[wpiutil] Change StringExtras split() to template (#7636)

It now calls back a function for each part rather than creating a SmallVector.

datalogtool/src/main/native/cpp/Exporter.cpp

@@ -102,7 +102,8 @@ static void RebuildEntryTree() {
     } else {
       parts.emplace_back(prefix);
     }
-    wpi::split(mainpart, parts, '/', -1, false);
+    wpi::split(mainpart, '/', -1, false,
+               [&](auto part) { parts.emplace_back(part); });
 
     // ignore a raw "/" key
     if (parts.empty()) {

glass/src/libnt/native/cpp/NetworkTables.cpp

@@ -1015,7 +1015,8 @@ void NetworkTablesModel::RebuildTreeImpl(std::vector<TreeNode>* tree,
       continue;
     }
     parts.clear();
-    wpi::split(entry->info.name, parts, '/', -1, false);
+    wpi::split(entry->info.name, '/', -1, false,
+               [&](auto part) { parts.emplace_back(part); });
 
     // ignore a raw "/" key
     if (parts.empty()) {

glass/src/libnt/native/cpp/NetworkTablesProvider.cpp

@@ -68,7 +68,8 @@ void NetworkTablesProvider::DisplayMenu() {
   wpi::SmallString<64> name;
   for (auto&& entry : m_viewEntries) {
     path.clear();
-    wpi::split(entry->name, path, '/', -1, false);
+    wpi::split(entry->name, '/', -1, false,
+               [&](auto name) { path.emplace_back(name); });
 
     bool fullDepth = true;
     int depth = 0;

glass/src/libnt/native/cpp/NetworkTablesSettings.cpp

@@ -12,7 +12,6 @@
 #include <imgui.h>
 #include <imgui_stdlib.h>
 #include <ntcore_cpp.h>
-#include <wpi/SmallVector.h>
 #include <wpi/StringExtras.h>
 
 #include "glass/Context.h"
@@ -69,12 +68,10 @@ void NetworkTablesSettings::Thread::Main() {
           (team = wpi::parse_integer<unsigned int>(serverTeam, 10))) {
         nt::SetServerTeam(m_inst, team.value(), m_port);
       } else {
-        wpi::SmallVector<std::string_view, 4> serverNames;
         std::vector<std::pair<std::string_view, unsigned int>> servers;
-        wpi::split(serverTeam, serverNames, ',', -1, false);
-        for (auto&& serverName : serverNames) {
+        wpi::split(serverTeam, ',', -1, false, [&](auto serverName) {
           servers.emplace_back(serverName, m_port);
-        }
+        });
         nt::SetServer(m_inst, servers);
       }
 

hal/src/main/native/athena/HAL.cpp

@@ -352,16 +352,15 @@ void InitializeTeamNumber(void) {
   std::string_view hostname{hostnameBuf, sizeof(hostnameBuf)};
 
   // hostname is frc-{TEAM}-roborio
-  // Split string around '-' (max of 2 splits), take the second element of the
-  // resulting array.
-  wpi::SmallVector<std::string_view> elements;
-  wpi::split(hostname, elements, "-", 2);
-  if (elements.size() < 3) {
-    teamNumber = 0;
-    return;
-  }
-
-  teamNumber = wpi::parse_integer<int32_t>(elements[1], 10).value_or(0);
+  // Split string around '-' (max of 2 splits), take the second element
+  teamNumber = 0;
+  int i = 0;
+  wpi::split(hostname, '-', 2, false, [&](auto part) {
+    if (i == 1) {
+      teamNumber = wpi::parse_integer<int32_t>(part, 10).value_or(0);
+    }
+    ++i;
+  });
 }
 
 int32_t HAL_GetTeamNumber(void) {

hal/src/main/native/athena/cpp/SerialHelper.cpp

@@ -245,7 +245,8 @@ void SerialHelper::QueryHubPaths(int32_t* status) {
 
       wpi::SmallVector<std::string_view, 16> pathSplitVec;
       // Split path into individual directories
-      wpi::split(path, pathSplitVec, '/', -1, false);
+      wpi::split(path, '/', -1, false,
+                 [&](auto part) { pathSplitVec.emplace_back(part); });
 
