Support subsets, implement getopt interface

This change also introduces support for selecting the downloaded font formats, selecting the subset of the fonts to be downloaded and selecting the output file (including stdout, which can be achieved using -). It also allows specifying the font names and weights on the command line instead of by editing the file.
neverpanic · Jun 13, 2015 · 1106f7b · 1106f7b
1 parent 1cbc294
commit 1106f7b
Showing 1 changed file with 203 additions and 51 deletions.
diff --git a/google-font-download b/google-font-download
@@ -8,8 +8,12 @@
 # With modifications by
 #  - Chris Jung, campino2k.de, and
 #  - Robert, github.com/rotx.
+#  - Thomas Papamichail, https://gist.github.com/pointergr
 #
 # Changelog:
+#   Version 1.2, 2015-06-13
+#     - Add getopt interface
+#     - Support font subsets, modifications by Thomas Papamichail
 #   Version 1.1.1, 2015-04-09
 #     - Switch user agent to IE 8 for WOF to fix problems
 #   Version 1.1, 2014-06-21
@@ -21,6 +25,9 @@
 #     - Added sed extended regex flag detection
 #   Version 1.0, 2014-03-19
 #
+# If you made modifications you'd like to see merged into this script, please mail me a patch to 'cl' at 'clang' dot
+# 'name' or leave a comment at https://neverpanic.de/blog/2014/03/19/downloading-google-web-fonts-for-local-hosting/.
+#
 # License:
 #   Copyright (c) 2014-2015, Clemens Lang
 #   All rights reserved.
@@ -43,49 +50,178 @@
 #   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #=======================================================================================================================
 
-#=======================================================================================================================
-# Place this script in the directory where you want the font files to be downloaded and a font.css file using the fonts
-# to be created. Adjust the list of fonts in the $families variable below with the names of the Google fonts you want to
-# use. If you like, you can adjust the $css variable to write the generated CSS to a different file. Not that this file
-# will be created and overwritten if it exists.
-#=======================================================================================================================
+# Ensure the bash version is new enough. If it isn't error out with a helpful error message rather than crashing later.
+if [ ${BASH_VERSINFO[0]} -lt 4 ]; then
+	echo "Error: This script needs Bash 4.x to run." >&2
+	exit 1
+fi
 
-declare -a families
-families+=('PT Sans:400')
-families+=('PT Sans:700')
-families+=('PT Sans:400italic')
-families+=('PT Sans:700italic')
-families+=('PT Serif:400')
-families+=('PT Serif:700')
-families+=('PT Serif:400italic')
-families+=('PT Serif:700italic')
-
-# Adjust this is you want the created file to have a different name. Note that this file will be overwritten!
+set -euo pipefail
+
+# Set up defaults
 css="font.css"
+lang="latin"
+format="all"
 url="http://fonts.googleapis.com/css"
 
-#=======================================================================================================================
-# No user-serviceable parts below this line. If you just want to use this script, you can stop reading here.
-#
-# If you made modifications you'd like to see merged into this script, please mail me a patch to 'cl' at 'clang' dot
-# 'name' or leave a comment at https://neverpanic.de/blog/2014/03/19/downloading-google-web-fonts-for-local-hosting/.
-#=======================================================================================================================
+# Usage message
+usage() {
+	cat >&2 <<-EOF
+		Usage
+		-----
+		  ${0:-google-font-download} [OPTION...] [FONT...]
 
-# Ensure the bash version is new enough. If it isn't error out with a helpful error message rather than crashing later.
-if [ ${BASH_VERSINFO[0]} -lt 4 ]; then
-	echo "Error: This script needs Bash 4.x to run." >&2
+		  google-font-download, a script to download a local copy of
+		  Google's webfonts and generate a CSS file to use them.
+
+		Options
+		-------
+		  -f FORMAT, --format=FORMAT
+		    Download the specified set of webfont formats from Google's
+		    servers. FORMAT is a comma-separated list of identifiers for
+		    webfont formats. Supported identifiers are eof, woff, svg,
+		    and ttf. Additionally, the special value \`all' expands to
+		    all supported formats. The default is \`$format'. Note that
+		    you may not really need all formats. In most cases, WOFF is
+		    enough. See http://caniuse.com/#search=woff for a current
+		    status.
+
+		  -h, --help
+		    Display this message and exit.
+
+		  -l LANGSPEC, --languages=LANGSPEC
+		    Download the specified subset of languages from Google's
+		    webfonts. LANGSPEC is a comma-separated list of idenfitiers
+		    for font subsets. Common identifiers are latin, latin-ext,
+		    cyrillic, cyrillic-ext, greek, greek-ext, etc. The default
+		    is \`$lang'.
+
+		  -o OUTPUT, --output=OUTPUT
+		    Write the generated CSS into OUTPUT. The file will be
+		    overwritten and will be created if it doesn't exist. The
+		    default is \`$css'.
+
+		Positional Arguments
+		--------------------
+		  This script accepts an arbitrary number of font specs. A font
+		  spec is any string that Google's servers will accept as
+		  famility URL parameter. Note that your font spec should *not*
+		  be URL-encoded and only one font weight is supported per font
+		  specification. If you want to download multiple font weights
+		  or styles, provide multiple font specs.
+
+		  For example, to download Open Sans in
+		   - light (300),
+		   - normal (400),
+		   - normal italic (400italic),
+		   - bold (700), and
+		   - bold italic (700italic),
+		  run:
+		    ${0:-google-font-download} "Open Sans:300" "Open Sans:400" \\
+		        "Open Sans:400italic" "Open Sans:700" \\
+		        "Open Sans:700italic"
+	EOF
 	exit 1
+}
+
+err_exit() {
+	echo >&2 "${0:-google-font-download}: Error: $1"
+	exit 2
+}
+
+misuse_exit() {
+	echo >&2 "${0:-google-font-download}: Error: $1"
+	echo >&2
+	usage
+}
+
+# Parse options
+temp=$(getopt -o f:hl:o: --long format:,help,languages:,output -n "${0:-google-font-download}" -- "$@")
+if [ $? != 0 ]; then
+	echo >&2
+	usage
+fi
+
+# Process arguments
+eval set -- "$temp"
+while true; do
+	case "$1" in
+		-f|--format)
+			format=$2
+			shift 2
+			;;
+		-h|--help)
+			usage
+			exit 1
+			;;
+		-l|--languages)
+			lang=$2
+			shift 2
+			;;
+		-o|--output)
+			css=$2
+			shift 2
+			;;
+		--)
+			shift
+			break
+			;;
+	esac
+done
+
+# Validate font family input
+declare -a families
+families=()
+for family do
+	families+=("$family")
+done
+if [ ${#families[@]} -eq 0 ]; then
+	misuse_exit "No font families given"
 fi
 
