Font

  • New in v1.16.18

This class represents a font as defined in MuPDF (fz_font_s structure). It is required for the new class TextWriter and the new Page.write_text(). Currently, it has no connection to how fonts are used in methods Page.insert_text() or Page.insert_textbox(), respectively.

A Font object also contains useful general information, like the font bbox, the number of defined glyphs, glyph names or the bbox of a single glyph.

Method / Attribute

Short Description

glyph_advance()

Width of a character

glyph_bbox()

Glyph rectangle

glyph_name_to_unicode()

Get unicode from glyph name

has_glyph()

Return glyph id of unicode

text_length()

Compute string length

char_lengths()

Tuple of char widths of a string

unicode_to_glyph_name()

Get glyph name of a unicode

valid_codepoints()

Array of supported unicodes

ascender

Font ascender

descender

Font descender

bbox

Font rectangle

buffer

Copy of the font’s binary image

flags

Collection of font properties

glyph_count

Number of supported glyphs

name

Name of font

is_writable

Font usable with TextWriter

Class API

class Font

  • __init__(self, fontname=None, fontfile=None,

    fontbuffer=None, script=0, language=None, ordering=-1, is_bold=0,

    is_italic=0, is_serif=0)

    Font constructor. The large number of parameters are used to locate font, which most closely resembles the requirements. Not all parameters are ever required – see the below pseudo code explaining the logic how the parameters are evaluated.

    • Parameters

      • fontname (str) –

        one of the PDF Base 14 Fonts or CJK fontnames. Also possible are a select few other names like (watch the correct spelling): “Arial”, “Times”, “Times Roman”.

        (Changed in v1.17.5)

        If you have installed pymupdf-fonts, there are also new “reserved” fontnames available, which are listed in fitz_fonts and in the table further down.

      • fontfile (str) – the filename of a fontfile somewhere on your system 1.

      • fontbuffer (bytes,bytearray,io.BytesIO) – a fontfile loaded in memory 1.

      • script (in) – the number of a UCDN script. Currently supported in PyMuPDF are numbers 24, and 32 through 35.

      • language (str) – one of the values “zh-Hant” (traditional Chinese), “zh-Hans” (simplified Chinese), “ja” (Japanese) and “ko” (Korean). Otherwise, all ISO 639 codes from the subsets 1, 2, 3 and 5 are also possible, but are currently documentary only.

      • ordering (int) – an alternative selector for one of the CJK fonts.

      • is_bold (bool) – look for a bold font.

      • is_italic (bool) – look for an italic font.

      • is_serif (bool) – look for a serifed font.

      Returns

      a MuPDF font if successful. This is the overall sequence of checks to determine an appropriate font:

      Argument

      Action

      fontfile?

      Create font from file, exception if failure.

      fontbuffer?

      Create font from buffer, exception if failure.

      ordering>=0

      Create universal font, always succeeds.

      fontname?

      Create a Base-14 font, universal font, or font provided by pymupdf-fonts. See table below.

    Note

    With the usual reserved names “helv”, “tiro”, etc., you will create fonts with the expected names “Helvetica”, “Times-Roman” and so on. However, and in contrast to Page.insert_font() and friends,

    • a font file will always be embedded in your PDF,

    • Greek and Cyrillic characters are supported without needing the encoding parameter.

    Using ordering >= 0, or fontnames “cjk”, “china-t”, “china-s”, “japan” or “korea” will always create the same “universal” font “Droid Sans Fallback Regular”. This font supports all Chinese, Japanese, Korean and Latin characters, including Greek and Cyrillic. This is a sans-serif font.

    Actually, you would rarely ever need another sans-serif font than “Droid Sans Fallback Regular”. Except that this font file is relatively large and adds about 1.65 MB (compressed) to your PDF file size. If you do not need CJK support, stick with specifying “helv”, “tiro” etc., and you will get away with about 35 KB compressed.

    If you know you have a mixture of CJK and Latin text, consider just using Font("cjk") because this supports everything and also significantly (by a factor of up to three) speeds up execution: MuPDF will always find any character in this single font and never needs to check fallbacks.

    But if you do use some other font, you will still automatically be able to also write CJK characters: MuPDF detects this situation and silently falls back to the universal font (which will then of course also be embedded in your PDF).

    (New in v1.17.5) Optionally, some new “reserved” fontname codes become available if you install pymupdf-fonts, pip install pymupdf-fonts. “Fira Mono” is a mono-spaced sans font set and FiraGO is another non-serifed “universal” font set which supports all Latin (including Cyrillic and Greek) plus Thai, Arabian, Hewbrew and Devanagari – but none of the CJK languages. The size of a FiraGO font is only a quarter of the “Droid Sans Fallback” size (compressed 400 KB vs. 1.65 MB) – and it provides the weights bold, italic, bold-italic – which the universal font doesn’t.

