Fixing markup escaping for slice2confluence.

3 files changed, 223 insertions, 52 deletions

diff --git a/cpp/src/slice2confluence/ConfluenceOutput.cpp b/cpp/src/slice2confluence/ConfluenceOutput.cpp
index 92ff6214aa1..2120541d025 100644
--- a/cpp/src/slice2confluence/ConfluenceOutput.cpp
+++ b/cpp/src/slice2confluence/ConfluenceOutput.cpp
@@ -1,6 +1,7 @@
 #include <ConfluenceOutput.h>
 #include <iostream>
 #include <sstream>
+#include <utility>
 #include <string.h>
 
 using namespace std;
@@ -17,6 +18,9 @@ namespace Confluence
 // ConfluenceOutput
 // ----------------------------------------------------------------------
 
+const string Confluence::ConfluenceOutput::TEMP_ESCAPER_START = "$$$$$$$";
+const string Confluence::ConfluenceOutput::TEMP_ESCAPER_END = "!!!!!!!";
+
 Confluence::ConfluenceOutput::ConfluenceOutput() :
     OutputBase(),
     _se(false),
@@ -70,22 +74,85 @@ Confluence::ConfluenceOutput::print(const char* s)
 string 
 Confluence::ConfluenceOutput::escapeComment(string comment)
 {
-    string escapeChars = "\\{}-*|[]"; //backslash ("\\") needs to be first because it gets added later
+    list< pair<unsigned int,unsigned int> > escaperLimits = getMarkerLimits(comment);
+    string escapeChars = "\\{}-*|[]";
+    
+    //for each escape character
     for (string::iterator i = escapeChars.begin(); i < escapeChars.end(); ++i)
     {
         string c(1, *i);
-        string replacement = "\\" + c;
+        string replacement;
+        
+        if (c == "\\")
+        {
+            replacement = "&#92;";
+        }
+        else if (c == "{")
+        {
+            replacement = "&#123;";
+        }
+        else if (c == "}")
+        {
+            replacement = "&#125;";
+        }
+        else if (c == "-")
+        {
+            replacement = "&#45;";
+        }
+        else if (c == "*")
+        {
+            replacement = "&#42;";
+        }
+        else if (c == "|")
+        {
+            replacement = "&#124;";
+        }
+        else if (c == "[")
+        {
+            replacement = "&#91;";
+        }
+        else if (c == "]")
+        {
+            replacement = "&#93;";
+        }
         
         size_t pos = comment.find(c);
+        
+        //for each position of a found escape character
         while (pos != string::npos)
         {
-            cout << "COMMENT: " << comment << endl;
-            comment.replace(pos, c.size(), replacement);
-            cout << "COMMENT AFTER: " << comment << endl;
+            pair<unsigned int,unsigned int> *region = NULL;
             
-            pos = comment.find(c, pos+replacement.size());
+            //is this pos in an escaped section?
+            for (list<pair<unsigned int,unsigned int> >::iterator i = escaperLimits.begin(); i != escaperLimits.end(); ++i)
+            {
+                if (pos >= i->first && pos <= i->second)
+                {
+                    region = &*i;
+                    break;
+                }
+            }
+            
+            if (region == NULL)
+            {
+                comment.replace(pos, c.size(), replacement);
+                pos = comment.find(c, pos+replacement.size());
+            }
+            else
+            {
+//                cout << "skipping ahead to: '" << comment.substr(region->second) << "'"<< endl;
+                pos = comment.find(c, region->second+1);
+            }
         }
     }
+    
+//    cout << "COMMENT: " << comment << endl;
+    comment = removeMarkers(comment);
+    size_t f = comment.find("This exception is raised if the ");
+    if (f != string::npos)
+    {
+        cout << "AFTER: " << comment << endl;
+    }
     return comment;
 }
 
