| | |
| | """pixelconversion.ipynb |
| | |
| | Automatically generated by Colaboratory. |
| | |
| | Original file is located at |
| | https://colab.research.google.com/drive/1hfdgkYOw8w6DdJqsZx8txmw8INUGxesl |
| | """ |
| |
|
| |
|
| | |
| |
|
| | """### Imports""" |
| |
|
| | import fitz |
| | from PIL import Image |
| | import numpy as np |
| | import cv2 |
| | import db |
| | import tsadropboxretrieval |
| | import fitz |
| | from io import BytesIO |
| |
|
| | |
| | |
| |
|
| | def is_content_rotated(docpage): |
| | docpage.draw_rect([250,250,docpage.mediabox.width-500,docpage.mediabox.height-500], color = (0,0,0), width = 1,fill=(0,0,0),fill_opacity=1) |
| | pix = docpage.get_pixmap() |
| | image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples) |
| | img = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2GRAY) |
| | kernel = np.ones((3,3),np.uint8) |
| | img = cv2.dilate(img,kernel, iterations=6) |
| | img=cv2.threshold(img, 0, 255, cv2.THRESH_BINARY_INV)[1] |
| | |
| | contours, _ = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) |
| | |
| | for contour in contours: |
| | rect = cv2.minAreaRect(contour) |
| | angle = rect[-1] |
| | |
| | if angle != 0.0 and angle != 90.0 and angle != 180.0 and angle!=-90.0 and angle!=-180.0: |
| | print("Page content appears visually rotated.") |
| | return True |
| | print("Page content does not appear visually rotated.") |
| | return False |
| |
|
| | def drawisrotated(data=0,pdf_content=0,dpi=0): |
| | if pdf_content: |
| | docPath = fitz.open(stream=pdf_content, filetype="pdf") |
| | else: |
| | docPath = fitz.open("pdf",data) |
| | pageDocPath=docPath[0] |
| | |
| | bool_T_F=is_content_rotated(pageDocPath) |
| |
|
| | if bool_T_F==True: |
| | if pdf_content: |
| | docPath = fitz.open(stream=pdf_content, filetype="pdf") |
| | else: |
| | docPath = fitz.open("pdf",data) |
| | pageDocPath=docPath[0] |
| | pageDocPath.draw_rect([250,250,pageDocPath.mediabox.width-500,pageDocPath.mediabox.height-500], color = (75/255,0,130/255), width = 1,fill=(75/255,0,130/255),fill_opacity=0.9) |
| | if dpi: |
| | pix=pageDocPath.get_pixmap(dpi=dpi) |
| |
|
| | newwidth_ratio=pix.width/pageDocPath.mediabox.width |
| | newheight_ratio=pix.height/pageDocPath.mediabox.height |
| | else: |
| | newwidth_ratio=1 |
| | newheight_ratio=1 |
| | area=((pageDocPath.mediabox.height-500-250)*newheight_ratio) * ((pageDocPath.mediabox.width-500-250)*newwidth_ratio) |
| | perimeter=(((pageDocPath.mediabox.width-500-250)*newwidth_ratio) *2) + ( ((pageDocPath.mediabox.height-500-250)*newheight_ratio) *2) |
| |
|
| | if pageDocPath.mediabox.height >pageDocPath.mediabox.width: |
| | rectText=fitz.Rect(300, 200, (pageDocPath.mediabox.width-80),( pageDocPath.mediabox.width+80) ) |
| | else: |
| | rectText=fitz.Rect(500, 200, (pageDocPath.mediabox.height-80),( pageDocPath.mediabox.height+80) ) |
| |
|
| | text = """Scale Document""" |
| | annot1=pageDocPath.add_freetext_annot(rectText, text, fontsize=45, fontname='helv', border_color=(1,1,1), text_color=(1,1,1), rotate= pageDocPath.rotation, align=1) |
| | annot1.update() |
| | |
| | return docPath,area,perimeter |
| | else: |
| | if pdf_content: |
| | if dpi: |
| | doc,area,perimeter=openDrawPDF(pdf_content=pdf_content,dpi=dpi) |
| | return doc,area,perimeter |
| | else: |
| | doc,area,perimeter=openDrawPDF(pdf_content=pdf_content) |
| | return doc,area,perimeter |
| | else: |
| | if dpi: |
| | doc,area,perimeter=openDrawPDF(data,dpi=dpi) |
| | return doc,area,perimeter |
| | else: |
| | doc,area,perimeter=openDrawPDF(data) |
| | return doc,area,perimeter |
| | |
| |
|
| | |
| |
|
| |
|
| | def openDrawPDF(data=0,pdf_content=0,dpi=0): |
| | |
| | if pdf_content: |
| | doc = fitz.open(stream=pdf_content, filetype="pdf") |
| | else: |
| | doc = fitz.open("pdf",data) |
| | page = doc[0] |
| | print(page.rotation) |
| |
|
| | if page.rect.height > page.rect.width: |
| | rectText = fitz.Rect(300, 200, (page.mediabox.width - 80), (page.mediabox.width + 80)) |
| | else: |
| | rectText = fitz.Rect(500, 200, (page.mediabox.height - 80), (page.mediabox.height + 80)) |
| | page.draw_rect([0 + 10, 0 + 10, page.mediabox.width - 10, page.mediabox.height - 10],color=(75 / 255, 0, 130 / 255), width=1, fill=(75 / 255, 0, 130 / 255), fill_opacity=0.9) |
| |
|
| | if dpi: |
| | pix = page.get_pixmap(dpi=dpi) |
| | newwidth_ratio = pix.width / page.mediabox.width |
| | newheight_ratio = pix.height / page.mediabox.height |
| | else: |
| | print('No DPI specified') |
| | newwidth_ratio = 1 |
| | newheight_ratio = 1 |
| |
|
| | |
| | area = ((page.mediabox.width - 10 - 10) * newwidth_ratio) * ((page.mediabox.height - 10 - 10) * newheight_ratio) |
| | perimeter = (((page.mediabox.width - 10 - 10) * 2) * newwidth_ratio) + (((page.mediabox.height - 10 - 10) * 2) * newheight_ratio) |
| |
|
| | |
| | text = """Scale Document""" |
| | annot1 = page.add_freetext_annot(rectText, text, fontsize=45, fontname='helv', border_color=(1, 1, 1),text_color=(1, 1, 1), rotate=page.rotation, align=1) |
| | annot1.update() |
| | |
| | |
| | return doc, area, perimeter |