def get_hocr(lang, title):
# FIXME, delete all no ocr and redo them with nb code lang.
if lang == 'nb':
lang = 'no'
if type(title) == type(u''):
title = title.encode('utf-8')
title = title.replace(' ', '_')
try:
if lang == 'bn':
title = unicode(title, 'utf-8')
page_nr = re.sub(u'^.*/([০-৯]+)$', '\\1', title)
book_name = re.sub(u'^(.*?)(/[০-৯]+)?$', '\\1', title)
book_name = book_name.encode('utf-8')
result = ord(page_nr[0]) - ord(u'০')
for ch in page_nr[1:]:
result *= 10
result += ord(ch) - ord(u'০')
page_nr = result
else:
page_nr = re.sub('^.*/([0-9]+)$', '\\1', title)
book_name = re.sub('^(.*?)(/[0-9]+)?$', '\\1', title)
page_nr = int(page_nr)
except:
return ret_val(1, "unable to extract page number from page: " + title)
path = cache_path(book_name, lang)
filename = path + 'page_%04d.hocr' % page_nr
# We support data built with different compress scheme than the one
# actually generated by the server
text = utils.uncompress_file(filename, ['bzip2', 'gzip', ''])
if text == None:
# not available, add a request to do this hocr so we build data
# lazilly but we filter here unsupported file type
if book_name.endswith('.djvu') or book_name.endswith('.pdf'):
import hocr_request
hocr_request.add_hocr_request(lang, book_name, True)
return ret_val(
1, "unable to locate file %s for page %s lang %s" %
(filename, book_name, lang))
# work-around https://code.google.com/p/tesseract-ocr/issues/detail?id=690&can=1&q=utf-8 a simple patch exists: https://code.google.com/p/tesseract-ocr/source/detail?r=736# but it's easier to do a double conversion to remove invalid utf8 rather than to maintain a patched version of tesseract.
text = unicode(text, 'utf-8', 'ignore')
text = text.encode('utf-8', 'ignore')
return ret_val(0, text)
def get_hocr(lang, title):
# FIXME, delete all no ocr and redo them with nb code lang.
if lang == 'nb':
lang = 'no'
if type(title) == type(u''):
title = title.encode('utf-8')
title = title.replace(' ', '_')
try:
if lang == 'bn':
title = unicode(title, 'utf-8')
page_nr = re.sub(u'^.*/([০-৯]+)$', '\\1', title)
book_name = re.sub(u'^(.*?)(/[০-৯]+)?$', '\\1', title)
book_name = book_name.encode('utf-8')
result = ord(page_nr[0]) - ord(u'০')
for ch in page_nr[1:]:
result *= 10
result += ord(ch) - ord(u'০')
page_nr = result
else:
page_nr = re.sub('^.*/([0-9]+)$', '\\1', title)
book_name = re.sub('^(.*?)(/[0-9]+)?$', '\\1', title)
page_nr = int(page_nr)
except:
return ret_val(1, "unable to extract page number from page: " + title)
path = cache_path(book_name, lang)
filename = path + 'page_%04d.hocr' % page_nr
# We support data built with different compress scheme than the one
# actually generated by the server
text = utils.uncompress_file(filename, [ 'bzip2', 'gzip', '' ])
if text == None:
# not available, add a request to do this hocr so we build data
# lazilly but we filter here unsupported file type
if book_name.endswith('.djvu') or book_name.endswith('.pdf'):
import hocr_request
hocr_request.add_hocr_request(lang, book_name, True)
return ret_val(1, "unable to locate file %s for page %s lang %s" % (filename, book_name, lang))
# work-around https://code.google.com/p/tesseract-ocr/issues/detail?id=690&can=1&q=utf-8 a simple patch exists: https://code.google.com/p/tesseract-ocr/source/detail?r=736# but it's easier to do a double conversion to remove invalid utf8 rather than to maintain a patched version of tesseract.
