class FontMetric(object):
def __init__(self, metric_file):
self.metric = AFM(open(metric_file))
@staticmethod
def _map_to_adobe(char):
c = ord(char)
if c in UNICODE_TO_ADOBE:
return UNICODE_TO_ADOBE[c]
return c
def get_width(self, char, size):
# sys.stderr.write(char)
# sys.stderr.write('\n')
c = self._map_to_adobe(char)
return size * (self.metric.get_width_char(c, isord=True) / 1000.0)
def get_kern_distance(self, c1, c2, size):
c1 = self._map_to_adobe(c1)
c2 = self._map_to_adobe(c2)
return size * (self.metric.get_kern_dist(c1, c2, isord=True) / 1000.0)
# https://hdbscan.readthedocs.io/en/latest/index.html # The method may be useful for non uniform samples, allowing automated measurement # of adhesion and modulus for different phases across the sample surface. # Import AFM class from afm import AFM # Import packages for clustering import hdbscan import numpy as np import matplotlib.pyplot as plt import seaborn as sns import sklearn.datasets as data # Run AFM functions expt_1 = AFM() expt_1.run() # Shape adhesion and modulus data _adhesion = expt_1.adhesion _modulus = expt_1.modulus _adhesion = _adhesion.reshape(-1, 1) _modulus = _modulus.reshape(-1, 1) data = np.append(_adhesion, _modulus, axis=1) # Find clusters clusterer = hdbscan.HDBSCAN(allow_single_cluster=True, min_cluster_size=10) clusterer.fit(data) # Plot condensed tree clusterer.condensed_tree_.plot(select_clusters=True,
def __init__(self, metric_file):
self.metric = AFM(open(metric_file))