From 288111986392ffbd733d71cd964f49437b899bdb Mon Sep 17 00:00:00 2001
From: Adam Dickmeiss <adam@indexdata.dk>
Date: Thu, 27 Jun 2013 14:33:01 +0200
Subject: [PATCH] HTMLParserEvent, attributes takes quoting sep

as argument, so that quoting (if any) is preserved.
---
 src/filter_http_rewrite.cpp |   11 +++++++----
 src/html_parser.cpp         |   18 ++++++++++++------
 src/html_parser.hpp         |    3 ++-
 src/test_filter_rewrite.cpp |    4 ++--
 src/test_html_parser.cpp    |   15 ++++++++-------
 5 files changed, 31 insertions(+), 20 deletions(-)

diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp
index 293886a..5a4ca83 100644
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_rewrite.cpp
@@ -85,7 +85,8 @@ namespace metaproxy_1 {
             void anyTagEnd(const char *tag, int tag_len, int close_it);
             void attribute(const char *tag, int tag_len,
                            const char *attr, int attr_len,
-                           const char *value, int val_len);
+                           const char *value, int val_len,
+                           const char *sep);
             void closeTag(const char *tag, int tag_len);
             void text(const char *value, int len);
             const Phase *m_phase;
@@ -304,7 +305,8 @@ void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 
 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
                                        const char *attr, int attr_len,
-                                       const char *value, int val_len)
+                                       const char *value, int val_len,
+                                       const char *sep)
 {
     std::list<Within>::const_iterator it = m_phase->within_list.begin();
     bool subst = false;
@@ -332,7 +334,8 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 
     wrbuf_putc(m_w, ' ');
     wrbuf_write(m_w, attr, attr_len);
-    wrbuf_puts(m_w, "=\"");
+    wrbuf_puts(m_w, "=");
+    wrbuf_puts(m_w, sep);
 
     std::string output;
     if (subst)
@@ -344,7 +347,7 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
         wrbuf_write(m_w, value, val_len);
     else
         wrbuf_puts(m_w, output.c_str());
-    wrbuf_puts(m_w, "\"");
+    wrbuf_puts(m_w, sep);
 }
 
 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
diff --git a/src/html_parser.cpp b/src/html_parser.cpp
index 1436553..ddbbbe0 100644
--- a/src/html_parser.cpp
+++ b/src/html_parser.cpp
@@ -44,7 +44,7 @@ namespace metaproxy_1 {
                      const char *cp);
         int skipAttribute(HTMLParserEvent &event,
                           const char *cp, int *attr_len,
-                          const char **value, int *val_len);
+                          const char **value, int *val_len, int *tr);
         Rep();
         ~Rep();
         int m_verbose;
@@ -99,7 +99,8 @@ static int skipName(const char *cp)
 
 int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
                                        const char *cp, int *attr_len,
-                                       const char **value, int *val_len)
+                                       const char **value, int *val_len,
+                                       int *tr)
 {
     int i = skipName(cp);
     *attr_len = i;
@@ -114,9 +115,9 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
         i += skipSpace(cp + i);
         if (cp[i] == '\"' || cp[i] == '\'')
         {
-            char tr = cp[i];
+            *tr = cp[i];
             v0 = ++i;
-            while (cp[i] != tr && cp[i])
+            while (cp[i] != *tr && cp[i])
                 i++;
             v1 = i;
             if (cp[i])
@@ -124,6 +125,7 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
         }
         else
         {
+            *tr = 0;
             v0 = i;
             while (cp[i] && !strchr(SPACECHR ">", cp[i]))
                 i++;
@@ -147,14 +149,18 @@ int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
         int attr_len;
         const char *value;
         int val_len;
-        int nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len);
+        int tr;
+        int nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len, &tr);
         i += nor;
 	if (nor)
 	{
+            char x[2];
+            x[0] = tr;
+            x[1] = 0;
             if (m_verbose)
                 printf ("------ attr %.*s=%.*s\n", attr_len, attr_name,
                         val_len, value);
-            event.attribute(name, len, attr_name, attr_len, value, val_len);
+            event.attribute(name, len, attr_name, attr_len, value, val_len, x);
 	}
         else
         {
diff --git a/src/html_parser.hpp b/src/html_parser.hpp
index 8eeb152..3e10daf 100644
--- a/src/html_parser.hpp
+++ b/src/html_parser.hpp
@@ -29,7 +29,8 @@ namespace metaproxy_1 {
                                    int close_it) = 0;
             virtual void attribute(const char *tag, int tag_len,
                                    const char *attr, int attr_len,
-                                   const char *value, int val_len) = 0;
+                                   const char *value, int val_len,
+                                   const char *sep) = 0;
             virtual void closeTag(const char *tag, int tag_len) = 0;
             virtual void text(const char *value, int len) = 0;
         };
diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp
index 2e4ae54..7df4d0b 100644
--- a/src/test_filter_rewrite.cpp
+++ b/src/test_filter_rewrite.cpp
@@ -146,7 +146,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 
         const char *resp_expected =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 522\r\n"
+            "Content-Length: 520\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"
@@ -164,7 +164,7 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "xified"
             "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
             "  An absolute link</a>"
-            "<a target=\"_blank\" href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
+            "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
             "  Another abs link</a>"
             "<a href=\"/docs/page4.html\"/>"
             "</body></html>";
diff --git a/src/test_html_parser.cpp b/src/test_html_parser.cpp
index cdb266c..6f5c134 100644
--- a/src/test_html_parser.cpp
+++ b/src/test_html_parser.cpp
@@ -42,16 +42,17 @@ public:
     void openTagStart(const char *tag, int tag_len) {
         out += "<";
         out.append(tag, tag_len);
-    } 
-    
+    }
+
     void attribute(const char *tag, int tag_len,
                    const char *attr, int attr_len,
-                   const char *value, int val_len) {
+                   const char *value, int val_len, const char *sep) {
         out += " ";
         out.append(attr, attr_len);
-        out += "=\"";
+        out += "=";
+        out += sep;
         out.append(value, val_len);
-        out += "\"";
+        out += sep;
     }
     void anyTagEnd(const char *tag, int tag_len, int close_it) {
         if (close_it)
@@ -76,7 +77,7 @@ BOOST_AUTO_TEST_CASE( test_html_parser_1 )
             "<html><body><a t1=v1 t2='v2' t3=\"v3\">some text</a>"
             "<hr><table ></table  ><a href=\"x\"/></body></html>";
         const char* expected =
-            "<html><body><a t1=\"v1\" t2=\"v2\" t3=\"v3\">some text</a>"
+            "<html><body><a t1=v1 t2='v2' t3=\"v3\">some text</a>"
             "<hr><table></table  ><a href=\"x\"/></body></html>";
         MyEvent e;
         hp.set_verbose(1);
@@ -88,7 +89,7 @@ BOOST_AUTO_TEST_CASE( test_html_parser_1 )
         std::cout << e.out << std::endl;
         BOOST_CHECK_EQUAL(std::string(expected), e.out);
     }
-    catch (std::exception & e) 
+    catch (std::exception & e)
     {
         std::cout << e.what();
         std::cout << std::endl;
-- 
1.7.10.4