text = unicode(text, 'utf-8', 'ignore')
text = text.encode('utf-8', 'ignore')
return ret_val(0, text)
def __set_archives_to_analyze(self, mailing_list, archives):
archives_to_analyze = []
for archive in archives:
# Always set Gmane archives to analyze
if archive.url.find(GMANE_DOMAIN) == -1:
# Check if already analyzed
status = self.db.check_compressed_file(archive.url)
this_month = find_current_month(archive.url)
# If the file is for the current month, re-import to update.
# If already visited, ignore it.
if status == self.db.VISITED and not this_month:
self.__print_output('Already analyzed %s' % archive.url)
continue
# If not, set visited
# (before uncompressing, otherwise the db will point towards
# the uncompressed temporary file)
today = datetime.datetime.today().strftime(datetimefmt)
self.db.set_visited_url(archive.url, mailing_list.location, today,
self.db.NEW)
if archive.is_compressed():
try:
# Uncompress and get the raw filepaths
filepaths = uncompress_file(archive.filepath,
archive.compressed_type,
mailing_list.mbox_dir)
uncompressed_mboxes = [
MBoxArchive(fp, archive.url) for fp in filepaths
]
archives_to_analyze.extend(uncompressed_mboxes)
except IOError, e:
# It could be a plain file, so let's give it a chance
self.__print_output(
" ***WARNING: Uncompressing file %s - %s" %
(archive.filepath, str(e)))
archives_to_analyze.append(archive)
else:
archives_to_analyze.append(archive)
def _update_log(outputDir, subjectDir, err):
log_file = get_log_file(outputDir)
with open(log_file, "wr") as output:
for i in os.listdir(subjectDir):
if i == DS_STORE: continue # CPC2018
if not utils.is_fasta(i): continue
fullpath = subjectDir + "/" + i
if not os.path.isfile(fullpath): continue
fafile = os.path.abspath(fullpath)
name = os.path.splitext(i)[0]
extension = os.path.splitext(i)[1]
#check for gz files from ncbi
# TODO: rename "file" vars
if extension == ".gz":
fafile = utils.uncompress_file(fafile, subjectDir + "/" + name)
fafileheader = ""
with open(fafile, "r") as fafileopen:
fafileheader = fafileopen.readline().strip()
#read the files and check for fasta files
if ">" not in fafileheader[0]:
raise Exception("ERROR: " + name + " is not a fasta file")
else:
target_name = _get_target_name(fafileheader)
fasta_record = "\t".join([target_name, fafile, name])
#if VERBOSE: err.write(fasta_record+"\n")
output.write(fasta_record + "\n")
return
def __analyze_non_remote(self, dirname):
"""Walk recursively the directory looking for files,
and uncompress them. Then __analyze_local_directory is called."""
# Check if directory to stored uncompressed files already exists
mbox_dir = os.path.join(self.MBOX_DIR, dirname.lstrip('/'))
if not os.path.exists(mbox_dir):
os.makedirs(mbox_dir)
# Compressed files are left in their original location,
# because they can be uncompressed from that location
filepaths = []
for root, dirs, files in os.walk(dirname):
filepaths += [os.path.join(root, filename) for filename in files]
