octorus 0.5.5

//! Patch parsing utilities for extracting line information from Git diff patches.
//!
//! This module provides functions to analyze patch content and extract:
//! - Line content without diff prefixes (+/-)
//! - Line type classification (Added, Removed, Context, Header)
//! - New file line numbers for suggestion positioning
//! - Unified diff parsing for splitting multi-file diffs

use std::collections::HashMap;
use tracing::warn;

/// Represents the type of a line in a diff patch
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum LineType {
    /// Line added in the new version (starts with +)
    Added,
    /// Line removed from the old version (starts with -)
    Removed,
    /// Context line, unchanged (starts with space)
    Context,
    /// Hunk header (@@ ... @@)
    Header,
    /// Metadata lines (diff --, +++, index, etc.)
    Meta,
}

/// Information extracted from a single line in a diff patch
#[derive(Debug, Clone)]
pub struct DiffLineInfo {
    /// The line content without the diff prefix (+/-/space)
    pub line_content: String,
    /// Classification of the line type
    pub line_type: LineType,
    /// Line number in the new file (None for removed lines and headers)
    pub new_line_number: Option<u32>,
    /// Position within the patch (1-based). Corresponds to GitHub API's `position` parameter.
    /// Meta lines (diff --git, ---, +++, index) are not counted.
    /// The first `@@` header is not counted; position 1 is the first line below it.
    /// Subsequent `@@` headers (multi-hunk) ARE counted as positions.
    /// None for meta lines and the first `@@` header.
    pub diff_position: Option<u32>,
}

/// Parse a hunk header to extract the starting line number for new file
/// Format: @@ -old_start,old_count +new_start,new_count @@
fn parse_hunk_header(line: &str) -> Option<u32> {
    // Find the +new_start part
    let plus_pos = line.find('+')?;
    let after_plus = &line[plus_pos + 1..];

    // Extract the number (stop at comma or space)
    let end_pos = after_plus.find([',', ' ']).unwrap_or(after_plus.len());
    let num_str = &after_plus[..end_pos];

    num_str.parse().ok()
}

/// Get information about a specific line in a patch
///
/// # Arguments
/// * `patch` - The full patch content
/// * `line_index` - Zero-based index of the line to analyze
///
/// # Returns
/// * `Some(DiffLineInfo)` - Information about the line if valid
/// * `None` - If the line index is out of bounds
pub fn get_line_info(patch: &str, line_index: usize) -> Option<DiffLineInfo> {
    let lines: Vec<&str> = patch.lines().collect();

    if line_index >= lines.len() {
        return None;
    }

    // Track the current new file line number
    let mut new_line_number: Option<u32> = None;
    // Track the position within the patch (1-based, skipping meta lines)
    let mut position_counter: Option<u32> = None;

    for (i, line) in lines.iter().enumerate() {
        let (line_type, content) = classify_line(line);

        // Update position counter and line number BEFORE checking target
        match line_type {
            LineType::Meta => {
                // Meta lines don't count toward position
            }
            LineType::Header => {
                new_line_number = parse_hunk_header(line);
                // First @@ initializes to 0 (not counted); subsequent @@ lines increment
                position_counter = Some(position_counter.map_or(0, |p| p + 1));
            }
            LineType::Added | LineType::Context => {
                position_counter = position_counter.map(|p| p + 1);
            }
            LineType::Removed => {
                position_counter = position_counter.map(|p| p + 1);
            }
        }

        if i == line_index {
            // For the target line, return the info
            let current_new_line = match line_type {
                LineType::Removed | LineType::Header | LineType::Meta => None,
                _ => new_line_number,
            };

            let current_position = match line_type {
                // Meta lines and the first @@ header (position 0) have no valid position
                LineType::Meta => None,
                LineType::Header if position_counter == Some(0) => None,
                _ => position_counter,
            };

            return Some(DiffLineInfo {
                line_content: content.to_string(),
                line_type,
                new_line_number: current_new_line,
                diff_position: current_position,
            });
        }

        // Update new_line_number for next iteration
        match line_type {
            LineType::Added | LineType::Context => {
                if let Some(n) = new_line_number {
                    new_line_number = Some(n + 1);
                }
            }
            _ => {}
        }
    }

    None
}

/// Classify a line and extract its content without the prefix
pub fn classify_line(line: &str) -> (LineType, &str) {
    if line.starts_with("@@") {
        (LineType::Header, line)
    } else if line.starts_with("+++")
        || line.starts_with("---")
        || line.starts_with("diff ")
        || line.starts_with("index ")
    {
        (LineType::Meta, line)
    } else if let Some(content) = line.strip_prefix('+') {
        (LineType::Added, content)
    } else if let Some(content) = line.strip_prefix('-') {
        (LineType::Removed, content)
    } else if let Some(content) = line.strip_prefix(' ') {
        (LineType::Context, content)
    } else {
        // Lines without prefix (shouldn't happen in valid patches, but handle gracefully)
        (LineType::Context, line)
    }
}

