Version 0.0.9 - Delete obsolete reference, use std::distance and handle separators.

Author: Tomas Hagenau Andersen

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.20)
 
-project(OS2DSRules VERSION 0.0.8)
+project(OS2DSRules VERSION 0.0.9)
 
 # Detect available compiler on the host system.
 set(gcc_like_cxx "$<COMPILE_LANG_AND_ID:CXX,ARMClang,AppleClang,Clang,GNU,LCC>")

include/cpr-detector.hpp

@@ -21,7 +21,8 @@ constexpr bool is_digit(char c) noexcept { return '0' <= c && c <= '9'; }
 
 const auto is_separator = make_predicate(' ', '-', '/', '\t');
 
-const auto is_previous_ok = make_predicate(char(0), ' ', '.', ',', '\n', '\t', '\0');
+const auto is_previous_ok =
+    make_predicate(char(0), ' ', '.', ',', '\n', '\t', '\0');
 
 constexpr bool is_space(const char c) noexcept { return c == ' '; }
 
@@ -49,10 +50,11 @@ class CPRDetector {
   char update(char, CPRDetectorState, CPRDetectorState &, Predicate) noexcept;
   bool check_day_month(const std::string &, CPRDetectorState &) noexcept;
   void check_leap_year(const std::string &, CPRDetectorState &) noexcept;
-  void check_and_append_cpr(std::string &, MatchResults &, size_t,
-                            size_t) noexcept;
+  void check_and_append_cpr(std::string &, MatchResults &, size_t, size_t,
+                            char) noexcept;
   bool check_mod11(const MatchResult &) noexcept;
   bool examine_context(const std::string &) noexcept;
+  [[nodiscard]] std::string format_cpr(std::string &, char) const noexcept;
 
 public:
   constexpr CPRDetector(bool check_mod11 = false,

lib/cpr-detector.cpp

@@ -1,7 +1,7 @@
 #include <algorithm>
 #include <array>
-#include <cstddef>
 #include <cctype>
+#include <cstddef>
 #include <iterator>
 #include <numeric>
 #include <string_view>
@@ -25,22 +25,24 @@ static constexpr auto blacklist_words = std::to_array<std::string_view>(
 static const auto blacklist_words_set = FrozenHashSet(blacklist_words);
 }; // namespace
 
-static bool find_blacklisted_words(const std::string &content, const std::array<std::size_t, 4> indices) noexcept {
+static bool
+find_blacklisted_words(const std::string &content,
+                       const std::array<std::size_t, 4> indices) noexcept {
 
   for (std::size_t i = 1; i < 4; ++i) {
     for (std::size_t j = 0; j < i; ++j) {
       auto begin = indices[j];
       auto end = indices[i] - begin;
 
       if (end > content.size())
-	end = content.size() - begin - 1;
+        end = content.size() - begin - 1;
 
       std::string target = content.substr(begin, end);
       std::transform(target.begin(), target.end(), target.begin(),
                      [](unsigned char c) { return std::tolower(c); });
 
       if (blacklist_words_set.contains(target))
-	return true;
+        return true;
     }
   }
 
@@ -104,14 +106,25 @@ void CPRDetector::check_leap_year(const std::string &cpr,
     reset(state);
 }
 
+std::string CPRDetector::format_cpr(std::string &cpr,
+                                    char separator = 0) const noexcept {
+  if (separator == 0) {
+    return cpr;
+  } else {
+    return std::string(cpr, 0, 6) + separator + std::string(cpr, 6, 4);
+  }
+}
+
 void CPRDetector::check_and_append_cpr(std::string &cpr, MatchResults &results,
-                                       size_t begin, size_t end) noexcept {
+                                       size_t begin, size_t end,
+                                       char separator = 0) noexcept {
   // Convert the 4 control digits to an int.
   int control = std::stoi(std::string(cpr, 6, 4));
 
   // We reject the control sequence '0000'.
   if (control > 0) {
-    MatchResult result(cpr, begin, end, CPRDetector::sensitivity);
+    MatchResult result(format_cpr(cpr, separator), begin, end,
+                       CPRDetector::sensitivity);
 
     if (check_mod11_ && !check_mod11(result))
       return;
@@ -146,11 +159,11 @@ bool CPRDetector::examine_context(const std::string &content) noexcept {
       indices[4 - spaces] = i;
       --spaces;
       if (spaces == 0) {
-	if (find_blacklisted_words(content, indices))
-	  return true;
-	
-	spaces = 3;
-	indices[0] = indices[3] + 1;
+        if (find_blacklisted_words(content, indices))
+          return true;
+
+        spaces = 3;
+        indices[0] = indices[3] + 1;
       }
     }
   }
@@ -176,8 +189,9 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
   CPRDetectorState state = CPRDetectorState::Empty;
   std::string cpr(10, 0);
   char previous = 0;
-  std::size_t count = 0;
+  char separator = 0;
   std::size_t begin = 0;
+  std::size_t end = 0;
   bool allow_separator, leap_year = false;
   Predicate is_acceptable = [](char) { return false; };
 
@@ -195,7 +209,8 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
 
       if (state == CPRDetectorState::First) {
         cpr[0] = *it;
-        begin = count;
+        begin =
+            static_cast<std::size_t>(std::distance(std::begin(content), it));
       }
 
       break;
@@ -209,13 +224,11 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
       } else {
         reset(state);
         previous = *it;
-        ++count;
         continue;
       }
 
       previous = cpr[1] =
           update(*it, CPRDetectorState::Second, state, is_acceptable);
-
       if (previous != 0)
         // Next time, we allow a space.
         allow_separator = true;
@@ -235,7 +248,6 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
       } else {
         reset(state);
         previous = 0;
-        ++count;
         continue;
       }
 
@@ -274,8 +286,8 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
     case CPRDetectorState::Sixth:
       if (allow_separator && is_separator(*it)) {
         // Skip one of the valid separator characters.
+        separator = *it;
         allow_separator = false;
-	++count;
         continue;
       }
 
@@ -304,17 +316,16 @@ MatchResults CPRDetector::find_matches(const std::string &content) noexcept {
       cpr[9] = update(*it, CPRDetectorState::Match, state, is_acceptable);
 
       auto ahead = it;
-      if (is_previous_ok(*(++ahead)))
-        check_and_append_cpr(cpr, results, begin, count);
-
+      if (is_previous_ok(*(++ahead))) {
+        end = static_cast<std::size_t>(std::distance(std::begin(content), it));
+        check_and_append_cpr(cpr, results, begin, end, separator);
+      }
       previous = *it;
       allow_separator = false;
       reset(state);
 
       break;
     }
-
-    ++count;
   }
 
   return results;

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "os2ds-rules"
-version = "0.0.8"
+version = "0.0.9"
 authors = [
   { name="HackTheOxidation", email="[email protected]" },
 ]