Lettering typographic features (#3466)

* add svg font to layers extension which saves glyph annotations into the glyph name
---------
Co-authored-by: Claudine

4 files changed, 187 insertions, 31 deletions

diff --git a/lib/lettering/font.py b/lib/lettering/font.py
index 0c02de84..a554183e 100644
--- a/lib/lettering/font.py
+++ b/lib/lettering/font.py
@@ -324,40 +324,69 @@ class Font(object):
 
         group = inkex.Group()
         group.label = line
+        if self.text_direction == 'rtl':
+            group.label = line[::-1]
         group.set("inkstitch:letter-group", "line")
         last_character = None
 
         words = line.split(" ")
         for word in words:
+
             word_group = inkex.Group()
-            word_group.label = word
+            label = word
+            if self.text_direction == 'rtl':
+                label = word[::-1]
+            word_group.label = label
             word_group.set("inkstitch:letter-group", "word")
 
-            for character in word:
-                if self.letter_case == "upper":
-                    character = character.upper()
-                elif self.letter_case == "lower":
-                    character = character.lower()
-
-                glyph = glyph_set[character]
+            if self.text_direction == 'rtl':
+                glyphs = self._get_word_glyphs(glyph_set, word[::-1])
+                glyphs = glyphs[::-1]
+            else:
+                glyphs = self._get_word_glyphs(glyph_set, word)
 
-                if glyph is None and self.default_glyph == " ":
+            last_character = None
+            for glyph in glyphs:
+                if glyph is None:
                     position.x += self.word_spacing
                     last_character = None
-                else:
-                    if glyph is None:
-                        glyph = glyph_set[self.default_glyph]
+                    continue
+                node = self._render_glyph(destination_group, glyph, position, glyph.name, last_character)
+                word_group.append(node)
+                last_character = glyph.name
+            group.append(word_group)
+            position.x += self.word_spacing
+        return group
 
-                    if glyph is not None:
-                        node = self._render_glyph(destination_group, glyph, position, character, last_character)
-                        word_group.append(node)
+    def _get_word_glyphs(self, glyph_set, word):
+        glyphs = []
+        skip = []
+        previous_is_binding = True
 
-                    last_character = character
-            position.x += self.word_spacing
-            last_character = None
-            group.append(word_group)
+        for i, character in enumerate(word):
+            if i in skip:
+                continue
 
-        return group
+            # forced letter case
+            if self.letter_case == "upper":
+                character = character.upper()
+            elif self.letter_case == "lower":
+                character = character.lower()
+
+            glyph, glyph_len, binding = glyph_set.get_next_glyph(word, i, previous_is_binding)
+            previous_is_binding = binding
+
+            skip = list(range(i, i+glyph_len))
+
+            if glyph is None and self.default_glyph == " ":
+                glyphs.append(None)
+            else:
+                if glyph is None:
+                    glyphs.append(glyph_set[self.default_glyph])
+                if glyph is not None:
+                    glyphs.append(glyph)
+
+        return glyphs
 
     def _render_glyph(self, destination_group, glyph, position, character, last_character):
         """Render a single glyph.
@@ -383,9 +412,17 @@ class Font(object):
         node = deepcopy(glyph.node)
         if last_character is not None:
             if self.text_direction != "rtl":
-                position.x += glyph.min_x - self.kerning_pairs.get(last_character + character, 0)
+                kerning = self.kerning_pairs.get(f'{last_character} {character}', None)
+                if kerning is None:
+                    # legacy kerning without space
+                    kerning = self.kerning_pairs.get(last_character + character, 0)
+                position.x += glyph.min_x - kerning
             else:
-                position.x += glyph.min_x - self.kerning_pairs.get(character + last_character, 0)
+                kerning = self.kerning_pairs.get(f'{character} {last_character}', None)
+                if kerning is None:
+                    # legacy kerning without space
+                    kerning = self.kerning_pairs.get(character + last_character, 0)
+                position.x += glyph.min_x - kerning
 
         transform = "translate(%s, %s)" % position.as_tuple()
         node.set('transform', transform)
diff --git a/lib/lettering/font_info.py b/lib/lettering/font_info.py
index cb47feea..74cc0967 100644
--- a/lib/lettering/font_info.py
+++ b/lib/lettering/font_info.py
@@ -3,6 +3,8 @@
 # Copyright (c) 2010 Authors
 # Licensed under the GNU GPL version 3.0 or later.  See the file LICENSE for details.
 
