The log shows that WeeWX is trying to load a service named "w", which is
not listed under [[Services]].

I think what's happening is that the service list "prep_services" contains
only one item, weewx.engine.StdTimeSync, which is then being treated as an
iterable. The first letter is 'w', which WeeWX tries to import. I recall
something like this happening when porting WeeWX to Python V3. However, you
are not running Python V3, so I'm at a loss why this is happening.

Let's try this: replace your copy of weeutil.py (you'll find it at
/usr/share/weewx/weeutil/weeutil.py) with the attached. Let us know what
happens.

-tk

On Fri, Jun 21, 2019 at 10:37 PM Michael Aschauer <[email protected]>
wrote:

>
>
> no problem thomas, thank you for coming back.
> here is the debug output:
>
> Jun 22 07:30:02 wetterraspi systemd[1]: Starting LSB: weewx weather
> system...
> Jun 22 07:30:02 wetterraspi weewx[4147]: engine: Initializing weewx
> version 3.9.1
> Jun 22 07:30:02 wetterraspi weewx[4147]: engine: Using Python 2.7.9
> (default, Sep 17 2016, 20:26:04) #012[GCC 4.9.2]
> Jun 22 07:30:02 wetterraspi weewx[4147]: engine: Platform
> Linux-4.9.35-v7+-armv7l-with-debian-8.0
> Jun 22 07:30:02 wetterraspi weewx[4147]: engine: Locale is 'de_AT.UTF-8'
> Jun 22 07:30:02 wetterraspi weewx[4147]: engine: pid file is
> /var/run/weewx.pid
> Jun 22 07:30:02 wetterraspi weewx[4137]: Starting weewx weather system:
> weewx.
> Jun 22 07:30:02 wetterraspi systemd[1]: Started LSB: weewx weather system.
> Jun 22 07:30:02 wetterraspi weewx[4151]: engine: Using configuration file
> /etc/weewx/weewx.conf
> Jun 22 07:30:02 wetterraspi weewx[4151]: engine: Debug is 1
> Jun 22 07:30:02 wetterraspi weewx[4151]: engine: Initializing engine
> Jun 22 07:30:02 wetterraspi weewx[4151]: engine: Loading station type
> Vantage (weewx.drivers.vantage)
> Jun 22 07:30:02 wetterraspi weewx[4151]: vantage: Driver version is 3.1.1
> Jun 22 07:30:02 wetterraspi weewx[4151]: vantage: Opened up serial port
> /dev/ttyS0; baud 19200; timeout 4.00
> Jun 22 07:30:06 wetterraspi weewx[4151]: vantage: Retry  #0 failed
> Jun 22 07:30:06 wetterraspi weewx[4151]: vantage: Gentle wake up of
> console successful
> Jun 22 07:30:06 wetterraspi weewx[4151]: vantage: Hardware type is 16
> Jun 22 07:30:06 wetterraspi weewx[4151]: vantage: ISS ID is 3
> Jun 22 07:30:06 wetterraspi weewx[4151]: vantage: Hardware name: Vantage
> Pro2
> Jun 22 07:30:06 wetterraspi weewx[4151]: engine: Loading service w
> Jun 22 07:30:06 wetterraspi weewx[4151]: engine: Caught unrecoverable
> exception in engine:
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****  Empty module name
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****  Traceback (most recent
> call last):
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****    File
> "/usr/share/weewx/weewx/engine.py", line 884, in main
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****      engine =
> engine_class(config_dict)
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****    File
> "/usr/share/weewx/weewx/engine.py", line 78, in __init__
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****
>  self.loadServices(config_dict)
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****    File
> "/usr/share/weewx/weewx/engine.py", line 142, in loadServices
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****
>  self.service_obj.append(weeutil.weeutil._get_object(svc)(self,
> config_dict))
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****    File
> "/usr/share/weewx/weeutil/weeutil.py", line 1130, in _get_object
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****      mod =
> __import__(module)
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****  ValueError: Empty
> module name
> Jun 22 07:30:06 wetterraspi weewx[4151]:     ****  Exiting.
>
> --
> You received this message because you are subscribed to the Google Groups
> "weewx-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/weewx-user/197c0840-29e8-499d-a119-445e5d654a80%40googlegroups.com
> <https://groups.google.com/d/msgid/weewx-user/197c0840-29e8-499d-a119-445e5d654a80%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google Groups 
"weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/weewx-user/CAPq0zEB5MU5dH7OP_J7Stmk%3DiRygC-ht9TkgWFCpQKdBdTY1FQ%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
# This Python file uses the following encoding: utf-8
#
#    Copyright (c) 2009-2018 Tom Keffer <[email protected]>
#
#    See the file LICENSE.txt for your full rights.
#
"""Various handy utilities that don't belong anywhere else.
   Works under Python 2 and Python 3.
"""

from __future__ import absolute_import
from __future__ import print_function

import calendar
import datetime
import math
import os
import struct
import shutil
import syslog
import time
import traceback
try:
    # Python 2
    from StringIO import StringIO
except ImportError:
    # Python 3
    from io import StringIO

# For backwards compatibility:
from weeutil import config
search_up        = config.search_up
accumulateLeaves = config.accumulateLeaves
merge_config     = config.merge_config
patch_config     = config.patch_config
comment_scalar   = config.comment_scalar
conditional_merge= config.conditional_merge


def convertToFloat(seq):
    """Convert a sequence with strings to floats, honoring 'Nones'"""

    if seq is None:
        return None
    res = [None if s in ('None', 'none') else float(s) for s in seq]
    return res


def option_as_list(option):
    if option is None:
        return None
    return [option] if not isinstance(option, list) else option


def list_as_string(option):
    """Returns the argument as a string.
    
    Useful for insuring that ConfigObj options are always returned
    as a string, despite the presence of a comma in the middle.
    
    Example:
    >>> print(list_as_string('a string'))
    a string
    >>> print(list_as_string(['a', 'string']))
    a, string
    >>> print(list_as_string('Reno, NV'))
    Reno, NV
    """
    # Check if it's already a string.
    if option is not None and not isinstance(option, str):
        return ', '.join(option)
    return option


def stampgen(startstamp, stopstamp, interval):
    """Generator function yielding a sequence of timestamps, spaced interval apart.
    
    The sequence will fall on the same local time boundary as startstamp. 