/// Check if a line at the given index can have a suggestion
/// Only Added and Context lines can have suggestions
#[allow(dead_code)]
pub fn can_suggest_at_line(patch: &str, line_index: usize) -> bool {
    get_line_info(patch, line_index)
        .map(|info| matches!(info.line_type, LineType::Added | LineType::Context))
        .unwrap_or(false)
}

/// Validate that all lines in `start..=end` are contiguous new-side lines within a single hunk.
///
/// Returns `true` when every line in the range is `Added` or `Context` and no `Header` line
/// appears between `start` and `end` (i.e. the range does not cross a hunk boundary).
pub fn validate_multiline_range(patch: &str, start: usize, end: usize) -> bool {
    let lines: Vec<&str> = patch.lines().collect();
    for idx in start..=end {
        let Some(line) = lines.get(idx) else {
            return false;
        };
        let (line_type, _) = classify_line(line);
        match line_type {
            LineType::Added | LineType::Context => {}
            // Removed, Header, or Meta lines inside the range → invalid
            _ => return false,
        }
    }
    true
}

/// Convert a file line number (new_line_number) to a patch position.
///
/// Used by AI Rally to convert line numbers from reviewer output to GitHub API positions.
/// Scans the entire patch to find the Added or Context line matching the target line number.
/// Position counting follows the same rules as `get_line_info`: meta lines are skipped,
/// the first `@@` is not counted (position 1 is the line below it), and subsequent `@@`
/// headers are counted.
///
/// Works with both GitHub API patches (starting with `@@`) and local diff patches
/// (starting with `diff --git` meta lines).
pub fn line_number_to_position(patch: &str, target_line: u32) -> Option<u32> {
    let mut new_line_number: Option<u32> = None;
    let mut position_counter: Option<u32> = None;

    for line in patch.lines() {
        let (line_type, _) = classify_line(line);

        match line_type {
            LineType::Meta => continue,
            LineType::Header => {
                new_line_number = parse_hunk_header(line);
                // First @@ initializes to 0 (not counted); subsequent @@ lines increment
                position_counter = Some(position_counter.map_or(0, |p| p + 1));
            }
            LineType::Added | LineType::Context => {
                position_counter = position_counter.map(|p| p + 1);
                if new_line_number == Some(target_line) {
                    return position_counter;
                }
                new_line_number = new_line_number.map(|n| n + 1);
            }
            LineType::Removed => {
                position_counter = position_counter.map(|p| p + 1);
            }
        }
    }
    None
}

/// Parse a unified diff output into a map of filename -> patch content
///
/// This function splits the output of `git diff` or `gh pr diff` into individual
/// file patches. The filenames are normalized (without `a/` or `b/` prefixes).
///
/// # Arguments
/// * `unified_diff` - The full unified diff output
///
/// # Returns
/// A HashMap mapping normalized filenames to their patch content
pub fn parse_unified_diff(unified_diff: &str) -> HashMap<String, String> {
    let mut result = HashMap::new();
    let lines: Vec<&str> = unified_diff.lines().collect();

    if lines.is_empty() {
        return result;
    }

    let mut current_filename: Option<String> = None;
    let mut current_patch_start: Option<usize> = None;
    let mut pending_minus_filename: Option<String> = None;

    for (i, line) in lines.iter().enumerate() {
        if line.starts_with("diff --git ") {
            // Save previous file's patch if any
            if let (Some(filename), Some(start)) = (&current_filename, current_patch_start) {
                let patch = lines[start..i].join("\n");
                if !patch.is_empty() {
                    result.insert(filename.clone(), patch);
                }
            }

            // Extract filename for new file
            current_filename = extract_filename(line);
            current_patch_start = Some(i);
            pending_minus_filename = None;
        } else if current_filename.is_none() && current_patch_start.is_some() {
            // diff --git line was ambiguous; fall back to +++ / --- lines.
            // Prefer +++ (new filename) over --- (old filename).
            // --- is only used as fallback for deleted files where +++ is /dev/null.
            if let Some(rest) = line.strip_prefix("+++ ") {
                if rest != "/dev/null" {
                    current_filename = strip_diff_prefix(rest);
                } else if let Some(ref pending) = pending_minus_filename {
                    current_filename = Some(pending.clone());
                    pending_minus_filename = None;
                }
            } else if let Some(rest) = line.strip_prefix("--- ") {
                if rest != "/dev/null" {
                    pending_minus_filename = strip_diff_prefix(rest);
                }
            }
        }
    }

    // Save last file's patch
    if let (Some(filename), Some(start)) = (current_filename, current_patch_start) {
        let patch = lines[start..].join("\n");
        if !patch.is_empty() {
            result.insert(filename, patch);
        }
    }

    result
}

/// Strip the single-char diff prefix (a/, b/, w/, etc.) from a --- or +++ path.
fn strip_diff_prefix(path: &str) -> Option<String> {
    if path.len() >= 2 && path.as_bytes()[1] == b'/' {
        Some(path[2..].to_string())
    } else {
        Some(path.to_string())
    }
}

