Rework regex processor to be more betterer

processor/regex.go

@@ -50,13 +50,13 @@ func (p *RegexProcessor) ToLua(L *lua.LState, data interface{}) error {
 	}
 
 	// Set variables for each capture group, starting from v1/s1 for the first capture
-	for i := 1; i < len(captures); i++ {
+	for i := 0; i < len(captures); i++ {
 		// Set string version (always available as s1, s2, etc.)
-		L.SetGlobal(fmt.Sprintf("s%d", i), lua.LString(captures[i]))
+		L.SetGlobal(fmt.Sprintf("s%d", i+1), lua.LString(captures[i]))
 
 		// Try to convert to number and set v1, v2, etc.
 		if val, err := strconv.ParseFloat(captures[i], 64); err == nil {
-			L.SetGlobal(fmt.Sprintf("v%d", i), lua.LNumber(val))
+			L.SetGlobal(fmt.Sprintf("v%d", i+1), lua.LNumber(val))
 		}
 	}
 
@@ -111,70 +111,86 @@ func (p *RegexProcessor) ProcessContent(content string, pattern string, luaExpr
 	luaExpr = BuildLuaScript(luaExpr)
 	fmt.Printf("Changing Lua expression from: %s to: %s\n", previous, luaExpr)
 
-	L := lua.NewState()
+	L, err := NewLuaState()
+	if err != nil {
+		return "", 0, 0, fmt.Errorf("error creating Lua state: %v", err)
+	}
 	defer L.Close()
 
 	// Initialize Lua environment
 	modificationCount := 0
-	matchCount := 0
-
-	// Load math library
-	L.Push(L.GetGlobal("require"))
-	L.Push(lua.LString("math"))
-	if err := L.PCall(1, 1, nil); err != nil {
-		return content, 0, 0, fmt.Errorf("error loading Lua math library: %v", err)
-	}
-
-	// Initialize helper functions
-	if err := InitLuaHelpers(L); err != nil {
-		return content, 0, 0, err
-	}
 
 	// Process all regex matches
-	result := compiledPattern.ReplaceAllStringFunc(content, func(match string) string {
-		matchCount++
-		captures := compiledPattern.FindStringSubmatch(match)
-		if len(captures) <= 1 {
-			// No capture groups, return unchanged
+	result := content
+	indices := compiledPattern.FindAllStringSubmatchIndex(content, -1)
+	// We walk backwards because we're replacing something with something else that might be longer
+	// And in the case it is longer than the original all indicces past that change will be fucked up
+	// By going backwards we fuck up all the indices to the end of the file that we don't care about
+	// Because there either aren't any (last match) or they're already modified (subsequent matches)
+	for i := len(indices) - 1; i >= 0; i-- {
+		matchIndices := indices[i]
+		// Why we're doing this whole song and dance of indices is to properly handle empty matches
+		// Plus it's a little cleaner to surgically replace our matches
+		// If we were to use string.replace and encountered an empty match there'd be nothing to replace
+		// But using indices an empty match would have its starting and ending indices be the same
+		// So when we're cutting open the array we say 0:7 + modified + 7:end
+		// As if concatenating in the middle of the array
+		// Plus it supports lookarounds
+		match := content[matchIndices[0]:matchIndices[1]]
+
+		groups := matchIndices[2:]
+		if len(groups) <= 0 {
 			fmt.Println("No capture groups for lua to chew on")
-			return match
+			continue
+		}
+		if len(groups)%2 == 1 {
+			fmt.Println("Odd number of indices of groups, what the fuck?")
+			continue
+		}
+
+		captures := make([]string, 0, len(groups)/2)
+		for j := 0; j < len(groups); j += 2 {
+			captures = append(captures, content[groups[j]:groups[j+1]])
 		}
 
 		if err := p.ToLua(L, captures); err != nil {
 			fmt.Println("Error setting Lua variables:", err)
-			return match
+			continue
 		}
 
-		// Execute the user's Lua code
 		if err := L.DoString(luaExpr); err != nil {
-			fmt.Println("Error executing Lua code:", err)
-			return match // Return unchanged on error
+			fmt.Printf("Error executing Lua code %s for group %s: %v", luaExpr, captures, err)
+			continue
 		}
 