       // Find each individual item index
       int findusb = -1;

hal/src/main/native/sim/Extensions.cpp

@@ -10,7 +10,6 @@
 #include <utility>
 #include <vector>
 
-#include <wpi/SmallVector.h>
 #include <wpi/StringExtras.h>
 #include <wpi/fs.h>
 #include <wpi/print.h>
@@ -100,7 +99,6 @@ int HAL_LoadOneExtension(const char* library) {
 
 int HAL_LoadExtensions(void) {
   int rc = 1;
-  wpi::SmallVector<std::string_view, 2> libraries;
   const char* e = std::getenv("HALSIM_EXTENSIONS");
   if (!e) {
     if (GetShowNotFoundMessage()) {
@@ -109,13 +107,11 @@ int HAL_LoadExtensions(void) {
     }
     return rc;
   }
-  wpi::split(e, libraries, DELIM, -1, false);
-  for (auto& library : libraries) {
-    rc = HAL_LoadOneExtension(std::string(library).c_str());
-    if (rc < 0) {
-      break;
+  wpi::split(e, DELIM, -1, false, [&](auto library) {
+    if (rc >= 0) {
+      rc = HAL_LoadOneExtension(std::string(library).c_str());
     }
-  }
+  });
   return rc;
 }
 

hal/src/main/native/systemcore/HAL.cpp

@@ -18,7 +18,6 @@
 #include <utility>
 
 #include <wpi/MemoryBuffer.h>
-#include <wpi/SmallString.h>
 #include <wpi/StringExtras.h>
 #include <wpi/fs.h>
 #include <wpi/mutex.h>
@@ -233,16 +232,15 @@ void InitializeTeamNumber(void) {
   std::string_view hostname{hostnameBuf, sizeof(hostnameBuf)};
 
   // hostname is frc-{TEAM}-roborio
-  // Split string around '-' (max of 2 splits), take the second element of the
-  // resulting array.
-  wpi::SmallVector<std::string_view> elements;
-  wpi::split(hostname, elements, "-", 2);
-  if (elements.size() < 3) {
-    teamNumber = 0;
-    return;
-  }
-
-  teamNumber = wpi::parse_integer<int32_t>(elements[1], 10).value_or(0);
+  // Split string around '-' (max of 2 splits), take the second element
+  teamNumber = 0;
+  int i = 0;
+  wpi::split(hostname, '-', 2, false, [&](auto part) {
+    if (i == 1) {
+      teamNumber = wpi::parse_integer<int32_t>(part, 10).value_or(0);
+    }
+    ++i;
+  });
 }
 
