Create text data file based on a template

usage: python -m pywws.template [options] data_dir template_file output_file
options are:
 --help    display this help
data_dir is the root directory of the weather data
template_file is the template text source file
output_file is the name of the text file to be created


This is probably the most difficult to use module in the weather station software collection. It generates text files based on a “template” file plus the raw, hourly, daily & monthly weather station data. The template processing goes beyond simple substitution of values to include loops, jumps forwards or backwards in the data, processing of the data and substitution of missing values.

A template file can be any sort of text file (plain text, xml, html, etc.) to which “processing instructions” have been added. These processing instructions are delimited by hash (‘#’) characters. They are not copied to the output, but cause something else to happen: either a data value is inserted or one of a limited number of other actions is carried out.

Before writing your own template files, it might be useful to look at some of the examples in the example_templates directory.

Text encoding

The [config] section of weather.ini has a template encoding entry that tells pywws what text encoding most of your template files use. The default value, iso-8859-1, is suitable for most western European languages, but may need changing if you use another language. It can be set to any text encoding recognised by the Python codecs module.

Make sure your templates use the text encoding you set. The iconv program can be used to transcode files.

New in version 16.04.0: the #encoding# processing instruction can be used to set the text encoding of a template file.

Processing instructions

Note that if the closing ‘#’ of a processing instruction is the last character on a line then the following line break is not outputted. This makes templates easier to edit as you can have a separate line for each processing instruction and still produce output with no line breaks. If you want to output a line break after a processing instruction, put a blank line immediately after it.

Processing instructions can be split across lines to improve readability. Split lines are joined together before processing, after removing any trailing newline characters.


output a single ‘#’ character.

#! comment text#

a comment, no output generated. comment text can be any text without a line break.


switch to “monthly” summary data. The index is reset to the most recent value.


switch to “daily” summary data. The index is reset to the most recent value.


switch to “hourly” summary data. The index is reset to the most recent value.


switch to “raw” data. The index is reset to the most recent value.

Changed in version 11.09: This now selects “calibrated” data. The directive name remains unchanged for backwards compatibility.


switch to “live” data. If the template is processed in the [live] section of weather.ini this will select the most up-to-date weather data, otherwise it will have the same effect as #raw#. Any #jump# will go to “raw” data.

#timezone name#

convert all datetime values to time zone name before output. Permitted values for name are utc or local.

#locale expr#

switch use of ‘locale’ on or off, according to expr. When locale is on floating point numbers may use a comma as the decimal separator instead of a point, depending on your localisation settings. Use "True" or "False" for expr.

#encoding expr#

New in version 16.04.0.

set the template text encoding to expr, e.g. ascii, utf8 or html. The html encoding is a special case. It writes ascii files but with non ASCII characters converted to HTML entities.

Any #encoding# directive should be placed near the beginning of the template file, before any non-ASCII characters are used.

#roundtime expr#

switch time rounding on or off, according to expr. When time rounding is on, 30 seconds is added to each time value used. This is useful if you are only printing out hours and minutes, e.g. with a “%H:%M” format, and want time values such as 10:23:58 to appear as “10:24”. Use "True" or "False" for expr.

#jump count#

jump count values. The data index is adjusted by count hours or days. Negative values jump back in time.

It is a good idea to put jumps within a loop at the end, just before the #endloop# instruction. The loop can then terminate cleanly if it has run out of data.

#goto date-time#

go to date-time. The data index is adjusted to the record immediately after date-time. This can be in UTC or your local time zone, according to the setting of timezone, and must exactly match the ISO date format, for example "2010-11-01 12:00:00" is noon on 1st November 2010.

Parts of date-time can be replaced with strftime style % format characters to specify the current loop index. For example, "%Y-%m-01 12:00:00" is noon on 1st of this month.

#loop count#

start a loop that will repeat count times. count must be one or more.


end a loop started by #loop count#. The template processing will go back to the line containing the #loop count# instruction. Don’t try to nest loops.