/// Extract filename from a "diff --git" line
///
/// Handles various formats:
/// - `diff --git a/src/foo.rs b/src/foo.rs` -> `src/foo.rs` (standard prefix)
/// - `diff --git c/src/foo.rs w/src/foo.rs` -> `src/foo.rs` (mnemonicPrefix)
/// - `diff --git a/file with spaces.rs b/file with spaces.rs` -> `file with spaces.rs`
///
/// For renamed files, returns the new filename (from the second path).
/// This matches the GitHub API convention where `ChangedFile.filename` is the new name.
///
/// Supports arbitrary single-char prefixes (`a/`, `b/`, `c/`, `w/`, `i/`, `o/`)
/// to handle `diff.mnemonicPrefix = true` in git config.
///
/// Returns `None` for ambiguous cases (e.g. paths with spaces + subdirs that create
/// false separator matches). Callers should fall back to `+++ `/ `--- ` lines.
fn extract_filename(git_diff_line: &str) -> Option<String> {
    // Format: "diff --git {x}/{path} {y}/{path}"
    // where x,y are single-char prefixes (a/b for standard, c/w/i/o for mnemonic)
    //
    // We extract the second path (new filename) to match GitHub API behavior
    // and parse_name_status_output() convention.

    let content = git_diff_line.strip_prefix("diff --git ")?;

    // Verify format: must start with a single char + '/'
    if content.len() < 2 || content.as_bytes()[1] != b'/' {
        warn!("Failed to parse git diff line: {}", git_diff_line);
        return None;
    }

    let first_prefix = content.as_bytes()[0];
    let first_path = &content[2..]; // skip "X/"

    // Strategy 1: Assume non-rename (path1 == path2).
    // Format after X/: "path1 Y/path2" where path1 == path2
    // Total length = 2*path_len + 3 (space + Y + /)
    let total_len = first_path.len();
    if total_len >= 3 && (total_len - 3) % 2 == 0 {
        let path_len = (total_len - 3) / 2;
        if path_len > 0 {
            let bytes = first_path.as_bytes();
            let sep = path_len;
            if bytes[sep] == b' ' && bytes[sep + 2] == b'/' {
                let path1 = &first_path[..path_len];
                let path2 = &first_path[sep + 3..];
                if path1 == path2 {
                    return Some(path2.to_string());
                }
            }
        }
    }

    // Strategy 2: For renames, use the expected second prefix character to reduce
    // false positives. Known prefix pairs: a→b, c→w, i→w, o→w.
    let second_prefix = match first_prefix {
        b'a' => b'b',
        b'c' | b'i' | b'o' => b'w',
        _ => {
            warn!(
                "Failed to parse git diff line (unknown prefix): {}",
                git_diff_line
            );
            return None;
        }
    };

    let bytes = first_path.as_bytes();
    let mut matches: Vec<usize> = Vec::new();
    for i in 0..bytes.len().saturating_sub(2) {
        if bytes[i] == b' ' && bytes[i + 1] == second_prefix && bytes[i + 2] == b'/' {
            matches.push(i);
        }
    }

    // Only return if there's exactly one match (unambiguous)
    if matches.len() == 1 {
        let path2 = &first_path[matches[0] + 3..];
        if !path2.is_empty() {
            return Some(path2.to_string());
        }
    }

    // Ambiguous or no match — caller should use +++ line fallback
    None
}

#[cfg(test)]
mod tests {
    use super::*;
    use insta::assert_snapshot;
    use std::collections::BTreeMap;

    fn format_parsed_diff(result: &HashMap<String, String>) -> String {
        let sorted: BTreeMap<&str, &str> = result
            .iter()
            .map(|(k, v)| (k.as_str(), v.as_str()))
            .collect();
        let mut output = String::new();
        for (i, (filename, patch)) in sorted.iter().enumerate() {
            if i > 0 {
                output.push_str("\n---\n");
            }
            output.push_str(&format!("[{}]\n{}", filename, patch));
        }
        output
    }

    const SAMPLE_PATCH: &str = r#"@@ -1,4 +1,5 @@
 line 1
-old line 2
+new line 2
+added line
 line 3"#;

    // Unified diff test data
    const UNIFIED_DIFF_SINGLE: &str = r#"diff --git a/src/main.rs b/src/main.rs
index 1234567..abcdefg 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,3 +1,4 @@
 fn main() {
+    println!("Hello");
 }
"#;

    const UNIFIED_DIFF_MULTIPLE: &str = r#"diff --git a/src/lib.rs b/src/lib.rs
index 1111111..2222222 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,2 +1,3 @@
 pub mod app;
+pub mod config;
diff --git a/src/app.rs b/src/app.rs
index 3333333..4444444 100644
--- a/src/app.rs
+++ b/src/app.rs
@@ -10,6 +10,7 @@
 struct App {
     name: String,
+    version: String,
 }
"#;

