content bytes for resources now go through a coersion/cleanup stage. Currently we're only doing HTML, since web-world requires well-formed HTML.

resource/coerse-content-bytes.rkt

@@ -0,0 +1,19 @@
+#lang racket/base
+
+(provide coerse-content-bytes)
+(require (planet neil/html-parsing)
+         (planet neil/html-writing))
+
+;; coerse-content-bytes: path bytes -> bytes
+;; Given some bytes, we may need to do some data validation or cleanup.
+;; In particular, when the file's HTML, we need to make sure we're
+;; storing syntactically valid HTML.
+(define (coerse-content-bytes a-path bytes)
+  (cond
+   [(regexp-match #rx"\\.html$"
+                  (path->string a-path))
+    (define op (open-output-bytes))
+    (write-html (html->xexp (open-input-bytes bytes)) op)
+    (get-output-bytes op)]
+   [else
+    bytes]))

resource/compile-time.rkt

@@ -4,6 +4,7 @@
                      racket/path
                      racket/port
                      syntax/parse
+                     "coerse-content-bytes.rkt"
                      "munge-path.rkt"
                      "record.rkt"))
                      
@@ -27,7 +28,8 @@
                                   (current-directory))
                               (syntax-e #'path)))]
             [munged-path (munge-path normal-path)]
-            [content (call-with-input-file normal-path port->bytes)])
+            [content (coerse-content-bytes normal-path
+                                           (call-with-input-file normal-path port->bytes))])
        (with-syntax ([normal-path normal-path]
                      [munged-path munged-path]
                      [content content])