kerning corrections (#1138)

lib/lettering/kerning.py

@@ -28,26 +28,68 @@ class FontKerning(object):
     def hkern(self):
         xpath = ".//svg:hkern[(@u1 or @g1) and (@u1 or @g1) and @k]/@*[contains(name(), '1') or contains(name(), '2') or name()='k']"
         hkern = self.svg.xpath(xpath, namespaces=NSS)
-        for index, glyph in enumerate(hkern):
+
+        # the kerning list now contains the kerning values as a list where every first value contains the first letter(s),
+        # every second value contains the second letter(s) and every third value contains the kerning
+        u_first = [k for k in hkern[0::3]]
+        u_second = [k for k in hkern[1::3]]
+        k = [int(x) for x in hkern[2::3]]
+
+        # sometimes a font file contains conflicting kerning value for a letter pair
+        # in this case the value which is specified as a single pair overrules the one specified in a list of letters
+        # therefore we want to sort our list by length of the letter values
+        kern_list = list(zip(u_first, u_second, k))
+        kern_list.sort(key=lambda x: len(x[0] + x[1]), reverse=True)
+
+        for index, kerning in enumerate(kern_list):
+            first, second, key = kerning
             # fontTools.agl will import fontTools.misc.py23 which will output a deprecation warning
             # ignore the warning for now - until the library fixed it
             if index == 0:
                 import warnings
                 with warnings.catch_warnings():
                     warnings.simplefilter("ignore")
+                    global toUnicode
                     from fontTools.agl import toUnicode
-            if len(glyph) > 1 and not (index + 1) % 3 == 0:
-                glyph_names = glyph.split(",")
-                # the glyph name is written in various languages, second is english. Let's look it up.
-                if len(glyph_names) == 1:
-                    hkern[index] = toUnicode(glyph)
-                else:
-                    hkern[index] = toUnicode(glyph_names[1])
-        k = [int(x) for x in hkern[2::3]]
-        u = [k + v for k, v in zip(hkern[0::3], hkern[1::3])]
-        hkern = dict(zip(u, k))
+            first = self.split_glyph_list(first)
+            second = self.split_glyph_list(second)
+            kern_list[index] = (first, second, key)
+
+        hkern = {}
+        for first, second, key in kern_list:
+            for f in first:
+                for s in second:
+                    hkern[f+s] = key
         return hkern
 
+    def split_glyph_list(self, glyph):
+        glyphs = []
+        if len(glyph) > 1:
+            # glyph names need to be converted to unicode
+            # we need to take into account, that there can be more than one first/second letter in the very same hkern element
+            # in this case they will be commas separated and each first letter needs to be combined with each next letter
+            # e.g. <hkern g1="A,Agrave,Aacute,Acircumflex,Atilde,Adieresis,Amacron,Abreve,Aogonek" g2="T,Tcaron" k="5" />
+            glyph_names = glyph.split(",")
+            for glyph_name in glyph_names:
+                # each glyph can have additional special markers, e.g. o.cmp
+                # toUnicode will not respect those and convert them to a simple letter
+                # this behaviour will generate a wrong spacing for this letter.
+                # Let's make sure to also transfer the separators and extensions to our json file
+                separators = [".", "_"]
+                used_separator = False
+                for separator in separators:
+                    glyph_with_separator = glyph_name.split(separator)
+                    if len(glyph_with_separator) == 2:
+                        glyphs.append("%s%s%s" % (toUnicode(glyph_with_separator[0]), separator, glyph_with_separator[1]))
+                        used_separator = True
+                        continue
+                # there is no extra separator
+                if not used_separator:
+                    glyphs.append(toUnicode(glyph_name))
+        else:
+            glyphs.append(glyph)
+        return glyphs
+
     # the space character
     def word_spacing(self):
         xpath = "string(.//svg:glyph[@glyph-name='space'][1]/@*[name()='horiz-adv-x'])"