tests/tests_HTMLparser_areBlanks.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>

#include <libxml/tree.h>
#include <libxml/parser.h>
#include <libxml/xmlmemory.h>
#include <string.h>
#include <stdlib.h>

/* Wrapper for the static function provided in the module */
extern int test_areBlanks(htmlParserCtxtPtr ctxt, const xmlChar *str, int len);

static htmlParserCtxtPtr make_ctx_with_input(const char *s) {
    return htmlCreateMemoryParserCtxt(s, (int)strlen(s));
}

void setUp(void) {
    xmlInitParser();
}

void tearDown(void) {
    /* No global cleanup here to avoid interfering between tests */
}

/* Non-whitespace in str should return -1, independent of ctxt */
void test_areBlanks_non_whitespace_returns_minus1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("");
    TEST_ASSERT_NOT_NULL(ctxt);

    const xmlChar *str = BAD_CAST "a";
    int res = test_areBlanks(ctxt, str, 1);
    TEST_ASSERT_EQUAL_INT(-1, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR == 0 (empty input) returns 1 for all-whitespace str */
void test_areBlanks_empty_input_cur0_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("");
    TEST_ASSERT_NOT_NULL(ctxt);

    const xmlChar *str = BAD_CAST " \t";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(1, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR != '<' returns 0 for all-whitespace str */
void test_areBlanks_cur_not_lt_returns0(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("x");
    TEST_ASSERT_NOT_NULL(ctxt);

    const xmlChar *str = BAD_CAST " \n";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(0, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR == '<' and name == NULL returns 1 */
void test_areBlanks_cur_lt_name_null_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = NULL;

    const xmlChar *str = BAD_CAST " \t";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(1, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR == '<' and name == "html" returns 1 */
void test_areBlanks_name_html_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "html";

    const xmlChar *str = BAD_CAST "  ";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(1, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR == '<' and name == "head" returns 1 */
void test_areBlanks_name_head_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "head";

    const xmlChar *str = BAD_CAST "\n\t ";
    int res = test_areBlanks(ctxt, str, 3);
    TEST_ASSERT_EQUAL_INT(1, res);

    htmlFreeParserCtxt(ctxt);
}

/* CUR == '<', name == "body", and HTML 4.01 DTD returns 1 */
void test_areBlanks_name_body_html401_dtd_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "body";

    xmlDocPtr doc = xmlNewDoc(BAD_CAST "1.0");
    TEST_ASSERT_NOT_NULL(doc);
    xmlCreateIntSubset(doc, BAD_CAST "html",
                       BAD_CAST "-//W3C//DTD HTML 4.01//EN",
                       NULL);
    ctxt->myDoc = doc;

    const xmlChar *str = BAD_CAST " \t ";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(1, res);

    xmlFreeDoc(doc);
    htmlFreeParserCtxt(ctxt);
}

/* node == NULL returns 0 (after previous early checks) */
void test_areBlanks_node_null_returns0(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "nothtml";
    ctxt->myDoc = NULL;
    ctxt->node = NULL;

    const xmlChar *str = BAD_CAST " \n";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(0, res);

    htmlFreeParserCtxt(ctxt);
}

/* lastChild == NULL and node is non-element with content -> returns 0 */
void test_areBlanks_lastChild_null_non_element_with_content_returns0(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "x_test_tag";

    xmlNodePtr textNode = xmlNewText(BAD_CAST "abc");
    TEST_ASSERT_NOT_NULL(textNode);
    ctxt->node = textNode;

    const xmlChar *str = BAD_CAST "  ";
    int res = test_areBlanks(ctxt, str, (int)xmlStrlen(str));
    TEST_ASSERT_EQUAL_INT(0, res);

    xmlFreeNode(textNode);
    htmlFreeParserCtxt(ctxt);
}

/* lastChild is a text node -> returns 0 */
void test_areBlanks_lastChild_is_text_returns0(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "x_test_tag";

