systemfonts C interface

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
library(systemfonts)

Most of the functionality in systemfonts is intended to be used from compiled code to help e.g. graphic devices to resolve font specifications to a font file prior to rendering. systemfonts provide key functionality to get called at the C level by putting systemfonts in the LinkingTo field in the description and adding #include <systemfonts.h> to your C code. Make sure systemfonts is loaded before using it, e.g. by having match_font() imported into your package namespace. The different functionality will be discussed below

Font matching

The C equivalent of the match_font() R function is locate_font() with the following signature:

int locate_font(
  const char *family, 
  int italic, 
  int bold, 
  char *path, 
  int max_path_length
)

It takes a UTF-8 encoded string with the font family name, an int setting both italic and bold styles along with a char pointer to be filled with the located path and the maximum length it can hold. The return value is an int giving the index of the font in the font file.

With the advent of systemfonts 0.3.0 fonts can now also have OpenType features attached to them through the use of register_font() or register_variant(). If you wish to support such features you can use an alternative to the above:

FontSettings locate_font_with_features(
  const char *family, 
  int italic, 
  int bold
)

The returned FontSettings struct will contain both the font location and index along with any OpenType feature settings. The struct (along with its FontFeature struct dependency) is shown below and is pretty self-documenting.

Do not cache the FontSettings struct as the features array may be cleared at any time after the call has ended. systemfonts itself takes care of caching so this is not something you should be concerned with in your code.

struct FontFeature {
  char feature[4];
  int setting;
};
struct FontSettings {
  char file[PATH_MAX + 1];
  unsigned int index;
  const FontFeature* features;
  int n_features;
};

Glyph metrics

The C equivalent of glyph_info() is glyph_metrics() with the following signature:

int glyph_metrics(
  uint32_t code, 
  const char* fontfile, 
  int index, 
  double size, 
  double res, 
  double* ascent, 
  double* descent, 
  double* width
)

It takes the glyph to measure as an int giving the UTF code of the glyph, along with a fontfile and index to identify the font to measure with. Further it takes a size in pt and a resolution in ppi. It will write the ascent, descent, and width in pts to the pointers passed in, and return 0 if the operation was successful.

String width

The C equivalent of the string_width() R function is also called string_width() with the following signature:

string_width(
  const char* string, 
  const char* fontfile, 
  int index, 
  double size, 
  double res, 
  int include_bearing, 
  double* width
)

This function calculates the width of a string, ignoring any newlines (these are automatically being handled by the graphic engine). It takes a UTF-8 encoded string, along with a fontfile and index identifying the font to use for the calculation. It also take a size in pt and a res in ppi for setting the size. In addition it takes an include_bearing flag to control whether the bearings of the first and last character should be taken into account (this is recommended by the graphic engine). It will write the width in pts to the passed in pointer and return 0 if successful.

String shape

A parred down version of shape_string() is accessible at the C level with string_shape(). It behaves more or less like string_width() above, but instead returns the location to write each glyph at relative to a (0, 0) origin.

string_shape(
  const char* string, 
  const char* fontfile, 
  int index, 
  double size, 
  double res, 
  double* x, 
  double* y, 
  unsigned int max_length
)

string_shape() behaves more or less like string_width() above, but instead returns the location to write each glyph at relative to a (0, 0) origin. It takes a UTF-8 encoded string, along with a fontfile and index identifying the font to use for the calculation. It also take a size in pt and a res in ppi for setting the size. In addition it takes an include_bearing flag to control whether the bearings of the first and last character should be taken into account (this is recommended by the graphic engine). It will write the x and y location of each glyph in pts to the passed in arrays, stopping before the provided max_length and return 0 if successful.

Retrieving cached freetype face

A heavy part of text layouting is reading and parsing font files. systemfonts contains its own cache to make sure that parsing is kept at a minimum. If you want to use this cache to load and cache freetype face object (FT_Face) you can use get_cached_face(). This resides in a separate header (systemfonts-ft.h) because it requires FreeType to be linked in your package, which the rest of the C api does not. It will look in the cache for a face and size that matches your request and return that if found. If not, it will load it for you and add it to the cache, before returning it to you. get_cached_face() sets the passed in error pointer to 0 if successful.

get_cached_face(
  const char* fontfile, 
  int index, 
  double size, 
  double res,
  int * error
)

Freetype uses reference counting to keep track of objects and the count is increased by a call to get_cached_face(). It is the responsiblity of the caller to decrease it once the face is no longer needed using FT_Done_Face().



Try the systemfonts package in your browser

Any scripts or data that you put into this service are public.

systemfonts documentation built on Oct. 23, 2020, 8 p.m.