    Example:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> startstamp = 1236560400
    >>> print(timestamp_to_string(startstamp))
    2009-03-08 18:00:00 PDT (1236560400)
    >>> stopstamp = 1236607200
    >>> print(timestamp_to_string(stopstamp))
    2009-03-09 07:00:00 PDT (1236607200)
    
    >>> for stamp in stampgen(startstamp, stopstamp, 10800):
    ...     print(timestamp_to_string(stamp))
    2009-03-08 18:00:00 PDT (1236560400)
    2009-03-08 21:00:00 PDT (1236571200)
    2009-03-09 00:00:00 PDT (1236582000)
    2009-03-09 03:00:00 PDT (1236592800)
    2009-03-09 06:00:00 PDT (1236603600)

    Note that DST started in the middle of the sequence and that therefore the
    actual time deltas between stamps is not necessarily 3 hours.
    
    startstamp: The start of the sequence in unix epoch time.
    
    stopstamp: The end of the sequence in unix epoch time. 
    
    interval: The time length of an interval in seconds.
    
    yields a sequence of timestamps between startstamp and endstamp, inclusive.
    """
    dt = datetime.datetime.fromtimestamp(startstamp)
    stop_dt = datetime.datetime.fromtimestamp(stopstamp)
    if interval == 365.25 / 12 * 24 * 3600:
        # Interval is a nominal month. This algorithm is 
        # necessary because not all months have the same length.
        while dt <= stop_dt:
            t_tuple = dt.timetuple()
            yield time.mktime(t_tuple)
            year = t_tuple[0]
            month = t_tuple[1]
            month += 1
            if month > 12:
                month -= 12
                year += 1
            dt = dt.replace(year=year, month=month)
    else:
        # This rather complicated algorithm is necessary (rather than just
        # doing some time stamp arithmetic) because of the possibility that DST
        # changes in the middle of an interval.
        delta = datetime.timedelta(seconds=interval)
        ts_last = 0
        while dt <= stop_dt:
            ts = int(time.mktime(dt.timetuple()))
            # This check is necessary because time.mktime() cannot
            # disambiguate between 2am ST and 3am DST. For example,
            #   time.mktime((2013, 3, 10, 2, 0, 0, 0, 0, -1)) and
            #   time.mktime((2013, 3, 10, 3, 0, 0, 0, 0, -1))
            # both give the same value (1362909600)
            if ts > ts_last:
                yield ts
                ts_last = ts
            dt += delta


def startOfInterval(time_ts, interval):
    """Find the start time of an interval.
    
    This algorithm assumes the day is divided up into
    intervals of 'interval' length. Given a timestamp, it
    figures out which interval it lies in, returning the start
    time.
    
    time_ts: A timestamp. The start of the interval containing this
    timestamp will be returned.
    
    interval: An interval length in seconds.
    
    Returns: A timestamp with the start of the interval.

    Examples:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 01:57:35", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  300))
    'Thu Jul  4 01:55:00 2013'
    >>> time.ctime(startOfInterval(start_ts,  300.0))
    'Thu Jul  4 01:55:00 2013'
    >>> time.ctime(startOfInterval(start_ts,  600))
    'Thu Jul  4 01:50:00 2013'
    >>> time.ctime(startOfInterval(start_ts,  900))
    'Thu Jul  4 01:45:00 2013'
    >>> time.ctime(startOfInterval(start_ts, 3600))
    'Thu Jul  4 01:00:00 2013'
    >>> time.ctime(startOfInterval(start_ts, 7200))
    'Thu Jul  4 00:00:00 2013'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 01:00:00", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  300))
    'Thu Jul  4 00:55:00 2013'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 01:00:01", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  300))
    'Thu Jul  4 01:00:00 2013'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 01:04:59", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  300))
    'Thu Jul  4 01:00:00 2013'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 00:00:00", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  300))
    'Wed Jul  3 23:55:00 2013'
    >>> start_ts = time.mktime(time.strptime("2013-07-04 07:51:00", "%Y-%m-%d %H:%M:%S"))
    >>> time.ctime(startOfInterval(start_ts,  60))
    'Thu Jul  4 07:50:00 2013'
    >>> start_ts += 0.1
    >>> time.ctime(startOfInterval(start_ts,  60))
    'Thu Jul  4 07:51:00 2013'
    """

    interval_m = int(interval // 60)
    interval_h = int(interval // 3600)
    time_tt = time.localtime(time_ts)
    m = int(time_tt.tm_min // interval_m * interval_m)
    h = int(time_tt.tm_hour // interval_h * interval_h) if interval_h > 1 else time_tt.tm_hour

    # Replace the hour, minute, and seconds with the start of the interval.
    # Everything else gets retained:
    start_interval_ts = time.mktime((time_tt.tm_year,
                                     time_tt.tm_mon,
                                     time_tt.tm_mday,
                                     h, m, 0,
                                     0, 0, time_tt.tm_isdst))
    # Weewx uses the convention that the interval is exclusive on left, inclusive
    # on the right. So, if the timestamp is at the beginning of the interval,
    # it actually belongs to the previous interval.
    if time_ts == start_interval_ts:
        start_interval_ts -= interval
    return start_interval_ts


def _ord_to_ts(_ord):
    d = datetime.date.fromordinal(_ord)
    t = int(time.mktime(d.timetuple()))
    return t


# ===============================================================================
# What follows is a bunch of "time span" routines. Generally, time spans
# are used when start and stop times fall on calendar boundaries
# such as days, months, years.  So, it makes sense to talk of "daySpans",
# "weekSpans", etc. They are generally not used between two random times. 
# ===============================================================================

class TimeSpan(tuple):
    """Represents a time span, exclusive on the left, inclusive on the right."""

    def __new__(cls, *args):
        if args[0] > args[1]:
            raise ValueError("start time (%d) is greater than stop time (%d)" % (args[0], args[1]))
        return tuple.__new__(cls, args)

    @property
    def start(self):
        return self[0]

    @property
    def stop(self):
        return self[1]

    @property
    def length(self):
        return self[1] - self[0]

    def includesArchiveTime(self, timestamp):
        """
        Returns True if the span includes the time timestamp, otherwise False.
        
        timestamp: The timestamp to be tested.
        """
        return self.start < timestamp <= self.stop

    def includes(self, span):

        return self.start <= span.start <= self.stop and self.start <= span.stop <= self.stop

    def __eq__(self, other):
        return self.start == other.start and self.stop == other.stop

    def __str__(self):
        return "[%s -> %s]" % (timestamp_to_string(self.start),
                               timestamp_to_string(self.stop))

    def __hash__(self):
        return hash(self.start) ^ hash(self.stop)

    def __cmp__(self, other):
        if self.start < other.start:
            return - 1
        return 0 if self.start == other.start else 1


def intervalgen(start_ts, stop_ts, interval):
    """Generator function yielding a sequence of time spans whose boundaries
    are on constant local time.
    
    Yields a sequence of TimeSpans. The start times of the timespans will
    be on the same local time boundary as the start of the sequence. See the
    example below.
    
    Example:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> startstamp = 1236477600
    >>> print(timestamp_to_string(startstamp))
    2009-03-07 18:00:00 PST (1236477600)
    >>> stopstamp = 1236538800
    >>> print(timestamp_to_string(stopstamp))
    2009-03-08 12:00:00 PDT (1236538800)
    
    >>> for span in intervalgen(startstamp, stopstamp, 10800):
    ...     print(span)
    [2009-03-07 18:00:00 PST (1236477600) -> 2009-03-07 21:00:00 PST (1236488400)]
    [2009-03-07 21:00:00 PST (1236488400) -> 2009-03-08 00:00:00 PST (1236499200)]
    [2009-03-08 00:00:00 PST (1236499200) -> 2009-03-08 03:00:00 PDT (1236506400)]
    [2009-03-08 03:00:00 PDT (1236506400) -> 2009-03-08 06:00:00 PDT (1236517200)]
    [2009-03-08 06:00:00 PDT (1236517200) -> 2009-03-08 09:00:00 PDT (1236528000)]
    [2009-03-08 09:00:00 PDT (1236528000) -> 2009-03-08 12:00:00 PDT (1236538800)]

    (Note how in this example the local time boundaries are constant, despite
    DST kicking in. The interval length is not constant.)
    
