Add new build tool.

bibliogra.py requires Python 2, which is a pain to deal with.  This
commit adds a Go tool that compiles CensorBib from BibTeX to HTML.  The
tool does the bare minimum and is quite strict in the BibTeX format it
expects.

src/decode.go

@@ -0,0 +1,55 @@
+package main
+
+import (
+	"log"
+	"strings"
+)
+
+type conversion struct {
+	from string
+	to   string
+}
+
+func decodeTitle(title string) string {
+	for _, convert := range []conversion{
+		{`\#`, "#"},
+		{`--`, `–`},
+		{"``", "“"},
+		{"''", "”"},
+		{"'", "’"},       // U+2019
+		{`$\cdot$`, `·`}, // U+00B7.
+	} {
+		title = strings.Replace(title, convert.from, convert.to, -1)
+	}
+
+	// Get rid of all curly brackets. We're displaying titles without changing
+	// their casing.
+	title = strings.ReplaceAll(title, "{", "")
+	title = strings.ReplaceAll(title, "}", "")
+
+	return title
+}
+
+func decodeAuthors(authors string) string {
+	for _, convert := range []conversion{
+		{"'", "’"},
+	} {
+		authors = strings.Replace(authors, convert.from, convert.to, -1)
+	}
+	// For simplicity, we expect authors to be formatted as "John Doe" instead
+	// of "Doe, John".
+	if strings.Contains(authors, ",") {
+		log.Fatalf("author %q contains a comma", authors)
+	}
+	authorSlice := strings.Split(authors, " and ")
+	return strings.Join(authorSlice, ", ")
+}
+
+func decodeProceedings(proceedings string) string {
+	for _, convert := range []conversion{
+		{`\&`, "&"},
+	} {
+		proceedings = strings.Replace(proceedings, convert.from, convert.to, -1)
+	}
+	return proceedings
+}