@@ -118,11 +185,11 @@ Confluence::ConfluenceOutput::convertCommentHTML(string comment)
         {
             if (!isEndTag)
             {
-                replacement = "{{{";
+                replacement = "{{";
             }
             else
             {
-                replacement = "}}}";
+                replacement = "}}";
             }
         }
         else if (!strcmp(tag.c_str(), "p"))
@@ -177,21 +244,57 @@ Confluence::ConfluenceOutput::convertCommentHTML(string comment)
             }
             //do nothing for end tag
         }
-        else 
+        else if (!strcmp(tag.c_str(), "dl"))
         {
             if (!isEndTag)
             {
-                replacement = "{{";
+                replacement = "\n";
             }
             else
             {
-                replacement = "}}";
+                replacement = "\n";
+            }
+        }
+        else if (!strcmp(tag.c_str(), "dt"))
+        {
+            if (!isEndTag)
+            {
+                replacement = "";
+            }
+            else
+            {
+                replacement = " ";
+            }
+        }
+        else if (!strcmp(tag.c_str(), "dd"))
+        {
+            if (!isEndTag)
+            {
+                replacement = "--- ";
+            }
+            else
+            {
+                replacement = "\n";
             }
         }
+        else if (!strcmp(tag.c_str(), "em"))
+        {
+            if (!isEndTag)
+            {
+                replacement = "_";
+            }
+            else
+            {
+                replacement = "_";
+            }
+        }
+        else 
+        {
+            replacement = "*{{UNRECOGNIZED MARKUP: " + tag + "}}*";
+        }
         
         //apply replacement
         comment.replace(tagStart, tagEnd + 1 - tagStart, replacement);
-        
         tagStart = comment.find("<");
     }
     return comment;
@@ -552,6 +655,62 @@ Confluence::ConfluenceOutput::getNavMarkup(const std::string& prevLink, const st
     return oss.str();
 }
 
+list< pair<unsigned int,unsigned int> > 
+Confluence::ConfluenceOutput::getMarkerLimits(const string& str)
+{
+    list< pair<unsigned int,unsigned int> > pairs;
+    
+    size_t start = str.find(TEMP_ESCAPER_START); 
+    size_t end;
+    while (start != string::npos)
+    {
+        end = str.find(TEMP_ESCAPER_END, start+TEMP_ESCAPER_START.size());
+        if (end != string::npos)
+        {
+            pair<unsigned int, unsigned int> p = make_pair((unsigned int)start, (unsigned int)end+TEMP_ESCAPER_END.size());
+//            cout << "adding pair (" << p.first << ", " << p.second << ") for '" << str << "'"  << endl; 
+            pairs.push_back(p);
+            start = str.find(TEMP_ESCAPER_START, end+TEMP_ESCAPER_END.size());
+        }
+        else
+        {
+            cerr << "getEscaperLimits FOUND START OF ESCAPER WITH NO MATCHING END IN STRING:" << endl << str.substr(start) << endl;
+            break;
+        }
+    }
+    
+    return pairs;
+}
+
+string 
+Confluence::ConfluenceOutput::removeMarkers(string str)
+{
+    //remove starts
+//    cout << "REMOVE STARTS FROM STR: " << str << endl;
+    size_t start = str.find(TEMP_ESCAPER_START);
+    while (start != string::npos)
+    {
+        str.erase(start, TEMP_ESCAPER_START.size());
+        start = str.find(TEMP_ESCAPER_START, start);
+    }
+    
+//    cout << "WITH STARTS REMOVED: " << str << endl;
+    
+    //remove ends
+    size_t end = str.find(TEMP_ESCAPER_END);
+    while (end != string::npos)
+    {
+        str.erase(end, TEMP_ESCAPER_END.size());
+        end = str.find(TEMP_ESCAPER_END, end);
+    }
+    size_t f = str.find("This exception is raised if the ");
+    if (f != string::npos)
+    {
+        cout << "WITH STOPS REMOVED?: " << str << endl;
+    }
+    return str;
+}
+
 void
 Confluence::ConfluenceOutput::attr(const string& name, const string& value)
 {
diff --git a/cpp/src/slice2confluence/ConfluenceOutput.h b/cpp/src/slice2confluence/ConfluenceOutput.h
index 70f68b24132..294b0fb3054 100644
--- a/cpp/src/slice2confluence/ConfluenceOutput.h
+++ b/cpp/src/slice2confluence/ConfluenceOutput.h
@@ -1,4 +1,6 @@
 #include <IceUtil/OutputUtil.h>
+#include <list>
+#include <utility>
 
