print 'ERR:: failed to delete f_image'
pass
# Make list of lines from ``crawled_poetry`` and ``ocr_text``
#print '--- ocr text ---'
#print ocr_text
ocr_lines = [x for x in ocr_text.split('\n') if len(x) > 0]
cp = html_to_plain_text(crawled_poetry)
#print '--- crawled poetry ---'
#print cp
crawled_lines = [x for x in cp.split('\n') if len(x) > 0]
# Correct the line order of ``crawled_poetry``
poetry_lines = correct_poetry_lines_order(crawled_lines, ocr_lines)
poetry = '\n'.join(poetry_lines)
#print '--- Final poetry ---'
#print poetry
# Return
return poetry
def refine_poetry(crawled_poetry, url, language):
'''
'''
# Get image of the poetry
f_image = download_image(get_poetry_img_url(url, language))
if f_image is False:
print 'ERR:: failed to download image'
return False
# Get text from poetry image : ocr_text
ret = image_to_text_for_reindeer(f_image, language)
if ret is False:
print 'ERR: falied to convert image to text'
return False
ocr_text = ret.strip()
# Delete the `f_image` file from the storage
try:
os.remove(f_image)
except:
print 'ERR:: failed to delete f_image'
pass
# Make list of lines from ``crawled_poetry`` and ``ocr_text``
#print '--- ocr text ---'
#print ocr_text
ocr_lines = [x for x in ocr_text.split('\n') if len(x) > 0]
cp = html_to_plain_text(crawled_poetry)
#print '--- crawled poetry ---'
#print cp
crawled_lines = [x for x in cp.split('\n') if len(x) > 0]
# Correct the line order of ``crawled_poetry``
poetry_lines = correct_poetry_lines_order(crawled_lines, ocr_lines)
poetry = '\n'.join(poetry_lines)
#print '--- Final poetry ---'
#print poetry
# Return
return poetry
def article_to_poetry(request):
'''
Create poetries in Repository from given RawArticles
'''
##
# Check the parameters passed in the URL and process accordingly
creator_id = request.GET.get('creator', None)
article_ids = request.GET.get('articles', None)
# Only AJAX POST request is allowed
if request.is_ajax() and request.method == "POST":
if creator_id is None or article_ids is None:
# Return failure
print "Error: article_to_poetry: No parameter(s) passed."
res = {}
res['result'] = 'failure'
res['count'] = 0
return JsonResponse(res)
article_ids = [x.strip() for x in article_ids.split(',')]
# remove comma from the
if not article_ids[-1]:
article_ids = article_ids[:-1]
creator_id = creator_id.strip()
print "article_to_poetry: creator=", creator_id, "articles=", article_ids
count = 0
try:
# Get the `Person` object having id=`creator_id`
# Return failure if not found
person = Person.objects.get(pk=creator_id)
if person:
# Add all 'articles' to Poetry with 'person' as 'creator' field
print "article_to_poetry: found ", person.name, " in person@repository"
ct = ContentType.objects.get(app_label="repository",
model="poetry")
for article_id in article_ids:
article = RawArticle.objects.get(pk=article_id)
if article:
# Create a new poetry
poetry = Poetry()
poetry.name = article.title
poetry.language = article.language
poetry.body = html_to_plain_text(article.content)
poetry.creator = person
poetry.same_as = article.source_url
poetry.added_by = request.user
poetry.modified_by = request.user
poetry.date_modified = timezone.now()
poetry.save()
print "article_to_poetry: RawArticle ", article_id, "-> Poetry ", poetry.id
# Make the raw_article valid
article.valid = True
article.save()
count += 1
##
# Send signal to log the action
sig_action.send(request.user,
timestamp=poetry.date_added,
verb=VERBS['ADDITION'],
content_type=ct,
object_id=poetry.pk,
object_repr=poetry.name,
change_message=None,
public=True)
# Return success with number of poetry made
print "article_to_poetry: Total", count
res = {}
res['result'] = 'success'
res['count'] = count
return JsonResponse(res)
else:
# Return failure, means You have to add the entry of Person first
print "No such creator/person found in person@repository"
res = {}
res['result'] = 'failure'
res['count'] = 0
return JsonResponse(res)
except:
print("Error: Unexpected error:", sys.exc_info()[0])
for frame in traceback.extract_tb(sys.exc_info()[2]):
fname, lineno, fn, text = frame
print("DBG:: Error in %s on line %d" % (fname, lineno))
res = {}
res['result'] = 'error'
res['count'] = 0
return JsonResponse(res)
return HttpResponseForbidden()
def article_to_poetry(request):
'''
Create poetries in Repository from given RawArticles
'''
##
# Check the parameters passed in the URL and process accordingly
creator_id = request.GET.get('creator', None)
article_ids = request.GET.get('articles', None)
# Only AJAX POST request is allowed
if request.is_ajax() and request.method == "POST":
if creator_id is None or article_ids is None:
# Return failure
print "Error: article_to_poetry: No parameter(s) passed."
res = {}
res['result'] = 'failure'
res['count'] = 0
return JsonResponse(res)
article_ids = [x.strip() for x in article_ids.split(',')]
# remove comma from the
if not article_ids[-1]:
article_ids = article_ids[:-1]
creator_id = creator_id.strip()
print "article_to_poetry: creator=", creator_id, "articles=", article_ids
count = 0
try:
# Get the `Person` object having id=`creator_id`
# Return failure if not found
person = Person.objects.get(pk=creator_id)
if person:
# Add all 'articles' to Poetry with 'person' as 'creator' field
print "article_to_poetry: found ", person.name, " in person@repository"
ct = ContentType.objects.get(app_label="repository", model="poetry")
for article_id in article_ids:
article = RawArticle.objects.get(pk=article_id)
if article:
# Create a new poetry
poetry = Poetry()
poetry.name = article.title
poetry.language = article.language
poetry.body = html_to_plain_text(article.content)
poetry.creator = person
poetry.same_as = article.source_url
poetry.added_by = request.user
poetry.modified_by = request.user
poetry.date_modified = timezone.now()
poetry.save()
print "article_to_poetry: RawArticle ", article_id, "-> Poetry ", poetry.id
# Make the raw_article valid
article.valid = True
article.save()
count += 1
##
# Send signal to log the action
sig_action.send(request.user,
timestamp = poetry.date_added,
verb = VERBS['ADDITION'],
content_type = ct,
object_id = poetry.pk,
object_repr = poetry.name,
change_message = None,
public=True)
# Return success with number of poetry made
print "article_to_poetry: Total", count
res = {}
res['result'] = 'success'
res['count'] = count
return JsonResponse(res)
else:
# Return failure, means You have to add the entry of Person first
print "No such creator/person found in person@repository"
res = {}
res['result'] = 'failure'
res['count'] = 0
return JsonResponse(res)
except:
print ("Error: Unexpected error:", sys.exc_info()[0])
for frame in traceback.extract_tb(sys.exc_info()[2]):
fname,lineno,fn,text = frame
print ("DBG:: Error in %s on line %d" % (fname, lineno))
res = {}
res['result'] = 'error'
res['count'] = 0
return JsonResponse(res)
return HttpResponseForbidden()
