summaryrefslogtreecommitdiff
path: root/tests/linkextractor_tests.c
diff options
context:
space:
mode:
authorAntti Ajanki <antti.ajanki@iki.fi>2013-08-06 09:07:49 +0300
committerAntti Ajanki <antti.ajanki@iki.fi>2013-08-06 09:07:49 +0300
commit7c81286a59639e139ac7e947378be24410701a5e (patch)
tree88e43b758dc2330e8711ebae80eee0039cc57322 /tests/linkextractor_tests.c
downloadvdr-plugin-webvideo-7c81286a59639e139ac7e947378be24410701a5e.tar.gz
vdr-plugin-webvideo-7c81286a59639e139ac7e947378be24410701a5e.tar.bz2
import to vdr-developer repo
Diffstat (limited to 'tests/linkextractor_tests.c')
-rw-r--r--tests/linkextractor_tests.c158
1 files changed, 158 insertions, 0 deletions
diff --git a/tests/linkextractor_tests.c b/tests/linkextractor_tests.c
new file mode 100644
index 0000000..2aa9166
--- /dev/null
+++ b/tests/linkextractor_tests.c
@@ -0,0 +1,158 @@
+#include <string.h>
+#include "linkextractor_tests.h"
+
+#define LINK_TEMPLATE_PATH TEST_DATA_DIR "/links"
+
+#define BASEURL "http://example.com/index.html"
+
+#define HTML1_HREF "http://example.com/test/link"
+#define HTML1_TITLE "Test link"
+#define HTML1 "<html><body>" \
+ "<a href=\"" HTML1_HREF "\">" HTML1_TITLE "</a>" \
+ "</body></html>"
+
+#define HTML2_HREF "http://example.com/test/link"
+#define HTML2_TITLE "Test link"
+#define HTML2 "<html><body>" \
+ "<a href=\"" HTML2_HREF "\">" HTML2_TITLE "</a>" \
+ "<a href=\"http://example.com/index.html\">This is an unrecognized link</a>" \
+ "</body></html>"
+
+#define HTML3_HEADER "<html><body>"
+#define HTML3_HREF "http://example.com/test/link"
+#define HTML3_TITLE "Test link"
+#define HTML3_BODY "<a href=\"" HTML3_HREF "\">" HTML3_TITLE "</a>"
+#define HTML3_FOOTER "</body></html>"
+#define HTML_INVALID "<a hrefxxx=0</a>"
+
+#define HTML4_HREF "/test/link"
+#define HTML4_HREF_ABSOLUTE "http://example.com" HTML4_HREF
+#define HTML4_TITLE "Relative link"
+#define HTML4 "<html><body>" \
+ "<a href=\"" HTML4_HREF "\">" HTML4_TITLE "</a>" \
+ "</body></html>"
+
+#define HTML5_HREF "http://example.com/test/link"
+#define HTML5_TITLE "Test link"
+#define HTML5 "<html><body>" \
+ "<a href=\"" HTML5_HREF "\"><span><b> Test</b></span> <span>link</span></a>" \
+ "</body></html>"
+
+
+void link_extractor_fixture_setup(LinkExtractorFixture *fixture,
+ gconstpointer test_data) {
+ fixture->templates = link_templates_create();
+ g_assert(fixture->templates != NULL);
+ link_templates_load(fixture->templates, LINK_TEMPLATE_PATH);
+
+ fixture->extractor = link_extractor_create(fixture->templates, BASEURL);
+ g_assert(fixture->extractor != NULL);
+}
+
+void link_extractor_fixture_teardown(LinkExtractorFixture *fixture,
+ gconstpointer test_data) {
+ link_extractor_delete(fixture->extractor);
+ link_templates_delete(fixture->templates);
+}
+
+void test_link_extractor_extract(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML1, strlen(HTML1));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 1);
+ const struct Link *link = g_ptr_array_index(links, 0);
+ const char *href = link_get_href(link);
+ g_assert(href);
+ g_assert(strcmp(href, HTML1_HREF) == 0);
+ const char *title = link_get_title(link);
+ g_assert(title);
+ g_assert(strcmp(title, HTML1_TITLE) == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_unrecognized_link(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML2, strlen(HTML2));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 1);
+ const struct Link *link = g_ptr_array_index(links, 0);
+ const char *href = link_get_href(link);
+ g_assert(href);
+ g_assert(strcmp(href, HTML2_HREF) == 0);
+ const char *title = link_get_title(link);
+ g_assert(title);
+ g_assert(strcmp(title, HTML2_TITLE) == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_append(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML3_HEADER, strlen(HTML3_HEADER));
+ link_extractor_append(fixture->extractor, HTML3_BODY, strlen(HTML3_BODY));
+ link_extractor_append(fixture->extractor, HTML3_FOOTER, strlen(HTML3_FOOTER));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 1);
+ const struct Link *link = g_ptr_array_index(links, 0);
+ const char *href = link_get_href(link);
+ g_assert(href);
+ g_assert(strcmp(href, HTML3_HREF) == 0);
+ const char *title = link_get_title(link);
+ g_assert(title);
+ g_assert(strcmp(title, HTML3_TITLE) == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_invalid_html(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML_INVALID, strlen(HTML_INVALID));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_relative_urls(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML4, strlen(HTML4));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 1);
+ const struct Link *link = g_ptr_array_index(links, 0);
+ const char *href = link_get_href(link);
+ g_assert(href);
+ g_assert(strcmp(href, HTML4_HREF_ABSOLUTE) == 0);
+ const char *title = link_get_title(link);
+ g_assert(title);
+ g_assert(strcmp(title, HTML4_TITLE) == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_html_title(LinkExtractorFixture *fixture,
+ G_GNUC_UNUSED gconstpointer test_data) {
+ GPtrArray *links;
+ link_extractor_append(fixture->extractor, HTML5, strlen(HTML5));
+ links = link_extractor_get_links(fixture->extractor);
+ g_assert(links);
+ g_assert(links->len == 1);
+ const struct Link *link = g_ptr_array_index(links, 0);
+ const char *href = link_get_href(link);
+ g_assert(href);
+ g_assert(strcmp(href, HTML5_HREF) == 0);
+ const char *title = link_get_title(link);
+ g_assert(title);
+ g_assert(strcmp(title, HTML5_TITLE) == 0);
+ g_ptr_array_free(links, TRUE);
+}
+
+void test_link_extractor_xml(LinkExtractorFixture *fixture,
+ gconstpointer test_data) {
+
+}
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-24 16:09:32 +0000