    const UNIFIED_DIFF_NEW_FILE: &str = r#"diff --git a/src/new_file.rs b/src/new_file.rs
new file mode 100644
index 0000000..1234567
--- /dev/null
+++ b/src/new_file.rs
@@ -0,0 +1,3 @@
+fn new_function() {
+    todo!()
+}
"#;

    const UNIFIED_DIFF_DELETED: &str = r#"diff --git a/src/old_file.rs b/src/old_file.rs
deleted file mode 100644
index 1234567..0000000
--- a/src/old_file.rs
+++ /dev/null
@@ -1,3 +0,0 @@
-fn old_function() {
-    todo!()
-}
"#;

    const UNIFIED_DIFF_RENAMED: &str = r#"diff --git a/src/old_name.rs b/src/new_name.rs
similarity index 95%
rename from src/old_name.rs
rename to src/new_name.rs
index 1234567..abcdefg 100644
--- a/src/old_name.rs
+++ b/src/new_name.rs
@@ -1,3 +1,3 @@
-fn old_name() {
+fn new_name() {
 }
"#;

    const UNIFIED_DIFF_BINARY: &str = r#"diff --git a/image.png b/image.png
new file mode 100644
index 0000000..1234567
Binary files /dev/null and b/image.png differ
"#;

    #[test]
    fn test_parse_hunk_header() {
        assert_eq!(parse_hunk_header("@@ -1,4 +1,5 @@"), Some(1));
        assert_eq!(parse_hunk_header("@@ -10,3 +15,7 @@"), Some(15));
        assert_eq!(parse_hunk_header("@@ -1 +1 @@"), Some(1));
    }

    #[test]
    fn test_get_line_info_header() {
        let info = get_line_info(SAMPLE_PATCH, 0).unwrap();
        assert_eq!(info.line_type, LineType::Header);
        assert!(info.new_line_number.is_none());
    }

    #[test]
    fn test_get_line_info_context() {
        let info = get_line_info(SAMPLE_PATCH, 1).unwrap();
        assert_eq!(info.line_type, LineType::Context);
        assert_eq!(info.line_content, "line 1");
        assert_eq!(info.new_line_number, Some(1));
    }

    #[test]
    fn test_get_line_info_removed() {
        let info = get_line_info(SAMPLE_PATCH, 2).unwrap();
        assert_eq!(info.line_type, LineType::Removed);
        assert_eq!(info.line_content, "old line 2");
        assert!(info.new_line_number.is_none());
    }

    #[test]
    fn test_get_line_info_added() {
        let info = get_line_info(SAMPLE_PATCH, 3).unwrap();
        assert_eq!(info.line_type, LineType::Added);
        assert_eq!(info.line_content, "new line 2");
        assert_eq!(info.new_line_number, Some(2));
    }

    #[test]
    fn test_can_suggest_at_line() {
        // Header - no
        assert!(!can_suggest_at_line(SAMPLE_PATCH, 0));
        // Context - yes
        assert!(can_suggest_at_line(SAMPLE_PATCH, 1));
        // Removed - no
        assert!(!can_suggest_at_line(SAMPLE_PATCH, 2));
        // Added - yes
        assert!(can_suggest_at_line(SAMPLE_PATCH, 3));
    }

    #[test]
    fn test_classify_line_no_prefix() {
        // diff プレフィックスなし → Context にフォールバック (L123-125)
        let (line_type, content) = classify_line("no prefix");
        assert_eq!(line_type, LineType::Context);
        assert_eq!(content, "no prefix");
    }

    #[test]
    fn test_classify_line_empty() {
        // 空文字列 → Context にフォールバック (L123-125)
        let (line_type, content) = classify_line("");
        assert_eq!(line_type, LineType::Context);
        assert_eq!(content, "");
    }

    #[test]
    fn test_parse_hunk_header_no_comma_no_space() {
        // "@@ -1 +42\ntest" → after_plus = "42" で find([',', ' ']) が None
        // → unwrap_or(after_plus.len()) に到達 (L46)
        let patch = "@@ -1 +42\ntest";
        let info = get_line_info(patch, 1).unwrap();
        assert_eq!(info.line_type, LineType::Context);
        assert_eq!(info.new_line_number, Some(42));
    }

    #[test]
    fn test_out_of_bounds() {
        assert!(get_line_info(SAMPLE_PATCH, 100).is_none());
    }

    // ============================================
    // Unified diff parser tests
    // ============================================

    #[test]
    fn test_extract_filename() {
        assert_eq!(
            extract_filename("diff --git a/src/foo.rs b/src/foo.rs"),
            Some("src/foo.rs".to_string())
        );
        assert_eq!(
            extract_filename("diff --git a/main.rs b/main.rs"),
            Some("main.rs".to_string())
        );
        assert_eq!(
            extract_filename("diff --git a/deep/nested/path/file.rs b/deep/nested/path/file.rs"),
            Some("deep/nested/path/file.rs".to_string())
        );
    }

