Merge pull request #104 from dpaulat/feature/re2

Use RE2 in favor of std::regex
dpaulat · Dec 13, 2023 · 269cbde · 269cbde
2 parents 949c983 + 2bd5ec8
commit 269cbde
Show file tree

Hide file tree

Showing 10 changed files with 56 additions and 58 deletions.
diff --git a/ACKNOWLEDGEMENTS.md b/ACKNOWLEDGEMENTS.md
@@ -35,6 +35,7 @@ Supercell Wx uses code from the following dependencies:
 | [nunicode](https://bitbucket.org/alekseyt/nunicode/src/master/) | [MIT License](https://spdx.org/licenses/MIT.html) | Modified for MapLibre Native |
 | [OpenSSL](https://www.openssl.org/) | [OpenSSL License](https://spdx.org/licenses/OpenSSL.html) |
 | [Qt](https://www.qt.io/) | [GNU Lesser General Public License v3.0 only](https://spdx.org/licenses/LGPL-3.0-only.html) | Qt Core, Qt GUI, Qt Multimedia, Qt Network, Qt OpenGL, Qt Positioning, Qt SQL, Qt SVG, Qt Widgets<br/>Additional Licenses: https://doc.qt.io/qt-6/licenses-used-in-qt.html |
+| [re2](https://github.com/google/re2) | [BSD 3-Clause "New" or "Revised" License](https://spdx.org/licenses/BSD-3-Clause.html) |
 | [spdlog](https://github.com/gabime/spdlog) | [MIT License](https://spdx.org/licenses/MIT.html) |
 | [SQLite](https://www.sqlite.org/) | Public Domain |
 | [stb](https://github.com/nothings/stb) | Public Domain |

diff --git a/conanfile.py b/conanfile.py
@@ -13,6 +13,7 @@ class SupercellWxConan(ConanFile):
                   "libcurl/8.4.0",
                   "libxml2/2.12.2",
                   "openssl/3.2.0",
+                  "re2/20231101",
                   "spdlog/1.12.0",
                   "sqlite3/3.44.2",
                   "vulkan-loader/1.3.243.0",

diff --git a/scwx-qt/source/scwx/qt/manager/update_manager.cpp b/scwx-qt/source/scwx/qt/manager/update_manager.cpp
@@ -2,10 +2,10 @@
 #include <scwx/util/logger.hpp>
 
 #include <mutex>
-#include <regex>
 
 #include <boost/json.hpp>
 #include <cpr/cpr.h>
+#include <re2/re2.h>
 
 namespace scwx
 {
@@ -61,16 +61,10 @@ std::string UpdateManager::latest_version() const
 std::string
 UpdateManager::Impl::GetVersionString(const std::string& releaseName)
 {
-   static const std::regex re {"\\d+\\.\\d+\\.\\d+"};
-   std::string             versionString {};
-   std::smatch             m;
+   static constexpr LazyRE2 re = {"(\\d+\\.\\d+\\.\\d+)"};
+   std::string              versionString {};
 
-   std::regex_search(releaseName, m, re);
-
-   if (!m.empty())
-   {
-      versionString = m[0].str();
-   }
+   RE2::PartialMatch(releaseName, *re, &versionString);
 
    return versionString;
 }

diff --git a/scwx-qt/source/scwx/qt/map/map_widget.cpp b/scwx-qt/source/scwx/qt/map/map_widget.cpp
@@ -23,7 +23,6 @@
 #include <scwx/util/logger.hpp>
 #include <scwx/util/time.hpp>
 
-#include <regex>
 #include <set>
 
 #include <backends/imgui_impl_opengl3.h>
@@ -33,6 +32,7 @@
 #include <boost/uuid/random_generator.hpp>
 #include <fmt/format.h>
 #include <imgui.h>
+#include <re2/re2.h>
 #include <QApplication>
 #include <QColor>
 #include <QDebug>
@@ -769,14 +769,14 @@ std::string MapWidgetImpl::FindMapSymbologyLayer()
       const std::string layer = qlayer.toStdString();
 
