Implement smartify filter

filters/filters.go

@@ -48,18 +48,12 @@ func AddJekyllFilters(e liquid.Engine, c config.Config) {
 	e.RegisterFilter("push", func(array []interface{}, item interface{}) interface{} {
 		return append(array, evaluator.MustConvertItem(item, array))
 	})
-	e.RegisterFilter("pop", func(array []interface{}) interface{} {
-		if len(array) == 0 {
-			return nil
-		}
+	e.RegisterFilter("pop", requireNonEmptyArray(func(array []interface{}) interface{} {
 		return array[0]
-	})
-	e.RegisterFilter("shift", func(array []interface{}) interface{} {
-		if len(array) == 0 {
-			return nil
-		}
+	}))
+	e.RegisterFilter("shift", requireNonEmptyArray(func(array []interface{}) interface{} {
 		return array[len(array)-1]
-	})
+	}))
 	e.RegisterFilter("unshift", func(array []interface{}, item interface{}) interface{} {
 		return append([]interface{}{evaluator.MustConvertItem(item, array)}, array...)
 	})
@@ -132,7 +126,7 @@ func AddJekyllFilters(e liquid.Engine, c config.Config) {
 	e.RegisterFilter("cgi_escape", unimplementedFilter("cgi_escape"))
 	e.RegisterFilter("uri_escape", unimplementedFilter("uri_escape"))
 	e.RegisterFilter("scssify", unimplementedFilter("scssify"))
-	e.RegisterFilter("smartify", unimplementedFilter("smartify"))
+	e.RegisterFilter("smartify", smartifyFilter)
 	e.RegisterFilter("xml_escape", func(s string) string {
 		// TODO can't handle maps
 		// eval https://github.com/clbanning/mxj
@@ -145,6 +139,17 @@ func AddJekyllFilters(e liquid.Engine, c config.Config) {
 	})
 }
 
+// helpers
+
+func requireNonEmptyArray(fn func([]interface{}) interface{}) func([]interface{}) interface{} {
+	return func(array []interface{}) interface{} {
+		if len(array) == 0 {
+			return nil
+		}
+		return fn(array)
+	}
+}
+
 func unimplementedFilter(name string) func(value interface{}) interface{} {
 	warned := false
 	return func(value interface{}) interface{} {
@@ -156,6 +161,8 @@ func unimplementedFilter(name string) func(value interface{}) interface{} {
 	}
 }
 
+// array filters
+
 func arrayToSentenceStringFilter(array []string, conjunction func(string) string) string {
 	conj := conjunction("and ")
 	switch len(array) {
@@ -254,3 +261,4 @@ func whereFilter(array []map[string]interface{}, key string, value interface{})
 	}
 	return out
 }
+

filters/filters_test.go

@@ -61,6 +61,12 @@ var filterTests = []struct{ in, expected string }{
 	{`{{ "The _config.yml file" | slugify: 'default' }}`, "the-config-yml-file"},
 	{`{{ "The _config.yml file" | slugify: 'pretty' }}`, "the-_config.yml-file"},
 
+	{`{{ "smartify single 'quotes' here" | smartify }}`, "smartify single ‘quotes’ here"},
+	{`{{ 'smartify double "quotes" here' | smartify }}`, "smartify double “quotes” here"},
+	{"{{ \"smartify ``backticks''\" | smartify }}", "smartify “backticks”"},
+	{`{{ "smartify it's they're" | smartify }}`, "smartify it’s they’re"},
+	{`{{ "smartify ... (c) (r) (tm) -- ---" | smartify }}`, "smartify … © ® ™ – —"},
+
 	{`{{ "1 < 2 & 3" | xml_escape }}`, "1 &lt; 2 &amp; 3"},
 	// {`{{ "http://foo.com/?q=foo, \bar?" | uri_escape }}`, "http://foo.com/?q=foo,%20%5Cbar?"},
 }

filters/smartify.go

@@ -0,0 +1,69 @@
+package filters
+
+// This file contains a very inefficient implementation, that scans the string multiple times.
+// Replace it by a transducer if it shows up in hot spots.
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+)
+
+var smartifyTransforms = []struct {
+	match *regexp.Regexp
+	repl  string
+}{
+	{regexp.MustCompile("(^|[^[:alnum:]])``(.+?)''"), "$1“$2”"},
+	{regexp.MustCompile(`(^|[^[:alnum:]])'`), "$1‘"},
+	{regexp.MustCompile(`'`), "’"},
+	{regexp.MustCompile(`(^|[^[:alnum:]])"`), "$1“"},
+	{regexp.MustCompile(`"($|[^[:alnum:]])`), "”$1"},
+	{regexp.MustCompile(`(^|\s)--($|\s)`), "$1–$2"},
+	{regexp.MustCompile(`(^|\s)---($|\s)`), "$1—$2"},
+}
+
+// replace these wherever they appear
+var smartifyReplaceSpans = map[string]string{
+	"...":  "…",
+	"(c)":  "©",
+	"(r)":  "®",
+	"(tm)": "™",
+}
+
+// replace these only if bounded by space or word boundaries
+var smartifyReplaceWords = map[string]string{
+// "---": "–",
+// "--":  "—",
+}
+
+var smartifyReplacements map[string]string
+var smartifyReplacementPattern *regexp.Regexp
+
+func init() {
+	smartifyReplacements = map[string]string{}
+	disjuncts := []string{}
+	regexQuoter := regexp.MustCompile(`[\(\)\.]`)
+	escape := func(s string) string {
+		return regexQuoter.ReplaceAllString(s, `\$0`)
+	}
+	for k, v := range smartifyReplaceSpans {
+		disjuncts = append(disjuncts, escape(k))
+		smartifyReplacements[k] = v
+	}
+	for k, v := range smartifyReplaceWords {
+		disjuncts = append(disjuncts, fmt.Sprintf(`(\b|\s|^)%s(\b|\s|$)`, escape(k)))
+		smartifyReplacements[k] = fmt.Sprintf("$1%s$2", v)
+	}
+	p := fmt.Sprintf(`(%s)`, strings.Join(disjuncts, `|`))
+	smartifyReplacementPattern = regexp.MustCompile(p)
+}
+
+func smartifyFilter(s string) string {
+	for _, rule := range smartifyTransforms {
+		s = rule.match.ReplaceAllString(s, rule.repl)
+	}
+	s = smartifyReplacementPattern.ReplaceAllStringFunc(s, func(w string) string {
+		return smartifyReplacements[w]
+	})
+	return s
+}