Extract some more find/replace logic to libaegisub

libaegisub/common/util.cpp

@@ -52,8 +52,24 @@ std::pair<size_t, size_t> find_range(std::string const& haystack, std::string co
 	return {match_start, match_start + needle.size()};
 }
 
+void parse_blocks(std::vector<std::pair<size_t, size_t>>& blocks, std::string const& str) {
+	blocks.clear();
+
+	size_t ovr_start = bad_pos;
+	size_t i = 0;
+	for (auto const& c : str) {
+		if (c == '{' && ovr_start == bad_pos)
+			ovr_start = i;
+		else if (c == '}' && ovr_start != bad_pos) {
+			blocks.emplace_back(ovr_start, i);
+			ovr_start = bad_pos;
+		}
+		++i;
+	}
 }
 
+} // anonymous namespace
+
 namespace agi { namespace util {
 
 std::string strftime(const char *fmt, const tm *tmptr) {
@@ -129,12 +145,60 @@ std::pair<size_t, size_t> ifind(std::string const& haystack, std::string const&
 	return ret;
 }
 
+std::string tagless_find_helper::strip_tags(std::string const& str, size_t s) {
+	parse_blocks(blocks, str);
+
+	std::string out;
+
+	size_t last = s;
+	for (auto const& block : blocks) {
+		if (block.second < s) continue;
+		if (block.first > last)
+			out.append(str.begin() + last, str.begin() + block.first);
+		last = block.second + 1;
+	}
+
+	if (last < str.size())
+		out.append(str.begin() + last, str.end());
+
+	start = s;
+	return out;
 }
 
+void tagless_find_helper::map_range(size_t &s, size_t &e) {
+	s += start;
+	e += start;
+
+	// Shift the start and end of the match to be relative to the unstripped
+	// match
+	for (auto const& block : blocks) {
+		// Any blocks before start are irrelevant as they're included in `start`
+		if (block.second < s) continue;
+		// Skip over blocks at the very beginning of the match
+		// < should only happen if the cursor was within an override block
+		// when the user started a search
+		if (block.first <= s) {
+			size_t len = block.second - std::max(block.first, s) + 1;
+			s += len;
+			e += len;
+			continue;
+		}
+
+		assert(block.first > s);
+		// Blocks after the match are irrelevant
+		if (block.first >= e) break;
+
+		// Extend the match to include blocks within the match
+		// Note that blocks cannot be partially within the match
+		e += block.second - block.first + 1;
+	}
+}
+} // namespace util
+
 #ifndef __APPLE__
 namespace osx {
 AppNapDisabler::AppNapDisabler(std::string reason) { }
 AppNapDisabler::~AppNapDisabler() { }
 }
 #endif
-}
+} // namespace agi

libaegisub/include/libaegisub/util.h

@@ -15,11 +15,11 @@
 #include <algorithm>
 #include <boost/range/irange.hpp>
 #include <string>
+#include <vector>
 
 struct tm;
 
-namespace agi {
-	namespace util {
+namespace agi { namespace util {
 	/// Clamp `b` to the range [`a`,`c`]
 	template<typename T>
 	static inline T mid(T a, T b, T c) {
@@ -45,6 +45,20 @@ namespace agi {
 	/// based on the unfolded length.
 	std::pair<size_t, size_t> ifind(std::string const& haystack, std::string const& needle);
 
+	class tagless_find_helper {
+		std::vector<std::pair<size_t, size_t>> blocks;
+		size_t start = 0;
+
+	public:
+		/// Strip ASS override tags at or after `start` in `str`, and initialize
+		/// state for mapping ranges back to the input string
+		std::string strip_tags(std::string const& str, size_t start);
+
+		/// Convert a range in the string returned by `strip_tags()` to a range
+		/// int the string last passed to `strip_tags()`
+		void map_range(size_t& start, size_t& end);
+	};
+
 	/// Set the name of the calling thread in the Visual Studio debugger
 	/// @param name New name for the thread
 	void SetThreadName(const char *name);
@@ -66,6 +80,4 @@ namespace agi {
 	auto range(Integer end) -> decltype(boost::irange<Integer>(0, end)) {
 		return boost::irange<Integer>(0, end);
 	}
-
-	} // namespace util
-} // namespace agi
+} } // namespace agi::util

src/search_replace_engine.cpp

@@ -73,76 +73,17 @@ public:
 
 class skip_tags_accessor {
 	boost::flyweight<std::string> AssDialogueBase::*field;
-	std::vector<std::pair<size_t, size_t>> blocks;
-	size_t start = 0;
-
-	void parse_str(std::string const& str) {
-		blocks.clear();
-
-		size_t ovr_start = bad_pos;
-		size_t i = 0;
-		for (auto const& c : str) {
-			if (c == '{' && ovr_start == bad_pos)
-				ovr_start = i;
-			else if (c == '}' && ovr_start != bad_pos) {
-				blocks.emplace_back(ovr_start, i);
-				ovr_start = bad_pos;
-			}
-			++i;
-		}
-	}
+	agi::util::tagless_find_helper helper;
 
 public:
 	skip_tags_accessor(SearchReplaceSettings::Field f) : field(get_dialogue_field(f)) { }
 
 	std::string get(const AssDialogue *d, size_t s) {
-		auto const& str = get_normalized(d, field);
-		parse_str(str);
-
-		std::string out;
-
-		size_t last = s;
-		for (auto const& block : blocks) {
-			if (block.second < s) continue;
-			if (block.first > last)
-				out.append(str.begin() + last, str.begin() + block.first);
-			last = block.second + 1;
-		}
-
-		if (last < str.size())
-			out.append(str.begin() + last, str.end());
-
-		start = s;
-		return out;
+		return helper.strip_tags(get_normalized(d, field), s);
 	}
 
 	MatchState make_match_state(size_t s, size_t e, boost::u32regex *r = nullptr) {
-		s += start;
-		e += start;
-
-		// Shift the start and end of the match to be relative to the unstripped
-		// match
-		for (auto const& block : blocks) {
-			// Any blocks before start are irrelevant as they're included in `start`
-			if (block.second < s) continue;
-			// Skip over blocks at the very beginning of the match
-			// < should only happen if the cursor was within an override block
-			// when the user started a search
-			if (block.first <= s) {
-				size_t len = block.second - std::max(block.first, s) + 1;
-				s += len;
-				e += len;
-				continue;
-			}
-
-			assert(block.first > s);
-			// Blocks after the match are irrelevant
-			if (block.first >= e) break;
-
-			// Extend the match to include blocks within the match
-			// Note that blocks cannot be partially within the match
-			e += block.second - block.first + 1;
-		}
+		helper.map_range(s, e);
 		return {r, s, e};
 	}
 };