# If the file is for the current month (MailMan filename
# YYYY-MMM.txt.gz) don't mark as visited, and download again
# Assuming this is run daily, it's better to take yesterday's date,
# to ensure we get all of last month's email when the month rolls over.
yesterday= datetime.datetime.today() + datetime.timedelta(days=-1)
this_month= yesterday.strftime(mailmanfmt)
files_to_analyze = {}
url_list = []
for filepath in filepaths:
# Check if already analyzed
status = self.db.check_compressed_file(filepath)
# If the file is for the current month, reimport
current_month = -1 != filepath.find(this_month)
if current_month:
self.__print_output('Found substring %s in URL %s...' % \
(this_month, filepath))
# If already visited, ignore, unless it's for the current month
if status == self.db.VISITED and not current_month:
self.__print_output('Already analyzed %s' % filepath)
continue
# If not, set visited
# (before uncompressing, otherwise the db will point towards
# the uncompressed temporary file)
today = datetime.datetime.today().strftime(datetimefmt)
self.db.set_visited_url(filepath, dirname, today, self.db.NEW)
# Check if compressed
extension = check_compressed_file(filepath)
if extension:
# If compressed, uncompress and get the raw filepath
filepaths = uncompress_file(filepath, extension, mbox_dir)
# __uncompress_file returns a list containing
# the path to all the uncompressed files
# (for instance, a tar file may contain more than one file)
files_to_analyze.setdefault(filepath, []).extend(filepaths)
else:
# File was not uncompressed, so there is only
# one file to append
files_to_analyze.setdefault(filepath, []).append(filepath)
url_list.append(filepath)
# The archives are usually retrieved in descending
# chronological order (because the newest archives are always
# shown on the top of the archives)
# So we will analyze the list of files in the order inversed
# to the order in they were retrieved
url_list.reverse()
return self.__analyze_list_of_files(dirname, url_list,
files_to_analyze)
def __analyze_remote(self, url):
"""Download the archives from the remote url, stores and parses them."""
# Check directories to stored the archives
target = re.sub('^(http|ftp)[s]{0,1}://', '', url)
compressed_dir = os.path.join(self.COMPRESSED_DIR,
target)
mbox_dir = os.path.join(self.MBOX_DIR, target)
if not os.path.exists(compressed_dir):
os.makedirs(compressed_dir)
if not os.path.exists(mbox_dir):
os.makedirs(mbox_dir)
# If the file is for the current month (MailMan filename
# YYYY-MMM.txt.gz) don't mark as visited, and download again
# Assuming this is run daily, it's better to take yesterday's date,
# to ensure we get all of last month's email when the month rolls over.
yesterday = datetime.datetime.today() + datetime.timedelta(days=-1)
this_month = yesterday.strftime(mailmanfmt)
# Get all the links listed in the URL
htmlparser = MyHTMLParser(url, self.web_user, self.web_password)
links = htmlparser.get_mboxes_links()
filepaths = []
for link in links:
basename = os.path.basename(link)
destfilename = os.path.join(compressed_dir, basename)
# If the URL is for the current month, always retrieve.
# Otherwise, check visited status & local files first
if link.find(this_month) >= 0:
self.__print_output('Found substring %s in URL %s...' %
(this_month, link))
self.__print_output('Retrieving %s...' % link)
retrieve_remote_file(link, destfilename, self.web_user,
self.web_password)
elif os.path.exists(destfilename): # Check if already downloaded
self.__print_output('Already downloaded %s' % link)
else:
self.__print_output('Retrieving %s...' % link)
retrieve_remote_file(link, destfilename, self.web_user,
self.web_password)
filepaths.append((link, destfilename))
files_to_analyze = {}
url_list = []
for link, filepath in filepaths:
# Check if already analyzed
status = self.db.check_compressed_file(filepath)
# If the file is for the current month, reimport
current_month = -1 != filepath.find(this_month)
if current_month:
self.__print_output('Found substring %s in URL %s...' % \
(this_month, filepath))
# If already visited, ignore, unless it's for the current month
if status == self.db.VISITED and not current_month:
self.__print_output('Already analyzed %s' % filepath)
continue
# If not, set visited
# (before uncompressing, otherwise the db will point towards
# the uncompressed temporary file)
today = datetime.datetime.today().strftime(datetimefmt)
self.db.set_visited_url(link, url, today, self.db.NEW)
# Check if compressed
extension = check_compressed_file(filepath)
if extension:
# If compressed, uncompress and get the raw filepath
filepaths = uncompress_file(filepath, extension, mbox_dir)
# __uncompress_file returns a list containing
# the path to all the uncompressed files
# (for instance, a tar file may contain more than one file)
files_to_analyze.setdefault(link, []).extend(filepaths)
else:
# File was not uncompressed, so there is only
# one file to append
files_to_analyze.setdefault(link, []).append(filepath)
url_list.append(link)
# The archives are usually retrieved in descending
# chronological order (because the newest archives are always
# shown on the top of the archives)
# So we will analyze the list of files in the order inversed
# to the order in they were retrieved
url_list.reverse()
return self.__analyze_list_of_files(url, url_list, files_to_analyze)
