From 6832f91958e6c2cc44a8c4a4e126ea20b8c8d5a3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Tue, 30 Aug 2022 02:19:56 +0200
Subject: Fix Unique Good, Bad. Simplify HTML parser.

Good and Bad should obviously do the exact opposite thing
to what they did.
---
 c/data.cpp | 76 +++++++++++++++++++++++++++++---------------------------------
 1 file changed, 36 insertions(+), 40 deletions(-)

(limited to 'c/data.cpp')

diff --git a/c/data.cpp b/c/data.cpp
index df5627a..7ae6ad1 100644
--- a/c/data.cpp
+++ b/c/data.cpp
@@ -14,50 +14,44 @@
 static Unique<HINTERNET, InternetCloseHandle> s_hi =
 	InternetOpenW(L"Episode Browser", INTERNET_OPEN_TYPE_DIRECT, nullptr, nullptr, 0);
 
-/* ParsedDoc downloads and parses an HTML document. */
-struct ParsedDoc
+Unique<htmlParserCtxtPtr, xmlFreeParserCtxt> RemoteParserCtxt(const wchar_t* wszUrl, const char* szUrl)
 {
-	Unique<HINTERNET, InternetCloseHandle> hiUrl;
-	Unique<htmlParserCtxtPtr, xmlFreeParserCtxt> ctx;
-	char bufI[1024];
+	if (s_hi.Bad(0))
+		throw Win32Error();
+
+	Unique<HINTERNET, InternetCloseHandle> hiUrl = InternetOpenUrlW(
+	    s_hi.v, wszUrl, nullptr, 0, INTERNET_FLAG_NO_UI, 0);
+	if (hiUrl.Bad(0))
+		throw InternetError();
+
 	char bufX[1024];
+	Unique<htmlParserCtxtPtr, xmlFreeParserCtxt> ctx = htmlCreatePushParserCtxt(
+	    nullptr, nullptr, bufX, sizeof(bufX), szUrl, XML_CHAR_ENCODING_UTF8);
+	if (ctx.Bad(0))
+		throw XmlError();
 
-	ParsedDoc(const wchar_t* wszUrl, const char* szUrl)
-	{
-		if (!s_hi.Bad(0))
-			throw Win32Error();
+	htmlCtxtUseOptions(ctx.v, HTML_PARSE_RECOVER|HTML_PARSE_NOERROR|HTML_PARSE_NOWARNING);
 
-		hiUrl = InternetOpenUrlW(s_hi.v, wszUrl, nullptr, 0, INTERNET_FLAG_NO_UI, 0);
-		if (!hiUrl.Bad(0))
+	BOOL r;
+	DWORD cbRead;
+	char bufI[1024];
+	while (r = InternetReadFile(hiUrl.v, bufI, sizeof(bufI), &cbRead), cbRead) {
+		if (!r)
 			throw InternetError();
-
-		ctx = htmlCreatePushParserCtxt(nullptr, nullptr, bufX, sizeof(bufX), szUrl, XML_CHAR_ENCODING_UTF8);
-		if (!ctx.Bad(0))
+		if (!htmlParseChunk(ctx.v, bufI, cbRead, 0))
 			throw XmlError();
-
-		htmlCtxtUseOptions(ctx.v, HTML_PARSE_RECOVER|HTML_PARSE_NOERROR|HTML_PARSE_NOWARNING);
-
-		BOOL r;
-		DWORD cbRead;
-		while (r = InternetReadFile(hiUrl.v, bufI, sizeof(bufI), &cbRead), cbRead) {
-			if (!r)
-				throw InternetError();
-			if (!htmlParseChunk(ctx.v, bufI, cbRead, 0))
-				throw XmlError();
-		}
-		htmlParseChunk(ctx.v, bufI, 0, 1); /* Stop parsing. */
 	}
+	htmlParseChunk(ctx.v, bufI, 0, 1); /* Stop parsing. */
 
-	operator htmlDocPtr() { return ctx.v->myDoc; }
-};
+	return ctx;
+}
 
 static inline void XmlFree(void* p) { xmlFree(p); }
 
 template <size_t N>
-bool WcharsFromXmlchars(wchar_t (&dst)[N], xmlChar* utf8_)
+bool WcharsFromXmlchars(wchar_t (&dst)[N], Unique<xmlChar*, XmlFree> utf8)
 {
-	Unique<xmlChar*, XmlFree> utf8 = utf8_;
-	if (!utf8.Bad(0))
+	if (utf8.Bad(0))
 		throw XmlError();
 
 	/* Truncate if source is larger than destination. */
@@ -167,17 +161,18 @@ void FetchData(unsigned char* sig)
 	 * specific XPath query. This is fragile theoretically, but
 	 * unlikely to break practically. */
 
-	ParsedDoc doc(L"https://www.detectiveconanworld.com/wiki/Anime",
-	    "https://www.detectiveconanworld.com/wiki/Anime");
+	Unique<htmlParserCtxtPtr, xmlFreeParserCtxt> ctx =
+		RemoteParserCtxt(L"https://www.detectiveconanworld.com/wiki/Anime",
+		    "https://www.detectiveconanworld.com/wiki/Anime");
 
-	Unique<xmlXPathContextPtr, xmlXPathFreeContext> xpathCtx = xmlXPathNewContext(doc);
-	if (!xpathCtx.Bad(0))
+	Unique<xmlXPathContextPtr, xmlXPathFreeContext> xpathCtx = xmlXPathNewContext(ctx.v->myDoc);
+	if (xpathCtx.Bad(0))
 		throw XmlError();
 
 	Unique<xmlXPathObjectPtr, xmlXPathFreeObject> xpathObj = xmlXPathEvalExpression(
 	    reinterpret_cast<const xmlChar*>("//tr[./td[1] != '' and ./td[3][@style='background:#f2fde9;']]"),
 	    xpathCtx.v);
-	if (!xpathObj.Bad(0))
+	if (xpathObj.Bad(0))
 		throw XmlError();
 
 	xmlNodeSetPtr nodes = xpathObj.v->nodesetval;
@@ -269,16 +264,17 @@ void FetchScreenwriters(unsigned char* sig)
 		Wcscpy(Buf(url)+Len(prefix), d.wiki);
 
 		/* Retrieve screenwriter from HTML. */
-		ParsedDoc doc(url, nullptr);
-		Unique<xmlXPathContextPtr, xmlXPathFreeContext> xpathCtx = xmlXPathNewContext(doc);
-		if (!xpathCtx.Bad(0))
+
+		Unique<htmlParserCtxtPtr, xmlFreeParserCtxt> ctx = RemoteParserCtxt(url, nullptr);
+		Unique<xmlXPathContextPtr, xmlXPathFreeContext> xpathCtx = xmlXPathNewContext(ctx.v->myDoc);
+		if (xpathCtx.Bad(0))
 			throw XmlError();
 
 		Unique<xmlXPathObjectPtr, xmlXPathFreeObject> xpathObj =
 			xmlXPathEvalExpression(reinterpret_cast<const xmlChar*>(
 			    "//th[contains(text(), 'Screenplay:')]/following-sibling::td"),
 			    xpathCtx.v);
-		if (!xpathObj.Bad(0))
+		if (xpathObj.Bad(0))
 			throw XmlError();
 
 		xmlNodeSetPtr nodes = xpathObj.v->nodesetval;
-- 
cgit v1.2.3