    “Space Mono” is another nice and small mono-spaced font from Google Fonts, which supports Latin Extended characters and comes with all 4 important weights.

    The following table maps a fontname code to the corresponding font. For the current content of the package please see its documentation:

    Code

    Fontname

    New in

    Comment

    figo

    FiraGO Regular

    v1.0.0

    narrower than Helvetica

    figbo

    FiraGO Bold

    v1.0.0

    figit

    FiraGO Italic

    v1.0.0

    figbi

    FiraGO Bold Italic

    v1.0.0

    fimo

    Fira Mono Regular

    v1.0.0

    fimbo

    Fira Mono Bold

    v1.0.0

    spacemo

    Space Mono Regular

    v1.0.1

    spacembo

    Space Mono Bold

    v1.0.1

    spacemit

    Space Mono Italic

    v1.0.1

    spacembi

    Space Mono Bold-Italic

    v1.0.1

    math

    Noto Sans Math Regular

    v1.0.2

    math symbols

    music

    Noto Music Regular

    v1.0.2

    musical symbols

    symbol1

    Noto Sans Symbols Regular

    v1.0.2

    replacement for “symb”

    symbol2

    Noto Sans Symbols2 Regular

    v1.0.2

    extended symbol set

    notos

    Noto Sans Regular

    v1.0.3

    alternative to Helvetica

    notosit

    Noto Sans Italic

    v1.0.3

    notosbo

    Noto Sans Bold

    v1.0.3

    notosbi

    Noto Sans BoldItalic

    v1.0.3

  • has_glyph(chr, language=None, script=0, fallback=False)

    Check whether the unicode chr exists in the font or (option) some fallback font. May be used to check whether any “TOFU” symbols will appear on output.

    • Parameters

      • chr (int) – the unicode of the character (i.e. ord()).

      • language (str) – the language – currently unused.

      • script (int) – the UCDN script number.

      • fallback (bool) – (new in v1.17.5) perform an extended search in fallback fonts or restrict to current font (default).

      Returns

      (changed in 1.17.7) the glyph number. Zero indicates no glyph found.

  • valid_codepoints()

    • New in v1.17.5

    Return an array of unicodes supported by this font.

    • Returns

      an array.array 2 of length at most Font.glyph_count. I.e. chr() of every item in this array has a glyph in the font without using fallbacks. This is an example display of the supported glyphs:

      1. >>> import fitz
      2. >>> font = fitz.Font("math")
      3. >>> vuc = font.valid_codepoints()
      4. >>> for i in vuc:
      5. print("%04X %s (%s)" % (i, chr(i), font.unicode_to_glyph_name(i)))
      6. 0000
      7. 000D (CR)
      8. 0020 (space)
      9. 0021 ! (exclam)
      10. 0022 " (quotedbl)
      11. 0023 # (numbersign)
      12. 0024 $ (dollar)
      13. 0025 % (percent)
      14. ...
      15. 00AC ¬ (logicalnot)
      16. 00B1 ± (plusminus)
      17. ...
      18. 21D0 ⇐ (arrowdblleft)
      19. 21D1 ⇑ (arrowdblup)
      20. 21D2 ⇒ (arrowdblright)
      21. 21D3 ⇓ (arrowdbldown)
      22. 21D4 ⇔ (arrowdblboth)
      23. ...
      24. 221E ∞ (infinity)
      25. ...

    Note

    This method only returns meaningful data for fonts having a CMAP (character map, charmap, the /ToUnicode PDF key). Otherwise, this array will have length 1 and contain zero only.

  • glyph_advance(chr, language=None, script=0, wmode=0)

    Calculate the “width” of the character’s glyph (visual representation).

    • Parameters

      • chr (int) – the unicode number of the character. Use ord(), not the character itself. Again, this should normally work even if a character is not supported by that font, because fallback fonts will be checked where necessary.

      • wmode (int) – write mode, 0 = horizontal, 1 = vertical.

  1. The other parameters are not in use currently.
  2. - Returns
  3. a float representing the glyphs width relative to **fontsize 1**.
  • glyph_name_to_unicode(name)

    Return the unicode value for a given glyph name. Use it in conjunction with chr() if you want to output e.g. a certain symbol.

    • Parameters

      name (str) – The name of the glyph.

      Returns

      The unicode integer, or 65533 = 0xFFFD if the name is unknown. Examples: font.glyph_name_to_unicode("Sigma") = 931, font.glyph_name_to_unicode("sigma") = 963. Refer to the Adobe Glyph List publication for a list of glyph names and their unicode numbers. Example:

      1. >>> font = fitz.Font("helv")
      2. >>> font.has_glyph(font.glyph_name_to_unicode("infinity"))
      3. True
  • glyph_bbox(chr, language=None, script=0)

    The glyph rectangle relative to fontsize 1.

