[Boost-commit] svn:boost r86687 - in trunk/tools/quickbook: src test

Subject: [Boost-commit] svn:boost r86687 - in trunk/tools/quickbook: src test
From: dnljms_at_[hidden]
Date: 2013-11-13 16:49:05

Author: danieljames
Date: 2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)
New Revision: 86687
URL: http://svn.boost.org/trac/boost/changeset/86687

Log:
Validate and process link values.

Text files modified: 
   trunk/tools/quickbook/src/actions.cpp         |    51 ++++++++++++++++++++++++++++++++++++--- 
   trunk/tools/quickbook/src/utils.cpp           |    30 +++++++++++++++++++---                  
   trunk/tools/quickbook/src/utils.hpp           |     7 +++++                                   
   trunk/tools/quickbook/test/link-1_7.gold      |     3 --                                      
   trunk/tools/quickbook/test/link-1_7.quickbook |     3 --                                      
   5 files changed, 79 insertions(+), 15 deletions(-)

Modified: trunk/tools/quickbook/src/actions.cpp
==============================================================================
--- trunk/tools/quickbook/src/actions.cpp	Wed Nov 13 16:48:40 2013	(r86686)
+++ trunk/tools/quickbook/src/actions.cpp	2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)	(r86687)
@@ -73,6 +73,38 @@
             state.anchors.push_back(placeholder);
             return placeholder;
         }
+
+        std::string validate_id(quickbook::state& state,
+                quickbook::value const& id_value)
+        {
+            bool valid = true;
+            std::string id = id_value.is_encoded() ?
+                id_value.get_encoded() : detail::to_s(id_value.get_quickbook());
+
+            // Special case since I use dollar ids for id placeholders.
+            if (id[0] == '$') { valid = false; id[0] = '_'; }
+
+            if (qbk_version_n >= 107u) {
+                char const* allowed_punctuation = "_.-";
+
+                BOOST_FOREACH(char c, id) {
+                    if (!std::isalnum(c) &&
+                            !std::strchr(allowed_punctuation, c))
+                        valid = false;
+                }
+            }
+
+            if (!valid) {
+                detail::outerr(id_value.get_file(), id_value.get_position())
+                    << "Invalid id: "
+                    << (id_value.is_encoded() ? id_value.get_encoded() :
+                        detail::to_s(id_value.get_quickbook()))
+                    << std::endl;
+                ++state.error_count;
+            }
+
+            return id;
+        }
     }
 
     bool quickbook_range::in_range() const {
@@ -1511,10 +1543,21 @@
         value content = values.consume();
         values.finish();
 
-        // Note: dst is never actually encoded as boostbook, which
-        // is why the result is called with 'print_string' later.
-        std::string dst = dst_value.is_encoded() ?
-            dst_value.get_encoded() : detail::to_s(dst_value.get_quickbook());
+        std::string dst;
+
+        if (link.get_tag() == phrase_tags::link) {
+            dst = validate_id(state, dst_value);
+        }
+        else {
+            dst = dst_value.is_encoded() ?
+                dst_value.get_encoded() :
+                detail::to_s(dst_value.get_quickbook());
+
+            // TODO: Might be better to have an error for some invalid urls.
+            if (link.get_tag() == phrase_tags::url) {
+                dst = detail::partially_escape_uri(dst);
+            }
+        } 
         
         state.phrase << markup.pre;
         detail::print_string(dst, state.phrase.get());

Modified: trunk/tools/quickbook/src/utils.cpp
==============================================================================
--- trunk/tools/quickbook/src/utils.cpp	Wed Nov 13 16:48:40 2013	(r86686)
+++ trunk/tools/quickbook/src/utils.cpp	2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)	(r86687)
@@ -66,25 +66,45 @@
         return static_cast<char>(std::tolower(static_cast<unsigned char>(ch)));
     }
 