    Another example, this one over the Fall DST boundary, and using 1 hour intervals:

    >>> startstamp = 1257051600
    >>> print(timestamp_to_string(startstamp))
    2009-10-31 22:00:00 PDT (1257051600)
    >>> stopstamp = 1257080400
    >>> print(timestamp_to_string(stopstamp))
    2009-11-01 05:00:00 PST (1257080400)
    >>> for span in intervalgen(startstamp, stopstamp, 3600):
    ...    print(span)
    [2009-10-31 22:00:00 PDT (1257051600) -> 2009-10-31 23:00:00 PDT (1257055200)]
    [2009-10-31 23:00:00 PDT (1257055200) -> 2009-11-01 00:00:00 PDT (1257058800)]
    [2009-11-01 00:00:00 PDT (1257058800) -> 2009-11-01 01:00:00 PDT (1257062400)]
    [2009-11-01 01:00:00 PDT (1257062400) -> 2009-11-01 02:00:00 PST (1257069600)]
    [2009-11-01 02:00:00 PST (1257069600) -> 2009-11-01 03:00:00 PST (1257073200)]
    [2009-11-01 03:00:00 PST (1257073200) -> 2009-11-01 04:00:00 PST (1257076800)]
    [2009-11-01 04:00:00 PST (1257076800) -> 2009-11-01 05:00:00 PST (1257080400)]
    
    start_ts: The start of the first interval in unix epoch time. In unix epoch time.
    
    stop_ts: The end of the last interval will be equal to or less than this.
    In unix epoch time.
    
    interval: The time length of an interval in seconds.
    
    yields: A sequence of TimeSpans. Both the start and end of the timespan
    will be on the same time boundary as start_ts"""

    dt1 = datetime.datetime.fromtimestamp(start_ts)
    stop_dt = datetime.datetime.fromtimestamp(stop_ts)

    if interval == 365.25 / 12 * 24 * 3600:
        # Interval is a nominal month. This algorithm is 
        # necessary because not all months have the same length.
        while dt1 < stop_dt:
            t_tuple = dt1.timetuple()
            year = t_tuple[0]
            month = t_tuple[1]
            month += 1
            if month > 12:
                month -= 12
                year += 1
            dt2 = min(dt1.replace(year=year, month=month), stop_dt)
            stamp1 = time.mktime(t_tuple)
            stamp2 = time.mktime(dt2.timetuple())
            yield TimeSpan(stamp1, stamp2)
            dt1 = dt2
    else:
        # This rather complicated algorithm is necessary (rather than just
        # doing some time stamp arithmetic) because of the possibility that DST
        # changes in the middle of an interval
        delta = datetime.timedelta(seconds=interval)
        last_stamp1 = 0
        while dt1 < stop_dt:
            dt2 = min(dt1 + delta, stop_dt)
            stamp1 = int(time.mktime(dt1.timetuple()))
            stamp2 = int(time.mktime(dt2.timetuple()))
            if stamp2 > stamp1 > last_stamp1:
                yield TimeSpan(stamp1, stamp2)
                last_stamp1 = stamp1
            dt1 = dt2


def archiveHoursAgoSpan(time_ts, hours_ago=0, grace=1):
    """Returns a TimeSpan for x hours ago
    
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = time.mktime(time.strptime("2013-07-04 01:57:35", "%Y-%m-%d %H:%M:%S"))
    >>> print(archiveHoursAgoSpan(time_ts, hours_ago=0))
    [2013-07-04 01:00:00 PDT (1372924800) -> 2013-07-04 02:00:00 PDT (1372928400)]
    >>> print(archiveHoursAgoSpan(time_ts, hours_ago=2))
    [2013-07-03 23:00:00 PDT (1372917600) -> 2013-07-04 00:00:00 PDT (1372921200)]
    >>> time_ts = time.mktime(datetime.date(2013, 7, 4).timetuple())
    >>> print(archiveHoursAgoSpan(time_ts, hours_ago=0))
    [2013-07-03 23:00:00 PDT (1372917600) -> 2013-07-04 00:00:00 PDT (1372921200)]
    >>> print(archiveHoursAgoSpan(time_ts, hours_ago=24))
    [2013-07-02 23:00:00 PDT (1372831200) -> 2013-07-03 00:00:00 PDT (1372834800)]
    """
    if time_ts is None:
        return None
    time_ts -= grace
    dt = datetime.datetime.fromtimestamp(time_ts)
    hour_start_dt = dt.replace(minute=0, second=0, microsecond=0)
    start_span_dt = hour_start_dt - datetime.timedelta(hours=hours_ago)
    stop_span_dt = start_span_dt + datetime.timedelta(hours=1)

    return TimeSpan(time.mktime(start_span_dt.timetuple()),
                    time.mktime(stop_span_dt.timetuple()))


def archiveSpanSpan(time_ts, time_delta=0, hour_delta=0, day_delta=0, week_delta=0, month_delta=0, year_delta=0):
    """ Returns a TimeSpan for the last xxx seconds where xxx equals
        time_delta sec + hour_delta hours + day_delta days + week_delta weeks + month_delta months + year_delta years
        Note: For month_delta, 1 month = 30 days, For year_delta, 1 year = 365 days
    
    Example:
    >>> os.environ['TZ'] = 'Australia/Brisbane'
    >>> time_ts = time.mktime(time.strptime("2015-07-21 09:05:35", "%Y-%m-%d %H:%M:%S"))
    >>> print(archiveSpanSpan(time_ts, time_delta=3600))
    [2015-07-21 08:05:35 AEST (1437429935) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, hour_delta=6))
    [2015-07-21 03:05:35 AEST (1437411935) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, day_delta=1))
    [2015-07-20 09:05:35 AEST (1437347135) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, time_delta=3600, day_delta=1))
    [2015-07-20 08:05:35 AEST (1437343535) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, week_delta=4))
    [2015-06-23 09:05:35 AEST (1435014335) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, month_delta=1))
    [2015-06-21 09:05:35 AEST (1434841535) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts, year_delta=1))
    [2014-07-21 09:05:35 AEST (1405897535) -> 2015-07-21 09:05:35 AEST (1437433535)]
    >>> print(archiveSpanSpan(time_ts))
    [2015-07-21 09:05:34 AEST (1437433534) -> 2015-07-21 09:05:35 AEST (1437433535)]
    
    Example over a DST boundary. Because Brisbane does not observe DST, we need to
    switch timezones.
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = 1457888400
    >>> print(timestamp_to_string(time_ts))
    2016-03-13 10:00:00 PDT (1457888400)
    >>> span = archiveSpanSpan(time_ts, day_delta=1)
    >>> print(span)
    [2016-03-12 10:00:00 PST (1457805600) -> 2016-03-13 10:00:00 PDT (1457888400)]
    