 int32_t HAL_GetTeamNumber(void) {

ntcore/src/main/native/cpp/networktables/NetworkTable.cpp

@@ -54,12 +54,10 @@ std::string_view NetworkTable::NormalizeKey(std::string_view key,
     buf.push_back(PATH_SEPARATOR_CHAR);
   }
   // for each path element, add it with a slash following
-  wpi::SmallVector<std::string_view, 16> parts;
-  wpi::split(key, parts, PATH_SEPARATOR_CHAR, -1, false);
-  for (auto i = parts.begin(); i != parts.end(); ++i) {
-    buf.append(i->begin(), i->end());
+  wpi::split(key, PATH_SEPARATOR_CHAR, -1, false, [&](auto part) {
+    buf.append(part.begin(), part.end());
     buf.push_back(PATH_SEPARATOR_CHAR);
-  }
+  });
   // remove trailing slash if the input key didn't have one
   if (!key.empty() && key.back() != PATH_SEPARATOR_CHAR) {
     buf.pop_back();
@@ -72,19 +70,17 @@ std::vector<std::string> NetworkTable::GetHierarchy(std::string_view key) {
   hierarchy.emplace_back(1, PATH_SEPARATOR_CHAR);
   // for each path element, add it to the end of what we built previously
   wpi::SmallString<128> path;
-  wpi::SmallVector<std::string_view, 16> parts;
-  wpi::split(key, parts, PATH_SEPARATOR_CHAR, -1, false);
-  if (!parts.empty()) {
-    for (auto i = parts.begin(); i != parts.end(); ++i) {
-      path += PATH_SEPARATOR_CHAR;
-      path += *i;
-      hierarchy.emplace_back(path.str());
-    }
-    // handle trailing slash
-    if (key.back() == PATH_SEPARATOR_CHAR) {
-      path += PATH_SEPARATOR_CHAR;
-      hierarchy.emplace_back(path.str());
-    }
+  bool any = false;
+  wpi::split(key, PATH_SEPARATOR_CHAR, -1, false, [&](auto part) {
+    any = true;
+    path += PATH_SEPARATOR_CHAR;
+    path += part;
+    hierarchy.emplace_back(path.str());
+  });
+  // handle trailing slash
+  if (any && key.back() == PATH_SEPARATOR_CHAR) {
+    path += PATH_SEPARATOR_CHAR;
+    hierarchy.emplace_back(path.str());
   }
   return hierarchy;
 }

simulation/halsim_ws_client/src/main/native/cpp/HALSimWS.cpp

@@ -9,7 +9,6 @@
 #include <string>
 
 #include <wpi/SmallString.h>
-#include <wpi/SmallVector.h>
 #include <wpi/StringExtras.h>
 #include <wpi/print.h>
 #include <wpinet/uv/util.h>
@@ -73,15 +72,8 @@ bool HALSimWS::Initialize() {
   const char* msgFilters = std::getenv("HALSIMWS_FILTERS");
   if (msgFilters != nullptr) {
     m_useMsgFiltering = true;
-
-    std::string_view filters(msgFilters);
-    filters = wpi::trim(filters);
-    wpi::SmallVector<std::string_view, 16> filtersSplit;
-
-    wpi::split(filters, filtersSplit, ',', -1, false);
-    for (auto val : filtersSplit) {
-      m_msgFilters[wpi::trim(val)] = true;
-    }
+    wpi::split(wpi::trim(msgFilters), ',', -1, false,
+               [&](auto val) { m_msgFilters[wpi::trim(val)] = true; });
   } else {
     m_useMsgFiltering = false;
   }

simulation/halsim_ws_server/src/main/native/cpp/HALSimWeb.cpp

@@ -83,15 +83,8 @@ bool HALSimWeb::Initialize() {
   const char* msgFilters = std::getenv("HALSIMWS_FILTERS");
   if (msgFilters != nullptr) {
     m_useMsgFiltering = true;
-
-    std::string_view filters(msgFilters);
-    filters = wpi::trim(filters);
-    wpi::SmallVector<std::string_view, 16> filtersSplit;
-
-    wpi::split(filters, filtersSplit, ',', -1, false);
-    for (auto val : filtersSplit) {
-      m_msgFilters[wpi::trim(val)] = true;
-    }
+    wpi::split(wpi::trim(msgFilters), ',', -1, false,
+               [&](auto val) { m_msgFilters[wpi::trim(val)] = true; });
   } else {
     m_useMsgFiltering = false;
   }

sysid/src/main/native/cpp/view/LogLoader.cpp

@@ -128,7 +128,8 @@ void LogLoader::RebuildEntryTree() {
     } else {
       parts.emplace_back(prefix);
     }
-    wpi::split(mainpart, parts, '/', -1, false);
+    wpi::split(mainpart, '/', -1, false,
+               [&](auto part) { parts.emplace_back(part); });
 
     // ignore a raw "/" key
     if (parts.empty()) {

wpinet/src/main/native/cpp/HttpUtil.cpp

@@ -98,9 +98,7 @@ std::string_view EscapeHTML(std::string_view str, SmallVectorImpl<char>& buf) {
 }
 
