patchloom/ops/doc/

mod.rs

//! Document format ops (JSON/YAML/TOML): detect, parse, serialize, preserve.
//!
//! size-waiver: accepted single-domain bulk (policy #1408). Format detect,
//! multi-document YAML parse/write (#1718), and CST-preserving serialize
//! entrypoints are one document surface; do not split for LOC alone.

use crate::selector;
use anyhow::Context;
use serde::Deserialize;
use std::borrow::Cow;
use std::collections::HashSet;
use std::path::Path;

mod navigate;
mod preserve;
pub mod query;
mod toml_preserve;
mod yaml_cst;
mod yaml_splice;

use navigate::reject_blank_merge_overlay;
pub use navigate::{
    deep_merge, delete_at_selector, delete_where, move_at_path, navigate_mut, set_at_path,
    update_matching,
};
use toml_preserve::apply_value_diff;
use yaml_cst::{apply_yaml_mapping_diff, apply_yaml_sequence_diff, try_remove_subsequence};

pub use yaml_splice::needs_yaml_quoting;

#[derive(Debug, Clone, Copy)]
pub enum FileFormat {
    Json,
    Yaml,
    Toml,
}

pub fn detect_format(path: &str) -> anyhow::Result<FileFormat> {
    match Path::new(path).extension().and_then(|e| e.to_str()) {
        Some("json") => Ok(FileFormat::Json),
        Some("yaml" | "yml") => Ok(FileFormat::Yaml),
        Some("toml") => Ok(FileFormat::Toml),
        Some(ext) => Err(crate::exit::InvalidInputError {
            msg: format!(
                "unsupported file extension: .{ext} (supported: .json, .yaml, .yml, .toml)"
            ),
        }
        .into()),
        None => Err(crate::exit::InvalidInputError {
            msg: "file has no extension; doc commands require .json, .yaml, .yml, or .toml".into(),
        }
        .into()),
    }
}

pub fn serialize_value(value: &serde_json::Value, format: &FileFormat) -> anyhow::Result<String> {
    match format {
        FileFormat::Json => {
            let mut s = serde_json::to_string_pretty(value)?;
            s.push('\n');
            Ok(s)
        }
        FileFormat::Yaml => Ok(serde_yaml_ng::to_string(value)?),
        FileFormat::Toml => {
            let s = toml_edit::ser::to_string_pretty(value).map_err(|e| {
                anyhow::Error::new(crate::exit::InvalidInputError {
                    msg: format!("TOML serialization error: {e}"),
                })
            })?;
            Ok(s)
        }
    }
}

/// Serialize a value back to its original format, preserving comments and
/// formatting for TOML and YAML files.
///
/// For TOML, the original text is re-parsed with `toml_edit::DocumentMut`
/// (which retains comments and whitespace), and only the paths that differ
/// between `old_value` and `new_value` are updated.  Untouched keys keep
/// their original formatting, inline comments, and section ordering.
///
/// For YAML, the original text is re-parsed with `yaml_edit::Document`
/// (a Rowan-based CST that retains comments and whitespace), and only the
/// paths that differ between `old_value` and `new_value` are updated.
///
/// JSON falls through to [`serialize_value`] (JSON has no comments).
/// True when write text changed presentation style (e.g. YAML block-sequence
/// indent collapsed) while values may still be correct (#2070 honesty).
///
/// Agents must not claim a pure surgical text edit when this is true.
pub fn presentation_style_changed(original: &str, new_text: &str, format: &FileFormat) -> bool {
    if original == new_text {
        return false;
    }
    match format {
        FileFormat::Yaml => {
            yaml_block_sequence_style_marks(original) != yaml_block_sequence_style_marks(new_text)
                || yaml_alias_identity_counts(original) != yaml_alias_identity_counts(new_text)
        }
        // JSON/TOML pretty-print drift is not flagged here (comment/order
        // preservation paths already minimize noise).
        FileFormat::Json | FileFormat::Toml => false,
    }
}

/// Path-based presentation honesty for library, plan/tx, and CLI doc writes.
///
/// Single entry for surfaces that know a path string but not a pre-parsed
/// [`FileFormat`]. Unknown/unsupported formats return false (no false alarms).
pub fn style_changed_for_path(path: &str, original: &str, new_text: &str) -> bool {
    let Ok(fmt) = detect_format(path) else {
        return false;
    };
    presentation_style_changed(original, new_text, &fmt)
}

