CET_to_py.py

"""
Generate Python versions for each of the colormaps provided in
http://peterkovesi.com/projects/colourmaps/CETperceptual_csv_0_1.zip

Also adds Glasbey colormaps created using: https://github.com/taketwo/glasbey
see https://github.com/pyviz/colorcet/issues/11 for more details
"""

import csv
import re
from pathlib import Path

csv_folders = ['CET', 'Glasbey']
output_file = '../colorcet/__init__.py'
header = '''\
"""
Python versions of the 256-color colormaps provided in
http://peterkovesi.com/projects/colourmaps/CETperceptual_csv_0_1.zip

Each of these colormaps can be accessed as a Bokeh palette or
Matplotlib colormap, either by string name:

  palette['name']
  cm['name']

or as Python attributes:

  palette.name
  cm.name

or as individually importable Python attributes:

  m_name
  b_name

All colormaps are named using Peter Kovesi\'s naming scheme:

<category>_<huesequence>_<lightnessrange>_c<meanchroma>[_s<colorshift>_[r<ifreversed>]]

but some have shorter, more convenient aliases, some of which are
named for the color ranges included and others
based on the qualitative appearance.  The colormaps with
shorter names tend to be the most useful subset, and for
cases like automatic population of a GUI widget these
colormaps are provided as a separate subset:

  palette_n['name'] or palette_n.name
  cm_n['name'] or cm_n.name

Also included are some sets of 256 Glasbey colors. These are available via the
same methods described above and are named:

  glasbey_<starting_palette>[_<min|max>c_<chroma_value>][_<min|max>l_<lightness_value>][_hue_<start>_<end>]

Some of the Glasbey sets are aliased to short names as explained in the User Guide.
"""

from colorcet.version import Version
__version__ = str(Version(fpath=__file__, archive_commit="$Format:%h$",reponame="datashader"))

from collections import OrderedDict
from itertools import chain


class AttrODict(OrderedDict):
    """Ordered dictionary with attribute access (e.g. for tab completion)"""
    def __dir__(self): return self.keys()
    def __delattr__(self, name): del self[name]
    def __getattr__(self, name):
        return self[name] if not name.startswith('_') else super(AttrODict, self).__getattr__(name)
    def __setattr__(self, name, value):
        if (name.startswith('_')): return super(AttrODict, self).__setattr__(name, value)
        self[name] = value


try:
    from matplotlib.colors import LinearSegmentedColormap, ListedColormap
    from matplotlib.cm import register_cmap
except:
    def LinearSegmentedColormap(colorlist,name): pass
    def ListedColormap(colorlist,name): pass
    def register_cmap(name,cmap): pass
    LinearSegmentedColormap.from_list=lambda n,c,N: None


def rgb_to_hex(r,g,b):
    return '#%02x%02x%02x' % (r,g,b)


def bokeh_palette(name,colorlist):
    palette[name] = [rgb_to_hex(int(r*255),int(g*255),int(b*255)) for r,g,b in colorlist]
    return palette[name]


def mpl_cm(name,colorlist):
    cm[name]      = LinearSegmentedColormap.from_list(name, colorlist, N=len(colorlist))
    register_cmap("cet_"+name, cmap=cm[name])
    return cm[name]


def mpl_cl(name,colorlist):
    cm[name]      = ListedColormap(colorlist, name)
    register_cmap("cet_"+name, cmap=cm[name])
    return cm[name]


def get_aliases(name):
    """Get the aliases for a given colormap name"""
    names = [name]

    def check_aliases(names, d,  k_position=-1, v_position=0):
        for name in [n for n in names]:
            for k, v in d.items():
                v = [v] if not isinstance(v, list) else v
                for vname in v:
                    if name == vname and k not in names:
                        if k_position == -2:
                            names.append(k)
                        else:
                            names.insert(k_position, k)
                    if name == k and vname not in names:
                        if v_position == -2:
                            names.append(vname)
                        else:
                            names.insert(v_position, vname)
        return names

    # Repeatedly look for new aliases until no new aliases are found
    n_names = len(names)
    while True:
        names = check_aliases(names, aliases, k_position=-2, v_position=0)
        names = check_aliases(names, cetnames_flipped, k_position=-2, v_position=-1)
        if len(names) == n_names:
            break
        n_names = len(names)

