re-jigger bin/meta-normalize-stdio to be {strict,self-documenting}

1 files changed, 164 insertions, 28 deletions

diff --git a/bin/meta-normalize-stdio b/bin/meta-normalize-stdio
index 901ca0b..c49dcd9 100755
--- a/bin/meta-normalize-stdio
+++ b/bin/meta-normalize-stdio
@@ -1,33 +1,169 @@
 #!/usr/bin/env ruby
+
+# First we define a bunch of code-generators, then at the end is a
+# very neat and readable definition of the format of the YAML files.
+
 require 'yaml'
 
-core_order = [ "username",
-               "fullname",
-               "email", # ordered list
-               "groups", # ordered list
-               "pgp_keyid",
-               "pgp_revoked_keyids", # unordered list
-               "ssh_keys", # unordered map
-               "extra" ] # unordered map
-
-extra_order = [ "alias",
-                "other_contact",
-                "roles",
-                "website",
-                "occupation",
-                "yob",
-                "location",
-                "languages",
-                "interests",
-                "favorite_distros" ]
-
-_core_order = Hash[[*core_order.map.with_index]]
-_extra_order = Hash[[*extra_order.map.with_index]]
-
-user = YAML::load(STDIN)
-user = Hash[user.sort_by{|k,v| _core_order[k]}]}
-user["pgp_revoked_keyids"] = user["pgp_revoked_keyids"].sort       if user["extra"]}
-user["ssh_keys"] = Hash[user["ssh_keys"].sort_by{|k,v| k}]         if user["ssh_keys"]}
-user["extra"] = Hash[user["extra"].sort_by{|k,v| _extra_order[k]}] if user["extra"]}
+def error(msg)
+	$stderr.puts "ERROR: #{msg}"
+	@err = 1
+end
+
+def warning(msg)
+	$stderr.puts "WARNING: #{msg}"
+end
+
+
+# Generic validators/formatters
+
+def semiordered_list(cnt, validator)
+	lambda {|name,ary|
+		if ary.class != Array
+			error "`#{name}' must be a list"
+		else
+			ary.each_index{|i| ary[i] = validator.call("#{name}[#{i}]", ary[i])}
+			ary = ary.first(cnt).concat(ary.last(ary.count-cnt).sort)
+		end
+		ary
+	}
+end
+
+def unordered_list(validator)
+	semiordered_list(0, validator)
+end
+
+def _unknown(map_name, key)
+	error "Unknown item: #{map_name}[#{key.inspect}]"
+	0
+end
+def unordered_map1(validator)
+	lambda {|name,hash|
+		if hash.class != Hash
+			error "`#{name}' must be a map"
+		else
+			order = Hash[[*validator.keys.map.with_index]]
+			hash = Hash[hash.sort_by{|k,v| order[k] || _unknown(name,k) }]
+			hash.keys.each{|k|
+				hash[k] = validator[k].call("#{name}[#{k.inspect}]", hash[k])
+			}
+		end
+		hash
+	}
+end
+
+def unordered_map2(key_validator, val_validator)
+	lambda {|name,hash|
+		if hash.class != Hash
+			error "`#{name}' must be a map"
+		else
+			hash = Hash[hash.sort_by{|k,v| k}]
+			hash.keys.each{|k|
+				key_validator.call("#{name} key #{k.inspect}", k)
+				hash[k] = val_validator.call("#{name}[#{k.inspect}]", hash[k])
+			}
+		end
+		hash
+	}
+end
+
+string = lambda {|name,str|
+	if str.class != String
+		error "`#{name}' must be a string"
+	else
+		str
+	end
+}
+
+# Regular Expression String
+def restring(re)
+	lambda {|name,str|
+		if str.class != String
+			error "`#{name}' must be a string"
+		else
+			unless re =~ str
+				error "`#{name}' does not match #{re.inspect}: #{str}"
+			end
+			str
+		end
+	}
+end
+
+
+# Specific validators/formatters
+
+year = lambda {|name, num|
+	if num.class != Fixnum
+		error "`#{name}' must be a year"
+	else
+		if (num < 1900 || num > 3000)
+			error "`#{name}' is a number, but doesn't look like a year"
+		end
+		num
+	end
+}
+
+# This regex is taken from http://www.w3.org/TR/html5/forms.html#valid-e-mail-address
+_email_regex = /^[a-zA-Z0-9.!\#$%&'*+\/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*$/
+email_list = lambda {|name, ary|
+	if ary.class != Array
+		error "`#{name}' must be a list"
+	elsif ary.count > 1
+		preserve = 1
+		if ary.first.end_with?("@parabola.nu") and ary.count > 2
+			preserve = 2
+		end
+		ary = semiordered_list(preserve, restring(_email_regex)).call(name, ary)
+	end
+	ary
+}
+
+shell = lambda {|name, sh|
+	if sh.class != String
+		error "`#{name}' must be a string"
+	else
+		@valid_shells ||= open("/etc/shells").read.split("\n")
+						  .find_all{|line| /^[^\#]/ =~ line}
+		unless @valid_shells.include?(sh)
+			warning "shell not listed in /etc/shells: #{sh}"
+		end
+	end
+	sh
+}
+
+
+# The format of the YAML files
+
+format = unordered_map1(
+	{
+		"username" => restring(/^[a-z][a-z0-9]*$/),
+		"fullname" => string,
+		"email" => email_list,
+		"groups" => semiordered_list(1, string),
+		"pgp_keyid" => restring(/^[0-9A-F]{40}$/),
+		"pgp_revoked_keyids" => unordered_list(restring(/^[0-9A-F]{40}$/)),
+		"ssh_keys" => unordered_map2(string, string),
+		"shell" => shell,
+		"extra" => unordered_map1(
+			{
+				"alias" => string,
+				"other_contact" => string,
+				"roles" => string,
+				"website" => string,
+				"occupation" => string,
+				"yob" => year,
+				"location" => string,
+				"languages" => string,
+				"interests" => string,
+				"favorite_distros" => string,
+			})
+	})
+
+
 
+@err = 0
+user = format.call("user", YAML::load(STDIN))
+if @err != 0
+	exit @err
+end
 print user.to_yaml