Add HttpMultipartScanner (#1197)

This is a non-blocking version of FindMultipartBoundary.

wpiutil/src/main/native/cpp/HttpUtil.cpp

@@ -333,4 +333,78 @@ bool HttpConnection::Handshake(const HttpRequest& request,
   return true;
 }
 
+void HttpMultipartScanner::SetBoundary(StringRef boundary) {
+  m_boundaryWith = "\n--";
+  m_boundaryWith += boundary;
+  m_boundaryWithout = "\n";
+  m_boundaryWithout += boundary;
+  m_dashes = kUnknown;
+}
+
+void HttpMultipartScanner::Reset(bool saveSkipped) {
+  m_saveSkipped = saveSkipped;
+  m_state = kBoundary;
+  m_posWith = 0;
+  m_posWithout = 0;
+  m_buf.resize(0);
+}
+
+StringRef HttpMultipartScanner::Execute(StringRef in) {
+  if (m_state == kDone) Reset(m_saveSkipped);
+  if (m_saveSkipped) m_buf += in;
+
+  size_t pos = 0;
+  if (m_state == kBoundary) {
+    for (char ch : in) {
+      ++pos;
+      if (m_dashes != kWithout) {
+        if (ch == m_boundaryWith[m_posWith]) {
+          ++m_posWith;
+          if (m_posWith == m_boundaryWith.size()) {
+            // Found the boundary; transition to padding
+            m_state = kPadding;
+            m_dashes = kWith;  // no longer accept plain 'boundary'
+            break;
+          }
+        } else if (ch == m_boundaryWith[0]) {
+          m_posWith = 1;
+        } else {
+          m_posWith = 0;
+        }
+      }
+
+      if (m_dashes != kWith) {
+        if (ch == m_boundaryWithout[m_posWithout]) {
+          ++m_posWithout;
+          if (m_posWithout == m_boundaryWithout.size()) {
+            // Found the boundary; transition to padding
+            m_state = kPadding;
+            m_dashes = kWithout;  // no longer accept '--boundary'
+            break;
+          }
+        } else if (ch == m_boundaryWithout[0]) {
+          m_posWithout = 1;
+        } else {
+          m_posWithout = 0;
+        }
+      }
+    }
+  }
+
+  if (m_state == kPadding) {
+    for (char ch : in.drop_front(pos)) {
+      ++pos;
+      if (ch == '\n') {
+        // Found the LF; return remaining input buffer (following it)
+        m_state = kDone;
+        if (m_saveSkipped) m_buf.resize(m_buf.size() - in.size() + pos);
+        return in.drop_front(pos);
+      }
+    }
+  }
+
+  // We consumed the entire input
+  return StringRef{};
+}
+
 }  // namespace wpi

wpiutil/src/main/native/include/wpi/HttpUtil.h

@@ -141,6 +141,49 @@ class HttpConnection {
   explicit operator bool() const { return stream && !is.has_error(); }
 };
 
+class HttpMultipartScanner {
+ public:
+  explicit HttpMultipartScanner(StringRef boundary, bool saveSkipped = false) {
+    Reset(saveSkipped);
+    SetBoundary(boundary);
+  }
+
+  // Change the boundary.  This is only safe to do when IsDone() is true (or
+  // immediately after construction).
+  void SetBoundary(StringRef boundary);
+
+  // Reset the scanner.  This allows reuse of internal buffers.
+  void Reset(bool saveSkipped = false);
+
+  // Execute the scanner.  Will automatically call Reset() on entry if IsDone()
+  // is true.
+  // @param in input data
+  // @return the input not consumed; empty if all input consumed
+  StringRef Execute(StringRef in);
+
+  // Returns true when the boundary has been found.
+  bool IsDone() const { return m_state == kDone; }
+
+  // Get the skipped data.  Will be empty if saveSkipped was false.
+  StringRef GetSkipped() const {
+    return m_saveSkipped ? StringRef{m_buf} : StringRef{};
+  }
+
+ private:
+  SmallString<64> m_boundaryWith, m_boundaryWithout;
+
+  // Internal state
+  enum State { kBoundary, kPadding, kDone };
+  State m_state;
+  size_t m_posWith, m_posWithout;
+  enum Dashes { kUnknown, kWith, kWithout };
+  Dashes m_dashes;
+
+  // Buffer
+  bool m_saveSkipped;
+  std::string m_buf;
+};
+
 }  // namespace wpi
 
