Highlight python-docx with regex and spacy

I want to highlight regex pattern in docx files in a folder using python-docx. I am able to achieve it through the normal regex code below.

Issue comes when I want to achieve the same through spacy nlp.

from docx import Document
from docx.enum.text import WD_COLOR_INDEX
import pandas as pd
import os
import re

import spacy
nlp = spacy.load("en_core_web_sm")

path = r"/home/coder/Documents/"
doc1 = Document('test.docx')
doc = nlp(doc1)

#re_highlight = re.compile(r"[1-9][0-9]*|0") # This one works.
re_highlight = [token for token in doc if tok.like_num == "TRUE"]

for filename in os.listdir(path):
    if filename.endswith(".docx"):
        file = "/home/writer/Documents/" + filename
        print(file)
        for para in doc.paragraphs:
            text = para.text
            if len(re_highlight.findall(text)) > 0:
                matches = re_highlight.finditer(text)
                para.text = ''
                p3 = 0
                for match in matches:
                    p1 = p3
                    p2, p3 = match.span()
                    para.add_run(text[p1:p2])
                    run = para.add_run(text[p2:p3])
                    run.font.highlight_color = WD_COLOR_INDEX.YELLOW
                para.add_run(text[p3:])
        doc.save(file)

Error:

>raise ValueError(Errors.E1041.format(type=type(doc_like)))
ValueError: [E1041] Expected a string, Doc, or bytes as input, but got: <class 'docx.document.Document'>

I realize the doc doesn't have doc.paragraphs being nlp element. How to sort this problem?

Kindly help.

You can't do nlp(doc1) with doc1 being a Document object, you have to extract the text parts and work with them. I'd suggest something like the following instead (worked here for a sample file):

import re
from pathlib import Path
import spacy
from docx import Document
from docx.enum.text import WD_COLOR_INDEX

nlp = spacy.load("en_core_web_sm")

def highlight(text):
    tokens = (token.text for token in nlp(text) if token.like_num)
    return re.compile("|".join(sorted(tokens, key=len, reverse=True)))

path_in = Path("/home/coder/Documents/")    # Input folder  
path_out = Path("/home/writer/Documents/")  # Output folder  
for file in path_in.glob("*.docx"):
    print(f"Processing file '{file}' ... ", end="")
    doc = Document(file)
    for para in doc.paragraphs:
        text = para.text
        para.text = ""
        p3 = 0
        for match in highlight(text).finditer(text):
            p1 = p3
            p2, p3 = match.span()
            para.add_run(text[p1:p2])
            run = para.add_run(text[p2:p3])
            run.font.highlight_color = WD_COLOR_INDEX.YELLOW
        para.add_run(text[p3:])
    doc.save(path_out / file.name)
    print("done.")

There's a chance of accidental highlighting. If that happens, you could try to use

def highlight(text):
    tokens = (token.text for token in nlp(text) if token.like_num)
    pat = r"\b(?:" + "|".join(sorted(tokens, key=len, reverse=True)) + r")\b"
    return re.compile(pat)

instead.