+from collections import defaultdict
+
 from fontTools.agl import toUnicode
 from inkex import NSS
 from lxml import etree
@@ -21,11 +23,35 @@ class FontFileInfo(object):
     # horiz_adv_x defines the width of specific letters (distance to next letter)
     def horiz_adv_x(self):
         # In XPath 2.0 we could use ".//svg:glyph/(@unicode|@horiz-adv-x)"
-        xpath = ".//svg:glyph[@unicode and @horiz-adv-x]/@*[name()='unicode' or name()='horiz-adv-x']"
-        hax = self.svg.xpath(xpath, namespaces=NSS)
-        if len(hax) == 0:
+        xpath = ".//svg:glyph"  # [@unicode and @horiz-adv-x and @glyph-name]/@*[name()='unicode' or name()='horiz-adv-x' or name()='glyph-name']"
+        glyph_definitions = self.svg.xpath(xpath, namespaces=NSS)
+        if len(glyph_definitions) == 0:
             return {}
-        return dict(zip(hax[0::2], [float(x) for x in hax[1::2]]))
+
+        horiz_adv_x_dict = defaultdict(list)
+        for glyph in glyph_definitions:
+            unicode_char = glyph.get('unicode', None)
+            if unicode_char is None:
+                continue
+            hax = glyph.get('horiz-adv-x', None)
+            if hax is None:
+                continue
+            else:
+                hax = float(hax)
+
+            glyph_name = glyph.get('glyph-name', None)
+            if glyph_name is not None:
+                glyph_name = glyph_name.split('.')
+                if len(glyph_name) == 2:
+                    typographic_feature = glyph_name[1]
+                    unicode_char += f'.{typographic_feature}'
+                else:
+                    arabic_form = glyph.get('arabic-form', None)
+                    if arabic_form is not None and len(arabic_form) > 4:
+                        typographic_feature = arabic_form[:4]
+                        unicode_char += f'.{typographic_feature}'
+            horiz_adv_x_dict[unicode_char] = hax
+        return horiz_adv_x_dict
 
     # kerning (specific distances of two specified letters)
     def hkern(self):
@@ -54,7 +80,7 @@ class FontFileInfo(object):
         for first, second, key in kern_list:
             for f in first:
                 for s in second:
-                    hkern[f+s] = key
+                    hkern[f'{f} {s}'] = key
         return hkern
 
     def split_glyph_list(self, glyph):
@@ -62,7 +88,7 @@ class FontFileInfo(object):
         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
+            # in this case they will be comma 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:
@@ -73,11 +99,12 @@ class FontFileInfo(object):
                 separators = [".", "_"]
                 used_separator = False
                 for separator in separators:
+                    if used_separator:
+                        continue
                     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]))
+                        glyphs.append(f"{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))
diff --git a/lib/lettering/font_variant.py b/lib/lettering/font_variant.py
index baf7c09f..5bedaa87 100644
--- a/lib/lettering/font_variant.py
+++ b/lib/lettering/font_variant.py
@@ -5,6 +5,7 @@
 
 import os
 from collections import defaultdict
+from unicodedata import normalize
 
 import inkex
 
@@ -73,7 +74,7 @@ class FontVariant(object):
             for layer in glyph_layers:
                 self._clean_group(layer)
                 layer.attrib[INKSCAPE_LABEL] = layer.attrib[INKSCAPE_LABEL].replace("GlyphLayer-", "", 1)
-                glyph_name = layer.attrib[INKSCAPE_LABEL]
+                glyph_name = normalize('NFKC', layer.attrib[INKSCAPE_LABEL])
                 try:
                     self.glyphs[glyph_name] = Glyph(layer)
                 except (AttributeError, ValueError):
@@ -134,6 +135,93 @@ class FontVariant(object):
 