 HttpQueryMap::HttpQueryMap(std::string_view query) {
-  SmallVector<std::string_view, 16> queryElems;
-  split(query, queryElems, '&', 100, false);
-  for (auto elem : queryElems) {
+  split(query, '&', 100, false, [&](auto elem) {
     auto [nameEsc, valueEsc] = split(elem, '=');
     SmallString<64> nameBuf;
     bool err = false;
@@ -109,7 +107,7 @@ HttpQueryMap::HttpQueryMap(std::string_view query) {
     if (!err) {
       m_elems.try_emplace(name, valueEsc);
     }
-  }
+  });
 }
 
 std::optional<std::string_view> HttpQueryMap::Get(
@@ -132,9 +130,7 @@ HttpPath::HttpPath(std::string_view path) {
     m_pathEnds.emplace_back(0);
     return;
   }
-  wpi::SmallVector<std::string_view, 16> pathElems;
-  split(path, pathElems, '/', 100, false);
-  for (auto elem : pathElems) {
+  split(path, '/', 100, false, [&](auto elem) {
     SmallString<64> buf;
     bool err = false;
     auto val = wpi::UnescapeURI(elem, buf, &err);
@@ -144,7 +140,7 @@ HttpPath::HttpPath(std::string_view path) {
     }
     m_pathBuf += val;
     m_pathEnds.emplace_back(m_pathBuf.size());
-  }
+  });
 }
 
 bool HttpPath::startswith(size_t start,

wpinet/src/main/native/cpp/WebSocketServer.cpp

@@ -5,6 +5,7 @@
 #include "wpinet/WebSocketServer.h"
 
 #include <memory>
+#include <string>
 #include <utility>
 
 #include <wpi/StringExtras.h>
@@ -31,14 +32,12 @@ WebSocketServerHelper::WebSocketServerHelper(HttpParser& req) {
       m_version = value;
     } else if (equals_lower(name, "sec-websocket-protocol")) {
       // Protocols are comma delimited, repeated headers add to list
-      SmallVector<std::string_view, 2> protocols;
-      split(value, protocols, ",", -1, false);
-      for (auto protocol : protocols) {
+      split(value, ',', -1, false, [&](auto protocol) {
         protocol = trim(protocol);
         if (!protocol.empty()) {
           m_protocols.emplace_back(protocol);
         }
-      }
+      });
     }
   });
   req.headersComplete.connect([&req, this](bool) {
@@ -53,7 +52,22 @@ std::pair<bool, std::string_view> WebSocketServerHelper::MatchProtocol(
   if (protocols.empty() && m_protocols.empty()) {
     return {true, {}};
   }
-  for (auto protocol : protocols) {
+  for (auto&& protocol : protocols) {
+    for (auto&& clientProto : m_protocols) {
+      if (protocol == clientProto) {
+        return {true, protocol};
+      }
+    }
+  }
+  return {false, {}};
+}
+
+std::pair<bool, std::string_view> WebSocketServerHelper::MatchProtocol(
+    std::span<const std::string> protocols) {
+  if (protocols.empty() && m_protocols.empty()) {
+    return {true, {}};
+  }
+  for (auto&& protocol : protocols) {
     for (auto&& clientProto : m_protocols) {
       if (protocol == clientProto) {
         return {true, protocol};
@@ -101,9 +115,7 @@ WebSocketServer::WebSocketServer(uv::Stream& stream,
     }
 
     // Negotiate sub-protocol
-    SmallVector<std::string_view, 2> protocols{m_protocols.begin(),
-                                               m_protocols.end()};
-    std::string_view protocol = m_helper.MatchProtocol(protocols).second;
+    std::string_view protocol = m_helper.MatchProtocol(m_protocols).second;
 
     // Disconnect our header reader
     m_dataConn.disconnect();

wpinet/src/main/native/include/wpinet/WebSocketServer.h

@@ -56,6 +56,18 @@ class WebSocketServerHelper {
   std::pair<bool, std::string_view> MatchProtocol(
       std::span<const std::string_view> protocols);
 
