Replace <regex> with a helper function

bazelbuild · Oct 4, 2024 · b621e65 · b621e65
1 parent 4cbb7a0
commit b621e65
Show file tree

Hide file tree

Showing 4 changed files with 106 additions and 31 deletions.
diff --git a/MODULE.bazel.lock b/MODULE.bazel.lock
diff --git a/src/main/tools/build-runfiles-windows.cc b/src/main/tools/build-runfiles-windows.cc
@@ -20,7 +20,6 @@
 
 #include <fstream>
 #include <iostream>
-#include <regex>
 #include <sstream>
 #include <string>
 #include <unordered_map>
@@ -42,10 +41,6 @@ using std::wstring;
 
 namespace {
 
-const std::regex kEscapedBackslash(R"(\\b)");
-const std::regex kEscapedNewline(R"(\\n)");
-const std::regex kEscapedSpace(R"(\\s)");
-
 const wchar_t* manifest_filename;
 const wchar_t* runfiles_base_dir;
 
@@ -129,6 +124,39 @@ bool ReadSymlink(const wstring& abs_path, wstring* target, wstring* error) {
   return false;
 }
 
+// Replaces \s, \n, and \b with their respective characters.
+std::string Unescape(const std::string &path) {
+  std::string result;
+  result.reserve(path.size());
+  for (size_t i = 0; i < path.size(); ++i) {
+    if (path[i] == '\\' && i + 1 < path.size()) {
+      switch (path[i + 1]) {
+        case 's': {
+          result.push_back(' ');
+          break;
+        }
+        case 'n': {
+          result.push_back('\n');
+          break;
+        }
+        case 'b': {
+          result.push_back('\\');
+          break;
+        }
+        default: {
+          result.push_back(path[i]);
+          result.push_back(path[i + 1]);
+          break;
+        }
+      }
+      ++i;
+    } else {
+      result.push_back(path[i]);
+    }
+  }
+  return result;
+}
+
 }  // namespace
 
 class RunfilesCreator {
@@ -172,14 +200,9 @@ class RunfilesCreator {
         if (idx == string::npos) {
           die(L"Missing separator in manifest line: %hs", line.c_str());
         }
-        std::string link_path = line.substr(1, idx - 1);
-        link_path = std::regex_replace(link_path, kEscapedSpace, " ");
-        link_path = std::regex_replace(link_path, kEscapedNewline, "\n");
-        link_path = std::regex_replace(link_path, kEscapedBackslash, "\\");
+        std::string link_path = Unescape(line.substr(1, idx - 1));
         link = blaze_util::CstringToWstring(link_path);
-        std::string target_path = line.substr(idx + 1);
-        target_path = std::regex_replace(target_path, kEscapedNewline, "\n");
-        target_path = std::regex_replace(target_path, kEscapedBackslash, " ");
+        std::string target_path = Unescape(line.substr(idx + 1));
         target = blaze_util::CstringToWstring(target_path);
       } else {
         string::size_type idx = line.find(' ');

diff --git a/src/main/tools/build-runfiles.cc b/src/main/tools/build-runfiles.cc
@@ -108,6 +108,39 @@ struct FileInfo {
 
 typedef std::map<std::string, FileInfo> FileInfoMap;
 
+// Replaces \s, \n, and \b with their respective characters.
+std::string Unescape(const std::string &path) {
+  std::string result;
+  result.reserve(path.size());
+  for (size_t i = 0; i < path.size(); ++i) {
+    if (path[i] == '\\' && i + 1 < path.size()) {
+      switch (path[i + 1]) {
+        case 's': {
+          result.push_back(' ');
+          break;
+        }
+        case 'n': {
+          result.push_back('\n');
+          break;
+        }
+        case 'b': {
+          result.push_back('\\');
+          break;
+        }
+        default: {
+          result.push_back(path[i]);
+          result.push_back(path[i + 1]);
+          break;
+        }
+      }
+      ++i;
+    } else {
+      result.push_back(path[i]);
+    }
+  }
+  return result;
+}
+
 class RunfilesCreator {
  public:
   explicit RunfilesCreator(const std::string &output_base)
@@ -170,13 +203,8 @@ class RunfilesCreator {
         if (!s) {
           DIE("missing field delimiter at line %d: '%s'\n", lineno, buf);
         }
-        link = std::string(buf + 1, s);
-        link = std::regex_replace(link, kEscapedSpace, " ");
-        link = std::regex_replace(link, kEscapedNewline, "\n");
-        link = std::regex_replace(link, kEscapedBackslash, "\\");
-        target = s + 1;
-        target = std::regex_replace(target, kEscapedNewline, "\n");
-        target = std::regex_replace(target, kEscapedBackslash, "\\");
+        link = Unescape(std::string(buf + 1, s));
+        target = Unescape(s + 1);
       } else {
         // The line is of the form "foo /target/path", with only a single space
         // in the link path.

diff --git a/tools/cpp/runfiles/runfiles_src.cc b/tools/cpp/runfiles/runfiles_src.cc
@@ -29,7 +29,6 @@
 #include <fstream>
 #include <functional>
 #include <map>
-#include <regex>
 #include <sstream>
 #include <vector>
 
@@ -50,10 +49,6 @@ using std::vector;
 
 namespace {
 
-const std::regex kEscapedBackslash("\\\\b");
-const std::regex kEscapedNewline("\\\\n");
-const std::regex kEscapedSpace("\\\\s");
-
 bool starts_with(const string& s, const char* prefix) {
   if (!prefix || !*prefix) {
     return true;
@@ -184,6 +179,39 @@ string GetEnv(const string& key) {
 #endif
 }
 
+// Replaces \s, \n, and \b with their respective characters.
+string Unescape(const string &path) {
+  string result;
+  result.reserve(path.size());
+  for (size_t i = 0; i < path.size(); ++i) {
+    if (path[i] == '\\' && i + 1 < path.size()) {
+      switch (path[i + 1]) {
+        case 's': {
+          result.push_back(' ');
+          break;
+        }
+        case 'n': {
+          result.push_back('\n');
+          break;
+        }
+        case 'b': {
+          result.push_back('\\');
+          break;
+        }
+        default: {
+          result.push_back(path[i]);
+          result.push_back(path[i + 1]);
+          break;
+        }
+      }
+      ++i;
+    } else {
+      result.push_back(path[i]);
+    }
+  }
+  return result;
+}
+
 string Runfiles::Rlocation(const string& path) const {
   return Rlocation(path, source_repository_);
 }
@@ -274,13 +302,8 @@ bool ParseManifest(const string& path, map<string, string>* result,
         }
         return false;
       }
-      source = line.substr(1, idx - 1);
-      source = std::regex_replace(source, kEscapedSpace, " ");
-      source = std::regex_replace(source, kEscapedNewline, "\n");
-      source = std::regex_replace(source, kEscapedBackslash, "\\");
-      target = line.substr(idx + 1);
-      target = std::regex_replace(target, kEscapedNewline, "\n");
-      target = std::regex_replace(target, kEscapedBackslash, "\\");
+      source = Unescape(line.substr(1, idx - 1));
+      target = Unescape(line.substr(idx + 1));
     } else {
       string::size_type idx = line.find(' ');
       if (idx == string::npos) {