    Note that there is not 24 hours of time over this span:
    >>> print((span.stop - span.start) / 3600.0)
    23.0
    """

    if time_ts is None:
        return None

    # Use a datetime.timedelta so that it can take DST into account:
    time_dt = datetime.datetime.fromtimestamp(time_ts)
    time_dt -= datetime.timedelta(weeks=week_delta, days=day_delta, hours=hour_delta, seconds=time_delta)

    # Now add the deltas for months and years. Because these can be variable in length,
    # some special arithmetic is needed. Start by calculating the number of
    # months since 0 AD:
    total_months = 12 * time_dt.year + time_dt.month - 1 - 12 * year_delta - month_delta
    # Convert back from total months since 0 AD to year and month:
    year = total_months // 12
    month = total_months % 12 + 1
    # Apply the delta to our datetime object
    start_dt = time_dt.replace(year=year, month=month)

    # Finally, convert to unix epoch time
    start_ts = int(time.mktime(start_dt.timetuple()))

    if start_ts == time_ts:
        start_ts -= 1
    return TimeSpan(start_ts, time_ts)


def isMidnight(time_ts):
    """Is the indicated time on a midnight boundary, local time?
    NB: This algorithm does not work in countries that switch to DST
    at midnight, such as Brazil.
    
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = time.mktime(time.strptime("2013-07-04 01:57:35", "%Y-%m-%d %H:%M:%S"))
    >>> print(isMidnight(time_ts))
    False
    >>> time_ts = time.mktime(time.strptime("2013-07-04 00:00:00", "%Y-%m-%d %H:%M:%S"))
    >>> print(isMidnight(time_ts))
    True
    """

    time_tt = time.localtime(time_ts)
    return time_tt.tm_hour == 0 and time_tt.tm_min == 0 and time_tt.tm_sec == 0


def isStartOfDay(time_ts):
    """Is the indicated time at the start of the day, local time?
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = time.mktime(time.strptime("2013-07-04 01:57:35", "%Y-%m-%d %H:%M:%S"))
    >>> print(isStartOfDay(time_ts))
    False
    >>> time_ts = time.mktime(time.strptime("2013-07-04 00:00:00", "%Y-%m-%d %H:%M:%S"))
    >>> print(isStartOfDay(time_ts))
    True
    >>> os.environ['TZ'] = 'America/Sao_Paulo'
    >>> time_ts = 1541300400
    >>> print(isStartOfDay(time_ts))
    True
    >>> print(isStartOfDay(time_ts - 1))
    False
    """

    # Test the date of the time against the date a tenth of a second before.
    # If they do not match, the time must have been the start of the day
    dt1 = datetime.date.fromtimestamp(time_ts)
    dt2 = datetime.date.fromtimestamp(time_ts - .1)
    return not dt1 == dt2


def archiveDaySpan(time_ts, grace=1, days_ago=0):
    """Returns a TimeSpan representing a day that includes a given time.
    
    Midnight is considered to actually belong in the previous day if
    grace is greater than zero.
    
    time_ts: The day will include this timestamp. 
    
    grace: This many seconds past midnight marks the start of the next
    day. Set to zero to have midnight be included in the
    following day.  [Optional. Default is 1 second.]
    
    days_ago: Which day we want. 0=today, 1=yesterday, etc.
    
    returns: A TimeSpan object one day long. 
    
    Example, which spans the end-of-year boundary
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = time.mktime(time.strptime("2014-01-01 01:57:35", "%Y-%m-%d %H:%M:%S"))
    
    As for today:
    >>> print(archiveDaySpan(time_ts))
    [2014-01-01 00:00:00 PST (1388563200) -> 2014-01-02 00:00:00 PST (1388649600)]

    Ask for yesterday:
    >>> print(archiveDaySpan(time_ts, days_ago=1))
    [2013-12-31 00:00:00 PST (1388476800) -> 2014-01-01 00:00:00 PST (1388563200)]

    Day before yesterday
    >>> print(archiveDaySpan(time_ts, days_ago=2))
    [2013-12-30 00:00:00 PST (1388390400) -> 2013-12-31 00:00:00 PST (1388476800)]
    """
    if time_ts is None:
        return None
    time_ts -= grace
    _day_date = datetime.date.fromtimestamp(time_ts)
    _day_ord = _day_date.toordinal()
    return TimeSpan(_ord_to_ts(_day_ord - days_ago), _ord_to_ts(_day_ord - days_ago + 1))


# For backwards compatibility. Not sure if anyone is actually using this
archiveDaysAgoSpan = archiveDaySpan


def archiveWeekSpan(time_ts, startOfWeek=6, grace=1, weeks_ago=0):
    """Returns a TimeSpan representing a week that includes a given time.
    
    The time at midnight at the end of the week is considered to
    actually belong in the previous week.
    
    time_ts: The week will include this timestamp. 
    
    startOfWeek: The start of the week (0=Monday, 1=Tues, ..., 6 = Sun).

    grace: This many seconds past midnight marks the start of the next
    week. Set to zero to have midnight be included in the
    following week.  [Optional. Default is 1 second.]
    
    weeks_ago: Which week we want. 0=this week, 1=last week, etc.
    
    returns: A TimeSpan object one week long that contains time_ts. It will
    start at midnight of the day considered the start of the week, and be
    one week long.
    
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = 1483429962
    >>> print(timestamp_to_string(time_ts))
    2017-01-02 23:52:42 PST (1483429962)
    >>> print(archiveWeekSpan(time_ts))
    [2017-01-01 00:00:00 PST (1483257600) -> 2017-01-08 00:00:00 PST (1483862400)]
    >>> print(archiveWeekSpan(time_ts, weeks_ago=1))
    [2016-12-25 00:00:00 PST (1482652800) -> 2017-01-01 00:00:00 PST (1483257600)]
    """
    if time_ts is None:
        return None
    time_ts -= grace
    _day_date = datetime.date.fromtimestamp(time_ts)
    _day_of_week = _day_date.weekday()
    _delta = _day_of_week - startOfWeek
    if _delta < 0:
        _delta += 7
    _sunday_date = _day_date - datetime.timedelta(days=(_delta + 7 * weeks_ago))
    _next_sunday_date = _sunday_date + datetime.timedelta(days=7)
    return TimeSpan(int(time.mktime(_sunday_date.timetuple())),
                    int(time.mktime(_next_sunday_date.timetuple())))


def archiveMonthSpan(time_ts, grace=1, months_ago=0):
    """Returns a TimeSpan representing a month that includes a given time.
    
    Midnight of the 1st of the month is considered to actually belong
    in the previous month.
    
    time_ts: The month will include this timestamp. 
    
    grace: This many seconds past midnight marks the start of the next
    month. Set to zero to have midnight be included in the
    following month.  [Optional. Default is 1 second.]
    
    months_ago: Which month we want. 0=this month, 1=last month, etc.
    
    returns: A TimeSpan object one month long that contains time_ts.
    It will start at midnight of the start of the month, and end at midnight
    of the start of the next month.
    