+# Validate font format input
+declare -a formats
+if [ "$format" = "all" ]; then
+	# Deal with the special "all" value
+	formats=("eot" "woff" "svg" "ttf")
+else
+	IFS=', ' read -a formats <<< "$format"
+	for f in ${formats[@]}; do
+		case "$f" in
+			eof|woff|svg|ttf)
+				;;
+			*)
+				err_exit "Unsupported font format \`${f}'"
+				;;
+		esac
+	done
+fi
+if [ ${#formats[@]} -eq 0 ]; then
+	misuse_exit "Empty list of font formats given"
+fi
+
+# Validate language input
+if [ "$lang" = "" ]; then
+	misuse_exit "Empty language given"
+fi
+
+# Sanity check on output file
+if [ "$css" = "" ]; then
+	misuse_exit "Output file is empty string"
+elif [ "$css" = "-" ]; then
+	# a single dash means output to stdout
+	css=/dev/stdout
+fi
+
+
 # Check whether sed is GNU or BSD sed, or rather, which parameter enables extended regex support. Note that GNU sed does
 # have -E as an undocumented compatibility option on some systems.
 if [ "$(echo "test" | sed -E 's/([st]+)$/xx\1/' 2>/dev/null)" == "texxst" ]; then
 	ESED="sed -E"
 elif [ "$(echo "test" | sed -r 's/([st]+)$/xx\1/' 2>/dev/null)" == "texxst" ]; then
 	ESED="sed -r"
 else
-	echo "Error: $(which sed) seems to lack extended regex support with -E or -r." >&2
-	exit 2
+	err_exit "$(which sed) seems to lack extended regex support with -E or -r"
 fi
 
 # Store the useragents we're going to use to trick Google's servers into serving us the correct CSS file.
@@ -100,54 +236,70 @@ useragent[ttf]='Mozilla/5.0 (Windows NT 6.1) AppleWebKit/534.54.16 (KHTML, like
 
 # Loop over the fonts, and download them one-by-one
 for family in "${families[@]}"; do
-	echo -n "Downloading ${family}... "
-	printf "@font-face {\n" >>"$css"
+	printf >&2 "Downloading %s..." "$family"
+
+	# Test whether the chosen combination of font and language subset
+	# exists; Google returns HTTP 400 if it doesn't
+	css_string=$(curl -sf --get --data-urlencode "family=$family" --data-urlencode "subset=$lang" "$url") || continue
 
 	# Extract name, Windows-safe filename, font style and font weight from the font family name
 	fontname=$(echo "$family" | awk -F : '{print $1}')
 	fontnameescaped=$(echo "$family" | $ESED 's/( |:)/_/g')
 	fontstyle=$(echo "$family" | awk -F : '{print $2}')
 	fontweight=$(echo "$fontstyle" | $ESED 's/italic$//g')
 
-	printf "\tfont-family: '%s';\n" "$fontname" >>"$css"
+	numfontweight=$(echo "$fontweight" | tr -Cd ',')
+	if [ ${#numfontweight} -ge 1 ]; then
+		printf >&2 "Font specification contains multiple weights; this is unsupported\n"
+		continue
+	fi
+
+	printf >>"$css" "@font-face {\n"
+	printf >>"$css" "\tfont-family: '%s';\n" "$fontname"
 