-    std::string escape_uri(std::string uri_param)
+    static std::string escape_uri_impl(std::string& uri_param, char const* mark)
     {
+        // Extra capital characters for validating percent escapes.
+        static char const hex[] = "0123456789abcdefABCDEF";
+
         std::string uri;
         uri.swap(uri_param);
 
         for (std::string::size_type n = 0; n < uri.size(); ++n)
         {
-            static char const mark[] = "-_.!~*'()?\\/";
-            if((!std::isalnum(static_cast<unsigned char>(uri[n])) || 127 < static_cast<unsigned char>(uri[n]))
-              && 0 == std::strchr(mark, uri[n]))
+            if (static_cast<unsigned char>(uri[n]) > 127 ||
+                    (!std::isalnum(static_cast<unsigned char>(uri[n])) &&
+                     !std::strchr(mark, uri[n])) ||
+                    (uri[n] == '%' && !(n + 2 < uri.size() &&
+                                        std::strchr(hex, uri[n+1]) &&
+                                        std::strchr(hex, uri[n+2]))))
             {
-                static char const hex[] = "0123456789abcdef";
                 char escape[] = { hex[uri[n] / 16], hex[uri[n] % 16] };
                 uri.insert(n + 1, escape, 2);
                 uri[n] = '%';
                 n += 2;
             }
+            else if (uri[n] == '%')
+            {
+                n += 2;
+            }
         }
 
         return uri;
     }
+
+    std::string escape_uri(std::string uri_param)
+    {
+        // TODO: I don't understand this choice of characters.....
+        return escape_uri_impl(uri_param, "-_.!~*'()?\\/");
+    }
+
+    std::string partially_escape_uri(std::string uri_param)
+    {
+        return escape_uri_impl(uri_param, "-_.!~*'()?\\/:&=#%");
+    }
 }}

Modified: trunk/tools/quickbook/src/utils.hpp
==============================================================================
--- trunk/tools/quickbook/src/utils.hpp	Wed Nov 13 16:48:40 2013	(r86686)
+++ trunk/tools/quickbook/src/utils.hpp	2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)	(r86687)
@@ -34,11 +34,18 @@
         return out_name;
     }
 
+    // URI escape string
     std::string escape_uri(std::string uri);
     inline std::string escape_uri(boost::string_ref uri) {
         return escape_uri(std::string(uri.begin(), uri.end()));
     }
 
+    // URI escape string, leaving characters generally used in URIs.
+    std::string partially_escape_uri(std::string uri);
+    inline std::string partially_escape_uri(boost::string_ref uri) {
+        return escape_uri(std::string(uri.begin(), uri.end()));
+    }
+
     inline std::string to_s(boost::string_ref x) {
         return std::string(x.begin(), x.end());
     }

Modified: trunk/tools/quickbook/test/link-1_7.gold
==============================================================================
--- trunk/tools/quickbook/test/link-1_7.gold	Wed Nov 13 16:48:40 2013	(r86686)
+++ trunk/tools/quickbook/test/link-1_7.gold	2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)	(r86687)
@@ -21,9 +21,6 @@
     <para>
       <link linkend="link">description</link>
     </para>
-    <para>
-      <link linkend="link[Hello]">description</link>
-    </para>
   </section>
   <section id="link_tests.side_by_side_links">
     <title><link linkend="link_tests.side_by_side_links">Side-by-side links</link></title>

Modified: trunk/tools/quickbook/test/link-1_7.quickbook
==============================================================================
--- trunk/tools/quickbook/test/link-1_7.quickbook	Wed Nov 13 16:48:40 2013	(r86686)
+++ trunk/tools/quickbook/test/link-1_7.quickbook	2013-11-13 16:49:05 EST (Wed, 13 Nov 2013)	(r86687)
@@ -28,9 +28,6 @@
 
 [link link[/ comment]description]
 
-[link link\[Hello\] description]
-
-
 [endsect]
 
 [section Side-by-side links]