    • Parameters

      chr (int) – ord() of the character.

      Returns

      a Rect.

  • unicode_to_glyph_name(ch)

    Show the name of the character’s glyph.

    • Parameters

      ch (int) – the unicode number of the character. Use ord(), not the character itself.

      Returns

      a string representing the glyph’s name. E.g. font.glyph_name(ord("#")) = "numbersign". For an invalid code “.notfound” is returned.

      Note

      (Changed in v1.18.0) This method and Font.glyph_name_to_unicode() no longer depend on a font and instead retrieve information from the Adobe Glyph List. Also available as fitz.unicode_to_glyph_name() and resp. fitz.glyph_name_to_unicode().

  • text_length(text, fontsize=11)

    Calculate the length in points of a unicode string.

    Note

    There is a functional overlap with get_text_length() for Base-14 fonts only.

    • Parameters

      • text (str) – a text string, UTF-8 encoded.

      • fontsize (float) – the fontsize.

      Return type

      float

      Returns

      the length of the string in points when stored in the PDF. If a character is not contained in the font, it will automatically be looked up in a fallback font.

      Note

      This method was originally implemented in Python, based on calling Font.glyph_advance(). For performance reasons, it has been rewritten in C for v1.18.14. To compute the width of a single character, you can now use either of the following without performance penalty:

      1. font.glyph_advance(ord("Ä")) * fontsize

      2. font.text_length("Ä", fontsize=fontsize)

      For multi-character strings, the method offers a huge performance advantage compared to the previous implementation: instead of about 0.5 microseconds for each character, only 12.5 nanoseconds are required for the second and subsequent ones.

  • char_lengths(text, fontsize=11)

    New in v1.18.14

    Sequence of character lengths in points of a unicode string.

    • Parameters

      • text (str) – a text string, UTF-8 encoded.

      • fontsize (float) – the fontsize.

      Return type

      tuple

      Returns

      the lengths in points of the characters of a string when stored in the PDF. It works like Font.text_length() broken down to single characters. This is a high speed method, used e.g. in TextWriter.fill_textbox(). The following is true (allowing rounding errors): font.text_length(text) == sum(font.char_lengths(text)).

      1. >>> font = fitz.Font("helv")
      2. >>> text = "PyMuPDF"
      3. >>> font.text_length(text)
      4. 50.115999937057495
      5. >>> fitz.get_text_length(text, fontname="helv")
      6. 50.115999937057495
      7. >>> sum(font.char_lengths(text))
      8. 50.115999937057495
      9. >>> pprint(font.char_lengths(text))
      10. (7.336999952793121, # P
      11. 5.5, # y
      12. 9.163000047206879, # M
      13. 6.115999937057495, # u
      14. 7.336999952793121, # P
      15. 7.942000031471252, # D
      16. 6.721000015735626) # F
  • buffer

    • New in v1.17.6

    Copy of the binary font file content.

    • Return type

      bytes

  • flags

    A dictionary with various font properties, each represented as bools. Example for Helvetica:

    1. >>> pprint(font.flags)
    2. {'bold': 0,
    3. 'fake-bold': 0,
    4. 'fake-italic': 0,
    5. 'invalid-bbox': 0,
    6. 'italic': 0,
    7. 'mono': 0,
    8. 'opentype': 0,
    9. 'serif': 1,
    10. 'stretch': 0,
    11. 'substitute': 0}
    • Return type

      dict

  • name

    • Return type

      str

    Name of the font. May be “” or “(null)”.

  • bbox

    The font bbox. This is the maximum of its glyph bboxes.

  • glyph_count

    • Return type

      int

    The number of glyphs defined in the font.

  • ascender

    • New in v1.18.0

    The ascender value of the font, see here) for details. Please note that there is a difference to the strict definition: our value includes everything above the baseline – not just the height difference between upper case “A” and and lower case “a”.

    • Return type

      float

  • descender

    • New in v1.18.0

    The descender value of the font, see here for details. This value always is negative and is the portion that some glyphs descend below the base line, for example “g” or “y”. As a consequence, the value ascender - descender is the total height, that every glyph of the font fits into. This is true at least for most fonts – as always, there are exceptions, especially for calligraphic fonts, etc.

    • Return type

      float

  • is_writable

    • New in v1.18.0

    Indicates whether this font can be used with TextWriter.

    • Return type

      bool

Footnotes

1(1,2)

MuPDF does not support all fontfiles with this feature and will raise exceptions like “mupdf: FT_New_Memory_Face((null)): unknown file format”, if it encounters issues. The TextWriter methods check Font.is_writable.

2

The built-in module array has been chosen for its speed and its compact representation of values.