mongoimport: support for escaped " in csv with either "" or \" SERVER-773

2 files changed, 34 insertions, 19 deletions

diff --git a/jstests/tool/csv1.js b/jstests/tool/csv1.js
index c57767454f1..edf9dc23d5f 100644
--- a/jstests/tool/csv1.js
+++ b/jstests/tool/csv1.js
@@ -4,7 +4,7 @@ t = new ToolTest( "csv1" )
 
 c = t.startDB( "foo" );
 
-base = { a : 1 , b : "foo,bar" , c: 5, 'd d': 6 };
+base = { a : 1 , b : "foo,bar\"baz,qux" , c: 5, 'd d': 6 };
 
 assert.eq( 0 , c.count() , "setup1" );
 c.insert( base );
diff --git a/tools/import.cpp b/tools/import.cpp
index a26dd392873..e34e73d07d7 100644
--- a/tools/import.cpp
+++ b/tools/import.cpp
@@ -76,28 +76,44 @@ class Import : public Tool {
             }
             pos++;
             
-            int skip = 1;
+            bool done = false;
+            string data;
             char * end;
             if ( _type == CSV && line[0] == '"' ){
-                line++;
-                end = strstr( line , "\"" );
-                skip = 2;
-            }
-            else {
+                line++; //skip first '"'
+
+                while (true) {
+                    end = strchr( line , '"' );
+                    if (!end){
+                        data += line;
+                        done = true;
+                        break;
+                    } else if (end[1] == '"') {
+                        // two '"'s get appended as one
+                        data.append(line, end-line+1); //include '"'
+                        line = end+2; //skip both '"'s
+                    } else if (end[-1] == '\\') {
+                        // "\\\"" gets appended as '"'
+                        data.append(line, end-line-1); //exclude '\\'
+                        data.append("\"");
+                        line = end+1; //skip the '"'
+                    } else {
+                        data.append(line, end-line);
+                        line = end+2; //skip '"' and ','
+                        break;
+                    }
+                }
+            } else {
                 end = strstr( line , _sep );
+                if ( ! end ){
+                    done = true;
+                    data = string( line );
+                } else {
+                    data = string( line , end - line );
+                    line = end+1;
+                }
             }
-            
-            bool done = false;
-            string data;
 
-            if ( ! end ){
-                done = true;
-                data = string( line );
-            }
-            else {
-                data = string( line , end - line );
-            }
-            
             if ( _headerLine ){
                 while ( isspace( data[0] ) )
                     data = data.substr( 1 );
@@ -108,7 +124,6 @@ class Import : public Tool {
             
             if ( done )
                 break;
-            line = end + skip;
         }
         return b.obj();
     }