"""

A Python interface to Adobe Font Metrics Files.

Although a number of other Python implementations exist, and may be more

complete than this, it was decided not to go with them because they were

either:

1) copyrighted or used a non-BSD compatible license

2) had too many dependencies and a free standing lib was needed

3) did more than needed and it was easier to write afresh rather than

figure out how to get just what was needed.

It is pretty easy to use, and has no external dependencies:

>>> import matplotlib as mpl

>>> from pathlib import Path

>>> afm_path = Path(mpl.get_data_path(), 'fonts', 'afm', 'ptmr8a.afm')

>>>

>>> from matplotlib._afm import AFM

>>> with afm_path.open('rb') as fh:

... afm = AFM(fh)

>>> afm.get_fontname()

'Times-Roman'

As in the Adobe Font Metrics File Format Specification, all dimensions

are given in units of 1/1000 of the scale factor (point size) of the font

being used.

"""

import inspect

import logging

import re

from typing import BinaryIO, NamedTuple, TypedDict, cast

from ._mathtext_data import uni2type1

from .ft2font import CharacterCodeType, GlyphIndexType

_log = logging.getLogger(__name__)

def _to_int(x: bytes | str) -> int:

# Some AFM files have floats where we are expecting ints -- there is

# probably a better way to handle this (support floats, round rather than

# truncate). But I don't know what the best approach is now and this

# change to _to_int should at least prevent Matplotlib from crashing on

# these. JDH (2009-11-06)

return int(float(x))

def _to_float(x: bytes | str) -> float:

# Some AFM files use "," instead of "." as decimal separator -- this

# shouldn't be ambiguous (unless someone is wicked enough to use "," as

# thousands separator...).

if isinstance(x, bytes):

# Encoding doesn't really matter -- if we have codepoints >127 the call

# to float() will error anyways.

x = x.decode('latin-1')

return float(x.replace(',', '.'))

def _to_str(x: bytes) -> str:

return x.decode('utf8')

def _to_list_of_ints(s: bytes) -> list[int]:

s = s.replace(b',', b' ')

return [_to_int(val) for val in s.split()]

def _to_list_of_floats(s: bytes | str) -> list[float]:

return [_to_float(val) for val in s.split()]

def _to_bool(s: bytes) -> bool:

if s.lower().strip() in (b'false', b'0', b'no'):

return False

else:

return True

class FontMetricsHeader(TypedDict, total=False):

StartFontMetrics: float

FontName: str

FullName: str

FamilyName: str

Weight: str

ItalicAngle: float

IsFixedPitch: bool

FontBBox: list[int]

UnderlinePosition: float

UnderlineThickness: float

Version: str

# Some AFM files have non-ASCII characters (which are not allowed by the spec).

# Given that there is actually no public API to even access this field, just return

# it as straight bytes.

Notice: bytes

EncodingScheme: str

CapHeight: float # Is the second version a mistake, or

Capheight: float # do some AFM files contain 'Capheight'? -JKS

XHeight: float

Ascender: float

Descender: float

StdHW: float

StdVW: float

StartCharMetrics: int

CharacterSet: str

Characters: int

def _parse_header(fh: BinaryIO) -> FontMetricsHeader:

"""

Read the font metrics header (up to the char metrics).

Returns

-------

dict

A dictionary mapping *key* to *val*. Dictionary keys are:

StartFontMetrics, FontName, FullName, FamilyName, Weight, ItalicAngle,

IsFixedPitch, FontBBox, UnderlinePosition, UnderlineThickness, Version,

Notice, EncodingScheme, CapHeight, XHeight, Ascender, Descender,

StartCharMetrics

*val* will be converted to the appropriate Python type as necessary, e.g.,:

* 'False' -> False

* '0' -> 0

* '-168 -218 1000 898' -> [-168, -218, 1000, 898]

"""

header_converters = {

bool: _to_bool,

bytes: lambda x: x,

float: _to_float,

int: _to_int,

list[int]: _to_list_of_ints,

str: _to_str,

}

header_value_types = inspect.get_annotations(FontMetricsHeader)

d: FontMetricsHeader = {}

first_line = True

for line in fh:

line = line.rstrip()

if line.startswith(b'Comment'):

continue

lst = line.split(b' ', 1)

key = lst[0]

if first_line:

# AFM spec, Section 4: The StartFontMetrics keyword

# [followed by a version number] must be the first line in

# the file, and the EndFontMetrics keyword must be the

# last non-empty line in the file. We just check the

# first header entry.

if key != b'StartFontMetrics':

raise RuntimeError('Not an AFM file')

first_line = False

if len(lst) == 2:

val = lst[1]

else:

val = b''

try:

key_str = _to_str(key)

value_type = header_value_types[key_str]

except (KeyError, UnicodeDecodeError):

_log.error("Found an unknown keyword in AFM header (was %r)", key)

continue

try:

converter = header_converters[value_type]

d[key_str] = converter(val) # type: ignore[literal-required]

except ValueError:

_log.error('Value error parsing header in AFM: %r, %r', key, val)

continue

if key == b'StartCharMetrics':

break

else:

raise RuntimeError('Bad parse')

return d

class CharMetrics(NamedTuple):

"""

Represents the character metrics of a single character.

