Search and replace for text within a pdf, in Python

Question

I am writing mailmerge software as part of a Python web app.

I have a template called letter.pdf which was generated from a MS Word file and includes the text {name} where the resident's name will go. I also have a list of c. 100 residents' names.

What I want to do is to read in letter.pdf do a search for "{name}" and replace it with the resident's name (for each resident) then write the result to another pdf. I then want to gather all these pdfs together into a big pdf (one page per letter) which my web app's users will print out to create their letters.

Are there any Python libraries that will do this? I've looked at pdfrw and pdfminer but I couldn't see where they would be able to do it.

(NB: I also have the MS Word file, so if there was another way of using that, and not going through a pdf, that would also do the job.)

Answer 1

This can be done with PyPDF2 package. The implementation may depend on the original PDF template structure. But if the template is stable enough and isn't changed very often the replacement code shouldn't be generic but rather simple.

I did a small sketch on how you could replace the text inside a PDF file. It replaces all occurrences of PDF tokens to DOC.

import os
import argparse
from PyPDF2 import PdfFileReader, PdfFileWriter
from PyPDF2.generic import DecodedStreamObject, EncodedStreamObject


def replace_text(content, replacements = dict()):
    lines = content.splitlines()

    result = ""
    in_text = False

    for line in lines:
        if line == "BT":
            in_text = True

        elif line == "ET":
            in_text = False

        elif in_text:
            cmd = line[-2:]
            if cmd.lower() == 'tj':
                replaced_line = line
                for k, v in replacements.items():
                    replaced_line = replaced_line.replace(k, v)
                result += replaced_line + "\n"
            else:
                result += line + "\n"
            continue

        result += line + "\n"

    return result


def process_data(object, replacements):
    data = object.getData()
    decoded_data = data.decode('utf-8')

    replaced_data = replace_text(decoded_data, replacements)

    encoded_data = replaced_data.encode('utf-8')
    if object.decodedSelf is not None:
        object.decodedSelf.setData(encoded_data)
    else:
        object.setData(encoded_data)


if __name__ == "__main__":
    ap = argparse.ArgumentParser()
    ap.add_argument("-i", "--input", required=True, help="path to PDF document")
    args = vars(ap.parse_args())

    in_file = args["input"]
    filename_base = in_file.replace(os.path.splitext(in_file)[1], "")

    # Provide replacements list that you need here
    replacements = { 'PDF': 'DOC'}

    pdf = PdfFileReader(in_file)
    writer = PdfFileWriter()

    for page_number in range(0, pdf.getNumPages()):

        page = pdf.getPage(page_number)
        contents = page.getContents()

        if isinstance(contents, DecodedStreamObject) or isinstance(contents, EncodedStreamObject):
            process_data(contents, replacements)
        elif len(contents) > 0:
            for obj in contents:
                if isinstance(obj, DecodedStreamObject) or isinstance(obj, EncodedStreamObject):
                    streamObj = obj.getObject()
                    process_data(streamObj, replacements)

        writer.addPage(page)

    with open(filename_base + ".result.pdf", 'wb') as out_file:
        writer.write(out_file)

The results are

UPDATE 2021-03-21:

Updated the code example to handle DecodedStreamObject and EncodedStreamObject which actually contian data stream with text to update.

Answer 2

If @Dmytrio solution do not alter final PDF

Dymitrio's updated code example to handle DecodedStreamObject and EncodedStreamObject which actually contain data stream with text to update could run fine, but with a file different from example, was not able to alter pdf text content.

According to EDIT 3, from How to replace text in a PDF using Python?:

By inserting page[NameObject("/Contents")] = contents.decodedSelf before writer.addPage(page), we force pyPDF2 to update content of the page object.

This way I was able to overcome this problem and replace text from pdf file.