 		// Get modifications from Lua
 		modResult, err := p.FromLua(L)
 		if err != nil {
 			fmt.Println("Error getting modifications:", err)
-			return match
+			continue
 		}
 
 		// Apply modifications to the matched text
 		modsMap, ok := modResult.(map[int]string)
 		if !ok || len(modsMap) == 0 {
 			fmt.Println("No modifications to apply")
-			return match // No changes
+			continue
 		}
 
 		// Apply the modifications to the original match
-		result := match
-		for i, newVal := range modsMap {
-			oldVal := captures[i+1]
-			result = strings.Replace(result, oldVal, newVal, 1)
+		replacement := match
+		for i := len(modsMap) - 1; i >= 0; i-- {
+			newVal := modsMap[i]
+			// Indices of the group are relative to content
+			// To relate them to match we have to subtract the match start index
+			groupStart := groups[i*2] - matchIndices[0]
+			groupEnd := groups[i*2+1] - matchIndices[0]
+			replacement = replacement[:groupStart] + newVal + replacement[groupEnd:]
 		}
 
 		modificationCount++
-		return result
-	})
+		result = result[:matchIndices[0]] + replacement + result[matchIndices[1]:]
+	}
 
-	return result, modificationCount, matchCount, nil
+	return result, modificationCount, len(indices), nil
 }

processor/regex_test.go

@@ -181,7 +181,7 @@ func TestArrayNotation(t *testing.T) {
 	}
 }
 
-func TestMultipleMatches(t *testing.T) {
+func TestMultipleNumericMatches(t *testing.T) {
 	content := `<data>
     <entry>50</entry>
     <entry>100</entry>
@@ -212,19 +212,21 @@ func TestMultipleMatches(t *testing.T) {
 	if result != expected {
 		t.Errorf("Expected content to be:\n%s\n\nGot:\n%s", expected, result)
 	}
+}
 
-	// Test string operations
-	content = `<data>
+func TestMultipleStringMatches(t *testing.T) {
+	content := `<data>
     <name>John</name>
     <name>Mary</name>
 </data>`
 
-	expected = `<data>
+	expected := `<data>
     <name>John_modified</name>
     <name>Mary_modified</name>
 </data>`
 
-	result, mods, matches, err = p.ProcessContent(content, `<name>([A-Za-z]+)</name>`, `s1 = s1 .. "_modified"`)
+	p := &RegexProcessor{}
+	result, mods, matches, err := p.ProcessContent(content, `<name>([A-Za-z]+)</name>`, `s1 = s1 .. "_modified"`)
 
 	if err != nil {
 		t.Fatalf("Error processing content: %v", err)
@@ -243,7 +245,7 @@ func TestMultipleMatches(t *testing.T) {
 	}
 }
 
-func TestStringOperations(t *testing.T) {
+func TestStringUpperCase(t *testing.T) {
 	content := `<users>
     <user>John</user>
     <user>Mary</user>
@@ -273,19 +275,21 @@ func TestStringOperations(t *testing.T) {
 	if result != expected {
 		t.Errorf("Expected content to be:\n%s\n\nGot:\n%s", expected, result)
 	}
+}
 
-	// Test string concatenation
-	content = `<products>
+func TestStringConcatenation(t *testing.T) {
+	content := `<products>
     <product>Apple</product>
     <product>Banana</product>
 </products>`
 
-	expected = `<products>
+	expected := `<products>
     <product>Apple_fruit</product>
     <product>Banana_fruit</product>
 </products>`
 
-	result, mods, matches, err = p.ProcessContent(content, `<product>([A-Za-z]+)</product>`, `s1 = s1 .. "_fruit"`)
+	p := &RegexProcessor{}
+	result, mods, matches, err := p.ProcessContent(content, `<product>([A-Za-z]+)</product>`, `s1 = s1 .. "_fruit"`)
 
 	if err != nil {
 		t.Fatalf("Error processing content: %v", err)