From fafb98623457b4b29bbe3a32af19e901d92106dc Mon Sep 17 00:00:00 2001
From: Adam Dickmeiss <adam@indexdata.dk>
Date: Fri, 5 Jul 2013 13:43:06 +0200
Subject: [PATCH] Honor within type="quoted-literal"

---
 src/filter_http_rewrite.cpp |  146 ++++++++++++++++++++++++++-----------------
 src/filter_http_rewrite.hpp |    4 +-
 src/test_filter_rewrite.cpp |   10 ++-
 3 files changed, 96 insertions(+), 64 deletions(-)

diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp
index 501c87e..80a876a 100644
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_rewrite.cpp
@@ -62,8 +62,11 @@ namespace metaproxy_1 {
             std::string header;
             std::string attr;
             std::string tag;
+            std::string type;
             bool reqline;
             RulePtr rule;
+            bool exec(std::map<std::string, std::string> &vars,
+                      std::string &txt, bool anchor) const;
         };
 
         class HttpRewrite::Content {
@@ -198,7 +201,7 @@ void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
         if (it->reqline)
         {
             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
-            if (it->rule->test_patterns(vars, path, true))
+            if (it->exec(vars, path, true))
             {
                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", path.c_str());
                 hreq->path = odr_strdup(o, path.c_str());
@@ -230,7 +233,7 @@ void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
                 sheader += ": ";
                 sheader += header->value;
 
-                if (it->rule->test_patterns(vars, sheader, true))
+                if (it->exec(vars, sheader, true))
                 {
                     size_t pos = sheader.find(": ");
                     if (pos == std::string::npos)
@@ -255,6 +258,8 @@ void yf::HttpRewrite::Phase::rewrite_body(
     int *content_len,
     std::map<std::string, std::string> & vars) const
 {
+    if (*content_len == 0)
+        return;
     std::list<Content>::const_iterator cit = content_list.begin();
     for (; cit != content_list.end(); cit++)
     {
@@ -267,18 +272,15 @@ void yf::HttpRewrite::Phase::rewrite_body(
     if (cit == content_list.end())
         return;
 
-    if (*content_buf)
-    {
-        int i;
-        for (i = 0; i < *content_len; i++)
-            if ((*content_buf)[i] == 0)
-                return;  // binary content. skip
-
-        std::string content(*content_buf, *content_len);
-        cit->parse(m_verbose, content, vars);
-        *content_buf = odr_strdup(o, content.c_str());
-        *content_len = strlen(*content_buf);
-    }
+    int i;
+    for (i = 0; i < *content_len; i++)
+        if ((*content_buf)[i] == 0)
+            return;  // binary content. skip
+
+    std::string content(*content_buf, *content_len);
+    cit->parse(m_verbose, content, vars);
+    *content_buf = odr_strdup(o, content.c_str());
+    *content_len = strlen(*content_buf);
 }
 
 yf::HttpRewrite::Event::Event(const Content *p,
@@ -383,7 +385,7 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
         if (subst)
         {
             std::string s(value, val_len);
-            it->rule->test_patterns(m_vars, s, true);
+            it->exec(m_vars, s, true);
             wrbuf_puts(m_w, s.c_str());
         }
         else
@@ -413,13 +415,72 @@ void yf::HttpRewrite::Event::text(const char *value, int len)
     if (it != m_content->within_list.end())
     {
         std::string s(value, len);
-        it->rule->test_patterns(m_vars, s, false);
+        it->exec(m_vars, s, false);
         wrbuf_puts(m_w, s.c_str());
     }
     else
         wrbuf_write(m_w, value, len);
 }
 
+static bool embed_quoted_literal(
+    std::string &content,
+    std::map<std::string, std::string> &vars,
+    mp::filter::HttpRewrite::RulePtr ruleptr)
+{
+    bool replace = false;
+    std::string res;
+    const char *cp = content.c_str();
+    const char *cp0 = cp;
+    while (*cp)
+    {
+        if (*cp == '"' || *cp == '\'')
+        {
+            int m = *cp;
+            cp++;
+            res.append(cp0, cp - cp0);
+            cp0 = cp;
+            while (*cp)
+            {
+                if (cp[-1] != '\\' && *cp == m)
+                    break;
+                if (*cp == '\n')
+                    break;
+                cp++;
+            }
+            if (!*cp)
+                break;
+            std::string s(cp0, cp - cp0);
+            if (ruleptr->test_patterns(vars, s, true))
+                replace = true;
+            cp0 = cp;
+            res.append(s);
+        }
+        else if (*cp == '/' && cp[1] == '/')
+        {
+            while (cp[1] && cp[1] != '\n')
+                cp++;
+        }
+        cp++;
+    }
+    res.append(cp0, cp - cp0);
+    content = res;
+    return replace;
+}
+
+bool yf::HttpRewrite::Within::exec(
+    std::map<std::string, std::string> & vars,
+    std::string & txt, bool anchor) const
+{
+    if (type == "quoted-literal")
+    {
+        return embed_quoted_literal(txt, vars, rule);
+    }
+    else
+    {
+        return rule->test_patterns(vars, txt, anchor);
+    }
+}
+
 bool yf::HttpRewrite::Rule::test_patterns(
     std::map<std::string, std::string> & vars,
     std::string & txt, bool anchor)
@@ -622,43 +683,9 @@ void yf::HttpRewrite::Content::quoted_literal(
     std::string &content,
     std::map<std::string, std::string> &vars) const
 {
-    std::string res;
-    const char *cp = content.c_str();
-    const char *cp0 = cp;
-    while (*cp)
-    {
-        if (*cp == '"' || *cp == '\'')
-        {
-            int m = *cp;
-            cp++;
-            res.append(cp0, cp - cp0);
-            cp0 = cp;
-            while (*cp)
-            {
-                if (cp[-1] != '\\' && *cp == m)
-                    break;
-                if (*cp == '\n')
-                    break;
-                cp++;
-            }
-            if (!*cp)
-                break;
-            std::list<Within>::const_iterator it = within_list.begin();
-            std::string s(cp0, cp - cp0);
-            if (it != within_list.end())
-                it->rule->test_patterns(vars, s, true);
-            cp0 = cp;
-            res.append(s);
-        }
-        else if (*cp == '/' && cp[1] == '/')
-        {
-            while (cp[1] && cp[1] != '\n')
-                cp++;
-        }
-        cp++;
-    }
-    res.append(cp0, cp - cp0);
-    content = res;
+    std::list<Within>::const_iterator it = within_list.begin();
+    if (it != within_list.end())
+        embed_quoted_literal(content, vars, it->rule);
 }
 