+  /**
+   * Try to find a match to the list of sub-protocols provided by the client.
+   * The list is priority ordered, so the first match wins.
+   * Only valid during and after the upgrade event.
+   * @param protocols Acceptable protocols
+   * @return Pair; first item is true if a match was made, false if not.
+   *         Second item is the matched protocol if a match was made, otherwise
+   *         is empty.
+   */
+  std::pair<bool, std::string_view> MatchProtocol(
+      std::span<const std::string> protocols);
+
   /**
    * Try to find a match to the list of sub-protocols provided by the client.
    * The list is priority ordered, so the first match wins.

wpiutil/src/main/native/thirdparty/llvm/cpp/llvm/StringExtras.cpp

@@ -22,7 +22,6 @@
 #include <string_view>
 
 #include "wpi/SmallString.h"
-#include "wpi/SmallVector.h"
 
 // strncasecmp() is not available on non-POSIX systems, so define an
 // alternative function here.
@@ -118,65 +117,6 @@ bool wpi::ends_with_lower(std::string_view str,
                            suffix.data(), suffix.size()) == 0;
 }
 
-void wpi::split(std::string_view str, SmallVectorImpl<std::string_view>& arr,
-                std::string_view separator, int maxSplit,
-                bool keepEmpty) noexcept {
-  std::string_view s = str;
-
-  // Count down from maxSplit. When maxSplit is -1, this will just split
-  // "forever". This doesn't support splitting more than 2^31 times
-  // intentionally; if we ever want that we can make maxSplit a 64-bit integer
-  // but that seems unlikely to be useful.
-  while (maxSplit-- != 0) {
-    auto idx = s.find(separator);
-    if (idx == std::string_view::npos) {
-      break;
-    }
-
-    // Push this split.
-    if (keepEmpty || idx > 0) {
-      arr.push_back(slice(s, 0, idx));
-    }
-
-    // Jump forward.
-    s = slice(s, idx + separator.size(), std::string_view::npos);
-  }
-
-  // Push the tail.
-  if (keepEmpty || !s.empty()) {
-    arr.push_back(s);
-  }
-}
-
-void wpi::split(std::string_view str, SmallVectorImpl<std::string_view>& arr,
-                char separator, int maxSplit, bool keepEmpty) noexcept {
-  std::string_view s = str;
-
-  // Count down from maxSplit. When maxSplit is -1, this will just split
-  // "forever". This doesn't support splitting more than 2^31 times
-  // intentionally; if we ever want that we can make maxSplit a 64-bit integer
-  // but that seems unlikely to be useful.
-  while (maxSplit-- != 0) {
-    size_t idx = s.find(separator);
-    if (idx == std::string_view::npos) {
-      break;
-    }
-
-    // Push this split.
-    if (keepEmpty || idx > 0) {
-      arr.push_back(slice(s, 0, idx));
-    }
-
-    // Jump forward.
-    s = slice(s, idx + 1, std::string_view::npos);
-  }
-
-  // Push the tail.
-  if (keepEmpty || !s.empty()) {
-    arr.push_back(s);
-  }
-}
-
 static unsigned GetAutoSenseRadix(std::string_view& str) noexcept {
   if (str.empty()) {
     return 10;

