utils.py

import base64
import codecs
import datetime
from io import BytesIO
import json
import re
from chroma import Color, Scale
from colour import Color as Colour
import geojson
from matplotlib.image import imsave
import requests

from .colors import color_ramps, common_html_colors
from .errors import SourceDataError, DateConversionError


def row_to_geojson(row, lon, lat, precision, date_format='epoch'):
    """Convert a pandas dataframe row to a geojson format object.  Converts all datetimes to epoch seconds.
    """

    # Let pandas handle json serialization
    row_json = json.loads(row.to_json(date_format=date_format, date_unit='s'))
    return geojson.Feature(geometry=geojson.Point((round(row_json[lon], precision), round(row_json[lat], precision))),
                           properties={key: row_json[key] for key in row_json.keys() if key not in [lon, lat]})


def df_to_geojson(df, properties=None, lat='lat', lon='lon', precision=6, date_format='epoch', filename=None):
    """Serialize a Pandas dataframe to a geojson format Python dictionary / file
    """

    if not properties:
        # if no properties are selected, use all properties in dataframe
        properties = [c for c in df.columns if c not in [lon, lat]]

    for prop in properties:
        # Check if list of properties exists in dataframe columns
        if prop not in list(df.columns):
            raise ValueError(
                'properties must be a valid list of column names from dataframe')
        if prop in [lon, lat]:
            raise ValueError(
                'properties cannot be the geometry longitude or latitude column')

    # convert dates/datetimes to preferred string format if specified
    df = convert_date_columns(df, date_format)

    if filename:
        with open(filename, 'w') as f:
            # Overwrite file if it already exists
            pass

        with open(filename, 'a+') as f:

            # Write out file to line
            f.write('{"type": "FeatureCollection", "features": [\n')
            # Iterate over enumerated iterrows as index from iterrows alone could be non-sequential
            for i, (index, row) in enumerate(df[[lon, lat] + properties].iterrows()):
                if i == 0:
                    f.write(geojson.dumps(row_to_geojson(row, lon, lat, precision, date_format)) + '\n')
                else:
                    f.write(',' + geojson.dumps(row_to_geojson(row, lon, lat, precision, date_format)) + '\n')
            f.write(']}')

            return {
                "type": "file",
                "filename": filename,
                "feature_count": df.shape[0]
            }
    else:
        features = []
        df[[lon, lat] + properties].apply(lambda x: features.append(
            row_to_geojson(x, lon, lat, precision, date_format)), axis=1)
        return geojson.FeatureCollection(features)


def geojson_to_dict_list(data):
    """Parse GeoJSON-formatted information in <data> to list of Python dicts"""
    
    # return data formatted as list or dict
    if type(data) in (list, dict):
        return data

    # read from data defined as local file address
    try:
        with open(data, 'r') as f:
            features = json.load(f)['features']

    # if data is defined as a URL, load JSON object from address
    except IOError:
        features = requests.get(data).json()['features']

    except:
        raise SourceDataError('MapViz data must be valid GeoJSON or JSON.  Please check your <data> parameter.')

    return [feature['properties'] for feature in features]


def gdf_to_geojson(gdf, date_format='epoch', properties=None, filename=None):
    """Serialize a GeoPandas dataframe to a geojson format Python dictionary / file
    """

    # convert dates/datetimes to preferred string format if specified
    gdf = convert_date_columns(gdf, date_format)

    gdf_out = gdf[['geometry'] + properties or []]

    geojson_str = gdf_out.to_json()

    if filename:
        with codecs.open(filename, "w", "utf-8-sig") as f:
            f.write(geojson_str)
        return None
    else:
        return json.loads(geojson_str)


def convert_date_columns(df, date_format='epoch'):
    """Convert dates/datetimes to preferred string format if specified
        i.e. '%Y-%m-%d', 'epoch', 'iso'
    """

    if date_format not in ['epoch', 'iso']:
        if '%' in date_format:
            try:
                datetime.datetime.now().strftime(date_format)
            except:
                raise DateConversionError('Error serializing dates in DataFrame using format {}.'.format(date_format))
            finally:
                for column, data_type in df.dtypes.to_dict().items():
                    if 'date' in str(data_type):
                        df[column] = df[column].dt.strftime(date_format)
        else:
            raise DateConversionError('Error serializing dates in DataFrame using format {}.'.format(date_format))

    return df


def scale_between(minval, maxval, numStops):
    """ Scale a min and max value to equal interval domain with
        numStops discrete values
    """

    scale = []

    if numStops < 2:
        return [minval, maxval]
    elif maxval < minval:
        raise ValueError()
    else:
        domain = maxval - minval
        interval = float(domain) / float(numStops)
        for i in range(numStops):
            scale.append(round(minval + interval * i, 2))
        return scale