    # Sort names as 1or0_underscores, CET, multiple_under_scores (alias, cetname, algorithmicname)
    def name_sortfn(name):
        if name.count("_") > 1:
            return 2
        if "CET" in name:
            return 1
        return 0

    return ',  '.join(sorted(names, key=name_sortfn))


def all_original_names(group=None, not_group=None, only_aliased=False, only_CET=False):
    """
    Returns a list (optionally filtered) of the names of the available colormaps
    Filters available:
    - group: only include maps whose name include the given string(s)
      (e.g. "'linear'" or "['linear','diverging']"). 
    - not_group: filter out any maps whose names include the given string(s)
    - only_aliased: only include maps with shorter/simpler aliases
    - only_CET: only include maps from CET
    """
    names = palette.keys()
    if group:
        groups = group if isinstance(group, list) else [group]
        names = [n for ns in [list(filter(lambda x: g in x, names)) for g in groups] for n in ns]
    if not_group:
        not_groups = not_group if isinstance(not_group, list) else [not_group]
        for g in not_groups:
            names = list(filter(lambda x: g not in x, names))
    if only_aliased:
        names = filter(lambda x: x in aliases.keys(), names)
    else:
        names = filter(lambda x: x not in chain.from_iterable(aliases.values()), names)
    if only_CET:
        names = filter(lambda x: x in cetnames_flipped.values(), names)
    else:
        names = filter(lambda x: x not in cetnames_flipped.values(), names)
    return sorted(list(names))


palette = AttrODict()
cm = AttrODict()
palette_n = AttrODict()
cm_n = AttrODict()

'''

footer = """
palette_n = AttrODict(sorted(palette_n.items()))
cm_n = AttrODict(sorted(cm_n.items()))
"""

# Here #mpl indicates a colormap name taken from Matplotlib
aliases = dict(
    circle_mgbm_67_c31                              = ['cyclic_isoluminant'],
    cyclic_mygbm_30_95_c78_s25                      = ['colorwheel'],
    diverging_bkr_55_10_c35                         = ['bkr'],
    diverging_bky_60_10_c30                         = ['bky'],
    diverging_bwr_40_95_c42                         = ['coolwarm'],  #mpl
    diverging_gwv_55_95_c39                         = ['gwv'],
    diverging_linear_bjy_30_90_c45                  = ['bjy'],
    diverging_protanopic_deuteranopic_bwy_60_95_c32 = ['bwy'],
    diverging_tritanopic_cwr_75_98_c20              = ['cwr'],
    glasbey_bw_minc_20                              = ['glasbey'],
    glasbey_bw_minc_20_hue_150_280                  = ['glasbey_cool'],
    glasbey_bw_minc_20_hue_330_100                  = ['glasbey_warm'],
    glasbey_bw_minc_20_maxl_70                      = ['glasbey_dark'],
    glasbey_bw_minc_20_minl_30                      = ['glasbey_light'],
    isoluminant_cgo_80_c38                          = ['isolum'],
    linear_bgy_10_95_c74                            = ['bgy'],
    linear_bgyw_15_100_c68                          = ['bgyw'],
    linear_blue_95_50_c20                           = ['blues'],  #mpl
    linear_bmw_5_95_c89                             = ['bmw'],
    linear_bmy_10_95_c78                            = ['bmy'],
    linear_grey_0_100_c0                            = ['gray'],  #mpl
    linear_grey_10_95_c0                            = ['dimgray'],
    linear_kbc_5_95_c73                             = ['kbc', 'linear_blue_5_95_c73'],
    linear_kbgoy_20_95_c57                          = ['gouldian'],
    linear_kbgyw_10_98_c63                          = ['kbgyw'],
    linear_kgy_5_95_c69                             = ['kgy', 'linear_green_5_95_c69'],
    linear_kryw_0_100_c71                           = ['fire'],
    linear_ternary_blue_0_44_c57                    = ['kb'],
    linear_ternary_green_0_46_c42                   = ['kg'],
    linear_ternary_red_0_50_c52                     = ['kr'],
    rainbow_bgyr_10_90_c83                          = ['rainbow4'],
    rainbow_bgyr_35_85_c73                          = ['rainbow'],
)