    #[test]
    fn test_extract_filename_renamed() {
        // For renamed files, we use the "b/" path (new name) to match
        // GitHub API's ChangedFile.filename and parse_name_status_output()
        assert_eq!(
            extract_filename("diff --git a/src/old_name.rs b/src/new_name.rs"),
            Some("src/new_name.rs".to_string())
        );
    }

    #[test]
    fn test_extract_filename_mnemonic_prefix() {
        // diff.mnemonicPrefix = true: c/ (committed) と w/ (working tree)
        assert_eq!(
            extract_filename("diff --git c/src/foo.rs w/src/foo.rs"),
            Some("src/foo.rs".to_string())
        );
        // i/ (index) と w/ (working tree)
        assert_eq!(
            extract_filename("diff --git i/src/bar.rs w/src/bar.rs"),
            Some("src/bar.rs".to_string())
        );
        // mnemonic prefix での rename
        assert_eq!(
            extract_filename("diff --git c/src/old.rs w/src/new.rs"),
            Some("src/new.rs".to_string())
        );
    }

    #[test]
    fn test_extract_filename_invalid() {
        assert_eq!(extract_filename("not a diff line"), None);
        assert_eq!(extract_filename("diff something else"), None);
    }

    #[test]
    fn test_extract_filename_no_separator() {
        // 2つ目のパスが見つからない場合 → None
        assert_eq!(extract_filename("diff --git a/file nob"), None);
    }

    #[test]
    fn test_extract_filename_spaces_with_subdir() {
        // Regression: paths with spaces + directory boundaries must not
        // be truncated by false separator matches inside the path.
        assert_eq!(
            extract_filename("diff --git a/my Folder/src/file.rs b/my Folder/src/file.rs"),
            Some("my Folder/src/file.rs".to_string())
        );
        // Space + single-char dir name that looks like a prefix
        assert_eq!(
            extract_filename("diff --git a/a b/c d/file.rs b/a b/c d/file.rs"),
            Some("a b/c d/file.rs".to_string())
        );
        // Deeply nested path with spaces
        assert_eq!(
            extract_filename(
                "diff --git a/docs/my project/sub b/notes.md b/docs/my project/sub b/notes.md"
            ),
            Some("docs/my project/sub b/notes.md".to_string())
        );
    }

    #[test]
    fn test_extract_filename_ambiguous_falls_back_to_none() {
        // Rename where both old and new paths contain " b/" — truly ambiguous
        // from the diff --git line alone; extract_filename should return None
        // and let parse_unified_diff fall back to +++ line.
        assert_eq!(
            extract_filename("diff --git a/x b/old.rs b/x b/new.rs"),
            None
        );
    }

    #[test]
    fn test_parse_unified_diff_plusplus_fallback() {
        // When extract_filename cannot determine the filename (ambiguous paths),
        // parse_unified_diff should fall back to the +++ line.
        let diff = "\
diff --git a/x b/old.rs b/x b/new.rs
index 1234567..abcdefg 100644
--- a/x b/old.rs
+++ b/x b/new.rs
@@ -1,3 +1,3 @@
 line1
-old
+new";
        let result = parse_unified_diff(diff);
        assert!(
            result.contains_key("x b/new.rs"),
            "expected key 'x b/new.rs', got: {:?}",
            result.keys().collect::<Vec<_>>()
        );
    }

    #[test]
    fn test_strip_diff_prefix() {
        assert_eq!(
            strip_diff_prefix("b/src/file.rs"),
            Some("src/file.rs".to_string())
        );
        assert_eq!(strip_diff_prefix("w/file.rs"), Some("file.rs".to_string()));
        assert_eq!(strip_diff_prefix("file.rs"), Some("file.rs".to_string()));
    }

    #[test]
    fn test_parse_single_file() {
        let result = parse_unified_diff(UNIFIED_DIFF_SINGLE);
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [src/main.rs]
        diff --git a/src/main.rs b/src/main.rs
        index 1234567..abcdefg 100644
        --- a/src/main.rs
        +++ b/src/main.rs
        @@ -1,3 +1,4 @@
         fn main() {
        +    println!("Hello");
         }
        "#);
    }

    #[test]
    fn test_parse_multiple_files() {
        let result = parse_unified_diff(UNIFIED_DIFF_MULTIPLE);
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [src/app.rs]
        diff --git a/src/app.rs b/src/app.rs
        index 3333333..4444444 100644
        --- a/src/app.rs
        +++ b/src/app.rs
        @@ -10,6 +10,7 @@
         struct App {
             name: String,
        +    version: String,
         }
        ---
        [src/lib.rs]
        diff --git a/src/lib.rs b/src/lib.rs
        index 1111111..2222222 100644
        --- a/src/lib.rs
        +++ b/src/lib.rs
        @@ -1,2 +1,3 @@
         pub mod app;
        +pub mod config;
        "#);
    }