def create_radius_stops(breaks, min_radius, max_radius):
    """Convert a data breaks into a radius ramp
    """
    num_breaks = len(breaks)
    radius_breaks = scale_between(min_radius, max_radius, num_breaks)
    stops = []

    for i, b in enumerate(breaks):
        stops.append([b, radius_breaks[i]])
    return stops


def create_weight_stops(breaks):
    """Convert data breaks into a heatmap-weight ramp
    """
    num_breaks = len(breaks)
    weight_breaks = scale_between(0, 1, num_breaks)
    stops = []

    for i, b in enumerate(breaks):
        stops.append([b, weight_breaks[i]])
    return stops


def create_numeric_stops(breaks, min_value, max_value):
    """Convert data breaks into a general numeric ramp (height, radius, weight, etc.)
    """
    weight_breaks = scale_between(min_value, max_value, len(breaks))
    return [list(x) for x in zip(breaks, weight_breaks)]


def create_color_stops(breaks, colors='RdYlGn', color_ramps=color_ramps):
    """Convert a list of breaks into color stops using colors from colorBrewer
    or a custom list of color values in RGB, RGBA, HSL, CSS text, or HEX format.
    See www.colorbrewer2.org for a list of color options to pass
    """

    num_breaks = len(breaks)
    stops = []

    if isinstance(colors, list):
        # Check if colors contain a list of color values
        if len(colors) == 0 or len(colors) != num_breaks:
            raise ValueError(
                'custom color list must be of same length as breaks list')

        for color in colors:
            # Check if color is valid string
            try:
                Colour(color)
            except:
                raise ValueError(
                    'The color code {color} is in the wrong format'.format(color=color))

        for i, b in enumerate(breaks):
            stops.append([b, colors[i]])

    else:
        if colors not in color_ramps.keys():
            raise ValueError('color does not exist in colorBrewer!')
        else:

            try:
                ramp = color_ramps[colors][num_breaks]
            except KeyError:
                raise ValueError("Color ramp {} does not have a {} breaks".format(
                    colors, num_breaks))

            for i, b in enumerate(breaks):
                stops.append([b, ramp[i]])

    return stops


def rgb_tuple_from_str(color_string):
    """Convert color in format 'rgb(RRR,GGG,BBB)', 'rgba(RRR,GGG,BBB,alpha)',  
    '#RRGGBB', or limited English color name (eg 'red') to tuple (RRR, GGG, BBB)
    """
    try:
        # English color names (limited)
        rgb_string = common_html_colors[color_string]
        return tuple([float(x) for x in re.findall(r'\d{1,3}', rgb_string)]) 
    
    except KeyError:
        try:
            # HEX color code
            hex_string = color_string.lstrip('#')
            return tuple(int(hex_string[i:i+2], 16) for i in (0, 2 ,4))
        
        except ValueError:
            # RGB or RGBA formatted strings
            return tuple([int(x) if float(x) > 1 else float(x) 
                          for x in re.findall(r"[-+]?\d*\.*\d+", color_string)])


def color_map(lookup, color_stops, default_color='rgb(122,122,122)'):
    """Return an rgb color value interpolated from given color_stops;
    assumes colors in color_stops provided as strings of form 'rgb(RRR,GGG,BBB)'
    or in hex: '#RRGGBB'
    """
    # if no color_stops, use default color
    if len(color_stops) == 0:
        return default_color
    
    # dictionary to lookup color from match-type color_stops
    match_map = dict((x, y) for (x, y) in color_stops)

    # if lookup matches stop exactly, return corresponding color (first priority)
    # (includes non-numeric color_stop "keys" for finding color by match)
    if lookup in match_map.keys():
        return match_map.get(lookup)

    # if lookup value numeric, map color by interpolating from color scale
    if isinstance(lookup, (int, float, complex)):

        # try ordering stops 
        try:
            stops, colors = zip(*sorted(color_stops))
        
        # if not all stops are numeric, attempt looking up as if categorical stops
        except TypeError:
            return match_map.get(lookup, default_color)

        # for interpolation, all stops must be numeric
        if not all(isinstance(x, (int, float, complex)) for x in stops):
            return default_color

        # check if lookup value in stops bounds
        if float(lookup) <= stops[0]:
            return colors[0]
        
        elif float(lookup) >= stops[-1]:
            return colors[-1]
        
        # check if lookup value matches any stop value
        elif float(lookup) in stops:
            return colors[stops.index(lookup)]
        
        # interpolation required
        else:

            rgb_tuples = [Color(rgb_tuple_from_str(x)) for x in colors]

            # identify bounding color stop values
            lower = max([stops[0]] + [x for x in stops if x < lookup])
            upper = min([stops[-1]] + [x for x in stops if x > lookup])
            