         return svg
 
+    def glyphs_start_with(self, character):
+        glyph_selection = [glyph_name for glyph_name, glyph_layer in self.glyphs.items() if glyph_name.startswith(character)]
+        return sorted(glyph_selection, key=lambda glyph: (len(glyph.split('.')[0]), len(glyph)), reverse=True)
+
+    def isbinding(self, character):
+        # after  a non binding letter a letter can only be in isol or fina shape.
+        # binding glyph only have  two shapes, isol and fina
+        non_binding_char = ['ا', 'أ', 'ﺇ', 'آ', 'د', 'ذ', 'ر', 'ز', 'و']
+        normalized_non_binding_char = [normalize('NFKC', letter) for letter in non_binding_char]
+        return not (character in normalized_non_binding_char)
+
+    def ispunctuation(self, character):
+        # punctuation sign  are  not considered as part of the word. They onnly have one shape
+        punctuation_signs = ['؟', '،', '.', ',', ';', '.', '!', ':', '؛']
+        normalized_punctuation_signs = [normalize('NFKC', letter) for letter in punctuation_signs]
+        return (character in normalized_punctuation_signs)
+
+    def get_glyph(self, character, word):
+        """
+        Returns the glyph for the given character, searching for combined glyphs first
+        This expects glyph annotations to be within the given word, for example: a.init
+
+        Returns glyph node and length of the glyph name
+        """
+        glyph_selection = self.glyphs_start_with(character)
+        for glyph in glyph_selection:
+            if word.startswith(glyph):
+                return self.glyphs[glyph], len(glyph)
+        return self.glyphs.get(self.default_glyph, None), 1
+
+    def get_next_glyph_shape(self, word, starting, ending, previous_is_binding):
+        # in arabic each letter (or ligature) may have up to 4 different shapes, hence 4 glyphs
+        # this computes the shape of the glyph that represents word[starting:ending+1]
+
+        # punctuation is not really part of the word
+        # they may appear at begining or end of words
+        # computes where the actual word begins and ends up
+        last_char_index = len(word)-1
+        first_char_index = 0
+
+        while self.ispunctuation(word[last_char_index]):
+            last_char_index = last_char_index - 1
+        while self.ispunctuation(word[first_char_index]):
+            first_char_index = first_char_index + 1
+
+        # first glyph is eithher isol or init depending wether it is also the last glyph of the actual word
+        if starting == first_char_index:
+            if not self.isbinding(word[ending]) or len(word) == 1:
+                shape = 'isol'
+            else:
+                shape = 'init'
+        # last glyph  is  final if previous is binding, isol otherwise
+        # a non binding  glyph behaves like the last glyph
+        elif ending == last_char_index or not self.isbinding(word[ending]):
+            if previous_is_binding:
+                shape = 'fina'
+            else:
+                shape = 'isol'
+        # in the middle of the actual word, the shape of a glyph is medi if previous glyph is bendinng,  init otherwise
+        elif previous_is_binding:
+            shape = 'medi'
+        else:
+            shape = 'init'
+
+        return shape
+
+    def get_next_glyph(self, word, i, previous_is_binding):
+        # search for the glyph of word that starts at i,taking  into acount the previous glyph binding status
+
+        # find  all the glyphs in tthe font that start with first letter of the glyph
+        glyph_selection = self.glyphs_start_with(word[i])
+
+        # find the longest glyph that match
+        for glyph in glyph_selection:
+            glyph_name = glyph.split('.')
+            if len(glyph_name) == 2 and glyph_name[1] in ['isol', 'init', 'medi', 'fina']:
+                is_binding = self.isbinding(glyph_name[0][-1])
+                if len(word) < i + len(glyph_name[0]):
+                    continue
+                shape = self.get_next_glyph_shape(word, i, i + len(glyph_name[0]) - 1, previous_is_binding)
+                if glyph_name[1] == shape and word[i:].startswith(glyph_name[0]):
+                    return self.glyphs[glyph], len(glyph_name[0]),  is_binding
+            elif word[i:].startswith(glyph):
+                return self.glyphs[glyph], len(glyph), True
+        # nothing was found
+        return self.glyphs.get(self.default_glyph, None), 1, True
+
     def __getitem__(self, character):
         if character in self.glyphs:
             return self.glyphs[character]
diff --git a/lib/lettering/glyph.py b/lib/lettering/glyph.py
index a3a9df71..fd1e1985 100644
--- a/lib/lettering/glyph.py
+++ b/lib/lettering/glyph.py
@@ -5,6 +5,7 @@
 
 from collections import defaultdict
 from copy import copy
+from unicodedata import normalize
 
 from inkex import paths, transforms, units
 
@@ -36,6 +37,9 @@ class Glyph(object):
             this Glyph.  Nested groups are allowed.
         """
 
+        self.name = group.label
+        if len(self.name) > 11:
+            self.name = normalize('NFKC', self.name[11:])
         self._process_baseline(group.getroottree().getroot())
         self.clips = self._process_clips(group)
         self.node = self._process_group(group)