From dd08d816dca127808b2343005ec8a728b6cb2a2a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Anton=20Luka=20=C5=A0ijanec?= <anton@sijanec.eu>
Date: Fri, 5 Jan 2024 16:44:26 +0100
Subject: support for suggested queries and query redirects

---
 src/api.c | 44 +++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 41 insertions(+), 3 deletions(-)

(limited to 'src/api.c')

diff --git a/src/api.c b/src/api.c
index a652e3a..b0ef96d 100644
--- a/src/api.c
+++ b/src/api.c
@@ -127,6 +127,7 @@ int sc_fix_url (char ** h) { /* fixes a (result) URL in-place (removes tracking
 enum sc_return sc_query_google (const char * s, /* breaking change: changed return type */
 		struct sc_cache * c,
 		struct sc_query * q,
+		char ** redirect, /* variable redirect will be set to a heap allocated string that must be freed by the caller if the upstream returned results for a different query. in that case the returned query object will be for a different search string! -- if NULL, request that upstream does not enable "results for" feature */
 		SC_OPT_TYPE opt) { /* check4cachedB4 */
 	/* query is in most cases NULL. then it will be allocated and put into sc_cache. otherwise response will be put into passed q. */
 	/* if query is not NULL, it MUST be initialized */
@@ -155,6 +156,9 @@ enum sc_return sc_query_google (const char * s, /* breaking change: changed retu
 	htmlDocPtr xmldoc = NULL;
 	char * txtdoc = NULL;
 	int qwasgiven = 0;
+	SC_LOG(SC_LOG_DEBUG, c, "%s called, redirect is %p", __func__, redirect);
+	if (redirect)
+		*redirect = NULL;
 	if (!s || !c) {
 		rs = SC_BADCALL;
 		goto rc;
@@ -166,7 +170,7 @@ enum sc_return sc_query_google (const char * s, /* breaking change: changed retu
 		qwasgiven++;
 	char * us = malloc(sizeof(char)*strlen(s)*3+1);
 	urlencode(us, s);
-	txtdoc = SC_CAPI(c, NULL, NULL, "http://wap.google.com/search?q=%s&num=100&ie=UTF-8%s", us, (opt&SC_OPT_IMAGE) ? "&tbm=isch" : "");
+	txtdoc = SC_CAPI(c, NULL, NULL, "http://wap.google.com/search?q=%s&num=100&ie=UTF-8%s%s", us, (opt&SC_OPT_IMAGE) ? "&tbm=isch" : "", redirect ? "" : "&nfpr=1");
 	// fprintf(stdout, "%s\n", txtdoc);
 	free(us);
 	if (!txtdoc) {
@@ -178,6 +182,7 @@ enum sc_return sc_query_google (const char * s, /* breaking change: changed retu
 		rs = SC_CAPTCHA;
 		goto rc;
 	}
+	char * resultsforclass = sc_find_class(txtdoc, "{color:#1967d2}");
 	if (opt & SC_OPT_IMAGE) {
 		imageclass = sc_find_class(txtdoc, "{font-family:Roboto,Helvetica,Arial,sans-serif}");
 		if (!imageclass) {
@@ -292,11 +297,42 @@ enum sc_return sc_query_google (const char * s, /* breaking change: changed retu
 			SC_CUE(c, c->queries_lock);
 		goto rc;
 	}
+	q->string = realloc(q->string, sl+1);
+	strcpy(q->string, s);
+	char * xpathsugg = NULL;
+	if (resultsforclass) {
+		xpathsugg = malloc(512+strlen(resultsforclass));
+		sprintf(xpathsugg, "//a[contains(@class, '%s')]", resultsforclass);
+		xmlNodePtr suggnode = nthNodeX(xmldoc, xpathsugg, 0);
+		if (suggnode && xmlHasProp(suggnode, BAD_CAST "href")) {
+			char * href = (char *) xmlGetProp(suggnode, BAD_CAST "href");
+			char * content = (char *) xmlNodeGetContent(suggnode);
+			if (href && strstr(href, "&spell=1&"))
+				strcpy((q->suggested = realloc(q->suggested, strlen(content)+1)), content);
+			xmlFree(href);
+			xmlFree(content);
+		} else {
+			free(q->suggested);
+			q->suggested = NULL;
+		}
+	} else {
+		free(q->suggested);
+		q->suggested = NULL;
+	}
+	xmlNodePtr first = nthNodeX(xmldoc, xpathsugg, 1);
+	if (redirect && xpathsugg && q->suggested && xmlHasProp(first, BAD_CAST "href")) {
+		char * href = (char *) xmlGetProp(first, BAD_CAST "href");
+		if (href && strstr(href, "&nfpr=1&")) {
+			*redirect = q->suggested;
+			q->suggested = NULL;
+			q->string = realloc(q->string, strlen(*redirect)+1);
+			strcpy(q->string, *redirect);
+		}
+		xmlFree(href);
+	}
 	q->cache = c;
 	q->lookup_time = time(NULL);
-	q->string = realloc(q->string, sl+1);
 	q->opt |= opt | SC_ENGINE_GOOGLE;
-	strcpy(q->string, s);
 	if (!qwasgiven) {
 		SC_CWLE(c, c->queries_lock);
 #ifdef SC_OLD_STORAGE
@@ -318,6 +354,8 @@ rc:
 	free(titleclass);
 	free(descclass);
 	free(imageclass);
+	free(resultsforclass);
 	free(xpath);
+	free(xpathsugg);
 	return rs;
 }
-- 
cgit v1.2.3