/// Counts of `&name`, `*name`, and `<<:` tokens, ignoring full-line and
/// trailing comments so comment-only edits do not flag style.
fn yaml_alias_identity_counts(text: &str) -> (usize, usize, usize) {
    let mut anchors = 0usize;
    let mut aliases = 0usize;
    let mut merges = 0usize;
    for line in text.lines() {
        let code = yaml_line_without_comment(line);
        anchors += count_yaml_prefixed_idents(code, '&');
        aliases += count_yaml_prefixed_idents(code, '*');
        merges += code.matches("<<:").count();
    }
    (anchors, aliases, merges)
}

fn yaml_line_without_comment(line: &str) -> &str {
    let trimmed = line.trim_start();
    if trimmed.starts_with('#') {
        return "";
    }
    match line.find(" #") {
        Some(i) => &line[..i],
        None => line,
    }
}

fn count_yaml_prefixed_idents(s: &str, prefix: char) -> usize {
    let chars: Vec<char> = s.chars().collect();
    let mut n = 0usize;
    let mut i = 0usize;
    while i < chars.len() {
        if chars[i] == prefix {
            let next_ok = chars
                .get(i + 1)
                .is_some_and(|c| c.is_ascii_alphanumeric() || *c == '_');
            if next_ok {
                n += 1;
                i += 2;
                while i < chars.len()
                    && (chars[i].is_ascii_alphanumeric() || chars[i] == '_' || chars[i] == '-')
                {
                    i += 1;
                }
                continue;
            }
        }
        i += 1;
    }
    n
}

/// `(dash_indent, spaces_after_dash)` for every block-sequence entry.
/// Post-dash offset catches `-   name:` vs `- name:` (#2276).
fn yaml_block_sequence_style_marks(text: &str) -> std::collections::BTreeSet<(usize, usize)> {
    text.lines()
        .filter_map(|line| {
            let trimmed = line.trim_start();
            let dash_indent = line.len() - trimmed.len();
            if trimmed == "-" {
                return Some((dash_indent, 0));
            }
            let rest = trimmed.strip_prefix('-')?;
            if rest.starts_with(' ') || rest.starts_with('\t') {
                let pad = rest.len() - rest.trim_start().len();
                Some((dash_indent, pad))
            } else {
                None
            }
        })
        .collect()
}

pub fn serialize_value_preserving(
    original_content: &str,
    old_value: &serde_json::Value,
    new_value: &serde_json::Value,
    format: &FileFormat,
) -> anyhow::Result<String> {
    match format {
        FileFormat::Toml => {
            let mut doc: toml_edit::DocumentMut = toml_source_for_parse(original_content)
                .parse()
                .map_err(|e| {
                anyhow::Error::new(crate::exit::ParseErrorError {
                    msg: format!("TOML re-parse for comment preservation: {e}"),
                })
            })?;
            apply_value_diff(doc.as_item_mut(), old_value, new_value);
            Ok(restore_toml_file_eol(original_content, doc.to_string()))
        }
        FileFormat::Yaml => {
            // Multi-document streams must stay multi-document on write. Falling
            // through to single-doc serialize turns `---` docs into a YAML
            // sequence (`- item:`), which breaks kubectl/kustomize manifests.
            if is_multi_document_yaml(original_content) {
                return serialize_multi_document_yaml(original_content, old_value, new_value);
            }
            if let Some(result) = try_preserve_yaml(original_content, old_value, new_value)? {
                return Ok(result);
            }
            // Fall back to non-preserving (with hoisted comments).
            if old_value == new_value {
                Ok(original_content.to_string())
            } else {
                let body = serialize_value(new_value, format)?;
                Ok(preserve::hoist_comments(original_content, &body))
            }
        }
        // JSON has no comments; return the original text when unchanged
        // to avoid spurious formatting diffs (e.g. array compaction changes).
        _ => {
            if old_value == new_value {
                Ok(original_content.to_string())
            } else {
                serialize_value(new_value, format)
            }
        }
    }
}

