From 2849f0ef14e97ed6ad5f8a8abec25270d1927c18 Mon Sep 17 00:00:00 2001
From: Matthew Butterick <mb@mbtype.com>
Date: Thu, 13 Feb 2014 17:27:58 -0800
Subject: [PATCH] start xexpr file

---
 hyphenate/xexpr.rkt | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 hyphenate/xexpr.rkt

diff --git a/hyphenate/xexpr.rkt b/hyphenate/xexpr.rkt
new file mode 100644
index 00000000..a90e5efc
--- /dev/null
+++ b/hyphenate/xexpr.rkt
@@ -0,0 +1,39 @@
+#lang racket/base
+(require "main.rkt")
+(require xml)
+
+
+
+#|
+The following grammar describes expressions that create X-expressions:
+
+  xexpr	 	=	 	string
+ 	 	|	 	(list symbol (list (list symbol string) ...) xexpr ...)
+ 	 	|	 	(cons symbol (list xexpr ...))
+ 	 	|	 	symbol
+ 	 	|	 	valid-char?
+ 	 	|	 	cdata
+ 	 	|	 	misc
+
+|#
+
+
+;; recursively hyphenate strings within xexpr
+;; todo: add exclusion #:only [only-proc (λ(x) x)]
+(define (hyphenate-xexpr x ) 
+  (define exclusions '(style script)) ; omit these from ever being hyphenated
+  
+  (cond
+    ; todo: the only-proc semantics are illogical.
+    ; main issue: keep it out of tags like <style> that parse as textual elements, but are not.
+    ; So two choices, opt-out or opt-in.
+    ; Problem with opt-out: is set of outlier tags like <style> well-defined?
+    ; Won't it make hyphenation naturally overinclusive?
+    ; Problem with opt-in: conceals a lot of tags that naturally live inside other tags
+    ; only reaches text at the "root level" of the tag.
+    [(tagged-xexpr? x) (if (and (only-proc x) (not ((car x) . in? . exclusions)))
+                           (map-xexpr-elements hyphenate x)
+                           (map-xexpr-elements (λ(x) (if (tagged-xexpr? x) (hyphenate x) x)) x))] ; only process subxexprs
+    
+    [(string? x) (hyphenate-string x)]
+    [else x]))