Final code should look like this:

import os
import argparse
from PyPDF2 import PdfFileReader, PdfFileWriter
from PyPDF2.generic import DecodedStreamObject, EncodedStreamObject, NameObject


def replace_text(content, replacements = dict()):
    lines = content.splitlines()

    result = ""
    in_text = False

    for line in lines:
        if line == "BT":
            in_text = True

        elif line == "ET":
            in_text = False

        elif in_text:
            cmd = line[-2:]
            if cmd.lower() == 'tj':
                replaced_line = line
                for k, v in replacements.items():
                    replaced_line = replaced_line.replace(k, v)
                result += replaced_line + "\n"
            else:
                result += line + "\n"
            continue

        result += line + "\n"

    return result


def process_data(object, replacements):
    data = object.getData()
    decoded_data = data.decode('utf-8')

    replaced_data = replace_text(decoded_data, replacements)

    encoded_data = replaced_data.encode('utf-8')
    if object.decodedSelf is not None:
        object.decodedSelf.setData(encoded_data)
    else:
        object.setData(encoded_data)


if __name__ == "__main__":
    ap = argparse.ArgumentParser()
    ap.add_argument("-i", "--input", required=True, help="path to PDF document")
    args = vars(ap.parse_args())

    in_file = args["input"]
    filename_base = in_file.replace(os.path.splitext(in_file)[1], "")

    # Provide replacements list that you need here
    replacements = { 'PDF': 'DOC'}

    pdf = PdfFileReader(in_file)
    writer = PdfFileWriter()

    for page_number in range(0, pdf.getNumPages()):

        page = pdf.getPage(page_number)
        contents = page.getContents()

        if isinstance(contents, DecodedStreamObject) or isinstance(contents, EncodedStreamObject):
            process_data(contents, replacements)
        elif len(contents) > 0:
            for obj in contents:
                if isinstance(obj, DecodedStreamObject) or isinstance(obj, EncodedStreamObject):
                    streamObj = obj.getObject()
                    process_data(streamObj, replacements)

        # Force content replacement
        page[NameObject("/Contents")] = contents.decodedSelf
        writer.addPage(page)

    with open(filename_base + ".result.pdf", 'wb') as out_file:
        writer.write(out_file)

Important: from PyPDF2.generic import NameObject

Answer 3

1. Decompress the pdf to make parsing easier (solves many of the issues in the previous answer). I use pdftk. (If this step fails, one hack to pre-process the pdf is to open the pdf in OSX Preview, print it, and then choose save as pdf from the print menu. Then retry the command below.)

pdftk original.pdf output uncompressed.pdf uncompress

2. Parse and replace using PyPDF2.

from PyPDF2 import PdfFileReader, PdfFileWriter

replacements = [
    ("old string", "new string")
]

pdf = PdfFileReader(open("uncompressed.pdf", "rb"))
writer = PdfFileWriter() 

for page in pdf.pages:
    contents = page.getContents().getData()
    for (a,b) in replacements:
        contents = contents.replace(a.encode('utf-8'), b.encode('utf-8'))
    page.getContents().setData(contents)
    writer.addPage(page)
    
with open("modified.pdf", "wb") as f:
     writer.write(f)

3. [Optional] Re-compress the pdf.

pdftk modified.pdf output recompressed.pdf compress

Answer 4

Here is a solution using the MS Word source file.

As trying to edit the pdf itself turned out to be too complicated for me because of the encoding errors, I went with the MS Word >> Pdf option.

1. Prepare MS Word template with {{input_fields}}
2. Fill in the template with data
3. Convert the filled in MS Word file to PDF

The DocxTemplate module uses jinja like syntax: {{variable_name}}

In my solution I use an intermediate temp file. I tried to get rid of this step using BytesIO/StringIO to virtualize this step only in memory, but haven't make that work yet.

Here is an easy and working solution to perform the required task:

import os
import comtypes.client
from pathlib import Path
from docxtpl import DocxTemplate
import random


# CFG
in_file_path = "files/template.docx"
temp_file_path = "files/"+str(random.randint(0,50))+".docx"
out_file_path = "files/output.pdf"


# Fill in text
data_to_fill = {'Field_name' : "John Tester",
                  'Field_ocupation' : "Test tester",
                  'Field_address' : "Test Address 123",
                  }

template = DocxTemplate(Path(in_file_path))
template.render(data_to_fill)
template.save(Path(temp_file_path))

# Convert to PDF
wdFormatPDF = 17

in_file = os.path.abspath(Path(temp_file_path))
out_file = os.path.abspath(Path(out_file_path))

word = comtypes.client.CreateObject('Word.Application')
doc = word.Documents.Open(in_file)
doc.SaveAs(out_file, FileFormat=wdFormatPDF)
doc.Close()
word.Quit()

# Get rid of the temp file
os.remove(Path(temp_file_path))