cetnames = {
    'CET-C1': 'cyclic_mrybm_35-75_c68',
    'CET-C1s': 'cyclic_mrybm_35-75_c68_s25',
    'CET-C2': 'cyclic_mygbm_30-95_c78',
    'CET-C2s': 'cyclic_mygbm_30-95_c78_s25',
    'CET-C3': 'cyclic_wrkbw_10_90_c43',
    'CET-C3s': 'cyclic_wrkbw_10_90_c43_s25',
    'CET-C4': 'cyclic_wrwbw_40-90_c42',
    'CET-C4s': 'cyclic_wrwbw_40-90_c42_s25',
    'CET-C5': 'cyclic_grey_15-85_c0',
    'CET-C5s': 'cyclic_grey_15-85_c0_s25',
    'CET-C6': 'cyclic_rygcbmr_50_90_c64',
    'CET-C6s': 'cyclic_rygcbmr_50_90_c64_s25',
    'CET-C7': 'cyclic_ymcgy_60_90_c67',
    'CET-C7s': 'cyclic_ymcgy_60_90_c67_s25',
    'CET-C8': 'cyclic_mygbm_50_90_c46',
    'CET-C8s': 'cyclic_mygbm_50_90_c46_s25',
    'CET-C9': 'cyclic_mybm_20_100_c48',
    'CET-C9s': 'cyclic_mybm_20_100_c48_s25',
    'CET-C10': 'circle_mgbm_67_c31',
    'CET-C10s': 'circle_mgbm_67_c31_s25',
    'CET-C11': 'cyclic_bgrmb_35_70_c75',
    'CET-C11s': 'cyclic_bgrmb_35_70_c75_s25',
    'CET-CBC1': 'cyclic-protanopic-deuteranopic_bwyk_16-96_c31',
    'CET-CBC2': 'cyclic-protanopic-deuteranopic_wywb_55-96_c33',
    'CET-CBD1': 'diverging-protanopic-deuteranopic_bwy_60-95_c32',
    'CET-CBD2': 'diverging_linear_protanopic_deuteranopic_bjy_57_89_c34',
    'CET-CBL1': 'linear-protanopic-deuteranopic_kbjyw_5-95_c25',
    'CET-CBL2': 'linear-protanopic-deuteranopic_kbw_5-98_c40',
    'CET-CBL3': 'linear_protanopic_deuteranopic_kbw_5_95_c34',
    'CET-CBL4': 'linear_protanopic_deuteranopic_kyw_5_95_c49',
    'CET-CBTC1': 'cyclic-tritanopic_cwrk_40-100_c20',
    'CET-CBTC2': 'cyclic-tritanopic_wrwc_70-100_c20',
    'CET-CBTD1': 'diverging-tritanopic_cwr_75-98_c20',
    'CET-CBTL1': 'linear-tritanopic_krjcw_5-98_c46',
    'CET-CBTL2': 'linear-tritanopic_krjcw_5-95_c24',
    'CET-CBTL3': 'linear_tritanopic_kcw_5_95_c22',
    'CET-CBTL4': 'linear_tritanopic_krw_5_95_c46',
    'CET-D1': 'diverging_bwr_40-95_c42',
    'CET-D1A': 'diverging_bwr_20-95_c54',
    'CET-D2': 'diverging_gwv_55-95_c39',
    'CET-D3': 'diverging_gwr_55-95_c38',
    'CET-D4': 'diverging_bkr_55-10_c35',
    'CET-D6': 'diverging_bky_60-10_c30',
    'CET-D7': 'diverging-linear_bjy_30-90_c45',
    'CET-D8': 'diverging-linear_bjr_30-55_c53',
    'CET-D9': 'diverging_bwr_55-98_c37',
    'CET-D10': 'diverging_cwm_80-100_c22',
    'CET-D11': 'diverging-isoluminant_cjo_70_c25',
    'CET-D12': 'diverging-isoluminant_cjm_75_c23',
    'CET-D13': 'diverging_bwg_20-95_c41',
    'CET-I1': 'isoluminant_cgo_70_c39',
    'CET-I2': 'isoluminant_cgo_80_c38',
    'CET-I3': 'isoluminant_cm_70_c39',
    'CET-L1': 'linear_grey_0-100_c0',
    'CET-L2': 'linear_grey_10-95_c0',
    'CET-L3': 'linear_kryw_0-100_c71',
    'CET-L4': 'linear_kry_0-97_c73',
    'CET-L5': 'linear_kgy_5-95_c69',
    'CET-L6': 'linear_kbc_5-95_c73',
    'CET-L7': 'linear_bmw_5-95_c86',
    'CET-L8': 'linear_bmy_10-95_c71',
    'CET-L9': 'linear_bgyw_20-98_c66',
    'CET-L10': 'linear_gow_60-85_c27',
    'CET-L11': 'linear_gow_65-90_c35',
    'CET-L12': 'linear_blue_95-50_c20',
    'CET-L13': 'linear_ternary-red_0-50_c52',
    'CET-L14': 'linear_ternary-green_0-46_c42',
    'CET-L15': 'linear_ternary-blue_0-44_c57',
    'CET-L16': 'linear_kbgyw_5-98_c62',
    'CET-L17': 'linear_worb_100-25_c53',
    'CET-L18': 'linear_wyor_100-45_c55',
    'CET-L19': 'linear_wcmr_100-45_c42',
    'CET-L20': 'linear_kbgoy_20_95_c57',
    'CET-R1': 'rainbow_bgyrm_35-85_c69',
    'CET-R2': 'rainbow_bgyr_35-85_c72',
    'CET-R3': 'diverging-rainbow_bgymr_45-85_c67',
    'CET-R4': 'rainbow_bgyr_10_90_c83',
}

