support text nodes, support comments, slightly better error messages

File.cpp

@@ -42,7 +42,8 @@ void File::reset() {
 
   if (_file) close(_file);
   _file = open(_path.c_str(), O_RDONLY);
-  if (_file < 0) throw XmlFileException(std::string("could not open ") + _path);
+  if (_file < 0)
+    throw XmlFileException(std::string("could not open file"), _path, 0, 0, 0);
 #ifdef __unix__
   posix_fadvise(_file, 0, 0, POSIX_FADV_SEQUENTIAL);
 #endif
@@ -93,6 +94,7 @@ bool File::next() {
 
   if (_s.hanging) _s.hanging--;
   _ret.name = 0;
+  _ret.text = emptyStr;
   _ret.attrs.clear();
   void* i;
   while (_lastBytes) {
@@ -107,10 +109,65 @@ bool File::next() {
             continue;
           } else {
             _s.s = IN_TEXT;
+            _ret.name = emptyStr;
+            _tmp = _c;
             continue;
           }
         case IN_TEXT:
-          throw XmlFileException("text nodes not yet supported");
+          i = memchr(_c, '<', _lastBytes - (_c - _buffer[_which]));
+          if (!i) {
+            _c = _buffer[_which] + _lastBytes;
+            continue;
+          } else {
+            _c = (char*)i;
+            *_c = 0;
+            _ret.text = _tmp;
+            _s.s = IN_TAG_TENTATIVE;
+            _c++;
+            return true;
+          }
+        case IN_COMMENT_TENTATIVE:
+          if (c == '-') {
+            _s.s = IN_COMMENT_TENTATIVE2;
+            continue;
+          }
+          throw XmlFileException("Expected comment", _path, _c, _buffer[_which],
+                                 _prevs.off);
+
+        case IN_COMMENT_TENTATIVE2:
+          if (c == '-') {
+            _s.s = IN_COMMENT;
+            continue;
+          }
+          throw XmlFileException("Expected comment", _path, _c, _buffer[_which],
+                                 _prevs.off);
+
+        case IN_COMMENT_CL_TENTATIVE:
+          if (c == '-') {
+            _s.s = IN_COMMENT_CL_TENTATIVE2;
+            continue;
+          }
+          _s.s = IN_COMMENT;
+          continue;
+
+        case IN_COMMENT_CL_TENTATIVE2:
+          if (c == '>') {
+            _s.s = NONE;
+            continue;
+          }
+          _s.s = IN_COMMENT;
+        // fall through, we are still in comment
+
+        case IN_COMMENT:
+          i = memchr(_c, '-', _lastBytes - (_c - _buffer[_which]));
+          if (!i) {
+            _c = _buffer[_which] + _lastBytes;
+            continue;
+          } else {
+            _c = (char*)i;
+            _s.s = IN_COMMENT_CL_TENTATIVE;
+            continue;
+          }
         case IN_TAG_TENTATIVE:
           if (c == '/') {
             _s.s = IN_TAG_NAME_CLOSE;
@@ -119,6 +176,9 @@ bool File::next() {
           } else if (c == '?') {
             _s.s = IN_TAG_NAME_META;
             continue;
+          } else if (c == '!') {
+            _s.s = IN_COMMENT_TENTATIVE;
+            continue;
           } else if (std::isalnum(c) || c == '-' || c == '_' || c == '.') {
             _s.s = IN_TAG_NAME;
             _ret.name = _c;
@@ -141,7 +201,8 @@ bool File::next() {
             _s.s = WS_SKIP;
             continue;
           } else {
-            throw XmlFileException("expected valid tag");
+            throw XmlFileException("Expected valid tag", _path, _c,
+                                   _buffer[_which], _prevs.off);
           }
 
         case IN_ATTRVAL_SQ:
@@ -182,7 +243,8 @@ bool File::next() {
             _tmp2 = _c + 1;
             continue;
           } else {
-            throw XmlFileException("expected attribute value");
+            throw XmlFileException("Expected attribute value", _path, _c,
+                                   _buffer[_which], _prevs.off);
           }
 
         case IN_ATTRKEY:
@@ -198,7 +260,8 @@ bool File::next() {
             continue;
           }
 
