lgcosta/paperless-ngx
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update: ocr pdf

Browse Source
This commit is contained in:
otxtan@gmail.com 2024-05-18 05:12:13 +07:00
parent bcf3e8db7d
commit dfaf59d347
1 changed files with 10 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/paperless_ocr_custom/parsers.py
View File

@ -247,30 +247,28 @@ class RasterisedDocumentParser(DocumentParser):
packet = io.BytesIO() packet = io.BytesIO()
can = canvas.Canvas(packet, pagesize=letter) can = canvas.Canvas(packet, pagesize=letter)
page_image = page.to_image()
page_image.save(packet, "JPG")
pdfmetrics.registerFont(TTFont('Arial', font_path)) pdfmetrics.registerFont(TTFont('Arial', font_path))
width_api_img = data["pages"][page_num]["dimensions"][1]
height_api_img = data["pages"][page_num]["dimensions"][0]
rolate_height = height_api_img /page_height
rolate_width = width_api_img /page_width
for block in data["pages"][page_num]["blocks"]: for block in data["pages"][page_num]["blocks"]:
for line in block.get("lines", []): for line in block.get("lines", []):
for word in line.get("words", []): for word in line.get("words", []):
x1 = word["bbox"][0][0] x1 = word["bbox"][0][0] / float(rolate_width)
y1 = word["bbox"][0][1] y1 = word["bbox"][0][1] / float(rolate_height)
x2 = word["bbox"][1][0] x2 = word["bbox"][1][0] / float(rolate_width)
y2 = word["bbox"][1][1] y2 = word["bbox"][1][1] / float(rolate_height)
value = word["value"] value = word["value"]
font_size = (y2-y1) * 72 / 96 font_size = float(y2-y1) * 72 / 96
x_center_coordinates =x2 - (x2-x1)/2 x_center_coordinates =x2 - (x2-x1)/2
y_center_coordinates =y2 - (y2-y1)/2 y_center_coordinates =y2 - (y2-y1)/2
w = can.stringWidth(value, font_name, font_size) w = can.stringWidth(value, font_name, font_size)
self.log.debug('w:', )
can.setFont('Arial', font_size) can.setFont('Arial', font_size)
can.drawString(x_center_coordinates - w/2 , int(page_height) - y_center_coordinates - (font_size/2) , value) can.drawString(x_center_coordinates - w/2 , int(page_height) - y_center_coordinates - (font_size/2) , value)
can.showPage() can.showPage()
can.save() can.save()
packet.seek(0) packet.seek(0)
new_pdf = PdfReader(packet) new_pdf = PdfReader(packet)
page.merge_page(new_pdf.pages[0]) page.merge_page(new_pdf.pages[0])