Read all toots you've written (excluding Boosts).

ReadMyToots.py

# -*- coding: utf-8 -*-
"""
    ReadMyToots.py - Download the toots you wrote, in chronological order.

    Author:  BobC - https://mastodon.hasameli.com/@BobC
    License: "CC BY-NC-SA 4.0" https://creativecommons.org/licenses/by-nc-sa/4.0/

    Usage: python ReadMyToots.py [UserName InstanceName LoginEmail Password]+

    Parameters: UserName        - Your handle on this instance (no '@')
                InstanceName    - The instance address (no 'http://')
                LoginEmail      - Yup
                Password        - Yup

    Parameters must form complete sets. Multiple sets are supported.

    WARNING: No parameter validation is done. Blood & guts fly everywhere.

    Output: Status and progress messages go to stdout.
            Toots go to file: "[User_Name]@[InstanceName].txt"

    Dependencies: Be sure you have Mastodon.py installed:

        pip install mastodon.py

    This ***SHOULD*** work on ALL major Python 3.x platforms (Win, Lin, Mac).

"""
DEBUG = False   # Set to True for so much fun!

import sys

if __name__ != "__main__":
    print("This program is NOT a library and must NOT be imported!", file=sys.stderr)
    sys.exit()

# And now we will continue with our regularly scheduled program...
import os
import re
import textwrap
import pathlib as pl
import html
from mastodon import Mastodon

if DEBUG is True:
    import code     ####DEBUG
    import pprint   ####DEBUG


## Local functions:

if DEBUG is True:
    def pdict(dct, indent=1, spaces=3):
        """ Print a dict that may contain other dicts """
        tabs = ' '*spaces*indent
        for key, val in dct.items():
            if not isinstance(val, dict):
                try:
                    print("%s%s = "%(tabs, key), val)
                except:
                    print("%s%s = [unprintable]"%(tabs, key))
            else:
                print("%s%s = {dict}"%(tabs, key))
                pdict(val, indent=indent+1)
        return


def witext(string, indent=1, width=80, spaces=3):
    """ Return a wrapped and indented string """
    tabs = ' '*spaces*indent
    wide = width - len(tabs)
    lines = string.split("\n")    # Paragraphs and line breaks
    lists = (textwrap.wrap(line, wide, initial_indent=tabs, subsequent_indent=tabs)
             for line in lines)
    body = "\n".join("\n".join(lst) for lst in lists)
    return body


def striphtml(data):
    """ Remove or replace HTML tags and replace symbols """
    data = html.unescape(data)                      # Replace symbols
    data = re.sub(r'<[/ ]*br[/ ]*>', '\n', data)    # Preserve line breaks
    data = re.sub(r'</*p>', '\n', data)             # Preserve paragraphs
    data = re.sub(r'<[^<]+?>', '', data)            # Strip all other tags
    return data


## Main code:

if DEBUG is True:
    p = pprint.PrettyPrinter().pprint   ####DEBUG

appName = os.path.splitext(os.path.basename(__file__))[0]
appArgNames = "UserName InstanceName LoginEmail Password"
appArgs = len(appArgNames.split())

args = sys.argv[1:] # Parameters follow program name
numArgs = len(args)
numAcct = 0

if numArgs > 0 and numArgs % appArgs == 0:  # Allow only complete parameter sets
    numAcct = int(numArgs / appArgs)
    # Read parameters as sets
    parameterSets = [args[(x*appArgs):(x+1)*appArgs] for x in range(numAcct)]
    numAcct = len(parameterSets)
    print("Read %d set(s) of user parameters.\n"%(numAcct))
else:
    print("Error: Argument[s] missing.")
    print("Usage: %s %s ..."%(appName, appArgNames))
    sys.exit()

# Repeat for each account
for pset in parameterSets:

    # Simplify parameter access
    me, instance, email, password = pset

    #TODO: How about some parameter validation?

    outFile = "%s@%s.txt"%(me, instance)

    # Prep filenames for persistent credentials
    cliCred = "%s_%s_clientcred.txt"%(appName, instance)
    usrCred = "%s_%s_usercred.txt"%(appName, me)

    # Register app - only once!
    url = "https://%s"%(instance)
    if pl.Path(cliCred).is_file():
        print("Using persistent client credentials.")
    else:
        print("Obtaining client credentials (persistent not present).")
        Mastodon.create_app(appName, api_base_url=url, to_file=cliCred)

    # Create mastodon client
    mastodon = Mastodon(api_base_url=url, client_id=cliCred)

    # User login - either every time, or use persisted
    if pl.Path(usrCred).is_file():  # Use persistent
        print("Using persistent user credentials.")
        numAcct = 1
    else:                           # Create fresh
        print("Obtaining user credentials (persistent not present).")
        # Login
        mastodon.log_in(username=email, password=password, to_file=usrCred)
        numAcct = 1

    # Create actual instance
    mastodon = Mastodon(api_base_url=url, client_id=cliCred, access_token=usrCred)

    # Get integer user ID
    user_id = mastodon.account_search(me)[0]['id']
    print("user_id for %s = %s"%(me, user_id))

    # Collect all toots for user
    print("\nGathering toots for @%s@%s "%(me, instance), end='')
    sys.stdout.flush()
    toots = []
    t = mastodon.account_statuses(user_id)
    while len(t) > 0:
        print('.', end='')
        sys.stdout.flush()
        toots += t
        t = mastodon.account_statuses(user_id, max_id=t[-1]['id'])
    print()

    print("Processing %d toots, id %d to %d\n"%(
        len(toots), toots[-1]['id'], toots[0]['id']))
    with open(outFile, 'w') as f:
        for toot in reversed(toots):    # Process from oldest to newest
            # Gather toots by me that aren't boosts
            if toot['reblog'] is None:
                print("\n%s\n"%(toot['created_at']),
                      witext(striphtml(toot['content'])), file=f)

    print("Toots written to: %s\n"%(outFile))


if DEBUG is True:
    # Go interactive only when stdout not redirected                ####DEBUG
    if os.fstat(0) == os.fstat(1):                                  ####DEBUG
        print("\nEntering Interactive Mode: ^Z or ^D when done.")   ####DEBUG
        code.interact(local=dict(globals(), **locals()))            ####DEBUG

print("Exiting!")