-          throw XmlFileException("expected attribute key char or =");
+          throw XmlFileException("Expected attribute key char or =", _path, _c,
+                                 _buffer[_which], _prevs.off);
 
         case AFTER_ATTRKEY:
           if (std::isspace(c))
@@ -207,8 +270,9 @@ bool File::next() {
             _s.s = AW_IN_ATTRVAL;
             continue;
           } else {
-            // TODO: error
-            continue;
+            throw XmlFileException(
+                std::string("Expected attribute value for '") + _tmp + "'.",
+                _path, _c, _buffer[_which], _prevs.off);
           }
 
         case IN_TAG_NAME:
@@ -249,7 +313,10 @@ bool File::next() {
           } else if (c == '>') {
             *_c = 0;
             if (_tmp != _s.tagStack.top()) {
-              throw XmlFileException("closing wrong tag");
+              throw XmlFileException(std::string("Closing wrong tag '<") + _tmp +
+                                         ">', expected close of '<" +
+                                         _s.tagStack.top() + ">'.",
+                                     _path, _c, _buffer[_which], _prevs.off);
             }
             _s.tagStack.pop();
             _s.s = NONE;
@@ -261,13 +328,17 @@ bool File::next() {
             continue;
           else if (c == '>') {
             if (_tmp != _s.tagStack.top()) {
-              throw XmlFileException("closing wrong tag");
+              throw XmlFileException(std::string("Closing wrong tag '<") + _tmp +
+                                         ">', expected close of '<" +
+                                         _s.tagStack.top() + ">'.",
+                                     _path, _c, _buffer[_which], _prevs.off);
             }
             _s.tagStack.pop();
             _s.s = NONE;
             continue;
           } else {
-            throw XmlFileException("expected '>'");
+            throw XmlFileException("Expected '>'", _path, _c, _buffer[_which],
+                                   _prevs.off);
           }
 
         case AW_CLOSING:
@@ -316,8 +387,8 @@ bool File::next() {
 
   if (_s.tagStack.size()) {
     if (_s.tagStack.top() != "[root]") {
-      // TODO error
-      throw XmlFileException("XML tree not complete");
+      throw XmlFileException("XML tree not complete", _path, _c,
+                             _buffer[_which], _prevs.off);
     } else {
       _s.tagStack.pop();
     }

File.h

@@ -7,6 +7,7 @@
 #include <cstring>
 #include <fstream>
 #include <map>
+#include <sstream>
 #include <stack>
 #include <string>
 
@@ -16,7 +17,12 @@ const static size_t BUFFER_S = 16 * 1024;
 
 class XmlFileException : public std::exception {
  public:
-  XmlFileException(std::string msg) : _msg(msg) {}
+  XmlFileException(std::string msg, std::string file, const char* p, char* buff,
+                   size_t offset) {
+    std::stringstream ss;
+    ss << file << " at position " << (offset + (p - buff)) << ": " << msg;
+    _msg = ss.str();
+  }
   ~XmlFileException() throw() {}
 
   virtual const char* what() const throw() { return _msg.c_str(); };
@@ -39,6 +45,11 @@ enum State {
   IN_ATTRVAL_SQ,
   IN_ATTRVAL_DQ,
   IN_TEXT,
+  IN_COMMENT_TENTATIVE,
+  IN_COMMENT_TENTATIVE2,
+  IN_COMMENT,
+  IN_COMMENT_CL_TENTATIVE,
+  IN_COMMENT_CL_TENTATIVE2,
   AW_CLOSING,
   WS_SKIP
 };
@@ -61,6 +72,7 @@ typedef std::map<const char*, const char*, AttrCmp> AttrMap;
 
 struct Tag {
   const char* name;
+  const char* text;
   AttrMap attrs;
 };
 
@@ -99,6 +111,7 @@ class File {
   Tag _ret;
 
   static size_t utf8(size_t cp, char* out);
+  const char* emptyStr = "";
 };
 }