Notes

-----

The fields do currently only describe a subset of character metrics

information defined in the AFM standard.

"""

width: float

name: str

bbox: tuple[int, int, int, int]

CharMetrics.width.__doc__ = """The character width (WX)."""

CharMetrics.name.__doc__ = """The character name (N)."""

CharMetrics.bbox.__doc__ = """

The bbox of the character (B) as a tuple (*llx*, *lly*, *urx*, *ury*)."""

def _parse_char_metrics(fh: BinaryIO) -> tuple[dict[CharacterCodeType, CharMetrics],

dict[str, CharMetrics]]:

"""

Parse the given filehandle for character metrics information.

It is assumed that the file cursor is on the line behind 'StartCharMetrics'.

Returns

-------

ascii_d : dict

A mapping "ASCII num of the character" to `.CharMetrics`.

name_d : dict

A mapping "character name" to `.CharMetrics`.

Notes

-----

This function is incomplete per the standard, but thus far parses

all the sample afm files tried.

"""

required_keys = {'C', 'WX', 'N', 'B'}

ascii_d: dict[CharacterCodeType, CharMetrics] = {}

name_d: dict[str, CharMetrics] = {}

for bline in fh:

# We are defensively letting values be utf8. The spec requires

# ascii, but there are non-compliant fonts in circulation

line = _to_str(bline.rstrip())

if line.startswith('EndCharMetrics'):

return ascii_d, name_d

# Split the metric line into a dictionary, keyed by metric identifiers

vals = dict(s.strip().split(' ', 1) for s in line.split(';') if s)

# There may be other metrics present, but only these are needed

if not required_keys.issubset(vals):

raise RuntimeError('Bad char metrics line: %s' % line)

num = _to_int(vals['C'])

wx = _to_float(vals['WX'])

name = vals['N']

bbox = tuple(map(int, _to_list_of_floats(vals['B'])))

if len(bbox) != 4:

raise RuntimeError(f'Bad parse: bbox has {len(bbox)} elements, should be 4')

metrics = CharMetrics(wx, name, bbox)

# Workaround: If the character name is 'Euro', give it the

# corresponding character code, according to WinAnsiEncoding (see PDF

# Reference).

if name == 'Euro':

num = 128

elif name == 'minus':

num = ord("\N{MINUS SIGN}") # 0x2212

if num != -1:

ascii_d[num] = metrics

name_d[name] = metrics

raise RuntimeError('Bad parse')

def _parse_kern_pairs(fh: BinaryIO) -> dict[tuple[str, str], float]:

"""

Return a kern pairs dictionary.

Returns

-------

dict

Keys are (*char1*, *char2*) tuples and values are the kern pair value. For

example, a kern pairs line like ``KPX A y -50`` will be represented as::

d['A', 'y'] = -50

"""

line = next(fh)

if not line.startswith(b'StartKernPairs'):

raise RuntimeError(f'Bad start of kern pairs data: {line!r}')

d: dict[tuple[str, str], float] = {}

for line in fh:

line = line.rstrip()

if not line:

continue

if line.startswith(b'EndKernPairs'):

next(fh) # EndKernData

return d

vals = line.split()

if len(vals) != 4 or vals[0] != b'KPX':

raise RuntimeError(f'Bad kern pairs line: {line!r}')

c1, c2, val = _to_str(vals[1]), _to_str(vals[2]), _to_float(vals[3])

d[(c1, c2)] = val

raise RuntimeError('Bad kern pairs parse')

class CompositePart(NamedTuple):

"""Represents the information on a composite element of a composite char."""

name: bytes

dx: float

dy: float

CompositePart.name.__doc__ = """Name of the part, e.g. 'acute'."""

CompositePart.dx.__doc__ = """x-displacement of the part from the origin."""

CompositePart.dy.__doc__ = """y-displacement of the part from the origin."""

def _parse_composites(fh: BinaryIO) -> dict[bytes, list[CompositePart]]:

"""

Parse the given filehandle for composites information.

It is assumed that the file cursor is on the line behind 'StartComposites'.

Returns

-------

dict

A dict mapping composite character names to a parts list. The parts

list is a list of `.CompositePart` entries describing the parts of

the composite.

Examples

--------

A composite definition line::

CC Aacute 2 ; PCC A 0 0 ; PCC acute 160 170 ;

will be represented as::

composites[b'Aacute'] = [CompositePart(name=b'A', dx=0, dy=0),

CompositePart(name=b'acute', dx=160, dy=170)]

"""

composites: dict[bytes, list[CompositePart]] = {}

for line in fh:

line = line.rstrip()

if not line:

continue

if line.startswith(b'EndComposites'):

return composites

vals = line.split(b';')

cc = vals[0].split()

name, _num_parts = cc[1], _to_int(cc[2])

if len(vals) != _num_parts + 2: # First element is 'CC', last is empty.

raise RuntimeError(f'Bad composites parse: expected {_num_parts} parts, '

f'but got {len(vals) - 2}')

pccParts = []

for s in vals[1:-1]:

pcc = s.split()

part = CompositePart(pcc[1], _to_float(pcc[2]), _to_float(pcc[3]))

pccParts.append(part)

composites[name] = pccParts

raise RuntimeError('Bad composites parse')

def _parse_optional(fh: BinaryIO) -> tuple[dict[tuple[str, str], float],

dict[bytes, list[CompositePart]]]:

"""

Parse the optional fields for kern pair data and composites.