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = 1483429962
    >>> print(timestamp_to_string(time_ts))
    2017-01-02 23:52:42 PST (1483429962)
    >>> print(archiveMonthSpan(time_ts))
    [2017-01-01 00:00:00 PST (1483257600) -> 2017-02-01 00:00:00 PST (1485936000)]
    >>> print(archiveMonthSpan(time_ts, months_ago=1))
    [2016-12-01 00:00:00 PST (1480579200) -> 2017-01-01 00:00:00 PST (1483257600)]
    """
    if time_ts is None:
        return None
    time_ts -= grace

    # First find the first of the month
    day_date = datetime.date.fromtimestamp(time_ts)
    start_of_month_date = day_date.replace(day=1)

    # Total number of months since 0AD
    total_months = 12 * start_of_month_date.year + start_of_month_date.month - 1

    # Adjust for the requested delta:
    total_months -= months_ago

    # Now rebuild the date
    start_year = total_months // 12
    start_month = total_months % 12 + 1
    start_date = datetime.date(year=start_year, month=start_month, day=1)

    # Advance to the start of the next month. This will be the end of the time span.
    total_months += 1
    stop_year = total_months // 12
    stop_month = total_months % 12 + 1
    stop_date = datetime.date(year=stop_year, month=stop_month, day=1)

    return TimeSpan(int(time.mktime(start_date.timetuple())),
                    int(time.mktime(stop_date.timetuple())))


def archiveYearSpan(time_ts, grace=1, years_ago=0):
    """Returns a TimeSpan representing a year that includes a given time.
    
    Midnight of the 1st of the January is considered to actually belong
    in the previous year.
    
    time_ts: The year will include this timestamp. 
    
    grace: This many seconds past midnight marks the start of the next
    year. Set to zero to have midnight be included in the
    following year.  [Optional. Default is 1 second.]
    
    years_ago: Which year we want. 0=this year, 1=last year, etc.
    
    returns: A TimeSpan object one year long that contains time_ts. It will
    begin and end at midnight 1-Jan.
    """
    if time_ts is None:
        return None
    time_ts -= grace
    _day_date = datetime.date.fromtimestamp(time_ts)
    return TimeSpan(int(time.mktime((_day_date.year - years_ago, 1, 1, 0, 0, 0, 0, 0, -1))),
                    int(time.mktime((_day_date.year - years_ago + 1, 1, 1, 0, 0, 0, 0, 0, -1))))


def archiveRainYearSpan(time_ts, sory_mon, grace=1):
    """Returns a TimeSpan representing a rain year that includes a given time.
    
    Midnight of the 1st of the month starting the rain year is considered to
    actually belong in the previous rain year.
    
    time_ts: The rain year will include this timestamp. 
    
    sory_mon: The month the rain year starts.
    
    grace: This many seconds past midnight marks the start of the next
    rain year. Set to zero to have midnight be included in the
    following rain year.  [Optional. Default is 1 second.]
    
    returns: A TimeSpan object one year long that contains time_ts. It will
    begin on the 1st of the month that starts the rain year.
    """
    if time_ts is None:
        return None
    time_ts -= grace
    _day_date = datetime.date.fromtimestamp(time_ts)
    _year = _day_date.year if _day_date.month >= sory_mon else _day_date.year - 1
    return TimeSpan(int(time.mktime((_year, sory_mon, 1, 0, 0, 0, 0, 0, -1))),
                    int(time.mktime((_year + 1, sory_mon, 1, 0, 0, 0, 0, 0, -1))))


def genHourSpans(start_ts, stop_ts):
    """Generator function that generates start/stop of hours in an inclusive range.

    Example:

    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> start_ts = 1204796460
    >>> stop_ts  = 1204818360

    >>> print(timestamp_to_string(start_ts))
    2008-03-06 01:41:00 PST (1204796460)
    >>> print(timestamp_to_string(stop_ts))
    2008-03-06 07:46:00 PST (1204818360)

    >>> for span in genHourSpans(start_ts, stop_ts):
    ...   print(span)
    [2008-03-06 01:00:00 PST (1204794000) -> 2008-03-06 02:00:00 PST (1204797600)]
    [2008-03-06 02:00:00 PST (1204797600) -> 2008-03-06 03:00:00 PST (1204801200)]
    [2008-03-06 03:00:00 PST (1204801200) -> 2008-03-06 04:00:00 PST (1204804800)]
    [2008-03-06 04:00:00 PST (1204804800) -> 2008-03-06 05:00:00 PST (1204808400)]
    [2008-03-06 05:00:00 PST (1204808400) -> 2008-03-06 06:00:00 PST (1204812000)]
    [2008-03-06 06:00:00 PST (1204812000) -> 2008-03-06 07:00:00 PST (1204815600)]
    [2008-03-06 07:00:00 PST (1204815600) -> 2008-03-06 08:00:00 PST (1204819200)]

    start_ts: A time stamp somewhere in the first day.

    stop_ts: A time stamp somewhere in the last day.

    yields: Instance of TimeSpan, where the start is the time stamp
    of the start of the day, the stop is the time stamp of the start
    of the next day.

    """
    _stop_dt = datetime.datetime.fromtimestamp(stop_ts)
    _start_hour = int(start_ts / 3600)
    _stop_hour = int(stop_ts / 3600)
    if (_stop_dt.minute, _stop_dt.second) == (0, 0):
        _stop_hour -= 1

    for _hour in range(_start_hour, _stop_hour + 1):
        yield TimeSpan(_hour * 3600, (_hour + 1) * 3600)


def genDaySpans(start_ts, stop_ts):
    """Generator function that generates start/stop of days in an inclusive range.
    
    Example:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> start_ts = 1204796460
    >>> stop_ts  = 1205265720
    
    >>> print(timestamp_to_string(start_ts))
    2008-03-06 01:41:00 PST (1204796460)
    >>> print(timestamp_to_string(stop_ts))
    2008-03-11 13:02:00 PDT (1205265720)
    
    >>> for span in genDaySpans(start_ts, stop_ts):
    ...   print(span)
    [2008-03-06 00:00:00 PST (1204790400) -> 2008-03-07 00:00:00 PST (1204876800)]
    [2008-03-07 00:00:00 PST (1204876800) -> 2008-03-08 00:00:00 PST (1204963200)]
    [2008-03-08 00:00:00 PST (1204963200) -> 2008-03-09 00:00:00 PST (1205049600)]
    [2008-03-09 00:00:00 PST (1205049600) -> 2008-03-10 00:00:00 PDT (1205132400)]
    [2008-03-10 00:00:00 PDT (1205132400) -> 2008-03-11 00:00:00 PDT (1205218800)]
    [2008-03-11 00:00:00 PDT (1205218800) -> 2008-03-12 00:00:00 PDT (1205305200)]
    
    Note that a daylight savings time change happened 8 March 2009.

    start_ts: A time stamp somewhere in the first day.
    
    stop_ts: A time stamp somewhere in the last day.
    
    yields: Instance of TimeSpan, where the start is the time stamp
    of the start of the day, the stop is the time stamp of the start
    of the next day.
    