       // Draw below layers defined in map style
-      auto it = std::find_if(
-         currentStyle_->drawBelow_.cbegin(),
-         currentStyle_->drawBelow_.cend(),
-         [&layer](const std::string& styleLayer) -> bool
-         {
-            std::regex re {styleLayer, std::regex_constants::icase};
-            return std::regex_match(layer, re);
-         });
+      auto it = std::find_if(currentStyle_->drawBelow_.cbegin(),
+                             currentStyle_->drawBelow_.cend(),
+                             [&layer](const std::string& styleLayer) -> bool
+                             {
+                                // Perform case-insensitive matching
+                                RE2 re {"(?i)" + styleLayer};
+                                return RE2::FullMatch(layer, re);
+                             });
 
       if (it != currentStyle_->drawBelow_.cend())
       {

diff --git a/scwx-qt/source/scwx/qt/settings/palette_settings.cpp b/scwx-qt/source/scwx/qt/settings/palette_settings.cpp
@@ -2,10 +2,9 @@
 #include <scwx/qt/settings/settings_variable.hpp>
 #include <scwx/qt/util/color.hpp>
 
-#include <regex>
-
 #include <boost/gil.hpp>
 #include <fmt/format.h>
+#include <re2/re2.h>
 
 namespace scwx
 {
@@ -134,8 +133,8 @@ class PaletteSettings::Impl
 
 bool PaletteSettings::Impl::ValidateColor(const std::string& value)
 {
-   static const std::regex re {"#[0-9A-Za-z]{8}"};
-   return std::regex_match(value, re);
+   static constexpr LazyRE2 re = {"#[0-9A-Fa-f]{8}"};
+   return RE2::FullMatch(value, *re);
 }
 
 PaletteSettings::PaletteSettings() :

diff --git a/test/source/scwx/qt/map/map_provider.test.cpp b/test/source/scwx/qt/map/map_provider.test.cpp
@@ -2,13 +2,12 @@
 #include <scwx/util/environment.hpp>
 #include <scwx/util/logger.hpp>
 
-#include <regex>
-
 #include <QCoreApplication>
 #include <QMapLibreGL/QMapLibreGL>
 #include <QTimer>
 
 #include <gtest/gtest.h>
+#include <re2/re2.h>
 
 namespace scwx
 {
@@ -108,14 +107,15 @@ TEST_P(ByMapProviderTest, MapProviderLayers)
             const std::string layer = qlayer.toStdString();
 
             // Draw below layers defined in map style
-            auto it = std::find_if(
-               mapStyle.drawBelow_.cbegin(),
-               mapStyle.drawBelow_.cend(),
-               [&layer](const std::string& styleLayer) -> bool
-               {
-                  std::regex re {styleLayer, std::regex_constants::icase};
-                  return std::regex_match(layer, re);
-               });
+            auto it =
+               std::find_if(mapStyle.drawBelow_.cbegin(),
+                            mapStyle.drawBelow_.cend(),
+                            [&layer](const std::string& styleLayer) -> bool
+                            {
+                               // Perform case insensitive matching
+                               RE2 re {"(?i)" + styleLayer};
+                               return RE2::FullMatch(layer, re);
+                            });
 
             if (it != mapStyle.drawBelow_.cend())
             {

diff --git a/wxdata/source/scwx/awips/text_product_message.cpp b/wxdata/source/scwx/awips/text_product_message.cpp
@@ -4,11 +4,11 @@
 #include <scwx/util/streams.hpp>
 
 #include <istream>
-#include <regex>
 #include <string>
 
 #include <boost/algorithm/string/replace.hpp>
 #include <boost/algorithm/string/trim.hpp>
+#include <re2/re2.h>
 
 namespace scwx
 {
@@ -24,7 +24,7 @@ static const auto        logger_    = scwx::util::Logger::Create(logPrefix_);
 // Segment Header only:
 // * <hhmm>_xM_<tz1>_day_mon_<dd>_year_/<hhmm>_xM_<tz2>_day_mon_<dd>_year/
 // Look for hhmm (xM|UTC) to key the date/time string
-static const std::regex reDateTimeString {"^[0-9]{3,4} ([AP]M|UTC)"};
+static constexpr LazyRE2 reDateTimeString = {"^[0-9]{3,4} ([AP]M|UTC)"};
 
 static void ParseCodedInformation(std::shared_ptr<Segment> segment,
                                   const std::string&       wfo);
@@ -333,7 +333,7 @@ void ParseCodedInformation(std::shared_ptr<Segment> segment,
       else if (codedMotionBegin != productContent.cend() &&
                codedMotionEnd == productContent.cend() &&
                !it->starts_with(" ") &&
-               !std::regex_search(*it, std::regex {"^[0-9]"})
+               !(it->length() > 0 && std::isdigit(it->at(0)))
                /* Continuation lines */)
       {
          codedMotionEnd = it;
@@ -448,7 +448,7 @@ std::vector<std::string> TryParseMndHeader(std::istream& is)
    }
 