/// Clean up whitespace artifacts left by `yaml_edit` CST mutations.
///
/// `Mapping::remove()` can leave trailing whitespace on lines and may
/// drop the final newline. This trims each line's trailing spaces and
/// ensures the output ends with exactly one newline.
///
/// Line endings match the input (`detect_eol`) so Windows CRLF YAML is not
/// rewritten to LF by the CST preserve path.
fn cleanup_yaml_cst_whitespace(text: &str) -> String {
    let eol = crate::write::detect_eol(text);
    let mut result: String = text
        .lines()
        .map(|line| line.trim_end())
        .collect::<Vec<_>>()
        .join(eol);
    // Ensure a final newline (LF or CRLF matching input).
    if result.is_empty() || !result.ends_with('\n') {
        result.push_str(eol);
    }
    result
}

/// Fix broken block-mapping indentation left by `yaml_edit` CST removals.
///
/// When `Mapping::remove()` deletes an entry in a block mapping, the
/// whitespace tokens surrounding the removed key may be absorbed by the
/// adjacent entry, giving it extra leading spaces. For example, removing
/// `name` from:
///
/// ```yaml
/// app:
///   name: "my-app"
///   version: "1.0.0"
///   port: "8080"
/// ```
///
/// can produce:
///
/// ```yaml
/// app:
///     version: "1.0.0"
///   port: "8080"
/// ```
///
/// or (when the middle key is removed):
///
/// ```yaml
/// app:
///   name: "my-app"
///     port: "8080"
/// ```
///
/// This function detects entries whose indentation exceeds their siblings
/// in the same block mapping and normalizes them to match.
#[cfg(test)]
fn fix_yaml_block_indentation(text: &str) -> String {
    fix_yaml_block_indentation_with_original(None, text)
}

fn fix_yaml_block_indentation_with_original(original: Option<&str>, text: &str) -> String {
    let lines: Vec<&str> = text.lines().collect();
    let mut result: Vec<String> = Vec::with_capacity(lines.len());
    // Probe once per finalize. Per-line `original.lines()` is O(n²) on CST retry.
    let original_trim_ends: Option<HashSet<&str>> =
        original.map(|orig| orig.lines().map(str::trim_end).collect());

    for i in 0..lines.len() {
        let line = lines[i];
        let trimmed = line.trim_start();

        // Skip empty lines; comments are handled below.
        if trimmed.is_empty() {
            result.push(line.to_string());
            continue;
        }

        // Fix comment lines: a comment that is indented more than the next
        // mapping entry at the same level has orphaned whitespace from a
        // removed sibling.  Re-indent it to match the next entry.
        if trimmed.starts_with('#') {
            let comment_indent = line.len() - trimmed.len();
            if comment_indent > 0
                && let Some(next_indent) = next_entry_indent(&lines, i)
                && next_indent < comment_indent
            {
                result.push(format!("{}{}", " ".repeat(next_indent), trimmed));
                continue;
            }
            result.push(line.to_string());
            continue;
        }

        let indent = line.len() - trimmed.len();
        // CST remove of `<<` on a dash line leaves `-     name:`. Collapse
        // extra spaces after `-` only on lines the original did not already
        // contain. Pre-existing `-   name:` is style, not an artifact (#2276).
        if is_yaml_sequence_item(trimmed)
            && let Some(rest) = trimmed.strip_prefix('-')
        {
            let rest_trim = rest.trim_start();
            if rest.len() - rest_trim.len() > 1
                && (rest_trim.contains(": ") || rest_trim.ends_with(':'))
                && !is_yaml_sequence_item(rest_trim)
                && original_trim_ends
                    .as_ref()
                    .is_some_and(|orig| !orig.contains(line.trim_end()))
            {
                result.push(format!("{}- {rest_trim}", " ".repeat(indent)));
                continue;
            }
        }
        // `- name: A` is a sequence item, not a mapping sibling of `value:`.
        let is_mapping_entry = indent > 0
            && !is_yaml_sequence_item(trimmed)
            && (trimmed.contains(": ") || trimmed.ends_with(':'));

        if is_mapping_entry
            && let Some(expected) = expected_sibling_indent(&lines, i, indent)
            && expected < indent
        {
            result.push(format!("{}{}", " ".repeat(expected), trimmed));
            continue;
        }

        result.push(line.to_string());
    }

    let eol = crate::write::detect_eol(text);
    let mut out = result.join(eol);
    if text.ends_with('\n') && !out.ends_with('\n') {
        out.push_str(eol);
    }
    out
}