#key fmt_string no_value_string conversion#

output a data value. key is the data key, e.g. temp_out for outdoor temperature. fmt_string is a printf-like format string (actually Python’s % operator) except for datetime values, when it is input to datetime’s strftime() method. no_value_string is output instead of fmt_string when the data value is absent, e.g. if the station lost contact with the outside sensor. conversion is a Python expression to convert the data, e.g. to convert wind speed from m/s to mph you could use "x * 3.6 / 1.609344", or the more convenient provided function "wind_mph(x)". See the pywws.conversions module for details of the available functions.

All these values need double quotes ” if they contain spaces or other potentially difficult characters. All except key are optional, but note that if you want to specify a conversion, you also need to specify fmt_string and no_value_string.

#calc expression fmt_string no_value_string conversion#

output a value computed from one or more data items. expression is any valid Python expression, e.g. "dew_point(data['temp_out'], data['hum_out'])" to compute the outdoor dew point. fmt_string, no_value_string and conversion are as described above. Note that it is probably more efficient to incorporate any conversion into expression.

In addition to the functions in the pywws.conversions module there are four more useful functions: rain_hour(data) returns the amount of rain in the last hour, rain_day(data) returns the amount of rain since midnight (local time), rain_24hr(data) returns the amount of rain in the last 24 hours, and hour_diff(data, key) returns the change in data item key over the last hour.


Here is an example snippet showing basic and advanced use of the template features. It is part of the 6hrs.txt example template file, which generates an HTML table of 7 hourly readings (which should span 6 hours).

#jump -6#
#loop 7#
    <td>#idx "%Y/%m/%d" "" "[None, x][x.hour == 0 or loop_count == 7]"#</td>
    <td>#idx "%H%M %Z"#</td>
    <td>#temp_out "%.1f °C"#</td>
    <td>#hum_out "%d%%"#</td>
    <td>#wind_dir "%s" "-" "winddir_text(x)"#</td>
    <td>#wind_ave "%.0f mph" "" "wind_mph(x)"#</td>
    <td>#wind_gust "%.0f mph" "" "wind_mph(x)"#</td>
    <td>#rain "%0.1f mm"#</td>
    <td>#rel_pressure "%.0f hPa"#, #pressure_trend "%s" "" "pressure_trend_text(x)"#</td>
#jump 1#

The first three lines of this snippet do the following: select hourly data, jump back 6 hours, start a loop with a count of 7. A jump forward of one hour appears just before the end of the repeated segment. As this last jump (of one hour) happens each time round the loop, a sequence of 7 data readings will be output. The last line marks the end of the loop — everything between the #loop 7# and #endloop# lines is output 7 times.

The #temp_out ...#, #hum_out ...#, #rain ...# and #rel_pressure ...# instructions show basic data output. They each use a fmt_string to format the data appropriately. The #wind_ave ...# and #wind_gust ...# instructions show how to use a conversion expression to convert m/s to mph.

The #wind_dir ...# and #pressure_trend ...# instructions show use of the built-in functions winddir_text and pressure_trend_text to convert numerical values into text.

Finally we get to datetime values. The #idx "%H%M"# instruction simply outputs the time (in HHMM format) of the data’s index. The #idx "%Y/%m/%d" "" "[None, x][x.hour == 0 or loop_count == 7]"# instruction is a bit more complicated. It outputs the date, but only on the first line or if the date has changed. It does this by indexing the array [None, x] with a boolean expression that is true when loop_count is 7 (i.e. on the first pass through the loop) or x.hour is zero (i.e. this is the first hour of the day).

Detailed API




Template(context[, use_locale])
class pywws.template.Computations(context)[source]

Bases: object

hour_diff(data, key)[source]
class pywws.template.Template(context, use_locale=True)[source]

Bases: object

process(live_data, template_file)[source]
make_text(template_file, live_data=None)[source]
make_file(template_file, output_file, live_data=None)[source]

Comments or questions? Please subscribe to the pywws mailing list and let us know.