Use embedded font with Identity-H encoding and write to other document or image

[terminalh2t3]

Hi all.

Working on a printing project so I need to make some measurements on fonts that are used in the design.
We need to use that embedded font subset write all the supported unique characters to the black and white image and do some processing to measure the font thickness.
But, I work perfectly with WinAnsiEncoding but not working for Identity-H encoding

Here is all fonts we have in the sample PDF doc:

all_fonts = page.get_fonts()
print(all_fonts)

[(14, 'ttf', 'Type0', 'IOXPLM+Arial-BoldMT', 'C2_0', 'Identity-H'),
(5, 'cff', 'Type1', 'IOXPLM+MyriadPro-BoldCond', 'T1_0', 'WinAnsiEncoding'),
(6, 'cff', 'Type1', 'IOXPLM+UTMBryantLGBold', 'T1_1', 'WinAnsiEncoding'),
...

With 'WinAnsiEncoding' font, there is no problem using that and display using ImageFont or fitz.Font.

# This example I am using ImageFont from PIL to draw the text on the image
font_info = all_fonts[1]        #  (5, 'cff', 'Type1', 'IOXPLM+MyriadPro-BoldCond', 'T1_0', 'WinAnsiEncoding')
xref_id = font_info[0]
buffer = doc.extract_font(xref_id)

font = ImageFont.truetype(BytesIO(buffer), 100)
text = "MPm"      # assume that those characters are embedded in font. I can verify using `get_text('dict') function`

# draw image
bbox = font.getbbox(text)
text_width = bbox[2] - bbox[0]
img = Image.new('L', (text_width, int(100 * 1.3)), color='white')
draw = ImageDraw.Draw(img)
draw.text((0, 0), text, font=font, fill='black')

img.convert('1')
display(img)

But if apply for Identity-H font, it is not working. font_info = all_fonts[0] => (14, 'ttf', 'Type0', 'IOXPLM+Arial-BoldMT', 'C2_0', 'Identity-H').

Checklist:

• Checked all characters used in PDF file, for example: "0135", "not", etc.
• Extracted cmap of this font:

cmap_xref = doc.xref_get_key(14, "ToUnicode")
cmap = doc.xref_stream(cmap_xref).decode()
# >>> '/CIDInit [/ProcSet](https://file+.vscode-resource.vscode-cdn.net/ProcSet) findresource begin\n12 dict begin\nbegincmap\n/CIDSystemInfo\n<< [/Registry](https://file+.vscode-resource.vscode-cdn.net/Registry) (Adobe)\n/Order

def parse_cmap(cmap_stream):
    cmap = {}
    bfchar_re = re.compile(r"<([0-9A-Fa-f]+)>\s*<([0-9A-Fa-f]+)>")
    for match in bfchar_re.findall(cmap_stream):
        cid_hex, unicode_hex = match
        cid = int(cid_hex, 16)
        unicode_char = chr(int(unicode_hex, 16))
        cmap[cid] = unicode_char
    return cmap

>>> {0: '\uffff',
 3: ' ',
 264: 'Đ',
 19: '0',
 20: '1',
 22: '3',
 24: '5',
 25: '6',
 27: '8',
 29: ':',
 75: 'h',
 76: 'i',
 81: 'n',
 82: 'o',
 87: 't',
 1189: 'ạ',
 1227: 'ệ'}

• I also tried to load the font using fitz.Font but nothing helped.

Here is my buffer of this font if anyone can help.
Many many thanks
buffer-Arial-MT.txt