EnlivenMinetest/utilities/pyissuesyncd

#!/usr/bin/env python3
'''
Title: pyissuesyncd
(c) 2021 Jake "Poikilos" Gustafson


Purpose:
This python-based daemon synchronizes issues (one-way) from one
repository to another.


License:
See the license file in the included EnlivenMinetest directory or at
[EnlivenMinetest](https://github.com/poikilos/EnlivenMinetest)


Outputs:
data_directory: The data directory for this service daemon is
os.path.join(profile, ".cache", "pyissuesyncd").


required arguments:
--dst-repo (or set the DST_REPO environment variable)
    Issues and dependent data will be overwritten at this API URL.


optional arguments:
Environment variables get be set, but a CLI argument will override the
corresponding variable noted below in all caps.

The two _CACHE directories below are used as the single_cache option
for the Repo (see enissue.py's Repo class for documentation).

--src-cache: Set the directory to store a cached version of the source repo's data.
* defaults to SRC_CACHE or os.path.join(data_directory, "source")

--dst-cache: Set the directory to store a cached version of the destination repo's data.
* defaults to DST_CACHE or os.path.join(data_directory, "destination")


Examples:
DST_REPO=https://example.com/git/repo pyissuesyncd
pyissuesyncd --dst-repo https://example.com/git/repo

'''
import os
import sys
import json
from datetime import datetime  #, timedelta
# see <https://stackoverflow.com/questions/5574702/how-to-print-to-stderr-in-python>
def error(*args, **kwargs):
    print(*args, file=sys.stderr, **kwargs)


myFilePath = os.path.realpath(__file__)
me = os.path.basename(__file__)
myDir = os.path.dirname(myFilePath)

try:
    import enissue
except ModuleNotFoundError as ex:
    PATH = os.environ.get("PATH")
    found_d = None
    if PATH is not None:
        more_paths = PATH.split(os.pathsep)
        # ^ pathsep such as ':' (NOT dirsep such as '/'!)
        more_paths
        for this_d in more_paths:
            tryF = os.path.join(this_d, "enissue.py")
            if os.path.isfile(tryF):
                found_d = this_d
                break
    if found_d is not None:
        sys.path.append(found_d)
        print("* detected enissue.py in {}".format(found_d))
        # print("path: {}".format(sys.path))

        try:
            import eni
        except ModuleNotFoundError as ex2:
            error(ex2)
            print("{} must be in the same directory as enissue.py or in"
                  " PATH".format(me))
            sys.exit(1)
    else:
        print("{} must be in the same directory as enissue.py or in"
              " PATH".format(me))
        raise ex

from enissue import Repo

data_directory = os.path.join(Repo.profile, ".cache", "pyissuesyncd")

def get_issue(repo, options, issue_no):
    results, err = repo.load_issues(
        options,
        issue_no=issue_no,
    )
    if results is None:
        if err is not None:
            if err.get('code') == 410:
                # error("The issue was deleted")
                pass
            elif err.get('code') == 404:
                # error("The issue doesn't exist")
                pass
            return None, err
        else:
            msg = ("Unknown error: Results should not be None unless"
                   " there is an error (issue_no={})."
                   "".format(issue_no))
            return (
                None,
                {
                    'reason': msg,
                }
            )
    elif not isinstance(results, list):
        raise RuntimeError("Results must be a list even if there is"
                           " only one result.")
    elif len(results) > 1:
        raise RuntimeError("Results should have"
                           " only one result.")
    issue = results[0]
    '''
    match = repo.get_match(
        mode,
        issue_no=issue_no,
        match_all_labels_lower=match_all_labels_lower,
    )
    matching_issue = match['issue']
    if matching_issue is not None:
        repo.show_issue(
            matching_issue,
            refresh=False,
            never_expire=options.get('never_expire') is True,
        )
    '''
    return issue, None


def start_issuesyncd(src_options, dst_options):
    # src_never_expire = src_options.get('never_expire') is True
    max_issue = src_options.get('max_issue')
    if max_issue is None:
        max_issue = 1000
        error("WARNING: SRC_MAX_ISSUE set to default: {}"
              "".format(max_issue))
    else:
        max_issue = int(max_issue)