    #[test]
    fn test_parse_new_file() {
        let result = parse_unified_diff(UNIFIED_DIFF_NEW_FILE);
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [src/new_file.rs]
        diff --git a/src/new_file.rs b/src/new_file.rs
        new file mode 100644
        index 0000000..1234567
        --- /dev/null
        +++ b/src/new_file.rs
        @@ -0,0 +1,3 @@
        +fn new_function() {
        +    todo!()
        +}
        "#);
    }

    #[test]
    fn test_parse_deleted_file() {
        let result = parse_unified_diff(UNIFIED_DIFF_DELETED);
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [src/old_file.rs]
        diff --git a/src/old_file.rs b/src/old_file.rs
        deleted file mode 100644
        index 1234567..0000000
        --- a/src/old_file.rs
        +++ /dev/null
        @@ -1,3 +0,0 @@
        -fn old_function() {
        -    todo!()
        -}
        "#);
    }

    #[test]
    fn test_parse_renamed_file() {
        let result = parse_unified_diff(UNIFIED_DIFF_RENAMED);
        // Uses new filename (from b/ path) for matching with GitHub API and local diff
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [src/new_name.rs]
        diff --git a/src/old_name.rs b/src/new_name.rs
        similarity index 95%
        rename from src/old_name.rs
        rename to src/new_name.rs
        index 1234567..abcdefg 100644
        --- a/src/old_name.rs
        +++ b/src/new_name.rs
        @@ -1,3 +1,3 @@
        -fn old_name() {
        +fn new_name() {
         }
        "#);
    }

    #[test]
    fn test_parse_binary_file() {
        let result = parse_unified_diff(UNIFIED_DIFF_BINARY);
        assert_snapshot!(format_parsed_diff(&result), @r#"
        [image.png]
        diff --git a/image.png b/image.png
        new file mode 100644
        index 0000000..1234567
        Binary files /dev/null and b/image.png differ
        "#);
    }

    #[test]
    fn test_parse_empty_diff() {
        let result = parse_unified_diff("");
        assert!(result.is_empty());
    }

    #[test]
    fn test_filename_matches_github_api_format() {
        // GitHub API returns filenames without "a/" or "b/" prefix
        // Our parser should return filenames in the same format
        let result = parse_unified_diff(UNIFIED_DIFF_SINGLE);
        let filename = result.keys().next().unwrap();

        // Should not have "a/" or "b/" prefix
        assert!(!filename.starts_with("a/"));
        assert!(!filename.starts_with("b/"));

        // Should match the format GitHub API returns
        assert_eq!(filename, "src/main.rs");
    }

    // ============================================
    // diff_position tests
    // ============================================

    #[test]
    fn test_diff_position_single_hunk() {
        // SAMPLE_PATCH starts with @@ (no meta lines)
        // GitHub position counts from the line BELOW the first @@:
        // Line 0: @@ header -> None (first @@ is not counted)
        // Line 1: context " line 1" -> position 1
        // Line 2: removed "-old line 2" -> position 2
        // Line 3: added "+new line 2" -> position 3
        // Line 4: added "+added line" -> position 4
        // Line 5: context " line 3" -> position 5
        let info = get_line_info(SAMPLE_PATCH, 0).unwrap();
        assert_eq!(info.diff_position, None);

        let info = get_line_info(SAMPLE_PATCH, 1).unwrap();
        assert_eq!(info.diff_position, Some(1));

        let info = get_line_info(SAMPLE_PATCH, 2).unwrap();
        assert_eq!(info.diff_position, Some(2));

        let info = get_line_info(SAMPLE_PATCH, 3).unwrap();
        assert_eq!(info.diff_position, Some(3));

        let info = get_line_info(SAMPLE_PATCH, 4).unwrap();
        assert_eq!(info.diff_position, Some(4));

        let info = get_line_info(SAMPLE_PATCH, 5).unwrap();
        assert_eq!(info.diff_position, Some(5));
    }

    #[test]
    fn test_diff_position_with_meta_lines() {
        // Patch with meta lines (diff --git, index, ---, +++)
        let patch = "diff --git a/foo.rs b/foo.rs\nindex 123..456 100644\n--- a/foo.rs\n+++ b/foo.rs\n@@ -1,2 +1,3 @@\n fn main() {\n+    println!(\"hello\");\n }";
        // Line 0: diff --git -> Meta, position None
        // Line 1: index -> Meta, position None
        // Line 2: --- -> Meta, position None
        // Line 3: +++ -> Meta, position None
        // Line 4: @@ -> Header, position None (first @@, not counted)
        // Line 5: " fn main()" -> Context, position 1
        // Line 6: "+    println..." -> Added, position 2
        // Line 7: " }" -> Context, position 3
        let info = get_line_info(patch, 0).unwrap();
        assert_eq!(info.line_type, LineType::Meta);
        assert_eq!(info.diff_position, None);

        let info = get_line_info(patch, 3).unwrap();
        assert_eq!(info.line_type, LineType::Meta);
        assert_eq!(info.diff_position, None);

        let info = get_line_info(patch, 4).unwrap();
        assert_eq!(info.line_type, LineType::Header);
        assert_eq!(info.diff_position, None);

        let info = get_line_info(patch, 5).unwrap();
        assert_eq!(info.line_type, LineType::Context);
        assert_eq!(info.diff_position, Some(1));

        let info = get_line_info(patch, 6).unwrap();
        assert_eq!(info.line_type, LineType::Added);
        assert_eq!(info.diff_position, Some(2));
    }