/// After a block sequence item is emptied, yaml-edit 0.3 `Sequence::set`
/// of `{}` / `[]` can drop the item newline and glue the next sibling
/// (`- {}  - name: B`, or `- {}- name:` at document root). Split that
/// so semantic_eq accepts the CST instead of dumping (#2274 / #2275).
fn repair_glued_block_sequence_items(text: &str) -> String {
    let eol = crate::write::detect_eol(text);
    let mut lines: Vec<String> = Vec::new();
    for line in text.lines() {
        let mut remaining = line.to_string();
        loop {
            match split_glued_block_sequence_line(&remaining) {
                Some((first, second)) => {
                    lines.push(first);
                    remaining = second;
                }
                None => {
                    lines.push(remaining);
                    break;
                }
            }
        }
    }
    let mut out = lines.join(eol);
    if text.ends_with('\n') && !out.ends_with('\n') {
        out.push_str(eol);
    }
    out
}

fn split_glued_block_sequence_line(line: &str) -> Option<(String, String)> {
    let indent_len = line.len() - line.trim_start().len();
    let trimmed = &line[indent_len..];
    let after_dash = trimmed.strip_prefix('-')?;
    let pad_len = after_dash.len() - after_dash.trim_start().len();
    if pad_len == 0 {
        return None;
    }
    let after_pad = after_dash.trim_start();
    let first_item = if after_pad.starts_with("{}") {
        "{}"
    } else if after_pad.starts_with("[]") {
        "[]"
    } else {
        return None;
    };
    let rest = &after_pad[first_item.len()..];
    let pad = &after_dash[..pad_len];
    let first = format!("{}-{pad}{first_item}", &line[..indent_len]);
    let rest_trim = rest.trim_start();
    // Nested items glue as `- {}  - name`. A document-root sequence
    // glues with no spaces (`- {}- name: Othello`).
    if is_yaml_sequence_item(rest_trim) {
        let between = &rest[..rest.len() - rest_trim.len()];
        if between.chars().all(|c| c == ' ' || c == '\t') {
            let second = if between.is_empty() {
                format!("{}{rest_trim}", &line[..indent_len])
            } else {
                format!("{between}{rest_trim}")
            };
            return Some((first, second));
        }
    }
    // Last-item empty: Sequence::set drops the newline before the next
    // mapping key (`- {}z: *x` at column 0, or `- {}  enabled: true`
    // when that key is indented).
    if !rest.is_empty()
        && !rest_trim.starts_with('#')
        && !is_yaml_sequence_item(rest_trim)
        && (rest_trim.contains(": ") || rest_trim.ends_with(':'))
    {
        return Some((first, rest.to_string()));
    }
    None
}

/// Find the indentation of the next non-empty, non-comment line after `i`.
/// Used to align orphaned comment lines with their associated mapping entry.
fn next_entry_indent(lines: &[&str], i: usize) -> Option<usize> {
    for line in &lines[i + 1..] {
        let t = line.trim();
        if !t.is_empty() && !t.starts_with('#') {
            return Some(line.len() - t.len());
        }
    }
    None
}

/// Determine the expected indentation for a mapping entry by inspecting
/// its neighbors. Returns `Some(indent)` if a sibling at a lower indent
/// is found (indicating the current line has orphaned extra whitespace).
fn expected_sibling_indent(lines: &[&str], i: usize, current_indent: usize) -> Option<usize> {
    let is_significant = |l: &&str| {
        let t = l.trim();
        !t.is_empty() && !t.starts_with('#')
    };

    // Strategy 1: look DOWN for a sibling with less indent.
    if let Some(next_line) = lines[i + 1..].iter().find(|l| is_significant(l)) {
        let nt = next_line.trim_start();
        let ni = next_line.len() - nt.len();
        let next_is_entry = !is_yaml_sequence_item(nt) && (nt.contains(": ") || nt.ends_with(':'));

        if ni < current_indent && ni > 0 && next_is_entry {
            // Verify: prev line must be a parent (indent < ni, ends ':')
            // or another sibling at the correct indent (== ni).
            let prev_ok = lines[..i]
                .iter()
                .rev()
                .find(|l| is_significant(l))
                .is_some_and(|l| {
                    let t = l.trim_start();
                    if is_yaml_sequence_item(t) {
                        return false;
                    }
                    let pi = l.len() - t.len();
                    (pi < ni && l.trim_end().ends_with(':')) || pi == ni
                });
            if prev_ok {
                return Some(ni);
            }
        }
    }