            # colors from bounding stops
            lower_color = rgb_tuples[stops.index(lower)]
            upper_color = rgb_tuples[stops.index(upper)]
            
            # generate color scale for mapping lookup value to interpolated color
            scale = Scale(Color(lower_color), Color(upper_color))

            # compute linear "relative distance" from lower bound color to upper bound color
            distance = (lookup - lower) / (upper - lower)

            # return string representing rgb color value
            return scale(distance).to_string().replace(', ', ',')

    # default color value catch-all
    return default_color


def numeric_map(lookup, numeric_stops, default=0.0):
    """Return a number value interpolated from given numeric_stops
    """
    # if no numeric_stops, use default
    if len(numeric_stops) == 0:
        return default
    
    # dictionary to lookup value from match-type numeric_stops
    match_map = dict((x, y) for (x, y) in numeric_stops)

    # if lookup matches stop exactly, return corresponding stop (first priority)
    # (includes non-numeric numeric_stop "keys" for finding value by match)
    if lookup in match_map.keys():
        return match_map.get(lookup)

    # if lookup value numeric, map value by interpolating from scale
    if isinstance(lookup, (int, float, complex)):

        # try ordering stops 
        try:
            stops, values = zip(*sorted(numeric_stops))
        
        # if not all stops are numeric, attempt looking up as if categorical stops
        except TypeError:
            return match_map.get(lookup, default)

        # for interpolation, all stops must be numeric
        if not all(isinstance(x, (int, float, complex)) for x in stops):
            return default

        # check if lookup value in stops bounds
        if float(lookup) <= stops[0]:
            return values[0]
        
        elif float(lookup) >= stops[-1]:
            return values[-1]
        
        # check if lookup value matches any stop value
        elif float(lookup) in stops:
            return values[stops.index(lookup)]
        
        # interpolation required
        else:

            # identify bounding stop values
            lower = max([stops[0]] + [x for x in stops if x < lookup])
            upper = min([stops[-1]] + [x for x in stops if x > lookup])
            
            # values from bounding stops
            lower_value = values[stops.index(lower)]
            upper_value = values[stops.index(upper)]
            
            # compute linear "relative distance" from lower bound to upper bound
            distance = (lookup - lower) / (upper - lower)

            # return interpolated value
            return lower_value + distance * (upper_value - lower_value)

    # default value catch-all
    return default


def img_encode(arr, **kwargs):
    """Encode ndarray to base64 string image data
    
    Parameters
    ----------
    arr: ndarray (rows, cols, depth)
    kwargs: passed directly to matplotlib.image.imsave
    """
    sio = BytesIO()
    imsave(sio, arr, **kwargs)
    sio.seek(0)
    img_format = kwargs['format'] if kwargs.get('format') else 'png'
    img_str = base64.b64encode(sio.getvalue()).decode()

    return 'data:image/{};base64,{}'.format(img_format, img_str)


def height_map(lookup, height_stops, default_height=0.0):
    """Return a height value (in meters) interpolated from given height_stops;
    for use with vector-based visualizations using fill-extrusion layers
    """
    # if no height_stops, use default height
    if len(height_stops) == 0:
        return default_height
    
    # dictionary to lookup height from match-type height_stops
    match_map = dict((x, y) for (x, y) in height_stops)

    # if lookup matches stop exactly, return corresponding height (first priority)
    # (includes non-numeric height_stop "keys" for finding height by match)
    if lookup in match_map.keys():
        return match_map.get(lookup)

    # if lookup value numeric, map height by interpolating from height scale
    if isinstance(lookup, (int, float, complex)):

        # try ordering stops 
        try:
            stops, heights = zip(*sorted(height_stops))
        
        # if not all stops are numeric, attempt looking up as if categorical stops
        except TypeError:
            return match_map.get(lookup, default_height)

        # for interpolation, all stops must be numeric
        if not all(isinstance(x, (int, float, complex)) for x in stops):
            return default_height

        # check if lookup value in stops bounds
        if float(lookup) <= stops[0]:
            return heights[0]
        
        elif float(lookup) >= stops[-1]:
            return heights[-1]
        
        # check if lookup value matches any stop value
        elif float(lookup) in stops:
            return heights[stops.index(lookup)]
        
        # interpolation required
        else:

            # identify bounding height stop values
            lower = max([stops[0]] + [x for x in stops if x < lookup])
            upper = min([stops[-1]] + [x for x in stops if x > lookup])
            
            # heights from bounding stops
            lower_height = heights[stops.index(lower)]
            upper_height = heights[stops.index(upper)]
            
            # compute linear "relative distance" from lower bound height to upper bound height
            distance = (lookup - lower) / (upper - lower)

            # return string representing rgb height value
            return lower_height + distance * (upper_height - lower_height)

    # default height value catch-all
    return default_height