    if (!mndHeader.empty() &&
-       !std::regex_search(mndHeader.back(), reDateTimeString))
+       !RE2::PartialMatch(mndHeader.back(), *reDateTimeString))
    {
       // MND Header should end with an Issuance Date/Time Line
       mndHeader.clear();
@@ -489,23 +489,23 @@ std::optional<SegmentHeader> TryParseSegmentHeader(std::istream& is)
    // UGC takes the form SSFNNN-NNN>NNN-SSFNNN-DDHHMM- (NWSI 10-1702)
    // Look for SSF(NNN)?[->] to key the UGC string
    // Look for DDHHMM- to end the UGC string
-   static const std::regex reUgcString {"^[A-Z]{2}[CZ]([0-9]{3})?[->]"};
-   static const std::regex reUgcExpiration {"[0-9]{6}-$"};
+   static constexpr LazyRE2 reUgcString     = {"^[A-Z]{2}[CZ]([0-9]{3})?[->]"};
+   static constexpr LazyRE2 reUgcExpiration = {"[0-9]{6}-$"};
 
    std::optional<SegmentHeader> header = std::nullopt;
    std::string                  line;
    std::streampos               isBegin = is.tellg();
 
    util::getline(is, line);
 
-   if (std::regex_search(line, reUgcString))
+   if (RE2::PartialMatch(line, *reUgcString))
    {
       header = SegmentHeader();
       header->ugcString_.push_back(line);
 
       // If UGC is multi-line, continue parsing
       while (!is.eof() && is.peek() != '\r' &&
-             !std::regex_search(line, reUgcExpiration))
+             !RE2::PartialMatch(line, *reUgcExpiration))
       {
          util::getline(is, line);
          header->ugcString_.push_back(line);
@@ -526,7 +526,7 @@ std::optional<SegmentHeader> TryParseSegmentHeader(std::istream& is)
       while (!is.eof() && is.peek() != '\r')
       {
          util::getline(is, line);
-         if (!std::regex_search(line, reDateTimeString))
+         if (!RE2::PartialMatch(line, *reDateTimeString))
          {
             header->ugcNames_.push_back(line);
          }
@@ -553,20 +553,20 @@ std::optional<Vtec> TryParseVtecString(std::istream& is)
    // P-VTEC takes the form /k.aaa.cccc.pp.s.####.yymmddThhnnZB-yymmddThhnnZE/
    // (NWSI 10-1703)
    // Look for /k. to key the P-VTEC string
-   static const std::regex rePVtecString {"^/[OTEX]\\."};
+   static constexpr LazyRE2 rePVtecString = {"^/[OTEX]\\."};
 
    // H-VTEC takes the form
    // /nwsli.s.ic.yymmddThhnnZB.yymmddThhnnZC.yymmddThhnnZE.fr/ (NWSI 10-1703)
    // Look for /nwsli. to key the H-VTEC string
-   static const std::regex reHVtecString {"^/[A-Z0-9]{5}\\."};
+   static constexpr LazyRE2 reHVtecString = {"^/[A-Z0-9]{5}\\."};
 
    std::optional<Vtec> vtec = std::nullopt;
    std::string         line;
    std::streampos      isBegin = is.tellg();
 
    util::getline(is, line);
 
-   if (std::regex_search(line, rePVtecString))
+   if (RE2::PartialMatch(line, *rePVtecString))
    {
       bool vtecValid;
 
@@ -577,7 +577,7 @@ std::optional<Vtec> TryParseVtecString(std::istream& is)
 
       util::getline(is, line);
 
-      if (std::regex_search(line, reHVtecString))
+      if (RE2::PartialMatch(line, *reHVtecString))
       {
          vtec->hVtec_.swap(line);
       }

diff --git a/wxdata/source/scwx/awips/ugc.cpp b/wxdata/source/scwx/awips/ugc.cpp
@@ -2,12 +2,12 @@
 #include <scwx/util/logger.hpp>
 
 #include <map>
-#include <regex>
 
 #include <boost/assign.hpp>
 #include <boost/bimap.hpp>
 #include <boost/bimap/unordered_set_of.hpp>
 #include <boost/tokenizer.hpp>
+#include <re2/re2.h>
 
 namespace scwx
 {
@@ -104,10 +104,10 @@ bool Ugc::Parse(const std::vector<std::string>& ugcString)
    bool dataValid = false;
 