 	# $fontstyle could be bolditalic, bold, italic, normal, or nothing.
 	case "$fontstyle" in
 		*italic)
-			printf "\tfont-style: italic;\n" >>"$css"
+			printf >>"$css" "\tfont-style: italic;\n"
 			;;
 		*)
-			printf "\tfont-style: normal;\n" >>"$css"
+			printf >>"$css" "\tfont-style: normal;\n"
 			;;
 	esac
 
 	# Either bold, a number, or empty. If empty, default to "normal".
-	printf "\tfont-weight: %s;\n" "${fontweight:-normal}" >>"$css"
-
-	printf "\tsrc:\n" >>"$css"
+	printf >>"$css" "\tfont-weight: %s;\n" "${fontweight:-normal}"
+	printf >>"$css" "\tsrc:\n"
 
 	# Determine the local names for the given fonts so we can use a locally-installed font if available.
-	local_name=$(curl -sf --get --data-urlencode "family=$family" "$url" | grep -E "src:" | $ESED "s/^.*src: local\\('([^']+)'\\),.*$/\\1/g")
-	local_postscript_name=$(curl -sf --get --data-urlencode "family=$family" "$url" | grep -E "src:" | $ESED "s/^.*, local\\('([^']+)'\\),.*$/\\1/g")
+	css_src_string=$(echo "$css_string" | grep "src:")
+	local_name=$(echo "$css_src_string" | $ESED "s/^.*src: local\\('([^']+)'\\),.*$/\\1/g")
+	local_postscript_name=$(echo "$css_src_string" | $ESED "s/^.*, local\\('([^']+)'\\),.*$/\\1/g")
 
 	# Some fonts don't have a local PostScript name.
-	printf "\t\tlocal('%s'),\n" "$local_name" >>"$css"
+	printf >>"$css" "\t\tlocal('%s'),\n" "$local_name"
 	if [ -n "$local_postscript_name" ]; then
-		printf "\t\tlocal('%s'),\n" "$local_postscript_name" >>"$css"
+		printf >>"$css" "\t\tlocal('%s'),\n" "$local_postscript_name"
 	fi
 
-	# For each font format, download the font file and print the corresponding CSS statements.
-	for uakey in eot woff ttf svg; do
-		echo -n "$uakey "
+	# For each requested font format, download the font file and print the corresponding CSS statements.
+	for ((uaidx=0; $uaidx < ${#formats[@]}; uaidx++)); do
+		uakey=${formats[$uaidx]}
+		if [ $uaidx -eq $(( ${#formats[@]} - 1 )) ]; then
+			terminator=";"
+		else
+			terminator=","
+		fi
+		printf >&2 "%s " "$uakey"
 
 		# Download Google's CSS and throw some regex at it to find the font's URL
 		if [ "$uakey" != "svg" ]; then
 			pattern="http:\\/\\/[^\\)]+\\.$uakey"
 		else
 			pattern="http:\\/\\/[^\\)]+"
 		fi
-		file=$(curl -sf -A "${useragent[$uakey]}" --get --data-urlencode "family=$family" "$url" | grep -Eo "$pattern" | sort -u)
-		printf "\t\t/* from %s */\n" "$file" >>"$css"
+		file=$(curl -sf -A "${useragent[$uakey]}" --get --data-urlencode "family=$family" --data-urlencode "subset=$lang" "$url" | grep -Eo "$pattern" | sort -u)
+		printf >>"$css" "\t\t/* from %s */\n" "$file"
 		if [ "$uakey" == "svg" ]; then
 			# SVG fonts need the font after a hash symbol, so extract the correct name from Google's CSS
 			svgname=$(echo "$file" | $ESED 's/^[^#]+#(.*)$/\1/g')
@@ -158,20 +310,20 @@ for family in "${families[@]}"; do
 		# Generate the CSS statements required to include the downloaded file.
 		case "$uakey" in
 			eot)
-				printf "\t\turl('%s?#iefix') format('embedded-opentype'),\n" "${fontnameescaped}.$uakey" >>"$css"
+				printf >>"$css" "\t\turl('%s?#iefix') format('embedded-opentype')%s\n" "${fontnameescaped}.$uakey" "${terminator}"
 				;;
 			woff)
-				printf "\t\turl('%s') format('woff'),\n" "${fontnameescaped}.$uakey" >>"$css"
+				printf >>"$css" "\t\turl('%s') format('woff')%s\n" "${fontnameescaped}.$uakey" "${terminator}"
 				;;
 			ttf)
-				printf "\t\turl('%s') format('truetype'),\n" "${fontnameescaped}.$uakey" >>"$css"
+				printf >>"$css" "\t\turl('%s') format('truetype')%s\n" "${fontnameescaped}.$uakey" "${terminator}"
 				;;
 			svg)
-				printf "\t\turl('%s#%s') format('svg');\n" "${fontnameescaped}.${uakey}" "$svgname" >>"$css"
+				printf >>"$css" "\t\turl('%s#%s') format('svg')%s\n" "${fontnameescaped}.${uakey}" "$svgname" "${terminator}"
 				;;
 		esac
 	done
 
-	printf "}\n" >>"$css"
-	echo
+	printf >>"$css" "}\n"
+	printf >&2 "\n"
 done