    issue_no = 0  # This is incremented to 1 before use.
    # issue_no = max_issue - 1  # debug only
    src_res_code = 0
    end_codes = [404, 403]
    # while src_res_code not in end_codes:
    while True:
    # while (issue_no + 1) <= max_issue: # for debug only
        issue_no += 1
        if max_issue is not None:
            if issue_no > max_issue:
                error("* ending due to setting: --src-max-issue={}"
                      " (can also be set by SRC_MAX_ISSUE env var)"
                      "".format({}))
                break
        src_repo = Repo(src_options)
        src_issue, err = get_issue(src_repo, src_options, issue_no)
        deleted = False
        if err is not None:
            '''
            error("Error accessing source issue {}: {}: {}"
                  "".format(issue_no, err.get('code'),
                            err.get('reason')))
            '''
            src_res_code = err.get('code')
            url = err.get('url')
            if src_res_code in end_codes:
                if src_res_code == 403:
                    error("#{}: stopping due to error {} ({})"
                          "".format(issue_no, err.get('code'),
                                    err.get('reason')))
                    # error("  * reason: {}".format())
                    # error("  * headers: {}".format(err.get('headers')))
                    break
                elif src_res_code == 404:
                    error("#{}: Error 404: There is no {}"
                          " so the end of the issues may have been"
                          " reached.".format(issue_no, url))
                    error("  * reason: {}".format(err.get('reason')))
                    # error("  * headers: {}".format(err.get('headers')))
                    continue
                elif src_res_code == 410:
                    error("#{}: The issue seems to have been deleted."
                          "".format(issue_no))
                    error("  * reason: {}".format(err.get('reason')))
                    # error("  * headers: {}".format(err.get('headers')))
                    deleted = False
                    # TODO: delete on dest (carefully!)
                    continue
                else:
                    error("#{}: stopping due to error code {}"
                          "".format(issue_no, src_res_code))
                break
            else:
                error("#{}: continuing anyway but got error code {}"
                      "".format(issue_no, src_res_code))


            if src_issue is None:
                if src_res_code not in end_codes:
                    error("#{}: Skipping due to unprocessed error {}"
                          "".format(issue_no, src_res_code))
                else:
                    error("#{}: Stopping due to unprocessed error {}"
                          "".format(issue_no, src_res_code))
                continue
            else:
                error("However, an issue was returned.")
        error("Got issue {}".format(issue_no))
        # Example: ~/.cache/pyissuesyncd/source/issues/1.json
        src_dt_parser = src_repo.options['default_dt_parser']
        src_created_dt_s = src_repo.getKnown(src_issue, 'created_at')
        src_updated_dt_s = src_repo.getKnown(src_issue, 'updated_at')
        src_updated_dt = src_dt_parser(src_updated_dt_s)
        src_updated_ts = int(src_updated_dt.strftime("%s"))
        # ^ See <https://stackoverflow.com/questions/19801727/convert-
        #   datetime-to-unix-timestamp-and-convert-it-back-in-python>
        '''
        print("* src_issue: {} updated: {} = {}"
              "".format(issue_no, src_updated_ts, src_updated_dt))
        '''
        # print(json.dumps(src_issue, indent=2))

        # enissue.set_verbose(True)
        dst_repo = Repo(dst_options)
        dst_issue, err = get_issue(dst_repo, dst_options, issue_no)
        if err is not None:
            dst_res_code = err.get('code')
            url = err.get('url')
            '''
            if dst_res_code in end_codes:
                if dst_res_code == 403:
                    error("* stopping due to: {}"
                          "".format(err.get('reason')))
                    break
                elif dst_res_code == 404:
                    error("* 404: There is no issue {} at {} so the end"
                          " of the issues may have been reached."
                          "".format(issue_no, url))
                    error("  * reason: {}".format(err.get('reason')))
                    # error("  * headers: {}".format(err.get('headers')))
                    continue
                elif dst_res_code == 410:
                    error(err.get('reason'))
                    error("* Issue {} seems to have been deleted."
                          "".format(issue_no))
                    continue
                break
            '''
            if dst_issue is None:
                # TODO: write the issue
                continue
        if dst_issue is None:
            raise RuntimeError("dst_issue shouldn't be None when error"
                               " is None.")