 void yf::HttpRewrite::Content::configure(
@@ -670,9 +697,9 @@ void yf::HttpRewrite::Content::configure(
             continue;
         if (!strcmp((const char *) ptr->name, "within"))
         {
-            static const char *names[6] =
-                { "header", "attr", "tag", "rule", "reqline", 0 };
-            std::string values[5];
+            static const char *names[7] =
+                { "header", "attr", "tag", "rule", "reqline", "type", 0 };
+            std::string values[6];
             mp::xml::parse_attr(ptr, names, values);
             Within w;
             w.header = values[0];
@@ -686,6 +713,13 @@ void yf::HttpRewrite::Content::configure(
                      "' in http_rewrite filter");
             w.rule = it->second;
             w.reqline = values[4] == "1";
+            w.type = values[5];
+            if (w.type.empty() || w.type == "quoted-literal")
+                ;
+            else
+                throw mp::filter::FilterException
+                    ("within type must be quoted-literal or none in "
+                     " in http_rewrite filter");
             within_list.push_back(w);
         }
     }
diff --git a/src/filter_http_rewrite.hpp b/src/filter_http_rewrite.hpp
index 85a66a2..bbff492 100644
--- a/src/filter_http_rewrite.hpp
+++ b/src/filter_http_rewrite.hpp
@@ -31,14 +31,14 @@ namespace metaproxy_1 {
             class Phase;
             class Content;
             class Replace;
-            class Rule;
             class Within;
             class Event;
-            typedef boost::shared_ptr<Rule> RulePtr;
             boost::scoped_ptr<Phase> req_phase;
             boost::scoped_ptr<Phase> res_phase;
             void configure_phase(const xmlNode *ptr, Phase &phase);
         public:
+            class Rule;
+            typedef boost::shared_ptr<Rule> RulePtr;
             HttpRewrite();
             ~HttpRewrite();
             void process(metaproxy_1::Package & package) const;
diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp
index 00065a4..f26feae 100644
--- a/src/test_filter_rewrite.cpp
+++ b/src/test_filter_rewrite.cpp
@@ -80,7 +80,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "  </content>\n"
             "  <content type=\"html\" mime=\"text/xml|text/html\">\n"
             "    <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
-            "    <within tag=\"script\" attr=\"#text\" rule=\"url\"/>\n"
+            "    <within tag=\"script\" attr=\"#text\" type=\"quoted-literal\" rule=\"url\"/>\n"
             "    <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
             "    <within attr=\"href,src\" rule=\"url\"/>\n"
             "  </content>\n"
@@ -115,7 +115,6 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 
         const char *resp_buf =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 441\r\n"
             "Content-Type: text/html\r\n"
             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
             "Link: </dir/file.xml>; rel=relative\r\n"
@@ -128,7 +127,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "</style>"
             "</head>"
             "<script>var jslink=\"http://targetsite/webservice.xml\";"
-            "var some=foo;"
+            "var some=\"foo\"; foo=1;"
             "</script>"
             "<body>"
             "<p>Welcome to our website. It doesn't make it easy to get pro"
@@ -144,7 +143,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 
         const char *resp_expected =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 564\r\n"
+            "Content-Length: 573\r\n"
             "Content-Type: text/html\r\n"
             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
             "Link: </dir/file.xml>; rel=relative\r\n"
@@ -157,7 +156,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "</style>"
             "</head>"
             "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
-            "var some=bar;"
+            "var some=\"bar\"; foo=1;"
             "</script>"
             "<body>"
             "<p>Welcome to our website. It doesn't make it easy to get pro"
@@ -301,7 +300,6 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
 
         const char *resp_buf =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 140\r\n"
             "Content-Type: application/javascript\r\n"
             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
             "Link: </dir/file.xml>; rel=relative\r\n"
-- 
1.7.10.4