diff --git a/src/dom/test/sgml-parser.c b/src/dom/test/sgml-parser.c
index 7c7f56ec..9dfe1f1a 100644
--- a/src/dom/test/sgml-parser.c
+++ b/src/dom/test/sgml-parser.c
@@ -21,7 +21,7 @@
static void
print_compressed_string(struct dom_string *string)
{
- unsigned char escape[2] = "\\";
+ unsigned char escape[3] = { '\\', '?', 0 };
size_t pos;
for (pos = 0; pos < string->length; pos++) {
@@ -142,8 +142,8 @@ struct dom_stack_context_info sgml_parser_test_context_info = {
/* DOM_NODE_ELEMENT */ sgml_parser_test_branch,
/* DOM_NODE_ATTRIBUTE */ sgml_parser_test_id_leaf,
/* DOM_NODE_TEXT */ sgml_parser_test_leaf,
- /* DOM_NODE_CDATA_SECTION */ sgml_parser_test_id_leaf,
- /* DOM_NODE_ENTITY_REFERENCE */ sgml_parser_test_id_leaf,
+ /* DOM_NODE_CDATA_SECTION */ sgml_parser_test_leaf,
+ /* DOM_NODE_ENTITY_REFERENCE */ sgml_parser_test_branch,
/* DOM_NODE_ENTITY */ sgml_parser_test_id_leaf,
/* DOM_NODE_PROC_INSTRUCTION */ sgml_parser_test_id_leaf,
/* DOM_NODE_COMMENT */ sgml_parser_test_leaf,
diff --git a/src/dom/test/test-sgml-parser-basic b/src/dom/test/test-sgml-parser-basic
index 8b40d51d..9c26420b 100755
--- a/src/dom/test/test-sgml-parser-basic
+++ b/src/dom/test/test-sgml-parser-basic
@@ -11,24 +11,89 @@ correctly in the DOM tree.
. ./libtest
+test_output_equals () {
+ desc="$1"
+ src="$2"
+ out="$3"
+
+ URI="test:$(echo "$desc" | sed '
+ s/^[ \t]*\[[^]]*\][ \t]*//;
+ s/[:., \t][:., \t]*/-/g;
+ s/_/-/g;
+ # *cough*
+ y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/;
+ s/[^a-zA-Z0-9-]//g;')"
+
+ sgml-parser --uri "$URI" --src "$src" | sed 's/^ //' > output
+ echo "#document: $URI" > expected
+ echo "$out" | sed -n '2,$p' >> expected
+
+ test_expect_success "$desc" 'cmp -b output expected'
+}
+
+
################################################################
-# init-db has been done in an empty repository.
-# make sure it is empty.
+# Parse various SGML node types.
-sgml-parser --uri "test:hello-world" \
- --src "Hello World!
" \
- > output
+test_output_equals \
+'Parse a small document.' \
+'Hello World!
' \
+'
+element: html
+ element: body
+ element: p
+ #text: Hello World!'
-cat > expected <' \
+'
+element: root
+ #comment: Hello World! '
-test_expect_success \
- 'Test a small document.' \
- 'cmp output expected'
+test_output_equals \
+'Parse an enclosed CDATA section.' \
+'...]]>' \
+'
+element: root
+ #cdata-section: ...] ]>...'
+
+test_output_equals \
+'Parse attributes.' \
+'' \
+'
+element: root
+ attribute: lang -> fr
+ attribute: attr ->
+ attribute: name -> value with &foo; &...*...&...copy;...&;...&#;' \
+'
+element: root
+ entity-reference: amp
+ #text: ...
+ entity-reference: #42
+ #text: ...
+ entity-reference: ...copy
+ #text: ...
+ #text: &;
+ #text: ...
+ entity-reference: #'
+
+# Test
+test_output_equals \
+'Parse processing instructions.' \
+'
+...
+' \
+'
+proc-instruction: xml -> encoding="UTF8"
+ attribute: encoding -> UTF8
+#text: \n...\n
+proc-instruction: ecmascript -> \nvar -> val=2;\n'
test_done