feat: TXT text reading - paragraph parsing, line/word stats

crates/zx_document_core/src/text.rs

@@ -0,0 +1,94 @@
+use crate::blocks::DocumentBlock;
+use crate::error::DocumentError;
+use uuid::Uuid;
+
+/// Read a TXT file and return its content as paragraph blocks.
+pub fn read_text(file_path: &str) -> Result<Vec<DocumentBlock>, DocumentError> {
+    let content = std::fs::read_to_string(file_path)?;
+    Ok(parse_text_content(&content))
+}
+
+/// Parse a string as text content, splitting by blank lines into paragraphs.
+pub fn parse_text_content(content: &str) -> Vec<DocumentBlock> {
+    let mut blocks = Vec::new();
+
+    for paragraph in content.split("\n\n") {
+        let text = paragraph.trim().to_string();
+        if text.is_empty() {
+            continue;
+        }
+        blocks.push(DocumentBlock::Paragraph {
+            id: Uuid::new_v4().to_string(),
+            text,
+        });
+    }
+
+    blocks
+}
+
+/// Return line count and word count for a text string.
+pub fn text_stats(content: &str) -> TextStats {
+    let line_count = content.lines().count() as u32;
+    let word_count = content.split_whitespace().count() as u32;
+    TextStats {
+        line_count,
+        word_count,
+    }
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct TextStats {
+    pub line_count: u32,
+    pub word_count: u32,
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_empty_text() {
+        let blocks = parse_text_content("");
+        assert!(blocks.is_empty());
+    }
+
+    #[test]
+    fn test_single_paragraph() {
+        let blocks = parse_text_content("Hello world.");
+        assert_eq!(blocks.len(), 1);
+        if let DocumentBlock::Paragraph { text, .. } = &blocks[0] {
+            assert_eq!(text, "Hello world.");
+        } else {
+            panic!("expected paragraph");
+        }
+    }
+
+    #[test]
+    fn test_multiple_paragraphs() {
+        let blocks = parse_text_content("First paragraph.\n\nSecond paragraph.\n\nThird paragraph.");
+        assert_eq!(blocks.len(), 3);
+    }
+
+    #[test]
+    fn test_trim_whitespace() {
+        let blocks = parse_text_content("  spaced text  \n\n  another  ");
+        assert_eq!(blocks.len(), 2);
+        if let DocumentBlock::Paragraph { text, .. } = &blocks[0] {
+            assert_eq!(text, "spaced text");
+        }
+    }
+
+    #[test]
+    fn test_stats() {
+        let content = "Hello world\nThis is a test.\n\nThird line.";
+        let stats = text_stats(content);
+        assert_eq!(stats.line_count, 4); // 3 text lines + 1 empty
+        assert_eq!(stats.word_count, 8);
+    }
+
+    #[test]
+    fn test_file_not_found() {
+        let result = read_text("/nonexistent/file.txt");
+        assert!(result.is_err());
+    }
+}