Abstract: A computer-implemented method technique can intercept, at a server including one or more processors, a first request for a source document representing a web page, the first request being transmitted from a computing device to a web server. The first request can include web browser information indicating web browser software executing on the computing device. The technique can intercept the source document being transmitted from the web server to the computing device in response to the first request. The source document can specify one or more fonts in which to display text in the source document. The technique can generate the web page, determine unique characters displayed at the web page, and obtain one or more font subsets. The technique can then transmit information to the computing device based on whether the web browser software is capable of displaying a font family having a plurality of fonts.

Abstract: A request for a font file including a first font table and a second font table is received. A first entropy encoder is selected, based on characteristics of the first font table, front among a plurality of entropy encoders. A second entropy encoder is selected, based on characteristics of the second font table, front among the plurality of entropy encoders. The first entropy encoder is applied to the first font table. The second entropy encoder is applied to the second font table. Compressed data corresponding to the first and second font tables are combined to generate a compressed font file. The compressed font file is transmitted.

Abstract: A compressed font file including compressed data is received. The compressed data is decompressed to generate decompressed data. An attempt to validate the decompressed data is made. Within the decompressed data, a predetermined pattern of invalid data corresponding to an element of a font file is detected. The font file is reconstructed by replacing at least some of the invalid data with the element.

Abstract: A computer-implemented method technique can intercept, at a server including one or more processors, a first request for a source document representing a web page, the first request being transmitted from a computing device to a web server. The first request can include web browser information indicating web browser software executing on the computing device. The technique can intercept the source document being transmitted from the web server to the computing device in response to the first request. The source document can specify one or more fonts in which to display text in the source document. The technique can generate the web page, determine unique characters displayed at the web page, and obtain one or more font subsets. The technique can then transmit information to the computing device based on whether the web browser software is capable of displaying a font family having a plurality of fonts.

Abstract: A request for a font file including a first font table and a second font table is received. A first entropy encoder is selected, based on characteristics of the first font table, front among a plurality of entropy encoders. A second entropy encoder is selected, based on characteristics of the second font table, front among the plurality of entropy encoders. The first entropy encoder is applied to the first font table. The second entropy encoder is applied to the second font table. Compressed data corresponding to the first and second font tables are combined to generate a compressed font file. The compressed font file is transmitted.

Abstract: A request for a font file including a first font table and a second font table is received. A first entropy encoder is selected, based on characteristics of the first font table, front among a plurality of entropy encoders. A second entropy encoder is selected, based on characteristics of the second font table, front among the plurality of entropy encoders. The first entropy encoder is applied to the first font table. The second entropy encoder is applied to the second font table. Compressed data corresponding to the first and second font tables are combined to generate a compressed font file. The compressed font file is transmitted.

Abstract: A request for a font comprising a font file is received from a device via a network. A shared initial probability chart associated with the font table is retrieved. The shared initial probability chart comprises an initial probability estimate based on a byte stream in the font table to be compressed. The byte stream is compressed based on the initial probability estimate to generate a compressed font table. The compressed font file is transmitted to the device. The device uses the shared initial probability chart to decompress the compressed font file.

Abstract: A compressed font file including compressed data is received. The compressed data is decompressed to generate decompressed data. An attempt to validate the decompressed data is made. Within the decompressed data, a predetermined pattern of invalid data corresponding to an element of a font file is detected. The font file is reconstructed by replacing at least some of the invalid data with the element.

Abstract: A request for a font file including a first font table and a second font table is received. A first entropy encoder is selected, based on characteristics of the first font table, front among a plurality of entropy encoders. A second entropy encoder is selected, based on characteristics of the second font table, front among the plurality of entropy encoders. The first entropy encoder is applied to the first font table. The second entropy encoder is applied to the second font table. Compressed data corresponding to the first and second font tables are combined to generate a compressed font file. The compressed font file is transmitted.