    // Strategy 2: look UP for a sibling with less indent (handles the case
    // where the corrupted line is the last entry in the mapping).
    if let Some(prev_line) = lines[..i].iter().rev().find(|l| is_significant(l)) {
        let pt = prev_line.trim_start();
        let pi = prev_line.len() - pt.len();

        // Prev must be a complete mapping entry (has a value after the colon,
        // so it is a sibling rather than a parent). A parent key ends with ':'
        // alone; a complete entry has ': <value>'. A sequence item
        // (`- name: A`) is not a mapping sibling of the next key.
        if pi < current_indent
            && pi > 0
            && pt.contains(": ")
            && !is_yaml_parent_line(pt)
            && !is_yaml_sequence_item(pt)
        {
            return Some(pi);
        }
    }

    None
}

/// Block sequence item (`- ` or a lone `-`), not a mapping sibling.
fn is_yaml_sequence_item(trimmed: &str) -> bool {
    trimmed == "-" || trimmed.starts_with("- ")
}

/// Check if a trimmed YAML line is a parent key (value is on the next line).
///
/// Parent patterns:
///   `key:`
///   `key: # comment`
///   `key:  `
///
/// Complete entry: `key: value`
fn is_yaml_parent_line(trimmed: &str) -> bool {
    if trimmed.ends_with(':') {
        return true;
    }
    if let Some(pos) = trimmed.find(": ") {
        let after = trimmed[pos + 2..].trim();
        return after.is_empty() || after.starts_with('#');
    }
    false
}

/// Attempt CST-preserving update for YAML. Returns Some(result) on success,
/// None if a text-level fallback (or plain serialize) is required.
fn try_preserve_yaml(
    original_content: &str,
    old_value: &serde_json::Value,
    new_value: &serde_json::Value,
) -> anyhow::Result<Option<String>> {
    use std::str::FromStr;

    let file = yaml_edit::YamlFile::from_str(original_content).map_err(|e| {
        anyhow::Error::new(crate::exit::ParseErrorError {
            msg: format!("YAML re-parse for comment preservation: {e}"),
        })
    })?;
    let promoted =
        yaml_cst::rewrite_yaml_alias_object_edits(original_content, &file, old_value, new_value)?;
    if let Some(spliced) = promoted.as_deref()
        && yaml_semantic_eq(spliced, new_value)
    {
        return Ok(Some(cleanup_yaml_cst_whitespace(spliced)));
    }
    // Leftover array growth: splice the alias-rewritten text. A yaml-edit
    // re-serialize of `- <<: *alias` plus overrides drops indent, then
    // parse_yaml_semantic fails and we dump.
    if let Some(spliced) = promoted.as_deref()
        && let Some(reparsed) = parse_yaml_semantic(spliced)
        && let Some(grown) = yaml_splice::splice_yaml_array_diffs(spliced, &reparsed, new_value)?
    {
        return Ok(Some(grown));
    }
    let (file, cst_old) = if let Some(spliced) = promoted.as_deref() {
        match yaml_file_after_partial_alias_splice(spliced) {
            Some(pair) => pair,
            None => return Ok(None),
        }
    } else {
        (file, old_value.clone())
    };

    if let Some(doc) = file.document() {
        if let Some(mapping) = doc.as_mapping() {
            if cst_old.is_object() && new_value.is_object() {
                return try_preserve_yaml_object(
                    promoted.as_deref().unwrap_or(original_content),
                    &file,
                    &mapping,
                    &cst_old,
                    new_value,
                );
            }
        } else if let Some(seq) = doc.as_sequence()
            && let (Some(old_arr), Some(new_arr)) = (cst_old.as_array(), new_value.as_array())
        {
            return try_preserve_yaml_array(
                &file,
                &seq,
                promoted.as_deref().unwrap_or(original_content),
                old_arr,
                new_arr,
                new_value,
            );
        }
    }
    Ok(None)
}

