Implement a MIME multipart parser

runtime/bin/mime_multipart_parser.dart

+// Copyright (c) 2012, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+
+class _MimeMultipartParser {
+  final int _START = 0;
+  final int _FIRST_BOUNDARY_ENDING = 111;
+  final int _FIRST_BOUNDARY_END = 112;
+  final int _BOUNDARY_ENDING = 1;
+  final int _BOUNDARY_END = 2;
+  final int _HEADER_START = 3;
+  final int _HEADER_FIELD = 4;
+  final int _HEADER_VALUE_START = 5;
+  final int _HEADER_VALUE = 6;
+  final int _HEADER_VALUE_FOLDING_OR_ENDING = 7;
+  final int _HEADER_VALUE_FOLD_OR_END = 8;
+  final int _HEADER_ENDING = 9;
+  final int _CONTENT = 10;
+  final int _LAST_BOUNDARY_DASH2 = 11;
+  final int _LAST_BOUNDARY_ENDING = 12;
+  final int _LAST_BOUNDARY_END = 13;
+  final int _DONE = 14;
+  final int _FAILURE = 15;
+
+  _MimeMultipartParser(String boundary) {
+    List<int> charCodes = boundary.charCodes();
+    _boundary = new List<int>(4 + charCodes.length);
+    // Set-up the matching boundary preceding it with CRLF and two
+    // dashes.
+    _boundary[0] = _CharCode.CR;
+    _boundary[1] = _CharCode.LF;
+    _boundary[2] = _CharCode.DASH;
+    _boundary[3] = _CharCode.DASH;
+    _boundary.setRange(4, charCodes.length, charCodes);
+    _state = _START;
+    _headerField = new StringBuffer();
+    _headerValue = new StringBuffer();
+  }
+
+  void update(List<int> buffer, int offset, int count) {
+    int index = offset;
+    int lastIndex = offset + count;
+    if (_state == _CONTENT && _boundaryIndex == 0) {
+      _contentStartIndex = 0;
+    } else {
+      _contentStartIndex = null;
+    }
+    // The data to parse might be "artificially" prefixed with a
+    // partial match of the boundary.
+    _boundaryPrefix = _boundaryIndex;
+
+    void reportData() {

[Mads Ager (google), 2012/05/25 08:11:58]

Having this local function defined here breaks the reading flow a bit. Can we
move this to the top of update so the code is not interrupted by a function
definition? Maybe add a short comment to update and reportData would make it
easier to read as well.

[Søren Gjesse, 2012/05/25 11:57:06]

Moved the function to the top. I still have to declare index and other state
variables before the function as it uses these.

+      if (partDataReceived == null) return;
+
+      var contentLength = _boundaryPrefix + index - _boundaryIndex;
+      if (_contentStartIndex < 0) {
+        if (contentLength <= _boundaryPrefix) {
+          partDataReceived(
+              _boundary.getRange(0, contentLength));
+        } else {
+          partDataReceived(
+              _boundary.getRange(0, _boundaryPrefix));
+          partDataReceived(
+              buffer.getRange(0, contentLength - _boundaryPrefix));
+        }
+      } else {
+        contentLength -= _contentStartIndex;

[Mads Ager (google), 2012/05/25 08:11:58]

So contentLength is the full length of contents seen so far and
contentStartIndex indicates how much you have already sent to the
partDataReceived callback. Comments on reportData would be good. :)

[Søren Gjesse, 2012/05/25 11:57:06]

Re-arranged the state variables. Most are now local variables to the update
function instead of member variables. Added comments to these state variables.

