pdf

pdf-and-word/miner.py

@@ -0,0 +1,33 @@
+# this is for doing some math operations
+import math
+# this is for handling the PDF operations
+import fitz
+# importing PhotoImage from tkinter
+from tkinter import PhotoImage
+
+class PDFMiner:
+    def __init__(self, filepath):
+        # creating the file path
+        self.filepath = filepath
+        # opening the pdf document
+        self.pdf = fitz.open(self.filepath)
+        # loading the first page of the pdf document
+        self.first_page = self.pdf.load_page(0)
+        # getting the height and width of the first page
+        self.width, self.height = self.first_page.rect.width, self.first_page.rect.height
+        # initializing the zoom values of the page
+        zoomdict = {800:0.8, 700:0.6, 600:1.0, 500:1.0}
+        # getting the width value
+        width = int(math.floor(self.width / 100.0) * 100)
+        # zooming the page
+        self.zoom = zoomdict[width]
+
+    # this will get the metadata from the document like 
+    # author, name of document, number of pages  
+    def get_metadata(self):
+        # getting metadata from the open PDF document
+        metadata = self.pdf.metadata
+        # getting number of pages from the open PDF document
+        numPages = self.pdf.page_count
+        # returning the metadata and the numPages
+        return metadata, numPages

pdf-and-word/pdfviewer.py

@@ -0,0 +1,107 @@
+# importing everything from tkinter
+from tkinter import *
+# importing ttk for styling widgets from tkinter
+from tkinter import ttk
+# importing filedialog from tkinter
+from tkinter import filedialog as fd
+# importing os module
+import os
+
+# creating a class called PDFViewer
+class PDFViewer:
+    # initializing the __init__ / special method
+    def __init__(self, master):
+        # creating the window
+        self.master = master
+        # gives title to the main window
+        self.master.title('PDF Viewer')
+        # gives dimensions to main window
+        self.master.geometry('580x520+440+180')
+        # this disables the minimize/maximize button on the main window
+        self.master.resizable(width = 0, height = 0)
+        # loads the icon and adds it to the main window
+        self.master.iconbitmap(self.master, 'pdf_file_icon.ico')
+
+        # creating the menu
+        self.menu = Menu(self.master)
+        # adding it to the main window
+        self.master.config(menu=self.menu)
+        # creating a sub menu
+        self.filemenu = Menu(self.menu)
+        # giving the sub menu a label
+        self.menu.add_cascade(label="File", menu=self.filemenu)
+        # adding a two buttons to the sub menus
+        self.filemenu.add_command(label="Open File")
+        self.filemenu.add_command(label="Exit", command=self.master.destroy)
+
+        # creating the top frame
+        self.top_frame = ttk.Frame(self.master, width=580, height=460)
+        # placing the frame using inside main window using grid()
+        self.top_frame.grid(row=0, column=0)
+        # the frame will not propagate
+        self.top_frame.grid_propagate(False)
+        # creating the bottom frame
+        self.bottom_frame = ttk.Frame(self.master, width=580, height=50)
+        # placing the frame using inside main window using grid()
+        self.bottom_frame.grid(row=1, column=0)
+        # the frame will not propagate
+        self.bottom_frame.grid_propagate(False)
+
+        # creating a vertical scrollbar
+        self.scrolly = Scrollbar(self.top_frame, orient=VERTICAL)
+        # adding the scrollbar
+        self.scrolly.grid(row=0, column=1, sticky=(N,S))
+        # creating a horizontal scrollbar
+        self.scrollx = Scrollbar(self.top_frame, orient=HORIZONTAL)
+        # adding the scrollbar
+        self.scrollx.grid(row=1, column=0, sticky=(W, E))
+
+        # creating the canvas for display the PDF pages
+        self.output = Canvas(self.top_frame, bg='#ECE8F3', width=560, height=435)
+        # inserting both vertical and horizontal scrollbars to the canvas
+        self.output.configure(yscrollcommand=self.scrolly.set, xscrollcommand=self.scrollx.set)
+        # adding the canvas
+        self.output.grid(row=0, column=0)
+        # configuring the horizontal scrollbar to the canvas
+        self.scrolly.configure(command=self.output.yview)
+        # configuring the vertical scrollbar to the canvas
+        self.scrollx.configure(command=self.output.xview)
+
+        # loading the button icons
+        self.uparrow_icon = PhotoImage(file='uparrow.png')
+        self.downarrow_icon = PhotoImage(file='downarrow.png')
+        # resizing the icons to fit on buttons
+        self.uparrow = self.uparrow_icon.subsample(3, 3)
+        self.downarrow = self.downarrow_icon.subsample(3, 3)
+        # creating an up button with an icon
+        self.upbutton = ttk.Button(self.bottom_frame, image=self.uparrow)
+        # adding the button
+        self.upbutton.grid(row=0, column=1, padx=(270, 5), pady=8)
+        # creating a down button with an icon
+        self.downbutton = ttk.Button(self.bottom_frame, image=self.downarrow)
+        # adding the button
+        self.downbutton.grid(row=0, column=3, pady=8)
+        # label for displaying page numbers
+        self.page_label = ttk.Label(self.bottom_frame, text='page')
+        # adding the label
+        self.page_label.grid(row=0, column=4, padx=5)
+
+        # path for the pdf doc
+        self.path = None
+        # state of the pdf doc, open or closed
+        self.fileisopen = None
+        # author of the pdf doc
+        self.author = None
+        # name for the pdf doc
+        self.name = None
+        # the current page for the pdf
+        self.current_page = 0
+        # total number of pages for the pdf doc
+        self.numPages = None
+
+# creating the root window using Tk() class
+root = Tk()
+# instantiating/creating object app for class PDFViewer
+app = PDFViewer(root)
+# calling the mainloop to run the app infinitely until user closes it
+root.mainloop()

pdf-and-word/read-pdf.py

@@ -0,0 +1,7 @@
+import PyPDF2
+
+pdfFileObj = open('pdf-and-word\meetingminutes.pdf', 'rb')
+
+pdfReader = PyPDF2.PdfReader(pdfFileObj)
+print(len(pdfReader.pages))
+