    #[test]
    fn test_diff_position_no_meta_lines() {
        // Patch starting with @@ (GitHub API format, no meta lines)
        let patch = "@@ -1,2 +1,3 @@\n fn main() {\n+    println!(\"hello\");\n }";
        let info = get_line_info(patch, 0).unwrap();
        assert_eq!(info.diff_position, None); // first @@ not counted

        let info = get_line_info(patch, 1).unwrap();
        assert_eq!(info.diff_position, Some(1));
    }

    #[test]
    fn test_diff_position_multi_hunk() {
        // Multi-hunk patch: position does NOT reset across hunks
        let patch = "@@ -1,3 +1,3 @@\n-old1\n+new1\n ctx\n@@ -10,3 +10,3 @@\n-old2\n+new2\n ctx2";
        // Line 0: @@ -> None (first @@, not counted)
        // Line 1: -old1 -> position 1
        // Line 2: +new1 -> position 2
        // Line 3: ctx -> position 3
        // Line 4: @@ -> position 4 (subsequent @@, counted)
        // Line 5: -old2 -> position 5
        // Line 6: +new2 -> position 6
        // Line 7: ctx2 -> position 7
        let info = get_line_info(patch, 0).unwrap();
        assert_eq!(info.diff_position, None);

        let info = get_line_info(patch, 4).unwrap();
        assert_eq!(info.line_type, LineType::Header);
        assert_eq!(info.diff_position, Some(4));

        let info = get_line_info(patch, 6).unwrap();
        assert_eq!(info.diff_position, Some(6));

        let info = get_line_info(patch, 7).unwrap();
        assert_eq!(info.diff_position, Some(7));
    }

    // ============================================
    // line_number_to_position tests
    // ============================================

    #[test]
    fn test_line_number_to_position_basic() {
        // SAMPLE_PATCH: @@ -1,4 +1,5 @@  (first @@, not counted)
        //   " line 1"        -> new_line=1, position=1
        //   "-old line 2"    -> removed, no new_line
        //   "+new line 2"    -> new_line=2, position=3
        //   "+added line"    -> new_line=3, position=4
        //   " line 3"        -> new_line=4, position=5
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 1), Some(1));
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 2), Some(3));
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 3), Some(4));
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 4), Some(5));
    }

    #[test]
    fn test_line_number_to_position_multi_hunk() {
        let patch = "@@ -1,3 +1,3 @@\n-old1\n+new1\n ctx\n@@ -10,2 +10,2 @@\n-old2\n+new2";
        // Hunk 1: new_line starts at 1 (first @@ not counted)
        //   +new1 -> new_line=1, position=2
        //   ctx   -> new_line=2, position=3
        // Hunk 2: new_line starts at 10 (second @@ counted as position=4)
        //   +new2 -> new_line=10, position=6
        assert_eq!(line_number_to_position(patch, 1), Some(2));
        assert_eq!(line_number_to_position(patch, 2), Some(3));
        assert_eq!(line_number_to_position(patch, 10), Some(6));
    }

    #[test]
    fn test_line_number_to_position_with_meta_lines() {
        let patch = "diff --git a/foo.rs b/foo.rs\nindex 123..456 100644\n--- a/foo.rs\n+++ b/foo.rs\n@@ -1,2 +1,3 @@\n fn main() {\n+    println!(\"hello\");\n }";
        // Meta lines skipped, first @@ not counted
        // " fn main()" -> new_line=1, position=1
        // "+    println..." -> new_line=2, position=2
        // " }" -> new_line=3, position=3
        assert_eq!(line_number_to_position(patch, 1), Some(1));
        assert_eq!(line_number_to_position(patch, 2), Some(2));
        assert_eq!(line_number_to_position(patch, 3), Some(3));
    }

    #[test]
    fn test_line_number_to_position_nonexistent_line() {
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 999), None);
        assert_eq!(line_number_to_position(SAMPLE_PATCH, 0), None);
    }

    // --- validate_multiline_range tests ---

    #[test]
    fn test_validate_multiline_range_valid_single_hunk() {
        let patch = "@@ -1,3 +1,4 @@\n context line\n+added line\n another context\n-removed line";
        // Lines 1..=2 are context + added → valid
        assert!(validate_multiline_range(patch, 1, 2));
        // Single line
        assert!(validate_multiline_range(patch, 1, 1));
    }

    #[test]
    fn test_validate_multiline_range_includes_removed_line() {
        let patch = "@@ -1,3 +1,4 @@\n context line\n+added line\n another context\n-removed line";
        // Lines 1..=4 includes a removed line at index 4 → invalid
        assert!(!validate_multiline_range(patch, 1, 4));
    }

