diff --git a/Makefile b/Makefile
index 3671a7fb3be5cfe9156823736902862a5349f172..18802f1579f65fb06b20450b78b409eb7b544f9a 100644
--- a/Makefile
+++ b/Makefile
@@ -39,8 +39,10 @@ install: all
cp -ra --no-preserve=ownership ccache/* $(DESTDIR)$(GUILE_CCACHE_DIR)
install -Dt ${DESTDIR}/usr/lib/ libguile-gumbo.so
+test_env=LD_LIBRARY_PATH=$$PWD GUILE_LOAD_PATH=$$PWD GUILE_LOAD_COMPILED_PATH=$$PWD/ccache
+
check: all
- env GUILE_LOAD_PATH=$$PWD GUILE_LOAD_COMPILED_PATH=$$PWD/ccache guile -s tests.scm
+ env ${test_env} guile -s tests.scm
clean:
-rm *.o
diff --git a/guile-gumbo.c b/guile-gumbo.c
index c85d6838d6bd9d2ca41e1d1a17802a7e68dec7eb..caa87093dd9fc888b7abc6addeb3939e3a149bf9 100644
--- a/guile-gumbo.c
+++ b/guile-gumbo.c
@@ -56,7 +56,7 @@ SCM_DEFINE_PUBLIC (scm_parse_html, "parse-html", 2, 0, 0,
#undef FUNC_NAME
SCM scm_parse_children (GumboVector* /* GumboNode */ children, options* opts) {
- SCM childvec =
+ SCM childvec =
scm_c_make_vector (
children->length,
SCM_UNSPECIFIED);
@@ -70,8 +70,18 @@ SCM scm_parse_children (GumboVector* /* GumboNode */ children, options* opts) {
}
SCM gumbo_element_to_sxml (GumboElement* el, options* opts) {
- SCM tag = scm_from_utf8_symbol
- (gumbo_normalized_tagname(el->tag));
+ SCM tag;
+ if (el->tag < GUMBO_TAG_UNKNOWN) {
+ tag = scm_from_utf8_symbol(gumbo_normalized_tagname(el->tag));
+ } else {
+ /* Unknown tag */
+ GumboStringPiece strp = el->original_tag;
+ gumbo_tag_from_original_text(&strp);
+ // assert strp->length > 0 ???
+ SCM str = scm_from_utf8_stringn(strp.data, strp.length);
+ str = scm_string_downcase(str);
+ tag = scm_string_to_symbol(str);
+ }
SCM scm_children = scm_parse_children (&el->children, opts);
GumboVector* attributes = &el->attributes;
diff --git a/tests.scm b/tests.scm
index 5b07d1950a1a91108847e483ea53b0df093fb916..ec15ca2520252d0783fc1d4be31043a6fb56b483 100644
--- a/tests.scm
+++ b/tests.scm
@@ -19,7 +19,7 @@
'(*TOP* (html (head) (body (p) (p))))
(html->sxml "<p></p> <p></p>"
#:trim-whitespace? #t))
-
+
(test-equal "Keep whitespace"
'(*TOP* (html (head) (body (p) " " (p))))
(html->sxml "<p></p> <p></p>"
@@ -29,12 +29,28 @@
'(*TOP* (doctype html) (html (head) (body "Hello")))
(html->sxml "<!doctype HTML>Hello"
#:full-document? #t))
-
+
(test-equal "HTML only"
'(html (head) (body "Hello"))
(html->sxml "<!doctype HTML>Hello"
#:full-document? #f))
+(test-equal "Unknown tag"
+ '(html (head) (body (unknown-tag)))
+ (html->sxml "<unknown-tag/>"
+ #:full-document? #f))
+
+(test-equal "unknown tag upper case"
+ '(html (head) (body (unknown-tag)))
+ (html->sxml "<UNKNOWN-tag/>"
+ #:full-document? #f))
+
+(test-equal "unknown tag weird characters"
+ '(html (head) (body (unknown-täg)))
+ (html->sxml "<unknown-TÄG/>"
+ #:full-document? #f))
+
+
(test-end "Gumbo HTML test")
(exit (test-runner-fail-count (test-runner-current)))