        dst_dt_parser = dst_repo.options['default_dt_parser']
        dst_created_dt_s = dst_repo.getKnown(dst_issue, 'created_at')
        dst_updated_dt_s = dst_repo.getKnown(dst_issue, 'updated_at')
        dst_updated_dt = dst_dt_parser(dst_updated_dt_s)
        dst_updated_ts = int(dst_updated_dt.strftime("%s"))
        # ^ See <https://stackoverflow.com/questions/19801727/convert-
        #   datetime-to-unix-timestamp-and-convert-it-back-in-python>
        '''
        print("* dst_issue: {} updated: {} = {}"
              "".format(issue_no, dst_updated_ts, dst_updated_dt))
        '''
        # Example: ~/.cache/pyissuesyncd/destination/issues/1.json
        # break  # for debug only

        continue  # for debug only
        # print("  * dst_issue:")
        # print(json.dumps(dst_issue, indent=2))
        if err is not None:
            if err.get('code') == 404:
                # dst_repo.create_issue(src_issue, src_repo)
                continue
            error("Error accessing destination issue {}: {}: {}"
                  "".format(issue_no, err.get('code'),
                            err.get('reason')))
            continue
        # if issue_differs:  # compare timestamp
        if True:  # for debug only
            pass
            # dst_repo.update_issue(src_issue, src_repo)


def usage():
    print(__doc__)

if __name__ == "__main__":
    src_options = {
        'repo_url': "https://github.com/poikilos/EnlivenMinetest",
        'never_expire': True,
        'quiet': True,
        'api_id': "GitHub",
    }
    dst_options = {
        'never_expire': True,
        'quiet': True,
        'api_id': "Gitea",
    }
    DST_REPO = os.environ.get('DST_REPO')
    if DST_REPO is not None:
        dst_options['repo_url'] = DST_REPO
        del DST_REPO
    SRC_REPO = os.environ.get('SRC_REPO')
    if DST_REPO is not None:
        src_options['repo_url'] = SRC_REPO
        del SRC_REPO

    SRC_CACHE = os.environ.get('SRC_CACHE')
    if SRC_CACHE is None:
        SRC_CACHE = os.path.join(data_directory, "source")

    DST_CACHE = os.environ.get('DST_CACHE')
    if DST_CACHE is None:
        DST_CACHE = os.path.join(data_directory, "destination")

    SRC_MAX_ISSUE = os.environ.get('SRC_MAX_ISSUE')

    prev_arg = None

    manual_args = ['--dst-repo', '--src-repo', '--src-cache',
                   '--dst-cache', '--src-max-issue']

    for arg in sys.argv[1:]:
        if prev_arg == "--dst-repo":
            dst_options['repo_url'] = arg
        elif prev_arg == "--src-repo":
            src_options['repo_url'] = arg
        elif prev_arg == "--src_cache":
            SRC_CACHE = arg
        elif prev_arg == "--dst_cache":
            DST_CACHE = arg
        elif prev_arg == "--src-max-issue":
            SRC_MAX_ISSUE = int(arg)
        elif arg in manual_args:
            pass
        else:
            usage()
            error("Error: The argument is not valid: {}".format(arg))
            sys.exit(1)
        prev_arg = arg

    src_options['single_cache'] = SRC_CACHE
    src_options['max_issue'] = SRC_MAX_ISSUE
    # ^ INFO: start_issuesyncd warns if SRC_MAX_ISSUE is None.
    dst_options['single_cache'] = DST_CACHE


    error("SRC_REPO (--src-repo) is {}"
          "".format(src_options.get('repo_url')))
    error("DST_REPO (--dst-repo) is {}"
          "".format(dst_options.get('repo_url')))
    if dst_options.get('repo_url') is None:
        error("Error: You must set DST_REPO in the environment or specify a url after --dst-repo")
        sys.exit(1)

    start_issuesyncd(src_options, dst_options)