Allow rescheduling from elasticsearch errors output

I have difficulties naming things.

ardumont/from-output-to-csv.py

+# Use sample:
+# cat mercurial.output.txt | \
+#     grep 'sensible-filter' | \
+#     python3 -m from-output-to-csv
+
+# The output of this command is designed to feed the
+# `swh.scheduler.cli task schedule` subcommand
+
+
+import ast
+import click
+import json
+import sys
+
+
+TYPES = {
+    'mercurial': 'origin-update-hg',
+    'mercurial-archive': 'origin-load-archive-hg',
+    'svn': 'origin-update-svn',
+    'svn-archive': 'swh-loader-mount-dump-and-load-svn-repository',
+    'pypi': 'origin-update-pypi',
+}
+
+
+@click.command()
+@click.option('--task-policy', default='oneshot', type=click.Choice([
+                  'oneshot', 'recurring']),
+              help="Task's policy")
+@click.option('--task-type', default='mercurial', type=click.Choice(
+    TYPES.keys()),
+              help="Task's type")
+def main(task_policy, task_type):
+    """Given an output from kibana_fetch_logs, transform the input in csv
+       format (; as delimiter) to ease back scheduling for those
+       origins.
+
+       The output format is of the form:
+       <task-type>;<task-policy>;task-args;task-kwargs
+
+       Then use of `swh.scheduler.cli schedule` subcommand.
+
+       cat <output-of-this-command> | \
+           python3 -m swh.mercurial.cli task schedule \
+               --columns type \
+               --columns policy \
+               --columns args \
+               --columns kwargs \
+               --delimiter ';' -
+
+    """
+    for line in sys.stdin:
+        line = line.rstrip()
+        data = ast.literal_eval(line)
+
+        _task_type = TYPES.get(task_type)
+        _task_args = json.dumps(data['args'])
+
+        kwargs = data['kwargs']
+
+        # HACK: Should have been set earlier...
+        if task_type in ['mercurial', 'mercurial-archive']:
+            if 'visit_date' not in kwargs:
+                kwargs['visit_date'] = 'Tue, 3 May 2016 17:16:32 +0200'
+
+        _task_kwargs = json.dumps(kwargs)
+
+        print(';'.join([_task_type, task_policy, _task_args, _task_kwargs]))
+
+
+if __name__ == '__main__':
+    main()