cetnames_flipped = {v.replace('-', '_'): k.replace('-', '_') for
                     k, v in cetnames.items()}

def create_alias(alias, base, output, cmtype='mpl_cm', is_name=True):
    output.write("{0} = b_{1}\n".format(alias,base))
    output.write("m_{0} = m_{1}\n".format(alias,base))
    output.write("m_{0}_r = m_{1}_r\n".format(alias,base))
    output.write("palette['{0}'] = b_{1}\n".format(alias,base))
    if is_name:
        output.write("palette_n['{0}'] = b_{1}\n".format(alias,base))
    output.write("cm['{0}'] = m_{1}\n".format(alias,base))
    output.write("cm['{0}_r'] = m_{1}_r\n".format(alias,base))
    if is_name:
        output.write("cm_n['{0}'] = {2}('{0}',{1})\n".format(alias,base,cmtype))
        output.write("cm_n['{0}_r'] = {2}('{0}_r',list(reversed({1})))\n".format(alias,base,cmtype))
    else:
        output.write("register_cmap('cet_{0}',m_{1})\n".format(alias,base))
        output.write("register_cmap('cet_{0}_r',m_{1}_r)\n".format(alias,base))


def get_csvs_in_order(output_file, csv_folders):
    """Get the CSV files to write to the __init__.py, keeping the order found in __init__.py"""

    # get order of existing maps in __init__.py
    init_cmap_order = []
    with open(output_file) as f:
        while line := f.readline():
            if match := re.match(r"(\w+) = \[  # cmap_def", line):
                init_cmap_order.append(match.groups()[0])
    new_order_i = len(init_cmap_order)  # index of next new map after those in int_map_order

    # get all csvs
    csv_paths = []
    for fld in csv_folders:
        csv_paths += list(Path(fld).glob("*.csv"))
    csv_order = [-1]*len(csv_paths)

    # Get a new ordering of the csvs from init_cmap_order
    for path_i, csv_path in enumerate(csv_paths):
        base = csv_path.stem.replace("-","_").replace("_n256","")
        try:
            csv_order[path_i] = init_cmap_order.index(base)
        except ValueError:
            # new csv not in the original order, so put it at the end
            csv_order[path_i] = new_order_i
            new_order_i += 1

    # Put the csv paths in the new order
    csv_paths_new = [Path()]*len(csv_paths)
    for path_i, order_i in enumerate(csv_order):
        csv_paths_new[order_i] = csv_paths[path_i]

    return csv_paths_new


def format_dict(name, d, tabs=0):
    t4 = " "*4
    tabs = t4*tabs
    s = tabs + "{} = {{\n".format(name)
    for k, v in d.items():
        v = "'{}'".format(v) if isinstance(v, str) else v
        s += tabs + t4 + "'{}': {},\n".format(k, v)
    s += tabs + "}\n"
    return s


def gen_init_py(output_file, csv_folders):
    csv_paths = get_csvs_in_order(output_file, csv_folders)

    cmaps = []
    with open(output_file, "w") as output:
        output.write(header)
        output.write(format_dict("aliases", aliases))
        output.write(format_dict("cetnames_flipped", cetnames_flipped))
        for csv_path in csv_paths:
            categorical = ('Glasbey' in [str(p) for p in csv_path.parents])
            cmtype = "mpl_cl" if categorical else 'mpl_cm'
            if csv_path.suffix == ".csv":
                base = csv_path.stem.replace("-","_").replace("_n256","")
                if base in cmaps:
                    continue
                output.write("\n\n")
                output.write("{0} = [  # cmap_def\n".format(base))
                with open(csv_path, 'r') as csvfile:
                    reader = csv.reader(csvfile)
                    for row in reader:
                        output.write("[{0}],\n".format(", ".join(row)))
                output.write("]\n")
                output.write("b_{0} = bokeh_palette('{0}',{0})\n".format(base))
                output.write("m_{0} = {1}('{0}',{0})\n".format(base, cmtype))
                output.write("m_{0}_r = {1}('{0}_r',list(reversed({0})))\n".format(base, cmtype))
                if base in aliases:
                    for alias in aliases[base]:
                        create_alias(alias, base, output, cmtype, is_name=True)
                if base in cetnames_flipped:
                    alias = cetnames_flipped[base]
                    create_alias(alias, base, output, cmtype, is_name=False)
                output.write("\n\n")
                cmaps.append(base)
        output.write(footer)


if __name__ == "__main__":
    gen_init_py(output_file, csv_folders)