/// After an alias splice that is not yet semantic-eq, reparse for leftover CST diffs.
/// Parse failure must dump (None), not CST the pre-splice document.
fn yaml_file_after_partial_alias_splice(
    spliced: &str,
) -> Option<(yaml_edit::YamlFile, serde_json::Value)> {
    use std::str::FromStr;

    let reparsed = yaml_edit::YamlFile::from_str(spliced).ok()?;
    let cst_old = parse_yaml_semantic(spliced)?;
    Some((reparsed, cst_old))
}

fn try_preserve_yaml_object(
    original: &str,
    file: &yaml_edit::YamlFile,
    mapping: &yaml_edit::Mapping,
    old_value: &serde_json::Value,
    new_value: &serde_json::Value,
) -> anyhow::Result<Option<String>> {
    let all_cst_applied = apply_yaml_mapping_diff(mapping, old_value, new_value)?;
    // yaml_edit's Mapping::remove() can leave trailing whitespace on the
    // line preceding the removed key and may shift the indentation of the
    // next sibling entry. Clean both artifacts so the output is valid YAML
    // that preserves original quote styles and key ordering.
    // After an emptied block item, Sequence::set of `{}` can drop the
    // item newline (#2274). Repair before semantic_eq or we dump and
    // lose anchors (#2275).
    let result = finalize_yaml_cst_text(original, &file.to_string());

    // Compare semantically (merge keys resolved) so CST results that keep
    // `<<: *anchor` / `*alias` form match parse_doc's flattened model.
    // Without this, verification always fails on merge-key documents and the
    // non-preserving fallback expands every anchor/alias into full copies.
    // Array growth applied as a new local key (inherited via `<<` only)
    // is already in the CST; do not require !has_array_growth or we dump.
    if yaml_semantic_eq(&result, new_value) {
        return Ok(Some(result));
    }

    // A same-length sequence can only Sequence::set one empty `{}` per
    // pass. Reparse the repaired text and apply leftover empties.
    if !all_cst_applied
        && let Some(finished) = retry_yaml_cst_empties(original, &result, new_value)?
    {
        return Ok(Some(finished));
    }

    // Array growth or structure mismatch: try splice.
    // If the CST produced invalid YAML (e.g., duplicated keys from
    // misinterpreted indentation, #972), serde_yaml_ng will fail to parse
    // it. In that case, skip the splice and fall through to the caller's
    // non-preserving fallback instead of propagating the error.
    //
    // Feed the splice path a merge-resolved view of `result` so array diffs
    // line up with `new_value` (also merge-resolved via parse_doc).
    if let Some(reparsed) = parse_yaml_semantic(&result)
        && let Some(spliced) = yaml_splice::splice_yaml_array_diffs(&result, &reparsed, new_value)?
    {
        return Ok(Some(spliced));
    }
    Ok(None)
}

fn finalize_yaml_cst_text(original: &str, cst: &str) -> String {
    fix_yaml_block_indentation_with_original(
        Some(original),
        &repair_glued_empty_flow_after_colon(&repair_glued_block_sequence_items(
            &cleanup_yaml_cst_whitespace(cst),
        )),
    )
}

/// yaml-edit 0.3.1 parent-set of `{}` / `[]` can omit the space after
/// colon (`key:{}`). serde_yaml_ng rejects that, so we dump and lose
/// sibling anchors. Insert the missing space.
fn repair_glued_empty_flow_after_colon(text: &str) -> String {
    let eol = crate::write::detect_eol(text);
    let mut lines: Vec<String> = Vec::new();
    for line in text.lines() {
        lines.push(repair_glued_empty_flow_line(line));
    }
    let mut out = lines.join(eol);
    if text.ends_with('\n') && !out.ends_with('\n') {
        out.push_str(eol);
    }
    out
}

fn repair_glued_empty_flow_line(line: &str) -> String {
    let indent_len = line.len() - line.trim_start().len();
    let trimmed = &line[indent_len..];
    if is_yaml_sequence_item(trimmed) {
        return line.to_string();
    }
    let Some((key, rest)) = trimmed.split_once(':') else {
        return line.to_string();
    };
    if key.is_empty() || key.contains('#') {
        return line.to_string();
    }
    if rest.starts_with("{}") || rest.starts_with("[]") {
        return format!("{}{key}: {rest}", &line[..indent_len]);
    }
    line.to_string()
}

fn retry_yaml_cst_empties(
    original: &str,
    start: &str,
    new_value: &serde_json::Value,
) -> anyhow::Result<Option<String>> {
    use std::str::FromStr;