 #include "HttpUtil.inl"

wpiutil/src/test/native/cpp/HttpUtilTest.cpp

@@ -0,0 +1,105 @@
+/*----------------------------------------------------------------------------*/
+/* Copyright (c) 2018 FIRST. All Rights Reserved.                             */
+/* Open Source Software - may be modified and shared by FRC teams. The code   */
+/* must be accompanied by the FIRST BSD license file in the root directory of */
+/* the project.                                                               */
+/*----------------------------------------------------------------------------*/
+
+#include "wpi/HttpUtil.h"  // NOLINT(build/include_order)
+
+#include "gtest/gtest.h"
+
+namespace wpi {
+
+TEST(HttpMultipartScannerTest, ExecuteExact) {
+  HttpMultipartScanner scanner("foo");
+  EXPECT_TRUE(scanner.Execute("abcdefg---\r\n--foo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_TRUE(scanner.GetSkipped().empty());
+}
+
+TEST(HttpMultipartScannerTest, ExecutePartial) {
+  HttpMultipartScanner scanner("foo");
+  EXPECT_TRUE(scanner.Execute("abcdefg--").empty());
+  EXPECT_FALSE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("-\r\n").empty());
+  EXPECT_FALSE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("--foo\r").empty());
+  EXPECT_FALSE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, ExecuteTrailing) {
+  HttpMultipartScanner scanner("foo");
+  EXPECT_EQ(scanner.Execute("abcdefg---\r\n--foo\r\nxyz"), "xyz");
+}
+
+TEST(HttpMultipartScannerTest, ExecutePadding) {
+  HttpMultipartScanner scanner("foo");
+  EXPECT_EQ(scanner.Execute("abcdefg---\r\n--foo    \r\nxyz"), "xyz");
+  EXPECT_TRUE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, SaveSkipped) {
+  HttpMultipartScanner scanner("foo", true);
+  scanner.Execute("abcdefg---\r\n--foo\r\n");
+  EXPECT_EQ(scanner.GetSkipped(), "abcdefg---\r\n--foo\r\n");
+}
+
+TEST(HttpMultipartScannerTest, Reset) {
+  HttpMultipartScanner scanner("foo", true);
+
+  scanner.Execute("abcdefg---\r\n--foo\r\n");
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_EQ(scanner.GetSkipped(), "abcdefg---\r\n--foo\r\n");
+
+  scanner.Reset(true);
+  EXPECT_FALSE(scanner.IsDone());
+  scanner.SetBoundary("bar");
+
+  scanner.Execute("--foo\r\n--bar\r\n");
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_EQ(scanner.GetSkipped(), "--foo\r\n--bar\r\n");
+}
+
+TEST(HttpMultipartScannerTest, WithoutDashes) {
+  HttpMultipartScanner scanner("foo", true);
+
+  EXPECT_TRUE(scanner.Execute("--\r\nfoo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, SeqDashesDashes) {
+  HttpMultipartScanner scanner("foo", true);
+  EXPECT_TRUE(scanner.Execute("\r\n--foo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("\r\n--foo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, SeqDashesNoDashes) {
+  HttpMultipartScanner scanner("foo", true);
+  EXPECT_TRUE(scanner.Execute("\r\n--foo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("\r\nfoo\r\n").empty());
+  EXPECT_FALSE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, SeqNoDashesDashes) {
+  HttpMultipartScanner scanner("foo", true);
+  EXPECT_TRUE(scanner.Execute("\r\nfoo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("\r\n--foo\r\n").empty());
+  EXPECT_FALSE(scanner.IsDone());
+}
+
+TEST(HttpMultipartScannerTest, SeqNoDashesNoDashes) {
+  HttpMultipartScanner scanner("foo", true);
+  EXPECT_TRUE(scanner.Execute("\r\nfoo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+  EXPECT_TRUE(scanner.Execute("\r\nfoo\r\n").empty());
+  EXPECT_TRUE(scanner.IsDone());
+}
+
+}  // namespace wpi