wpiutil/src/main/native/thirdparty/llvm/include/wpi/StringExtras.h

@@ -17,6 +17,7 @@
 
 #pragma once
 
+#include <concepts>
 #include <iterator>
 #include <limits>
 #include <optional>
@@ -541,42 +542,95 @@ constexpr std::pair<std::string_view, std::string_view> rsplit(
 /**
  * Splits @p str into substrings around the occurrences of a separator string.
  *
- * Each substring is stored in @p arr. If @p maxSplit is >= 0, at most
- * @p maxSplit splits are done and consequently <= @p maxSplit + 1
- * elements are added to arr.
- * If @p keepEmpty is false, empty strings are not added to @p arr. They
- * still count when considering @p maxSplit
+ * Each substring is passed to the callback @p func. If @p maxSplit is >= 0, at
+ * most @p maxSplit splits are done and consequently <= @p maxSplit + 1
+ * elements are provided to the callback.
+ * If @p keepEmpty is false, empty strings are not included. They
+ * still count when considering @p maxSplit.
  * An useful invariant is that
- * separator.join(arr) == str if maxSplit == -1 and keepEmpty == true
+ * separator.join(all callbacks) == str if keepEmpty == true.
  *
- * @param arr Where to put the substrings.
  * @param separator The string to split on.
  * @param maxSplit The maximum number of times the string is split.
  * @param keepEmpty True if empty substring should be added.
+ * @param func Function to call for each substring.
  */
-void split(std::string_view str, SmallVectorImpl<std::string_view>& arr,
-           std::string_view separator, int maxSplit = -1,
-           bool keepEmpty = true) noexcept;
+template <std::invocable<std::string_view> F>
+void split(std::string_view str, std::string_view separator, int maxSplit,
+           bool keepEmpty, F&& func) {
+  std::string_view s = str;
+
+  // Count down from maxSplit. When maxSplit is -1, this will just split
+  // "forever". This doesn't support splitting more than 2^31 times
+  // intentionally; if we ever want that we can make maxSplit a 64-bit integer
+  // but that seems unlikely to be useful.
+  while (maxSplit-- != 0) {
+    auto idx = s.find(separator);
+    if (idx == std::string_view::npos) {
+      break;
+    }
+
+    // Provide this split.
+    if (keepEmpty || idx > 0) {
+      func(slice(s, 0, idx));
+    }
+
+    // Jump forward.
+    s = slice(s, idx + separator.size(), std::string_view::npos);
+  }
+
+  // Provide the tail.
+  if (keepEmpty || !s.empty()) {
+    func(s);
+  }
+}
 
 /**
  * Splits @p str into substrings around the occurrences of a separator
  * character.
  *
- * Each substring is stored in @p arr. If @p maxSplit is >= 0, at most
- * @p maxSplit splits are done and consequently <= @p maxSplit + 1
- * elements are added to arr.
- * If @p keepEmpty is false, empty strings are not added to @p arr. They
- * still count when considering @p maxSplit
+ * Each substring is passed to the callback @p func. If @p maxSplit is >= 0, at
+ * most @p maxSplit splits are done and consequently <= @p maxSplit + 1
+ * elements are provided to the callback.
+ * If @p keepEmpty is false, empty strings are not included. They
+ * still count when considering @p maxSplit.
  * An useful invariant is that
- * separator.join(arr) == str if maxSplit == -1 and keepEmpty == true
+ * separator.join(all callbacks) == str if keepEmpty == true.
  *
- * @param arr Where to put the substrings.
  * @param separator The character to split on.
  * @param maxSplit The maximum number of times the string is split.
  * @param keepEmpty True if empty substring should be added.
+ * @param func Function to call for each substring.
  */
-void split(std::string_view str, SmallVectorImpl<std::string_view>& arr,
-           char separator, int maxSplit = -1, bool keepEmpty = true) noexcept;
+template <std::invocable<std::string_view> F>
+void split(std::string_view str, char separator, int maxSplit, bool keepEmpty,
+           F&& func) {
+  std::string_view s = str;
+
+  // Count down from maxSplit. When maxSplit is -1, this will just split
+  // "forever". This doesn't support splitting more than 2^31 times
+  // intentionally; if we ever want that we can make maxSplit a 64-bit integer
+  // but that seems unlikely to be useful.
+  while (maxSplit-- != 0) {
+    size_t idx = s.find(separator);
+    if (idx == std::string_view::npos) {
+      break;
+    }
+
+    // Provide this split.
+    if (keepEmpty || idx > 0) {
+      func(slice(s, 0, idx));
+    }
+
+    // Jump forward.
+    s = slice(s, idx + 1, std::string_view::npos);
+  }
+
+  // Provide the tail.
+  if (keepEmpty || !s.empty()) {
+    func(s);
+  }
+}
 
 /**
  * Returns @p str with consecutive @p ch characters starting from the