    let mut text = start.to_string();
    for _ in 0..32 {
        let Some(current) = parse_yaml_semantic(&text) else {
            return Ok(None);
        };
        if current == *new_value {
            return Ok(Some(text));
        }
        let Ok(file) = yaml_edit::YamlFile::from_str(&text) else {
            return Ok(None);
        };
        let Some(doc) = file.document() else {
            return Ok(None);
        };
        if let Some(mapping) = doc.as_mapping() {
            apply_yaml_mapping_diff(&mapping, &current, new_value)?;
        } else if let Some(seq) = doc.as_sequence()
            && let (Some(old_arr), Some(new_arr)) = (current.as_array(), new_value.as_array())
        {
            apply_yaml_sequence_diff(&seq, old_arr, new_arr)?;
        } else {
            return Ok(None);
        }
        let next = finalize_yaml_cst_text(original, &file.to_string());
        if yaml_semantic_eq(&next, new_value) {
            return Ok(Some(next));
        }
        if next == text {
            return Ok(None);
        }
        text = next;
    }
    Ok(None)
}

fn try_preserve_yaml_array(
    file: &yaml_edit::YamlFile,
    seq: &yaml_edit::Sequence,
    original_content: &str,
    old_arr: &[serde_json::Value],
    new_arr: &[serde_json::Value],
    new_value: &serde_json::Value,
) -> anyhow::Result<Option<String>> {
    let applied = if old_arr.len() == new_arr.len() {
        apply_yaml_sequence_diff(seq, old_arr, new_arr)?
    } else if new_arr.len() < old_arr.len() {
        try_remove_subsequence(seq, old_arr, new_arr)
    } else {
        false
    };
    // Same-length empties apply one `{}` per pass. Finalize the partial
    // CST (glue + indent) and retry leftover empties like mapping-root.
    if old_arr.len() == new_arr.len() || applied {
        let result = finalize_yaml_cst_text(original_content, &file.to_string());
        if yaml_semantic_eq(&result, new_value) {
            return Ok(Some(result));
        }
        if !applied
            && let Some(finished) = retry_yaml_cst_empties(original_content, &result, new_value)?
        {
            return Ok(Some(finished));
        }
    }

    // Growth or failure: text splice.
    if new_arr.len() > old_arr.len()
        && let Some(spliced) =
            yaml_splice::splice_yaml_root_sequence(original_content, old_arr, new_arr)?
        && yaml_semantic_eq(&spliced, new_value)
        && spliced.parse::<yaml_edit::YamlFile>().is_ok()
    {
        return Ok(Some(spliced));
    }
    Ok(None)
}

/// Parse YAML into the same semantic model as [`parse_doc`]: serde expands
/// aliases, then merge keys (`<<`) are flattened.
fn parse_yaml_semantic(text: &str) -> Option<serde_json::Value> {
    let mut value: serde_json::Value = serde_yaml_ng::from_str(text).ok()?;
    resolve_yaml_merge_keys(&mut value);
    Some(value)
}

/// True when `text` parses to the same semantic value as `expected`
/// (aliases expanded, merge keys resolved). Used to accept CST-preserving
/// writes that keep `&anchor` / `*alias` / `<<: *ref` form.
pub(super) fn yaml_semantic_eq(text: &str, expected: &serde_json::Value) -> bool {
    parse_yaml_semantic(text).is_some_and(|v| v == *expected)
}

/// TOML 1.0 only names LF and CRLF. `toml_edit` rejects a lone CR.
/// Normalize those to LF for parse; restore with [`restore_toml_file_eol`].
fn toml_source_for_parse(content: &str) -> Cow<'_, str> {
    let bytes = content.as_bytes();
    let mut i = 0;
    let mut lone = false;
    while i < bytes.len() {
        if bytes[i] == b'\r' {
            if i + 1 < bytes.len() && bytes[i + 1] == b'\n' {
                i += 2;
            } else {
                lone = true;
                break;
            }
        } else {
            i += 1;
        }
    }
    if !lone {
        return Cow::Borrowed(content);
    }
    let mut out = Vec::with_capacity(bytes.len());
    i = 0;
    while i < bytes.len() {
        if bytes[i] == b'\r' {
            if i + 1 < bytes.len() && bytes[i + 1] == b'\n' {
                out.extend_from_slice(b"\r\n");
                i += 2;
            } else {
                out.push(b'\n');
                i += 1;
            }
        } else {
            out.push(bytes[i]);
            i += 1;
        }
    }
    Cow::Owned(String::from_utf8(out).expect("input was UTF-8"))
}