    xmlDocPtr doc = xmlNewDoc(BAD_CAST "1.0");
    TEST_ASSERT_NOT_NULL(doc);
    xmlNodePtr root = xmlNewNode(NULL, BAD_CAST "div");
    TEST_ASSERT_NOT_NULL(root);
    xmlDocSetRootElement(doc, root);
    xmlAddChild(root, xmlNewText(BAD_CAST "hi"));
    ctxt->myDoc = doc;
    ctxt->node = root;

    const xmlChar *str = BAD_CAST " ";
    int res = test_areBlanks(ctxt, str, 1);
    TEST_ASSERT_EQUAL_INT(0, res);

    xmlFreeDoc(doc);
    htmlFreeParserCtxt(ctxt);
}

/* lastChild is a comment only; it is skipped -> behaves like lastChild == NULL; expect 1 */
void test_areBlanks_comment_only_child_skipped_then_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "unlikely_tag_name_123";

    xmlDocPtr doc = xmlNewDoc(BAD_CAST "1.0");
    TEST_ASSERT_NOT_NULL(doc);
    xmlNodePtr root = xmlNewNode(NULL, BAD_CAST "container");
    TEST_ASSERT_NOT_NULL(root);
    xmlDocSetRootElement(doc, root);

    xmlNodePtr comment = xmlNewComment(BAD_CAST "just a comment");
    TEST_ASSERT_NOT_NULL(comment);
    xmlAddChild(root, comment);

    ctxt->myDoc = doc;
    ctxt->node = root;

    const xmlChar *str = BAD_CAST "\t ";
    int res = test_areBlanks(ctxt, str, 2);
    TEST_ASSERT_EQUAL_INT(1, res);

    xmlFreeDoc(doc);
    htmlFreeParserCtxt(ctxt);
}

/* lastChild is an element with a name unlikely in allowPCData -> expect 1 */
void test_areBlanks_lastChild_element_not_in_pcdata_returns1(void) {
    htmlParserCtxtPtr ctxt = make_ctx_with_input("<");
    TEST_ASSERT_NOT_NULL(ctxt);
    ctxt->name = BAD_CAST "unlikely_parent_name_456";

    xmlDocPtr doc = xmlNewDoc(BAD_CAST "1.0");
    TEST_ASSERT_NOT_NULL(doc);
    xmlNodePtr root = xmlNewNode(NULL, BAD_CAST "rootx");
    TEST_ASSERT_NOT_NULL(root);
    xmlDocSetRootElement(doc, root);

    xmlNodePtr child = xmlNewNode(NULL, BAD_CAST "zzz_not_pcdata");
    TEST_ASSERT_NOT_NULL(child);
    xmlAddChild(root, child);

    ctxt->myDoc = doc;
    ctxt->node = root;

    const xmlChar *str = BAD_CAST " ";
    int res = test_areBlanks(ctxt, str, 1);
    TEST_ASSERT_EQUAL_INT(1, res);

    xmlFreeDoc(doc);
    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_areBlanks_non_whitespace_returns_minus1);
    RUN_TEST(test_areBlanks_empty_input_cur0_returns1);
    RUN_TEST(test_areBlanks_cur_not_lt_returns0);
    RUN_TEST(test_areBlanks_cur_lt_name_null_returns1);
    RUN_TEST(test_areBlanks_name_html_returns1);
    RUN_TEST(test_areBlanks_name_head_returns1);
    RUN_TEST(test_areBlanks_name_body_html401_dtd_returns1);
    RUN_TEST(test_areBlanks_node_null_returns0);
    RUN_TEST(test_areBlanks_lastChild_null_non_element_with_content_returns0);
    RUN_TEST(test_areBlanks_lastChild_is_text_returns0);
    RUN_TEST(test_areBlanks_comment_only_child_skipped_then_returns1);
    RUN_TEST(test_areBlanks_lastChild_element_not_in_pcdata_returns1);
    return UNITY_END();
}