    """
    _start_dt = datetime.datetime.fromtimestamp(start_ts)
    _stop_dt = datetime.datetime.fromtimestamp(stop_ts)

    _start_ord = _start_dt.toordinal()
    _stop_ord = _stop_dt.toordinal()
    if (_stop_dt.hour, _stop_dt.minute, _stop_dt.second) == (0, 0, 0):
        _stop_ord -= 1

    for _ord in range(_start_ord, _stop_ord + 1):
        yield TimeSpan(_ord_to_ts(_ord), _ord_to_ts(_ord + 1))


def genMonthSpans(start_ts, stop_ts):
    """Generator function that generates start/stop of months in an
    inclusive range.
    
    Example:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> start_ts = 1196705700
    >>> stop_ts  = 1206101100
    >>> print("start time is %s" % timestamp_to_string(start_ts))
    start time is 2007-12-03 10:15:00 PST (1196705700)
    >>> print("stop time is  %s" % timestamp_to_string(stop_ts))
    stop time is  2008-03-21 05:05:00 PDT (1206101100)
    
    >>> for span in genMonthSpans(start_ts, stop_ts):
    ...   print(span)
    [2007-12-01 00:00:00 PST (1196496000) -> 2008-01-01 00:00:00 PST (1199174400)]
    [2008-01-01 00:00:00 PST (1199174400) -> 2008-02-01 00:00:00 PST (1201852800)]
    [2008-02-01 00:00:00 PST (1201852800) -> 2008-03-01 00:00:00 PST (1204358400)]
    [2008-03-01 00:00:00 PST (1204358400) -> 2008-04-01 00:00:00 PDT (1207033200)]
    
    Note that a daylight savings time change happened 8 March 2009.

    start_ts: A time stamp somewhere in the first month.
    
    stop_ts: A time stamp somewhere in the last month.
    
    yields: Instance of TimeSpan, where the start is the time stamp
    of the start of the month, the stop is the time stamp of the start
    of the next month.
    """
    if None in (start_ts, stop_ts):
        return
    _start_dt = datetime.date.fromtimestamp(start_ts)
    _stop_date = datetime.datetime.fromtimestamp(stop_ts)

    _start_month = 12 * _start_dt.year + _start_dt.month
    _stop_month = 12 * _stop_date.year + _stop_date.month

    if (_stop_date.day, _stop_date.hour, _stop_date.minute, _stop_date.second) == (1, 0, 0, 0):
        _stop_month -= 1

    for month in range(_start_month, _stop_month + 1):
        _this_yr, _this_mo = divmod(month, 12)
        _next_yr, _next_mo = divmod(month + 1, 12)
        yield TimeSpan(time.mktime((_this_yr, _this_mo, 1, 0, 0, 0, 0, 0, -1)),
                       time.mktime((_next_yr, _next_mo, 1, 0, 0, 0, 0, 0, -1)))


def genYearSpans(start_ts, stop_ts):
    if None in (start_ts, stop_ts):
        return
    _start_date = datetime.date.fromtimestamp(start_ts)
    _stop_dt = datetime.datetime.fromtimestamp(stop_ts)

    _start_year = _start_date.year
    _stop_year = _stop_dt.year

    if (_stop_dt.month, _stop_dt.day, _stop_dt.hour,
        _stop_dt.minute, _stop_dt.second) == (1, 1, 0, 0, 0):
            _stop_year -= 1

    for year in range(_start_year, _stop_year + 1):
        yield TimeSpan(time.mktime((year, 1, 1, 0, 0, 0, 0, 0, -1)),
                       time.mktime((year + 1, 1, 1, 0, 0, 0, 0, 0, -1)))


def startOfDay(time_ts):
    """Calculate the unix epoch time for the start of a (local time) day.
    
    time_ts: A timestamp somewhere in the day for which the start-of-day
    is desired.
    
    returns: The timestamp for the start-of-day (00:00) in unix epoch time.
    
    """
    _time_tt = time.localtime(time_ts)
    _bod_ts = time.mktime((_time_tt.tm_year,
                           _time_tt.tm_mon,
                           _time_tt.tm_mday,
                           0, 0, 0, 0, 0, -1))
    return int(_bod_ts)


def startOfGregorianDay(date_greg):
    """Given a Gregorian day, returns the start of the day in unix epoch time.
    
    date_greg: A date as an ordinal Gregorian day.
    
    returns: The local start of the day as a unix epoch time.

    Example:
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> date_greg = 735973  # 10-Jan-2016
    >>> print(startOfGregorianDay(date_greg))
    1452412800
    """
    date_dt = datetime.datetime.fromordinal(date_greg)
    date_tt = date_dt.timetuple()
    sod_ts = int(time.mktime(date_tt))
    return sod_ts


def toGregorianDay(time_ts):
    """Return the Gregorian day a timestamp belongs to.
    
    time_ts: A time in unix epoch time.
    
    returns: The ordinal Gregorian day that contains that time
    
    Example:
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> time_ts = 1452412800  # Midnight, 10-Jan-2016
    >>> print(toGregorianDay(time_ts))
    735972
    >>> time_ts = 1452412801  # Just after midnight, 10-Jan-2016
    >>> print(toGregorianDay(time_ts))
    735973
    """

    date_dt = datetime.datetime.fromtimestamp(time_ts)
    date_greg = date_dt.toordinal()
    if date_dt.hour == date_dt.minute == date_dt.second == 0:
        # Midnight actually belongs to the previous day
        date_greg -= 1
    return date_greg


def startOfDayUTC(time_ts):
    """Calculate the unix epoch time for the start of a UTC day.
    
    time_ts: A timestamp somewhere in the day for which the start-of-day
    is desired.
    
    returns: The timestamp for the start-of-day (00:00) in unix epoch time.
    
    """
    _time_tt = time.gmtime(time_ts)
    _bod_ts = calendar.timegm((_time_tt.tm_year,
                               _time_tt.tm_mon,
                               _time_tt.tm_mday,
                               0, 0, 0, 0, 0, -1))
    return int(_bod_ts)


def startOfArchiveDay(time_ts, grace=1):
    """Given an archive time stamp, calculate its start of day.
    
    similar to startOfDay(), except that an archive stamped at midnight
    actually belongs to the *previous* day.

    time_ts: A timestamp somewhere in the day for which the start-of-day
    is desired.
    
    grace: The number of seconds past midnight when the following
    day is considered to start [Optional. Default is 1 second]
    
    returns: The timestamp for the start-of-day (00:00) in unix epoch time."""

    return startOfDay(time_ts - grace)


def getDayNightTransitions(start_ts, end_ts, lat, lon):
    """Return the day-night transitions between the start and end times.