    #[test]
    fn test_validate_multiline_range_crosses_hunk_boundary() {
        let patch = "@@ -1,2 +1,2 @@\n line1\n+new line2\n@@ -10,2 +10,2 @@\n line10\n+new line11";
        // Lines 1..=4 crosses the hunk header at index 3 → invalid
        assert!(!validate_multiline_range(patch, 1, 4));
        // Within first hunk only
        assert!(validate_multiline_range(patch, 1, 2));
        // Within second hunk only
        assert!(validate_multiline_range(patch, 4, 5));
    }

    #[test]
    fn test_validate_multiline_range_starts_at_header() {
        let patch = "@@ -1,2 +1,2 @@\n line1\n+added";
        // Starting at hunk header → invalid
        assert!(!validate_multiline_range(patch, 0, 1));
    }

    #[test]
    fn test_validate_multiline_range_out_of_bounds() {
        let patch = "@@ -1,2 +1,2 @@\n line1";
        // Index 10 doesn't exist
        assert!(!validate_multiline_range(patch, 1, 10));
    }

    #[test]
    fn test_validate_multiline_range_removed_lines_in_middle() {
        // Removed lines scattered between added/context lines
        let patch = "@@ -1,5 +1,4 @@\n context1\n+added1\n-removed_mid\n context2\n+added2";
        // Range 1..=4 includes removed line at index 3 → invalid
        assert!(!validate_multiline_range(patch, 1, 4));
        // Range 1..=2 is context+added only → valid
        assert!(validate_multiline_range(patch, 1, 2));
        // Range 4..=5 is context+added only → valid
        assert!(validate_multiline_range(patch, 4, 5));
    }

    #[test]
    fn test_validate_multiline_range_all_removed() {
        let patch = "@@ -1,3 +0,0 @@\n-removed1\n-removed2\n-removed3";
        // All lines are removed → invalid
        assert!(!validate_multiline_range(patch, 1, 3));
    }

    /// Validate that new-side line numbers are contiguous for a valid multiline range.
    /// This ensures the GitHub API will receive a valid start_line..line range.
    #[test]
    fn test_multiline_range_new_side_lines_contiguous() {
        let patch = "@@ -1,4 +1,5 @@\n context1\n+added1\n+added2\n context2\n+added3";
        // Valid range: indices 1..=4 → all are Added or Context
        assert!(validate_multiline_range(patch, 1, 4));

        // Verify new-side line numbers are contiguous: 1, 2, 3, 4
        let start_info = get_line_info(patch, 1).unwrap();
        let end_info = get_line_info(patch, 4).unwrap();
        assert_eq!(start_info.new_line_number, Some(1));
        assert_eq!(end_info.new_line_number, Some(4));
        // All intermediate lines should also have contiguous new_line_number
        for idx in 1..=4 {
            let info = get_line_info(patch, idx).unwrap();
            assert!(info.new_line_number.is_some());
        }
    }

    /// Test single-line vs multiline dispatch logic.
    /// When start and end new-side line numbers are equal, start_line should be None (single-line).
    /// When start < end, start_line should be Some (multiline API call).
    #[test]
    fn test_single_line_vs_multiline_dispatch() {
        let patch = "@@ -1,3 +1,4 @@\n context1\n+added1\n+added2\n context2";

        // Single line selection: start == end → should dispatch as single-line comment
        let info = get_line_info(patch, 2).unwrap();
        assert_eq!(info.new_line_number, Some(2));
        // start_line_number == line_number → start_line = None (single-line)
        let start_line = if info.new_line_number == info.new_line_number {
            None
        } else {
            info.new_line_number
        };
        assert_eq!(start_line, None);

        // Multiline selection: start=1, end=3 → should dispatch as multiline comment
        let start_info = get_line_info(patch, 1).unwrap();
        let end_info = get_line_info(patch, 3).unwrap();
        let start_ln = start_info.new_line_number.unwrap();
        let end_ln = end_info.new_line_number.unwrap();
        // start_line_number < end_line_number → start_line = Some (multiline API)
        let start_line = if start_ln < end_ln {
            Some(start_ln)
        } else {
            None
        };
        assert_eq!(start_line, Some(1));
        assert_eq!(end_ln, 3);
    }

    /// Test that validate_multiline_range correctly rejects meta lines in range.
    #[test]
    fn test_validate_multiline_range_meta_lines() {
        // A patch starting with diff --git meta lines
        let patch = "diff --git a/f.rs b/f.rs\nindex abc..def 100644\n--- a/f.rs\n+++ b/f.rs\n@@ -1,2 +1,3 @@\n context1\n+added1\n+added2";
        // Meta lines (indices 0..=3) are not commentable
        assert!(!validate_multiline_range(patch, 0, 5));
        // Valid range within the hunk (indices 5..=7)
        assert!(validate_multiline_range(patch, 5, 7));
    }
}