 #ifndef CONFLUENCE_OUTPUT
 #define CONFLUENCE_OUTPUT
@@ -41,6 +43,20 @@ public:
     
     std::string currentElement() const;
     
+    /**
+     * Wrap sections in these markers to prevent them from being confluence-escaped.
+     * The regular confluence-escaping process will remove these markers.
+     */
+    const static std::string TEMP_ESCAPER_START; // wrap sections
+    const static std::string TEMP_ESCAPER_END; // wrap sections
+    
+    /**
+     * Gets the start and end positions of all TEMP_ESCAPED sections of the given string.
+     */
+    std::list<std::pair<unsigned int,unsigned int> > getMarkerLimits(const std::string&);
+    
+    std::string removeMarkers(std::string);
+    
 private:
     
     std::string escape(const ::std::string&) const;
diff --git a/cpp/src/slice2confluence/Gen.cpp b/cpp/src/slice2confluence/Gen.cpp
index 455ab1d1841..bd141698902 100755
--- a/cpp/src/slice2confluence/Gen.cpp
+++ b/cpp/src/slice2confluence/Gen.cpp
@@ -426,8 +426,8 @@ Slice::GeneratorBase::removeNewlines(string str)
 string
 Slice::GeneratorBase::trim(string str)
 {
-    string out = str.erase( str.find_last_not_of(" ") + 1);
-    out = out.erase(0 , out.find_first_not_of(" "));
+    string out = str.erase( str.find_last_not_of(" \n\r\t") + 1);
+    out = out.erase(0 , out.find_first_not_of(" \n\r\t"));
     return out;
 }
 
@@ -462,7 +462,7 @@ Slice::GeneratorBase::printComment(const ContainedPtr& p, const ContainerPtr& co
     {
         comment.erase(pos + 1);
         _out.zeroIndent();
-        _out << _out.convertCommentHTML(removeNewlines(comment));
+        _out << removeNewlines(comment);
         _out.restoreIndent();
         _out << "\n";
     }
@@ -729,7 +729,7 @@ Slice::GeneratorBase::getSummary(const ContainedPtr& p, const ContainerPtr& modu
     }
     
     string summary = getComment(p, container, true, forIndex);
-    oss << _out.convertCommentHTML(removeNewlines(summary));
+    oss << removeNewlines(summary);
     
     if(deprecated)
     {
@@ -1123,6 +1123,7 @@ Slice::GeneratorBase::getComment(const ContainedPtr& contained, const ContainerP
         //
         if(s[i] == '\\' && i + 1 < s.size() && s[i + 1] == '[')
         {
+//            comment += Confluence::ConfluenceOutput::TEMP_ESCAPER_START;
             comment += '[';
             ++summarySize;
             ++i;
@@ -1140,7 +1141,9 @@ Slice::GeneratorBase::getComment(const ContainedPtr& contained, const ContainerP
                 literal += s[i];
             }
             size_t sz = 0;
+            comment += Confluence::ConfluenceOutput::TEMP_ESCAPER_START;
             comment += toString(literal, container, false, forIndex, summary ? &sz : 0);
+            comment += Confluence::ConfluenceOutput::TEMP_ESCAPER_END;
             summarySize += sz;
 
             //
@@ -1158,6 +1161,8 @@ Slice::GeneratorBase::getComment(const ContainedPtr& contained, const ContainerP
         {
             static const string atLink = "{@link";
             string::size_type pos = s.find(atLink, i);
+            
+            comment += Confluence::ConfluenceOutput::TEMP_ESCAPER_START;
             if(pos != i)
             {
                 comment += '{';
@@ -1172,6 +1177,7 @@ Slice::GeneratorBase::getComment(const ContainedPtr& contained, const ContainerP
             string literal = s.substr(pos + atLink.size(), endpos - pos - atLink.size());
             size_t sz = 0;
             comment += toString(toSliceID(literal, contained->file()), container, false, forIndex, summary ? &sz : 0);
+            comment += Confluence::ConfluenceOutput::TEMP_ESCAPER_END;
             summarySize += sz;
             i = static_cast<unsigned int>(endpos);
         }
@@ -1193,8 +1199,7 @@ Slice::GeneratorBase::getComment(const ContainedPtr& contained, const ContainerP
         cerr << contained->file() << ": warning: summary size (" << summarySize << ") exceeds " << _warnSummary
              << " characters: `" << comment << "'" << endl;
     }
-
-    return removeNewlines(comment);
+    return removeNewlines(_out.convertCommentHTML(comment));
 }
 
