Generate cover: If the default font cannot render characters in the metadata (for example for east asian languages) try to automatically find a font on the system that is capable of rendering the characters

src/calibre/ebooks/__init__.py

@@ -178,18 +178,41 @@ def normalize(x):
 
 def calibre_cover(title, author_string, series_string=None,
         output_format='jpg', title_size=46, author_size=36, logo_path=None):
+    from calibre.utils.config_base import tweaks
     title = normalize(title)
     author_string = normalize(author_string)
     series_string = normalize(series_string)
     from calibre.utils.magick.draw import create_cover_page, TextLine
-    lines = [TextLine(title, title_size), TextLine(author_string, author_size)]
+    text = title + author_string + (series_string or u'')
+    font_path = tweaks['generate_cover_title_font']
+    if font_path is None:
+        font_path = P('fonts/liberation/LiberationSerif-Bold.ttf')
+
+    from calibre.utils.fonts.utils import get_font_for_text
+    font = open(font_path, 'rb').read()
+    c = get_font_for_text(text, font)
+    cleanup = False
+    if c is not None and c != font:
+        from calibre.ptempfile import PersistentTemporaryFile
+        pt = PersistentTemporaryFile('.ttf')
+        pt.write(c)
+        pt.close()
+        font_path = pt.name
+        cleanup = True
+
+    lines = [TextLine(title, title_size, font_path=font_path),
+            TextLine(author_string, author_size, font_path=font_path)]
     if series_string:
-        lines.append(TextLine(series_string, author_size))
+        lines.append(TextLine(series_string, author_size, font_path=font_path))
     if logo_path is None:
         logo_path = I('library.png')
+    try:
         return create_cover_page(lines, logo_path, output_format='jpg',
             texture_opacity=0.3, texture_data=I('cover_texture.png',
                 data=True))
+    finally:
+        if cleanup:
+            os.remove(font_path)
 
 UNIT_RE = re.compile(r'^(-*[0-9]*[.]?[0-9]*)\s*(%|em|ex|en|px|mm|cm|in|pt|pc)$')
 

src/calibre/utils/fonts/__init__.py

@@ -60,6 +60,52 @@ class Fonts(object):
             ans[ft] = (ext, name, open(f, 'rb').read())
         return ans
 
+    def find_font_for_text(self, text, allowed_families={'serif', 'sans-serif'},
+            preferred_families=('serif', 'sans-serif', 'monospace', 'cursive', 'fantasy')):
+        '''
+        Find a font on the system capable of rendering the given text.
+
+        Returns a font family (as given by fonts_for_family()) that has a
+        "normal" font and that can render the supplied text. If no such font
+        exists, returns None.
+
+        :return: (family name, faces) or None, None
+        '''
+        from calibre.utils.fonts.free_type import FreeType, get_printable_characters, FreeTypeError
+        from calibre.utils.fonts.utils import panose_to_css_generic_family, get_font_characteristics
+        ft = FreeType()
+        found = {}
+        if not isinstance(text, unicode):
+            raise TypeError(u'%r is not unicode'%text)
+        text = get_printable_characters(text)
+
+        def filter_faces(faces):
+            ans = {}
+            for k, v in faces.iteritems():
+                try:
+                    font = ft.load_font(v[2])
+                except FreeTypeError:
+                    continue
+                if font.supports_text(text, has_non_printable_chars=False):
+                    ans[k] = v
+            return ans
+
+        for family in sorted(self.find_font_families()):
+            faces = filter_faces(self.fonts_for_family(family))
+            if 'normal' not in faces:
+                continue
+            panose = get_font_characteristics(faces['normal'][2])[5]
+            generic_family = panose_to_css_generic_family(panose)
+            if generic_family in allowed_families or generic_family == preferred_families[0]:
+                return (family, faces)
+            elif generic_family not in found:
+                found[generic_family] = (family, faces)
+
+        for f in preferred_families:
+            if f in found:
+                return found[f]
+        return None, None
+
 fontconfig = Fonts()
 
 def test():

src/calibre/utils/fonts/free_type.py

@@ -7,7 +7,7 @@ __license__   = 'GPL v3'
 __copyright__ = '2012, Kovid Goyal <kovid at kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
-import threading
+import threading, unicodedata
 from functools import wraps
 from future_builtins import map
 
@@ -20,6 +20,10 @@ class ThreadingViolation(Exception):
                 'You cannot use the MTP driver from a thread other than the '
                 ' thread in which startup() was called')
 
+def get_printable_characters(text):
+    return u''.join(x for x in unicodedata.normalize('NFC', text)
+            if unicodedata.category(x)[0] not in {'C', 'Z', 'M'})
+
 def same_thread(func):
     @wraps(func)
     def check_thread(self, *args, **kwargs):
@@ -28,6 +32,8 @@ def same_thread(func):
         return func(self, *args, **kwargs)
     return check_thread
 
+FreeTypeError = getattr(plugins['freetype'][0], 'FreeTypeError', Exception)
+
 class Face(object):
 
     def __init__(self, face):
@@ -42,9 +48,14 @@ class Face(object):
             setattr(self, x, val)
 
     @same_thread
-    def supports_text(self, text):
+    def supports_text(self, text, has_non_printable_chars=True):
+        '''
+        Returns True if all the characters in text have glyphs in this font.
+        '''
         if not isinstance(text, unicode):
             raise TypeError('%r is not a unicode object'%text)
+        if has_non_printable_chars:
+            text = get_printable_characters(text)
         chars = tuple(frozenset(map(ord, text)))
         return self.face.supports_text(chars)
 