    // UGC takes the form SSFNNN-NNN>NNN-SSFNNN-DDHHMM- (NWSI 10-1702)
-   static const std::regex reStart {"[A-Z]{2}[CZ]([0-9]{3}|ALL)"};
-   static const std::regex reAnyFipsId {"([0-9]{3}|ALL)"};
-   static const std::regex reSpecificFipsId {"(?!0{3})[0-9]{3}"};
-   static const std::regex reProductExpiration {"[0-9]{6}"};
+   static constexpr LazyRE2 reStart          = {"[A-Z]{2}[CZ]([0-9]{3}|ALL)"};
+   static constexpr LazyRE2 reAnyFipsId      = {"([0-9]{3}|ALL)"};
+   static constexpr LazyRE2 reSpecificFipsId = {"[0-9]{3}"};
+   static constexpr LazyRE2 reProductExpiration = {"[0-9]{6}"};
 
    std::stringstream ugcStream;
    for (auto& line : ugcString)
@@ -131,7 +131,7 @@ bool Ugc::Parse(const std::vector<std::string>& ugcString)
    for (auto& token : tokens)
    {
       // Product Expiration is the final token
-      if (std::regex_match(token, reProductExpiration))
+      if (RE2::FullMatch(token, *reProductExpiration))
       {
          p->productExpiration_ = token;
          dataValid             = true;
@@ -153,7 +153,7 @@ bool Ugc::Parse(const std::vector<std::string>& ugcString)
 
       // Look for the start of the UGC string (may be multiple per UGC string
       // for multiple states, territories, or marine area)
-      if (std::regex_match(firstToken, reStart))
+      if (RE2::FullMatch(firstToken, *reStart))
       {
          currentState  = firstToken.substr(0, 2);
          currentFormat = ugcFormatMap_.right.at(firstToken.at(2));
@@ -167,7 +167,7 @@ bool Ugc::Parse(const std::vector<std::string>& ugcString)
       }
       // Look for additional FIPS IDs in the UGC string
       else if (!currentState.empty() &&
-               std::regex_match(firstToken, reAnyFipsId))
+               RE2::FullMatch(firstToken, *reAnyFipsId))
       {
          firstFipsId = firstToken;
       }
@@ -188,7 +188,8 @@ bool Ugc::Parse(const std::vector<std::string>& ugcString)
       {
          std::string secondToken {(++tokenIt).current_token()};
 
-         if (std::regex_match(secondToken, reSpecificFipsId))
+         if (RE2::FullMatch(secondToken, *reSpecificFipsId) &&
+             secondToken != "000")
          {
             secondFipsId = secondToken;
          }

diff --git a/wxdata/source/scwx/provider/warnings_provider.cpp b/wxdata/source/scwx/provider/warnings_provider.cpp
@@ -3,7 +3,6 @@
 #include <scwx/util/logger.hpp>
 
 #include <ranges>
-#include <regex>
 #include <shared_mutex>
 
 #if defined(_MSC_VER)
@@ -13,6 +12,7 @@
 #define LIBXML_HTML_ENABLED
 #include <cpr/cpr.h>
 #include <libxml/HTMLparser.h>
+#include <re2/re2.h>
 
 #if !defined(_MSC_VER)
 #   include <date/date.h>
@@ -77,7 +77,7 @@ WarningsProvider::ListFiles(std::chrono::system_clock::time_point newerThan)
    using namespace date;
 #endif
 
-   static const std::regex reWarningsFilename {
+   static constexpr LazyRE2 reWarningsFilename = {
       "warnings_[0-9]{8}_[0-9]{2}.txt"};
    static const std::string dateTimeFormat {"warnings_%Y%m%d_%H.txt"};
 
@@ -101,7 +101,7 @@ WarningsProvider::ListFiles(std::chrono::system_clock::time_point newerThan)
          [](auto& record)
          {
             return record.type_ == std::filesystem::file_type::regular &&
-                   std::regex_match(record.filename_, reWarningsFilename);
+                   RE2::FullMatch(record.filename_, *reWarningsFilename);
          });
 
    std::unique_lock lock(p->filesMutex_);

diff --git a/wxdata/wxdata.cmake b/wxdata/wxdata.cmake
@@ -5,6 +5,7 @@ project(scwx-data)
 find_package(Boost)
 find_package(cpr)
 find_package(LibXml2)
+find_package(re2)
 find_package(spdlog)
 
 if (NOT MSVC)
@@ -268,6 +269,7 @@ target_link_libraries(wxdata PUBLIC aws-cpp-sdk-core
                                     aws-cpp-sdk-s3
                                     cpr::cpr
                                     LibXml2::LibXml2
+                                    re2::re2
                                     spdlog::spdlog
                                     units::units)
 target_link_libraries(wxdata INTERFACE Boost::iostreams