 string
@@ -1294,34 +1299,17 @@ Slice::GeneratorBase::getLinkPath(const SyntaxTreeBasePtr& p, const ContainerPtr
         
         target.pop_front();
     }
-
-    if(forEnum)
-    {
-        if(!path.empty())
-        {
-//            path = "../" + path;
-        }
-        else
-        {
-            path = "";
-        }
-    }
     
-    if (forIndex && path == parent)
+    if ((forIndex && path == parent) || (parent.empty() && path.find("-") == string::npos) || path == parent)
     {
         //link to parent, add suffix
         path += MODULE_SUFFIX;
-//        path = parent + MODULE_SUFFIX;
     }
-    else if (path.find("-") == string::npos && path != INDEX_NAME && path != "" && parent != "")
+    else if (path.find("-") == string::npos && path != INDEX_NAME && !path.empty() && !parent.empty() && path != parent)
     {
         //intra-package links need package name, unlike with html dir structure
         path = parent + "-" + path;
     }
-//    else if (path.find("-") == string::npos && path != "") {
-//        //link to a top level page, add suffix
-//        path += MODULE_SUFFIX;
-//    }
     
     return path;
 }
@@ -2151,7 +2139,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), false);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), false));
             if (!summary.empty())
             {
                 start("dd");
@@ -2189,7 +2177,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2220,7 +2208,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2253,7 +2241,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2286,7 +2274,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2319,7 +2307,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2352,7 +2340,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2385,7 +2373,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2418,7 +2406,7 @@ Slice::ModuleGenerator::visitContainer(const ContainerPtr& p)
             end();
             
             string metadata;
-            string summary = getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true);
+            string summary = trim(getSummary(*q, p, (*q)->findMetaData("deprecate", metadata), true));
             if (!summary.empty())
             {
                 start("dd");
@@ -2614,7 +2602,7 @@ Slice::ExceptionGenerator::generate(const ExceptionPtr& e)
             end();
             
             string metadata;
-            string summary = getSummary(*q, e, (*q)->findMetaData("deprecate", metadata), false);
+            string summary = trim(getSummary(*q, e, (*q)->findMetaData("deprecate", metadata), false));
             if (!summary.empty())
             {
                 start("dd");
@@ -2753,7 +2741,7 @@ Slice::ClassGenerator::generate(const ClassDefPtr& c)
             start("dt", "Symbol");
             _out << toString(*q, c, false);
             end();
-            string summary = getSummary(*q, c, (*q)->findMetaData("deprecate", metadata), false);
+            string summary = trim(getSummary(*q, c, (*q)->findMetaData("deprecate", metadata), false));
             if (!summary.empty())
             {
                 start("dd");
@@ -2782,7 +2770,7 @@ Slice::ClassGenerator::generate(const ClassDefPtr& c)
             start("dt", "Symbol");
             _out << toString(*q, c, false);
             end();
-            string summary = getSummary(*q, c, (*q)->findMetaData("deprecate", metadata), false);
+            string summary = trim(getSummary(*q, c, (*q)->findMetaData("deprecate", metadata), false));
             if (!summary.empty())
             {
                 start("dd");
@@ -2959,10 +2947,18 @@ Slice::StructGenerator::generate(const StructPtr& s)
             start("dt", "Symbol");
             _out << toString(*q, s, false);
             end();
-            start("dd");
             string metadata;
-            printSummary(*q, s, (*q)->findMetaData("deprecate", metadata), false);
-            end();
+            string summary = trim(getSummary(*q, s, (*q)->findMetaData("deprecate", metadata), false));
+            if (!summary.empty())
+            {
+                start("dd");
+                printSummary(*q, s, (*q)->findMetaData("deprecate", metadata), false);
+                end();
+            }
+            else
+            {
+                _out << "\n";
+            }
         }
         end();
     }