def read_document(path,
lang='eng',
spellcheck=False,
event="ocr_progress_bar"):
"""Call Tesseract OCR to extract the text from a document."""
from PIL import Image
import requests
import tesserocr
if path is None:
return None
if not lang_available(lang):
frappe.msgprint(frappe._(
"The selected language is not available. Please contact your administrator."
),
raise_exception=True)
frappe.publish_realtime(event, {"progress": "0"}, user=frappe.session.user)
if path.startswith('/assets/'):
# from public folder
fullpath = os.path.abspath(path)
elif path.startswith('/files/'):
# public file
fullpath = frappe.get_site_path() + '/public' + path
elif path.startswith('/private/files/'):
# private file
fullpath = frappe.get_site_path() + path
elif path.startswith('/'):
# local file (mostly for tests)
fullpath = os.path.abspath(path)
else:
# external link
fullpath = requests.get(path, stream=True).raw
ocr = frappe.get_doc("OCR Settings")
text = " "
with tesserocr.PyTessBaseAPI(lang=lang) as api:
if path.endswith('.pdf'):
from wand.image import Image as wi
# https://stackoverflow.com/questions/43072050/pyocr-with-tesseract-runs-out-of-memory
with wi(filename=fullpath, resolution=ocr.pdf_resolution) as pdf:
pdf_image = pdf.convert('jpeg')
i = 0
size = len(pdf_image.sequence) * 3
for img in pdf_image.sequence:
with wi(image=img) as img_page:
image_blob = img_page.make_blob('jpeg')
frappe.publish_realtime(event, {"progress": [i, size]},
user=frappe.session.user)
i += 1
recognized_text = " "
image = Image.open(io.BytesIO(image_blob))
api.SetImage(image)
frappe.publish_realtime(event, {"progress": [i, size]},
user=frappe.session.user)
i += 1
recognized_text = api.GetUTF8Text()
text = text + recognized_text
frappe.publish_realtime(event, {"progress": [i, size]},
user=frappe.session.user)
i += 1
else:
image = Image.open(fullpath)
api.SetImage(image)
frappe.publish_realtime(event, {"progress": [33, 100]},
user=frappe.session.user)
text = api.GetUTF8Text()
frappe.publish_realtime(event, {"progress": [66, 100]},
user=frappe.session.user)
if spellcheck:
text = get_spellchecked_text(text, lang)
frappe.publish_realtime(event, {"progress": [100, 100]},
user=frappe.session.user)
return text
def test_666_lang_available(self):
self.assertFalse(lang_available("666"))
def test_osd_lang_available(self):
self.assertTrue(lang_available("osd"))
def test_equ_lang_available(self):
self.assertTrue(lang_available("equ"))
def test_eng_lang_available(self):
self.assertTrue(lang_available("eng"))
