From 3add6b2f9e87258e06ecdf0120529a096e9ea8d3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Sat, 20 Aug 2022 12:45:46 +0200
Subject: Test XML parsing.

---
 c/test.cpp | 58 +++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 55 insertions(+), 3 deletions(-)

diff --git a/c/test.cpp b/c/test.cpp
index 8d43a46..78f7017 100644
--- a/c/test.cpp
+++ b/c/test.cpp
@@ -1,5 +1,8 @@
 #include <windows.h>
 #include <wininet.h>
+#include <libxml/HTMLparser.h>
+#include <libxml/HTMLtree.h>
+#include <libxml/xpath.h>
 
 #include "data.h"
 #include "episodelistview.h"
@@ -226,9 +229,10 @@ TESTS
 	TEST(Internet)
 	{
 		HINTERNET hi, hiUrl;
-		unsigned char buf[2048];
+		static unsigned char buf[8'388'608] = {0};
 		DWORD cbRead;
 
+		/* Download HTML. */
 		hi = InternetOpen(L"Episode Browser",
 		    INTERNET_OPEN_TYPE_DIRECT, nullptr, nullptr,
 		    /*INTERNET_FLAG_ASYNC*/0);
@@ -240,10 +244,58 @@ TESTS
 		if (!hiUrl)
 			goto b;
 
-		if (!InternetReadFile(hiUrl, &buf, sizeof(buf), &cbRead))
-			goto c;
+		cbRead = 1;
+		while (cbRead)
+			if (!InternetReadFile(hiUrl, &buf, sizeof(buf), &cbRead))
+				goto c;
 
+		//printf("%s\n", buf);
+
+		InternetCloseHandle(hiUrl);
+		InternetCloseHandle(hi);
+
+		/* Parse HTML. */
+		LIBXML_TEST_VERSION;
+		htmlDocPtr doc;
+		doc = htmlReadMemory(reinterpret_cast<const char*>(buf), sizeof(buf),
+		    "https://www.detectiveconanworld.com/wiki/Anime", nullptr,
+		    HTML_PARSE_RECOVER|HTML_PARSE_NOERROR|HTML_PARSE_NOWARNING);
+		if (!doc)
+			goto z;
+
+		xmlXPathContextPtr xpathCtx;
+		xmlXPathObjectPtr xpathObj;
+
+		xpathCtx = xmlXPathNewContext(doc);
+		if (!xpathCtx)
+			goto y;
+
+		xpathObj = xmlXPathEvalExpression(
+		    reinterpret_cast<const xmlChar*>("//tr"),
+		    xpathCtx);
+		if (!xpathObj)
+			goto x;
+
+		xmlNodeSetPtr nodes;
+		int cNodes;
+		nodes = xpathObj->nodesetval;
+		cNodes = nodes? nodes->nodeNr: 0;
+
+		printf("%d nodes\n", cNodes);
+		for (int i = 0; i < cNodes; i++) {
+			xmlNodePtr node = nodes->nodeTab[i];
+			printf("node \"%s\": type %d\n", node->name, node->type);
+		}
+
+		xmlXPathFreeObject(xpathObj);
+		xmlXPathFreeContext(xpathCtx);
+		xmlFreeDoc(doc);
 		return;
+
+	x:	xmlXPathFreeContext(xpathCtx);
+	y:	xmlFreeDoc(doc);
+	z:	FAIL(xmlGetLastError()->message);
+
 	c:	InternetCloseHandle(hiUrl);
 	b:	InternetCloseHandle(hi);
 	a:	FAIL(Win32Error{}.what());
-- 
cgit v1.2.3