    start_ts: A timestamp (UTC) indicating the beginning of the period

    end_ts: A timestamp (UTC) indicating the end of the period

    returns: indication of whether the period from start to first transition
    is day or night, plus array of transitions (UTC).
    """
    from weeutil import Sun

    first = None
    values = []
    for t in range(start_ts - 3600 * 24, end_ts + 3600 * 24 + 1, 3600 * 24):
        x = startOfDayUTC(t)
        x_tt = time.gmtime(x)
        y, m, d = x_tt[:3]
        (sunrise_utc, sunset_utc) = Sun.sunRiseSet(y, m, d, lon, lat)
        daystart_ts = calendar.timegm((y, m, d, 0, 0, 0, 0, 0, -1))
        sunrise_ts = int(daystart_ts + sunrise_utc * 3600.0 + 0.5)
        sunset_ts = int(daystart_ts + sunset_utc * 3600.0 + 0.5)

        if start_ts < sunrise_ts < end_ts:
            values.append(sunrise_ts)
            if first is None:
                first = 'night'
        if start_ts < sunset_ts < end_ts:
            values.append(sunset_ts)
            if first is None:
                first = 'day'
    return first, values


def secs_to_string(secs):
    """Convert seconds to a string with days, hours, and minutes"""
    str_list = []
    for (label, interval) in (('day', 86400), ('hour', 3600), ('minute', 60)):
        amt = int(secs / interval)
        plural = '' if amt == 1 else 's'
        str_list.append("%d %s%s" % (amt, label, plural))
        secs %= interval
    ans = ', '.join(str_list)
    return ans


def timestamp_to_string(ts, format_str="%Y-%m-%d %H:%M:%S %Z"):
    """Return a string formatted from the timestamp
    
    Example:

    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> print(timestamp_to_string(1196705700))
    2007-12-03 10:15:00 PST (1196705700)
    >>> print(timestamp_to_string(None))
    ******* N/A *******     (    N/A   )
    """
    if ts is not None:
        return "%s (%d)" % (time.strftime(format_str, time.localtime(ts)), ts)
    else:
        return "******* N/A *******     (    N/A   )"


def timestamp_to_gmtime(ts):
    """Return a string formatted for GMT
    
    >>> print(timestamp_to_gmtime(1196705700))
    2007-12-03 18:15:00 UTC (1196705700)
    >>> print(timestamp_to_gmtime(None))
    ******* N/A *******     (    N/A   )
    """
    if ts:
        return "%s (%d)" % (time.strftime("%Y-%m-%d %H:%M:%S UTC", time.gmtime(ts)), ts)
    else:
        return "******* N/A *******     (    N/A   )"


def utc_to_ts(y, m, d, hrs_utc):
    """Converts from a UTC tuple-time to unix epoch time.
    
    y,m,d: The year, month, day for which the conversion is desired.
    
    hrs_tc: Floating point number with the number of hours since midnight in UTC.
    
    Returns: The unix epoch time.
    
    >>> print(utc_to_ts(2009, 3, 27, 14.5))
    1238164200
    """
    # Construct a time tuple with the time at midnight, UTC:
    daystart_utc_tt = (y, m, d, 0, 0, 0, 0, 0, -1)
    # Convert the time tuple to a time stamp and add on the number of seconds since midnight:
    time_ts = int(calendar.timegm(daystart_utc_tt) + hrs_utc * 3600.0 + 0.5)
    return time_ts


def utc_to_local_tt(y, m, d, hrs_utc):
    """Converts from a UTC time to a local time.
    
    y,m,d: The year, month, day for which the conversion is desired.
    
    hrs_tc: Floating point number with the number of hours since midnight in UTC.
    
    Returns: A timetuple with the local time.
    
    >>> os.environ['TZ'] = 'America/Los_Angeles'
    >>> tt=utc_to_local_tt(2009, 3, 27, 14.5)
    >>> print(tt.tm_year, tt.tm_mon, tt.tm_mday, tt.tm_hour, tt.tm_min)
    2009 3 27 7 30
    """
    # Get the UTC time:
    time_ts = utc_to_ts(y, m, d, hrs_utc)
    # Convert to local time:
    time_local_tt = time.localtime(time_ts)
    return time_local_tt


def latlon_string(ll, hemi, which, format_list=None):
    """Decimal degrees into a string for degrees, and one for minutes.
    ll: The decimal latitude or longitude
    hemi: A tuple holding strings representing positive or negative values. E.g.: ('N', 'S')
    which: 'lat' for latitude, 'long' for longitude
    format_list: A list or tuple holding the format strings to be used. These are [whole degrees latitude, 
                 whole degrees longitude, minutes]
                 
    Returns:
    A 3-way tuple holding (latlon whole degrees, latlon minutes, hemisphere designator). 
    Example: (022, 08.3, 'N') """
    labs = abs(ll)
    (frac, deg) = math.modf(labs)
    minutes = frac * 60.0
    if format_list is None:
        format_list = ["%02d", "%03d", "%05.2f"]
    return ((format_list[0] if which == 'lat' else format_list[1]) % (deg,), format_list[2] % (minutes,),
            hemi[0] if ll >= 0 else hemi[1])


def log_traceback(prefix='', loglevel=syslog.LOG_INFO):
    """Log the stack traceback into syslog."""
    sfd = StringIO()
    traceback.print_exc(file=sfd)
    sfd.seek(0)
    for line in sfd:
        syslog.syslog(loglevel, prefix + line)


def _get_object(module_class):
    """Given a string with a module class name, it imports and returns the class."""
    # Split the path into its parts
    parts = module_class.split('.')
    # Strip off the classname:
    module = '.'.join(parts[:-1])
    # Import the top level module
    mod = __import__(module)
    # Recursively work down from the top level module to the class name.
    # Be prepared to catch an exception if something cannot be found.
    try:
        for part in parts[1:]:
            mod = getattr(mod, part)
    except AttributeError:
        # Can't find something. Give a more informative error message:
        raise AttributeError(
            "Module '%s' has no attribute '%s' when searching for '%s'" % (mod.__name__, part, module_class))
    return mod


class GenWithPeek(object):
    """Generator object which allows a peek at the next object to be returned.
    
    Sometimes Python solves a complicated problem with such elegance! This is
    one of them.
    
    Example of usage:
    >>> # Define a generator function:
    >>> def genfunc(N):
    ...     for j in range(N):
    ...        yield j
    >>>
    >>> # Now wrap it with the GenWithPeek object:
    >>> g_with_peek = GenWithPeek(genfunc(5))
    >>> # We can iterate through the object as normal:
    >>> for i in g_with_peek:
    ...    print(i)
    ...    # Every second object, let's take a peek ahead
    ...    if i%2:
    ...        # We can get a peek at the next object without disturbing the wrapped generator:
    ...        print("peeking ahead, the next object will be: %s" % g_with_peek.peek())
    0
    1
    peeking ahead, the next object will be: 2
    2
    3
    peeking ahead, the next object will be: 4
    4
    """

    def __init__(self, generator):
        """Initialize the generator object.
        
        generator: A generator object to be wrapped
        """
        self.generator = generator
        self.have_peek = False
        self.peek_obj = None

    def __iter__(self):
        return self

    def next(self):
        """Advance to the next object"""
        if self.have_peek:
            self.have_peek = False
            return self.peek_obj
        else:
            return next(self.generator)

    def peek(self):
        """Take a peek at the next object"""
        if not self.have_peek:
            self.peek_obj = next(self.generator)
            self.have_peek = True
        return self.peek_obj

    # For Python 3 compatiblity
    __next__ = next


def tobool(x):
    """Convert an object to boolean.
    
