new function for splitting game data into segments

src/lib.rs

@@ -108,9 +108,34 @@ fn transform_line_endings(input: String, mode: TransformMode) -> String {
     String::from_utf8(output.into_inner()).unwrap()
 }
 
+fn segments_from_string(string: String) -> Vec<String> {
+    let mut output:Vec<String> = Vec::new();
+    // are we inside `"""\n...\n"""`? if so, ignore empty lines
+    let mut inside_escaped_block = false;
+    let mut current_segment : Vec<String> = Vec::new();
+    let lines: Vec<&str> = string.lines().collect();
+
+    for line in lines {
+        if line == "\"\"\"" {
+            inside_escaped_block = ! inside_escaped_block;
+        }
+
+        if line == "" && !inside_escaped_block {
+            output.push(current_segment.join("\n"));
+            current_segment = Vec::new();
+        } else {
+            current_segment.push(line.to_string());
+        }
+    }
+
+    output.push(current_segment.join("\n"));
+
+    output
+}
+
 #[cfg(test)]
 mod test {
-    use crate::{from_base36, ToBase36, optional_data_line, mock};
+    use crate::{from_base36, ToBase36, optional_data_line, mock, segments_from_string};
 
     #[test]
     fn test_from_base36() {
@@ -129,4 +154,20 @@ mod test {
         let output = optional_data_line("NAME", mock::item().name);
         assert_eq!(output, "\nNAME door".to_string());
     }
+
+    #[test]
+    fn test_string_to_segments() {
+        let output = segments_from_string(
+            include_str!("./test-resources/segments").to_string()
+        );
+
+        let expected = vec![
+            "\"\"\"\nthe first segment is a long bit of text\n\n\nit contains empty lines\n\n\"\"\"".to_string(),
+            "this is a new segment\nthis is still the second segment\nblah\nblah".to_string(),
+            "DLG SEGMENT_3\n\"\"\"\nthis is a short \"long\" bit of text\n\"\"\"".to_string(),
+            "this is the last segment".to_string(),
+        ];
+
+        assert_eq!(output, expected);
+    }
 }

src/test-resources/segments

@@ -0,0 +1,19 @@
+"""
+the first segment is a long bit of text
+
+
+it contains empty lines
+
+"""
+
+this is a new segment
+this is still the second segment
+blah
+blah
+
+DLG SEGMENT_3
+"""
+this is a short "long" bit of text
+"""
+
+this is the last segment