mainpdfparser.py

from pdfminer.layout import LAParams, LTTextBox, LTChar,  LTTextLine, LTText, LTImage, LTCurve, LTFigure, LTRect, LTAnno
from pdfminer.pdfpage import PDFPage
from pdfminer.pdfinterp import PDFResourceManager
from pdfminer.pdfinterp import PDFPageInterpreter
from pdfminer.converter import PDFPageAggregator

#PDFPlumber
import numpy as np
import pdfplumber



#helper function
def is_letter(char):
    if (char.isalpha() == True) or (char.isdigit() == True) or char == "’" or char == "'":
        return True
    else:
        return False


#IMPORTANT: enter main file name here as a path to a pdf. Include the .pdf ending. It is a string.
#sample path to file: "C:/Users/aland/Documents/Rice Documents/Rice Freshman Year/PDF Parsing/bubbleballot01.pdf"
main_file = "/Users/arianawang/Documents/Rice/Sample/CHIL bubbleballot01 (1).pdf"





#Main Code for PDFMINER TEXT LOCATION. GO TO LINE 97 for CURVE DETECTION


fp = open(main_file, 'rb')
# fp = open("C:/Users/aland/Documents/Rice Documents/Rice Freshman Year/PDF Parsing/ESS_paper_ballot.pdf", 'rb')
# fp = open("C:/Users/aland/Documents/Rice Documents/Rice Freshman Year/PDF Parsing/Hart_paper_ballot.pdf", 'rb')

rsrcmgr = PDFResourceManager()
laparams = LAParams()
device = PDFPageAggregator(rsrcmgr, laparams=laparams)
interpreter = PDFPageInterpreter(rsrcmgr, device)
pages = PDFPage.get_pages(fp)

for page in pages:
    print('Processing next page...')
    interpreter.process_page(page)
    layout = device.get_result()
    
    #go through every container
    for lobj in layout:
        #if container is textbox,
        if isinstance(lobj, LTTextBox):
            """ x, y, text = lobj.bbox[0], lobj.bbox[3], lobj.get_text()
            print('At %r is text: %s' % ((x, y), text)) """
            #go through everything in the textbox.
            for box_obj in lobj:
                #if in the textbox, there is a textline, go through it.
                if isinstance(box_obj, LTTextLine):
                    """ x, y, text = box_obj.bbox[0], box_obj.bbox[3], box_obj.get_text()
                    print('At %r is text: %s' % ((x, y), text)) """
                    #preliminary flag to start process of finding a singular word
                    first_char_found = False
                    #go through all the characters inthe textline.
                    for char_obj in box_obj:
                        #if we haven't found the first character in whatever word, initialize variables to start (or start over)
                        if first_char_found == False:
                            string_word = ""
                            xlocation_first_letter = None
                            ylocation_first_letter = None
                        #after checking first_char_found flag, now check if this current char is a letter and not a word ending char
                        if isinstance(char_obj, LTText) and (is_letter(char_obj.get_text()) == True):
                            #if this is the first letter in a word, flip the first_char_found flag, concat to string, and record the location of this first letter.
                            if first_char_found == False:
                                first_char_found = True
                                string_word += char_obj.get_text()
                                xlocation_first_letter = char_obj.bbox[0]
                                ylocation_first_letter = char_obj.bbox[3]
                            #or else, this is just a letter in the middle of the word. So just concat to string.
                            else:
                                string_word += char_obj.get_text()
                                xlocation_last_letter = char_obj.bbox[0]
                                ylocation_last_letter = char_obj.bbox[3]
                        #if this current character is a word ending character (space, comma, etc add more!), then now we print word and location, and start again.
                        # elif first_char_found == True and is_word_ending_char(char_obj.get_text()):
                        elif (isinstance(char_obj, LTText)) and (first_char_found == True):
                            xlocation_average = (xlocation_first_letter + xlocation_last_letter) / 2
                            ylocation_average = (ylocation_first_letter + ylocation_last_letter) / 2
                            print("At %r is the word: %s" % ((xlocation_average, ylocation_average), string_word))
                            first_char_found = False





#PDFPLUMBER CURVE DETECTION
# Loading the file
# sample_ballot = "/Users/arianawang/Documents/Rice/Sample/CHIL bubbleballot01 (1).pdf"

sample_ballot = main_file

# Choosing page of ballot to run the code. Input different index for diff pages.
report = pdfplumber.open(sample_ballot).pages[0]
# Create the image to save
im = report.to_image()

# Colors that the coordinates will be mapped in
colors = [ "gray", "red", "blue", "green" ]

# Iterate through the detected curves
for i, curve in enumerate(report.curves):
    # If the curve detected is not a line (i.e. two points on the curve detected)
    if len(curve["points"]) != 2:
        # The color chosen
        stroke = colors[i%len(colors)]
        x_values = 0
        y_values = 0
        # Sums up the values of the curves
        for points in curve["points"]:
            x_values += points[0]
            y_values += points[1] 
        # Calculates the midpoint of the bubble
        midpoint = [(round((x_values/len(curve["points"]))), round((y_values/len(curve["points"]))))]
        # Marks the curve
        im.draw_circles(midpoint, radius=3, stroke=stroke, fill="white")

# Saves the new ballot
im.save("/Users/arianawang/Documents/Rice/Sample/CHIL bubbleballot01 testthree.png", format="PNG")

# Github link: https://github.com/jsvine/pdfplumber/blob/stable/examples/notebooks/ag-energy-roundup-curves.ipynb