    Examples:
    >>> print(tobool('TRUE'))
    True
    >>> print(tobool(True))
    True
    >>> print(tobool(1))
    True
    >>> print(tobool('FALSE'))
    False
    >>> print(tobool(False))
    False
    >>> print(tobool(0))
    False
    >>> print(tobool('Foo'))
    Traceback (most recent call last):
    ValueError: Unknown boolean specifier: 'Foo'.
    >>> print(tobool(None))
    Traceback (most recent call last):
    ValueError: Unknown boolean specifier: 'None'.
    """

    try:
        if x.lower() in ['true', 'yes']:
            return True
        elif x.lower() in ['false', 'no']:
            return False
    except AttributeError:
        pass
    try:
        return bool(int(x))
    except (ValueError, TypeError):
        pass
    raise ValueError("Unknown boolean specifier: '%s'." % x)


to_bool = tobool


def to_int(x):
    """Convert an object to an integer, unless it is None
    
    Examples:
    >>> print(to_int(123))
    123
    >>> print(to_int('123'))
    123
    >>> print(to_int(-5.2))
    -5
    >>> print(to_int(None))
    None
    """
    if isinstance(x, str) and x.lower() == 'none':
        x = None
    try:
        return int(x) if x is not None else None
    except ValueError:
        # Perhaps it's a string, holding a floating point number?
        return int(float(x))


def to_float(x):
    """Convert an object to a float, unless it is None
    
    Examples:
    >>> print(to_float(12.3))
    12.3
    >>> print(to_float('12.3'))
    12.3
    >>> print(to_float(None))
    None
    """
    if isinstance(x, str) and x.lower() == 'none':
        x = None
    return float(x) if x is not None else None


def to_unicode(string, encoding='utf8'):
    u"""Convert to Unicode, unless string is None
    
    Example:
    >>> print(to_unicode(u"degree sign from UTF8: °".encode('utf8')))
    degree sign from UTF8: °
    >>> print(to_unicode(u"degree sign from Unicode: °"))
    degree sign from Unicode: °
    >>> print(to_unicode(None))
    None
    """
    try:
        return unicode(string, encoding) if string is not None else None
    except NameError:
        # No unicode function. Probably running under Python 3
        try:
            return str(string, encoding) if string is not None else None
        except TypeError:
            # The string is already in Unicode. Just return it.
            return string
    except TypeError:
        # The string is already in Unicode. Just return it.
        return string


def min_with_none(x_seq):
    """Find the minimum in a (possibly empty) sequence, ignoring Nones"""
    xmin = None
    for x in x_seq:
        if xmin is None:
            xmin = x
        elif x is not None:
            xmin = min(x, xmin)
    return xmin


def max_with_none(x_seq):
    """Find the maximum in a (possibly empty) sequence, ignoring Nones.

    While this function is not necessary under Python 2, under Python 3 it is.
    """
    xmax = None
    for x in x_seq:
        if xmax is None:
            xmax = x
        elif x is not None:
            xmax = max(x, xmax)
    return xmax


def move_with_timestamp(filepath):
    """Save a file to a path with a timestamp."""
    import shutil
    # Sometimes the target has a trailing '/'. This will take care of it:
    filepath = os.path.normpath(filepath)
    newpath = filepath + time.strftime(".%Y%m%d%H%M%S")
    # Check to see if this name already exists
    if os.path.exists(newpath):
        # It already exists. Stick a version number on it:
        version = 1
        while os.path.exists(newpath + '-' + str(version)):
            version += 1
        newpath = newpath + '-' + str(version)
    shutil.move(filepath, newpath)
    return newpath


class ListOfDicts(dict):
    """A list of dictionaries, that are searched in order.
    
    It assumes only that any inserted dictionaries support a keyed
    lookup using the syntax obj[key].
    
    Example:

    # Try an empty dictionary:
    >>> lod = ListOfDicts()
    >>> print(lod['b'])
    Traceback (most recent call last):
    KeyError: 'b'
    >>> # Now initialize it with a starting dictionary:
    >>> lod = ListOfDicts({'a':1, 'b':2, 'c':3})
    >>> print(lod['b'])
    2
    >>> # Look for a non-existent key
    >>> print(lod['d'])
    Traceback (most recent call last):
    KeyError: 'd'
    >>> # Now extend the dictionary:
    >>> lod.extend({'d':4, 'e':5})
    >>> # And try the lookup:
    >>> print(lod['d'])
    4
    >>> # Explicitly add a new key to the dictionary:
    >>> lod['f'] = 6
    >>> # Try it:
    >>> print(lod['f'])
    6
    """

    def __init__(self, starting_dict=None):
        if starting_dict:
            super(ListOfDicts, self).__init__(starting_dict)
        self.dict_list = []

    def __getitem__(self, key):
        for this_dict in self.dict_list:
            try:
                return this_dict[key]
            except KeyError:
                pass
        return dict.__getitem__(self, key)

    def get(self, key, default=None):
        try:
            return self[key]
        except KeyError:
            return default

    def extend(self, new_dict):
        self.dict_list.append(new_dict)


class KeyDict(dict):
    """A dictionary that returns the key for an unsuccessful lookup."""
    def __missing__(self, key):
        return key


def to_sorted_string(rec):
    return ", ".join(["%s: %s" % (k, rec.get(k)) for k in sorted(rec, key=str.lower)])


# Define an "input" function that works for both Python 2 and 3:
# An exception will be raised in Python 3, but not Python 2
try:
    input = raw_input
except NameError:
    # Python 3
    pass


def y_or_n(msg, noprompt):
    """Prompt and look for a 'y' or 'n' response"""

    # If noprompt is truthy, always return 'y'
    if noprompt:
        return 'y'

    ans = None
    while ans not in ['y', 'n']:
        ans = input(msg)
    return ans


def int2byte(x):
    """Convert integer argument to signed byte string, under both Python 2 and 3"""
    return struct.pack('>b', x)


def deep_copy_path(path, dest_dir):
    """Copy a path to a destination, making any subdirectories along the way.
    The source path is relative to the current directory.

    Returns the number of files copied
    """

    ncopy = 0
    # Are we copying a directory?
    if os.path.isdir(path):
        # Yes. Walk it
        for dirpath, _, filenames in os.walk(path):
            for f in filenames:
                # For each source file found, call myself recursively:
                ncopy += deep_copy_path(os.path.join(dirpath, f), dest_dir)
    else:
        # path is a file. Get the directory it's in.
        d = os.path.dirname(os.path.join(dest_dir, path))
        # Make the destination directory, wrapping it in a try block in
        # case it already exists:
        try:
            os.makedirs(d)
        except OSError:
            pass
        # This version of copy does not copy over modification time,
        # so it will look like a new file, causing it to be (for
        # example) ftp'd to the server:
        shutil.copy(path, d)
        ncopy += 1
    return ncopy

if __name__ == '__main__':
    import doctest

    if not doctest.testmod().failed:
        print("PASSED")

Reply via email to