@@ -71,6 +82,17 @@ def test():
     if font.supports_text('abc'):
         raise RuntimeError('Incorrectly claiming that text is supported')
 
+def test_find_font():
+    from calibre.utils.fonts import fontconfig
+    abcd = '诶比西迪'
+    family = fontconfig.find_font_for_text(abcd)[0]
+    print ('Family for Chinese text:', family)
+    family = fontconfig.find_font_for_text(abcd)[0]
+    abcd = 'لوحة المفاتيح العربية'
+    print ('Family for Arabic text:', family)
+
+
 if __name__ == '__main__':
     test()
+    test_find_font()
 

src/calibre/utils/fonts/utils.py

@@ -38,8 +38,8 @@ def get_table(raw, name):
 
 def get_font_characteristics(raw):
     '''
-    Return (weight, is_italic, is_bold, is_regular, fs_type). These values are taken
-    from the OS/2 table of the font. See
+    Return (weight, is_italic, is_bold, is_regular, fs_type, panose). These
+    values are taken from the OS/2 table of the font. See
     http://www.microsoft.com/typography/otspec/os2.htm for details
     '''
     os2_table = get_table(raw, 'os/2')[0]
@@ -54,7 +54,6 @@ def get_font_characteristics(raw):
             family_class) = struct.unpack_from(common_fields, os2_table)
     offset = struct.calcsize(common_fields)
     panose = struct.unpack_from(b'>10B', os2_table, offset)
-    panose
     offset += 10
     (range1,) = struct.unpack_from(b'>L', os2_table, offset)
     offset += struct.calcsize(b'>L')
@@ -69,7 +68,21 @@ def get_font_characteristics(raw):
     is_italic = (selection & 0b1) != 0
     is_bold = (selection & 0b100000) != 0
     is_regular = (selection & 0b1000000) != 0
-    return weight, is_italic, is_bold, is_regular, fs_type
+    return weight, is_italic, is_bold, is_regular, fs_type, panose
+
+def panose_to_css_generic_family(panose):
+    proportion = panose[3]
+    if proportion == 9:
+        return 'monospace'
+    family_type = panose[0]
+    if family_type == 3:
+        return 'cursive'
+    if family_type == 4:
+        return 'fantasy'
+    serif_style = panose[1]
+    if serif_style in (11, 12, 13):
+        return 'sans-serif'
+    return 'serif'
 
 def decode_name_record(recs):
     '''
@@ -225,13 +238,33 @@ def remove_embed_restriction(raw):
     verify_checksums(raw)
     return raw
 
+def get_font_for_text(text, candidate_font_data=None):
+    ok = False
+    if candidate_font_data is not None:
+        from calibre.utils.fonts.free_type import FreeType, FreeTypeError
+        ft = FreeType()
+        try:
+            font = ft.load_font(candidate_font_data)
+            ok = font.supports_text(text)
+        except FreeTypeError:
+            ok = True
+    if not ok:
+        from calibre.utils.fonts import fontconfig
+        family, faces = fontconfig.find_font_for_text(text)
+        if family is not None:
+            f = faces.get('bold', faces['normal'])
+            candidate_font_data = f[2]
+    return candidate_font_data
+
 def test():
     import sys, os
     for f in sys.argv[1:]:
         print (os.path.basename(f))
         raw = open(f, 'rb').read()
         print (get_font_names(raw))
-        print (get_font_characteristics(raw))
+        characs = get_font_characteristics(raw)
+        print (characs)
+        print (panose_to_css_generic_family(characs[5]))
         verify_checksums(raw)
         remove_embed_restriction(raw)
 

src/calibre/utils/magick/draw.py

@@ -10,7 +10,7 @@ import os
 from calibre.utils.magick import Image, DrawingWand, create_canvas
 from calibre.constants import __appname__, __version__
 from calibre.utils.config import tweaks
-from calibre import fit_image
+from calibre import fit_image, force_unicode
 
 def _data_to_image(data):
     if isinstance(data, Image):
@@ -166,11 +166,8 @@ def add_borders_to_image(img_data, left=0, top=0, right=0, bottom=0,
     return canvas.export(fmt)
 
 def create_text_wand(font_size, font_path=None):
-    if font_path is None:
-        font_path = tweaks['generate_cover_title_font']
-        if font_path is None:
-            font_path = P('fonts/liberation/LiberationSerif-Bold.ttf')
     ans = DrawingWand()
+    if font_path is not None:
         ans.font = font_path
     ans.font_size = font_size
     ans.gravity = 'CenterGravity'
@@ -238,6 +235,17 @@ class TextLine(object):
     def __init__(self, text, font_size, bottom_margin=30, font_path=None):
         self.text, self.font_size, = text, font_size
         self.bottom_margin = bottom_margin
+        if font_path is None:
+            if not isinstance(text, unicode):
+                text = force_unicode(text)
+            from calibre.utils.fonts.utils import get_font_for_text
+            fd = get_font_for_text(text)
+            if fd is not None:
+                from calibre.ptempfile import PersistentTemporaryFile
+                pt = PersistentTemporaryFile('.ttf')
+                pt.write(fd)
+                pt.close()
+                font_path = pt.name
         self.font_path = font_path
 
     def __repr__(self):