How to fix ''UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 29815: character maps to ''?

Question

At the moment, I am trying to get a Python 3 program to do some manipulations with a text file filled with information, through the Spyder IDE/GUI. However, when trying to read the file I get the following error:

  File "<ipython-input-13-d81e1333b8cd>", line 77, in <module>
    parser(f)

  File "<ipython-input-13-d81e1333b8cd>", line 18, in parser
    data = infile.read()

  File "C:\ProgramData\Anaconda3\lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]

UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 29815: character maps to <undefined>

The code of the program is as follows:

import os

os.getcwd()

import glob
import re
import sqlite3
import csv

def parser(file):

    # Open a TXT file. Store all articles in a list. Each article is an item
    # of the list. Split articles based on the location of such string as
    # 'Document PRN0000020080617e46h00461'

    articles = []
    with open(file, 'r') as infile:
        data = infile.read()
    start = re.search(r'\n HD\n', data).start()
    for m in re.finditer(r'Document [a-zA-Z0-9]{25}\n', data):
        end = m.end()
        a = data[start:end].strip()
        a = '\n   ' + a
        articles.append(a)
        start = end

    # In each article, find all used Intelligence Indexing field codes. Extract
    # content of each used field code, and write to a CSV file.

    # All field codes (order matters)
    fields = ['HD', 'CR', 'WC', 'PD', 'ET', 'SN', 'SC', 'ED', 'PG', 'LA', 'CY', 'LP',
              'TD', 'CT', 'RF', 'CO', 'IN', 'NS', 'RE', 'IPC', 'IPD', 'PUB', 'AN']

    for a in articles:
        used = [f for f in fields if re.search(r'\n   ' + f + r'\n', a)]
        unused = [[i, f] for i, f in enumerate(fields) if not re.search(r'\n   ' + f + r'\n', a)]
        fields_pos = []
        for f in used:
            f_m = re.search(r'\n   ' + f + r'\n', a)
            f_pos = [f, f_m.start(), f_m.end()]
            fields_pos.append(f_pos)
        obs = []
        n = len(used)
        for i in range(0, n):
            used_f = fields_pos[i][0]
            start = fields_pos[i][2]
            if i < n - 1:
                end = fields_pos[i + 1][1]
            else:
                end = len(a)
            content = a[start:end].strip()
            obs.append(content)
        for f in unused:
            obs.insert(f[0], '')
        obs.insert(0, file.split('/')[-1].split('.')[0])  # insert Company ID, e.g., GVKEY
        # print(obs)
        cur.execute('''INSERT INTO articles
                       (id, hd, cr, wc, pd, et, sn, sc, ed, pg, la, cy, lp, td, ct, rf,
                       co, ina, ns, re, ipc, ipd, pub, an)
                       VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?,
                       ?, ?, ?, ?, ?, ?, ?, ?)''', obs)

# Write to SQLITE
conn = sqlite3.connect('factiva.db')
with conn:
    cur = conn.cursor()
    cur.execute('DROP TABLE IF EXISTS articles')
    # Mirror all field codes except changing 'IN' to 'INC' because it is an invalid name
    cur.execute('''CREATE TABLE articles
                   (nid integer primary key, id text, hd text, cr text, wc text, pd text,
                   et text, sn text, sc text, ed text, pg text, la text, cy text, lp text,
                   td text, ct text, rf text, co text, ina text, ns text, re text, ipc text,
                   ipd text, pub text, an text)''')
    for f in glob.glob('*.txt'):
        print(f)
        parser(f)

# Write to CSV to feed Stata
with open('factiva.csv', 'w', newline='') as csvfile:
    writer = csv.writer(csvfile)
    with conn:
        cur = conn.cursor()
        cur.execute('SELECT * FROM articles WHERE hd IS NOT NULL')
        colname = [desc[0] for desc in cur.description]
        writer.writerow(colname)
        for obs in cur.fetchall():
            writer.writerow(obs)

Answer 1

As you see from https://en.wikipedia.org/wiki/Windows-1252, the code 0x9D is not defined in CP1252.

The "error" is e.g. in your open function: you do not specify the encoding, so python (just in windows) will use some system encoding. In general, if you read a file that maybe was not create in the same machine, it is really better to specify the encoding.

I recommend to put also a coding also on your open for writing the csv. It is really better to be explicit.

I do no know the original file format, but adding to open , encoding='utf-8' is usually a good thing (and it is the default in Linux and MacOs).

Answer 2

Add encoding in the open statement For example:

f=open("filename.txt","r",encoding='utf-8')

Answer 3

The above did not work for me, try this instead: , errors='ignore' Worked wonders!

Answer 4

You can also try file = open(filename, 'rb') 'rb' translates to read binary if you wouldn't need to decode it. Say if you just want to upload to a website

Answer 5

errors='ignore' solved my headache in:

how to find the word "coma"in directories and subdirectories =

import os
rootdir=('K:\\0\\000.THU.EEG.nedc_tuh_eeg\\000edf.01_tcp_ar\\01_tcp_ar\\')
for folder, dirs, files in os.walk(rootdir):
    for file in files:
        if file.endswith('.txt'):
            fullpath = os.path.join(folder, file)
            with open(fullpath, 'r', errors='ignore') as f:
                for line in f:
                    if "coma" in line:
                        print(fullpath)
                        break

Answer 6

I dont believe that coding <errors=='ignore'> would be a good idea even if it works. Since you dont know what else it could be ignored, you should search for ways to bypass this problem without cutting off pieces of file.

I had this problem too when i attempted appending html as text into a file. You can try as i did, first return the content as bytes type, then convert it to string by decoding using 'utf-8'

converted_file = binary_file.decode('utf-8')