+        partDataReceived(
+            buffer.getRange(_contentStartIndex, contentLength));
+      }
+    }
+
+    while ((index < lastIndex) && _state != _FAILURE && _state != _DONE) {
+      int byte;
+      if (index < 0) {
+        byte = _boundary[_boundaryPrefix + index];
+      } else {
+        byte = buffer[index];
+      }
+      switch (_state) {
+        case _START:
+          if (_toLowerCase(byte) == _toLowerCase(_boundary[_boundaryIndex])) {
+            _boundaryIndex++;
+            if (_boundaryIndex == _boundary.length) {
+              _state = _FIRST_BOUNDARY_ENDING;
+              _boundaryIndex = 0;
+            }
+          } else {
+            // Restart matching of the boundary.
+            index = index - _boundaryIndex;
+            _boundaryIndex = 0;
+          }
+          break;
+
+        case _FIRST_BOUNDARY_ENDING:
+          if (byte == _CharCode.CR) {
+            _state = _FIRST_BOUNDARY_END;
+          } else {
+            _expectWS(byte);
+          }
+          break;
+
+        case _FIRST_BOUNDARY_END:
+          _expect(byte, _CharCode.LF);
+          _state = _HEADER_START;
+          break;
+
+        case _BOUNDARY_ENDING:
+          if (byte == _CharCode.CR) {
+            _state = _BOUNDARY_END;
+          } else if (byte == _CharCode.DASH) {
+            _state = _LAST_BOUNDARY_DASH2;
+          } else {
+            _expectWS(byte);
+          }
+          break;
+
+        case _BOUNDARY_END:
+          _expect(byte, _CharCode.LF);
+          if (partEnd != null) {
+            partEnd(false);
+          }
+          _state = _HEADER_START;
+          break;
+
+        case _HEADER_START:
+          if (byte == _CharCode.CR) {
+            _state = _HEADER_ENDING;
+            } else {
+              // Start of new header field.
+              _headerField.addCharCode(_toLowerCase(byte));
+              _state = _HEADER_FIELD;
+            }
+            break;
+
+          case _HEADER_FIELD:
+            if (byte == _CharCode.COLON) {
+              _state = _HEADER_VALUE_START;
+            } else {
+              if (!_isTokenChar(byte)) {
+                throw new HttpParserException("Invalid header field name");
+              }
+              _headerField.addCharCode(_toLowerCase(byte));
+            }
+            break;
+
+          case _HEADER_VALUE_START:
+            if (byte == _CharCode.CR) {
+              _state = _HEADER_VALUE_FOLDING_OR_ENDING;
+            } else if (byte != _CharCode.SP && byte != _CharCode.HT) {
+              // Start of new header value.
+              _headerValue.addCharCode(byte);
+              _state = _HEADER_VALUE;
+            }
+            break;
+
+          case _HEADER_VALUE:
+            if (byte == _CharCode.CR) {
+              _state = _HEADER_VALUE_FOLDING_OR_ENDING;
+            } else {
+              _headerValue.addCharCode(byte);
+            }
+            break;
+
+          case _HEADER_VALUE_FOLDING_OR_ENDING:
+            _expect(byte, _CharCode.LF);
+            _state = _HEADER_VALUE_FOLD_OR_END;
+            break;
+
+          case _HEADER_VALUE_FOLD_OR_END:
+            if (byte == _CharCode.SP || byte == _CharCode.HT) {
+              _state = _HEADER_VALUE_START;
+            } else {
+              String headerField = _headerField.toString();
+              String headerValue =_headerValue.toString();
+              if (headerReceived != null) {
+                headerReceived(headerField, headerValue);
+              }
+              _headerField.clear();
+              _headerValue.clear();
+              if (byte == _CharCode.CR) {
+                _state = _HEADER_ENDING;
+              } else {
+                // Start of new header field.
+                _headerField.addCharCode(_toLowerCase(byte));
+                _state = _HEADER_FIELD;
+              }
+            }
+            break;
+
+          case _HEADER_ENDING:
+            _expect(byte, _CharCode.LF);
+            if (headersComplete != null) headersComplete();
+            _state = _CONTENT;
+            _contentStartIndex = index + 1;
+            break;
+
+          case _CONTENT:
+            if (_toLowerCase(byte) == _toLowerCase(_boundary[_boundaryIndex])) {
+              _boundaryIndex++;
+              if (_boundaryIndex == _boundary.length) {
+                if (_contentStartIndex != null) {
+                  reportData();
+                }
+                _boundaryIndex = 0;
+                _state = _BOUNDARY_ENDING;
+              }
+            } else {
+              // Restart matching of the boundary.
+              index = index - _boundaryIndex;
+              if (_contentStartIndex == null) _contentStartIndex = index;
+              _boundaryIndex = 0;
+            }
+            break;
+
+        case _LAST_BOUNDARY_DASH2:
+          _expect(byte, _CharCode.DASH);
+          _state = _LAST_BOUNDARY_ENDING;
+          break;
+
+        case _LAST_BOUNDARY_ENDING:
+          if (byte == _CharCode.CR) {
+            _state = _LAST_BOUNDARY_END;
+          } else {
+            _expectWS(byte);
+          }
+          break;
+
+        case _LAST_BOUNDARY_END:
+          _expect(byte, _CharCode.LF);
+          if (partEnd != null) {
+            partEnd(true);
+          }
+          _state = _DONE;
+          break;
+
+        default:
+          // Should be unreachable.
+          assert(false);
+          break;
+      }
+
+      // Move to the next byte.
+      index++;
+    }
+
+    // Report any known content.
+    if (_state == _CONTENT && _contentStartIndex != null) {
+      reportData();
+    }
+    return index - offset;
+  }
+
+  bool _isTokenChar(int byte) {
+    return byte > 31 && byte < 128 && _Const.SEPARATORS.indexOf(byte) == -1;
+  }
+
+  int _toLowerCase(int byte) {
+    final int aCode = "A".charCodeAt(0);
+    final int zCode = "Z".charCodeAt(0);
+    final int delta = "a".charCodeAt(0) - aCode;
+    return (aCode <= byte && byte <= zCode) ? byte + delta : byte;
+  }
+
+  void _expect(int val1, int val2) {
+    if (val1 != val2) {
+      throw new MimeParserException("Failed to parse multipart mime 1");
+    }
+  }
+
+  void _expectWS(int byte) {
+    if (byte != _CharCode.SP && byte != _CharCode.HT) {
+      throw new MimeParserException("Failed to parse multipart mime 2");
+    }
+  }
+
+  List<int> _boundary;
+  int _state;
+  int _boundaryIndex = 0;
+  // Number of boundary bytes to artificially place before the supplied data.
+  int _boundaryPrefix = 0;
+  int _contentStartIndex = -1;
+
+  StringBuffer _headerField;
+  StringBuffer _headerValue;
+
+  Function partStart;
+  Function headerReceived;
+  Function headersComplete;
+  Function partDataReceived;
+  Function partEnd;
+}
+
+
+class MimeParserException implements Exception {
+  const MimeParserException([String this.message = ""]);
+  String toString() => "MimeParserException: $message";
+  final String message;
+}