Add CDATA tests and the infrastructure to support them.

svn path=/trunk/hubbub/; revision=4410

6 files changed, 52 insertions, 0 deletions

diff --git a/src/tokeniser/tokeniser.c b/src/tokeniser/tokeniser.c
index 1c73e9d..2fe00ac 100644
--- a/src/tokeniser/tokeniser.c
+++ b/src/tokeniser/tokeniser.c
@@ -361,6 +361,9 @@ hubbub_error hubbub_tokeniser_setopt(hubbub_tokeniser *tokeniser,
 	case HUBBUB_TOKENISER_CONTENT_MODEL:
 		tokeniser->content_model = params->content_model.model;
 		break;
+	case HUBBUB_TOKENISER_PROCESS_CDATA:
+		tokeniser->process_cdata_section = params->process_cdata;
+		break;
 	}
 
 	return HUBBUB_OK;
diff --git a/src/tokeniser/tokeniser.h b/src/tokeniser/tokeniser.h
index 20bbe20..2292cd7 100644
--- a/src/tokeniser/tokeniser.h
+++ b/src/tokeniser/tokeniser.h
@@ -27,6 +27,7 @@ typedef enum hubbub_tokeniser_opttype {
 	HUBBUB_TOKENISER_BUFFER_HANDLER,
 	HUBBUB_TOKENISER_ERROR_HANDLER,
 	HUBBUB_TOKENISER_CONTENT_MODEL,
+	HUBBUB_TOKENISER_PROCESS_CDATA
 } hubbub_tokeniser_opttype;
 
 /**
@@ -51,6 +52,8 @@ typedef union hubbub_tokeniser_optparams {
 	struct {
 		hubbub_content_model model;
 	} content_model;
+
+	bool process_cdata;
 } hubbub_tokeniser_optparams;
 
 /* Create a hubbub tokeniser */
diff --git a/test/data/tokeniser2/INDEX b/test/data/tokeniser2/INDEX
index 00c5e01..8c0bc41 100644
--- a/test/data/tokeniser2/INDEX
+++ b/test/data/tokeniser2/INDEX
@@ -9,3 +9,4 @@ test4.test		html5lib tests (part 4)
 contentModelFlags.test	html5lib content model tests
 entities.test		html5lib entity tests
 escapeFlag.test		html5lib escape flag tests
+cdata.test		CDATA section tests
diff --git a/test/data/tokeniser2/cdata.test b/test/data/tokeniser2/cdata.test
new file mode 100644
index 0000000..fb4fa8a
--- /dev/null
+++ b/test/data/tokeniser2/cdata.test
@@ -0,0 +1,23 @@
+{"tests": [
+
+{"description":"Basic CDATA test",
+"processCDATA":true,
+"input":"<![CDATA[test]]>",
+"output":[["Character", "test"]]},
+
+{"description":"Unfinished CDATA test",
+"processCDATA":true,
+"input":"<![CDAT",
+"output":["ParseError", ["Comment", "[CDAT"]]},
+
+{"description":"EOF in CDATA chunk",
+"processCDATA":true,
+"input":"<![CDATA[aa",
+"output":[["Character", "aa"]]},
+
+{"description":"False end in CDATA chunk",
+"processCDATA":true,
+"input":"<![CDATA[aa]]aa",
+"output":[["Character", "aa]]aa"]]},
+
+]}
diff --git a/test/tokeniser2.c b/test/tokeniser2.c
index 2ac508c..2054c6b 100644
--- a/test/tokeniser2.c
+++ b/test/tokeniser2.c
@@ -25,6 +25,7 @@ typedef struct context {
 
 	const char *last_start_tag;
 	struct array_list *content_model;
+	bool process_cdata;
 } context;
 
 static void run_test(context *ctx);
@@ -97,6 +98,9 @@ int main(int argc, char **argv)
 			} else if (strcmp(key, "contentModelFlags") == 0) {
 				ctx.content_model =
 						json_object_get_array(val);
+			} else if (strcmp(key, "processCDATA") == 0) {
+				ctx.process_cdata =
+					json_object_get_boolean(val);
 			}
 		}
 
@@ -153,6 +157,13 @@ void run_test(context *ctx)
 			assert(hubbub_tokeniser_run(tok) == HUBBUB_OK);
 		}
 
+		if (ctx->process_cdata) {
+			params.process_cdata = ctx->process_cdata;
+			assert(hubbub_tokeniser_setopt(tok,
+					HUBBUB_TOKENISER_PROCESS_CDATA,
+					&params) == HUBBUB_OK);
+		}
+
 		params.buffer_handler.handler = buffer_handler;
 		params.buffer_handler.pw = ctx;
 		assert(hubbub_tokeniser_setopt(tok,
diff --git a/test/tokeniser3.c b/test/tokeniser3.c
index 7b16ba0..523bfcd 100644
--- a/test/tokeniser3.c
+++ b/test/tokeniser3.c
@@ -25,6 +25,7 @@ typedef struct context {
 
 	const char *last_start_tag;
 	struct array_list *content_model;
+	bool process_cdata;
 } context;
 
 static void run_test(context *ctx);
@@ -97,6 +98,9 @@ int main(int argc, char **argv)
 			} else if (strcmp(key, "contentModelFlags") == 0) {
 				ctx.content_model =
 						json_object_get_array(val);
+			} else if (strcmp(key, "processCDATA") == 0) {
+				ctx.process_cdata =
+					json_object_get_boolean(val);
 			}
 		}
 
@@ -154,6 +158,13 @@ void run_test(context *ctx)
 			assert(hubbub_tokeniser_run(tok) == HUBBUB_OK);
 		}
 
+		if (ctx->process_cdata) {
+			params.process_cdata = ctx->process_cdata;
+			assert(hubbub_tokeniser_setopt(tok,
+					HUBBUB_TOKENISER_PROCESS_CDATA,
+					&params) == HUBBUB_OK);
+		}
+
 		params.buffer_handler.handler = buffer_handler;
 		params.buffer_handler.pw = ctx;
 		assert(hubbub_tokeniser_setopt(tok,