Returns

-------

kern_data : dict

A dict containing kerning information. May be empty.

See `._parse_kern_pairs`.

composites : dict

A dict containing composite information. May be empty.

See `._parse_composites`.

"""

kern_data: dict[tuple[str, str], float] = {}

composites: dict[bytes, list[CompositePart]] = {}

for line in fh:

line = line.rstrip()

if not line:

continue

match line.split()[0]:

case b'StartKernData':

kern_data = _parse_kern_pairs(fh)

case b'StartComposites':

composites = _parse_composites(fh)

return kern_data, composites

class AFM:

def __init__(self, fh: BinaryIO):

"""Parse the AFM file in file object *fh*."""

self._header = _parse_header(fh)

self._metrics, self._metrics_by_name = _parse_char_metrics(fh)

self._kern, self._composite = _parse_optional(fh)

def get_str_bbox_and_descent(self, s: str) -> tuple[int, int, float, int, int]:

"""Return the string bounding box and the maximal descent."""

if not len(s):

return 0, 0, 0, 0, 0

total_width = 0.0

namelast = ''

miny = 1_000_000_000

maxy = 0

left = 0

for c in s:

if c == '\n':

continue

name = uni2type1.get(ord(c), f"uni{ord(c):04X}")

try:

wx, _, bbox = self._metrics_by_name[name]

except KeyError:

name = 'question'

wx, _, bbox = self._metrics_by_name[name]

total_width += wx + self._kern.get((namelast, name), 0)

l, b, w, h = bbox

left = min(left, l)

miny = min(miny, b)

maxy = max(maxy, b + h)

namelast = name

return left, miny, total_width, maxy - miny, -miny

def get_glyph_name(self, # For consistency with FT2Font.

glyph_ind: GlyphIndexType) -> str:

"""Get the name of the glyph, i.e., ord(';') is 'semicolon'."""

return self._metrics[cast(CharacterCodeType, glyph_ind)].name

def get_char_index(self, # For consistency with FT2Font.

c: CharacterCodeType) -> GlyphIndexType:

"""

Return the glyph index corresponding to a character code point.

Note, for AFM fonts, we treat the glyph index the same as the codepoint.

"""

return cast(GlyphIndexType, c)

def get_width_char(self, c: CharacterCodeType) -> float:

"""Get the width of the character code from the character metric WX field."""

return self._metrics[c].width

def get_width_from_char_name(self, name: str) -> float:

"""Get the width of the character from a type1 character name."""

return self._metrics_by_name[name].width

def get_kern_dist_from_name(self, name1: str, name2: str) -> float:

"""

Return the kerning pair distance (possibly 0) for chars *name1* and *name2*.

"""

return self._kern.get((name1, name2), 0)

def get_fontname(self) -> str:

"""Return the font name, e.g., 'Times-Roman'."""

return self._header['FontName']

@property

def postscript_name(self) -> str: # For consistency with FT2Font.

return self.get_fontname()

def get_fullname(self) -> str:

"""Return the font full name, e.g., 'Times-Roman'."""

name = self._header.get('FullName')

if name is None: # use FontName as a substitute

name = self._header['FontName']

return name

def get_familyname(self) -> str:

"""Return the font family name, e.g., 'Times'."""

name = self._header.get('FamilyName')

if name is not None:

return name

# FamilyName not specified so we'll make a guess

name = self.get_fullname()

extras = (r'(?i)([ -](regular|plain|italic|oblique|bold|semibold|'

r'light|ultralight|extra|condensed))+$')

return re.sub(extras, '', name)

@property

def family_name(self) -> str: # For consistency with FT2Font.

"""The font family name, e.g., 'Times'."""

return self.get_familyname()

def get_weight(self) -> str:

"""Return the font weight, e.g., 'Bold' or 'Roman'."""

return self._header['Weight']

def get_angle(self) -> float:

"""Return the fontangle as float."""

return self._header['ItalicAngle']

def get_ascender(self) -> float:

"""Return the ascent as float."""

return self._header['Ascender']

def get_capheight(self) -> float:

"""Return the cap height as float."""

return self._header['CapHeight']

def get_descender(self) -> float:

"""Return the descent as float."""

return self._header['Descender']

def get_xheight(self) -> float:

"""Return the xheight as float."""

return self._header['XHeight']

def get_underline_thickness(self) -> float:

"""Return the underline thickness as float."""

return self._header['UnderlineThickness']