/// After a TOML CST write, put CR-only files back on CR (EditorConfig `cr`).
fn restore_toml_file_eol(original: &str, rendered: String) -> String {
    if crate::write::detect_eol(original) == "\r" {
        crate::write::normalize_eol(&rendered, crate::write::EolMode::Cr).into_owned()
    } else {
        rendered
    }
}

pub fn parse_doc(content: &str, format: &FileFormat) -> anyhow::Result<serde_json::Value> {
    // Notepad/VS/Out-File prefix. JSON rejects BOM; YAML multi-doc `---`
    // after U+FEFF is not a document marker (parse_error at `a:`).
    let content = crate::ops::file::strip_utf8_bom(content);
    match format {
        // Empty / whitespace-only files: treat as empty object so `doc set`
        // can bootstrap a new document (YAML/TOML already accept empty input;
        // serde_json rejects EOF — fixrealloop 2026-07-15).
        FileFormat::Json => {
            if content.trim().is_empty() {
                Ok(serde_json::json!({}))
            } else {
                serde_json::from_str(content).map_err(|e| {
                    anyhow::Error::new(crate::exit::ParseErrorError { msg: e.to_string() })
                })
            }
        }
        FileFormat::Yaml => {
            if is_multi_document_yaml(content) {
                parse_multi_document_yaml(content).map_err(|e| {
                    // Multi-doc path may already be typed; re-wrap plain anyhow.
                    if crate::exit::is_parse_error(&e) {
                        e
                    } else {
                        anyhow::Error::new(crate::exit::ParseErrorError { msg: e.to_string() })
                    }
                })
            } else {
                let mut val: serde_json::Value = serde_yaml_ng::from_str(content).map_err(|e| {
                    anyhow::Error::new(crate::exit::ParseErrorError { msg: e.to_string() })
                })?;
                resolve_yaml_merge_keys(&mut val);
                Ok(val)
            }
        }
        FileFormat::Toml => toml_edit::de::from_str(&toml_source_for_parse(content))
            .map_err(|e| anyhow::Error::new(crate::exit::ParseErrorError { msg: e.to_string() })),
    }
}

/// Parse for read-only queries (keys/len/get/has).
///
/// Empty / whitespace-only / BOM-ZWSP-only YAML is `{}` so keys/len at `.`
/// match empty JSON/TOML. Write bootstrap still uses [`parse_doc`] (empty
/// YAML stays Null). Comment-only and `---` preamble stay parsed YAML.
pub fn parse_doc_for_query(
    content: &str,
    format: &FileFormat,
) -> anyhow::Result<serde_json::Value> {
    // Blank-text check is on the source, not the parsed value: ZWSP-only
    // YAML deserializes as a string, not Null, and trim() leaves it intact.
    if matches!(format, FileFormat::Yaml) && crate::containment::is_blank_text(content) {
        return Ok(serde_json::json!({}));
    }
    parse_doc(content, format)
}

/// Load and parse a JSON/YAML/TOML file for read-only queries.
///
/// Load first so a missing path peels as `not_found` even when the
/// extension is absent or unsupported. Empty YAML is `{}` via
/// [`parse_doc_for_query`]. Write paths keep [`parse_doc`].
pub fn load_for_query(path: &Path) -> anyhow::Result<serde_json::Value> {
    let display = path.to_string_lossy();
    let content = crate::files::load_text_strict(path, &display)?;
    let format = detect_format(&display)?;
    parse_doc_for_query(&content, &format).with_context(|| format!("parsing {display}"))
}

/// Check whether YAML content contains multiple documents.
///
/// A `---` on its own line is the YAML document separator. A single leading
/// `---` is standard YAML preamble, not multi-document. We look for a second
/// `---` separator after stripping the optional leading one.
pub(crate) fn is_multi_document_yaml(content: &str) -> bool {
    // Skip the optional leading document marker.
    let rest = content.strip_prefix("---").map_or(content, |after| {
        // The leading `---` must be followed by whitespace, a comment, newline,
        // or EOF to count as a document marker rather than part of a value.
        if is_after_yaml_marker(after) {
            skip_to_next_line(after)
        } else {
            content
        }
    });

    // Check if rest itself starts with a `---` separator (consecutive markers).
    if rest.starts_with("---") && is_after_yaml_marker(&rest[3..]) {
        return true;
    }

    // Look for `\n---` followed by whitespace, comment, newline, or EOF.
    for (i, _) in rest.match_indices("\n---") {
        let after_marker = &rest[i + 4..];
        if is_after_yaml_marker(after_marker) {
            return true;
        }
    }
    false
}

/// Check whether the text after `---` constitutes a valid document marker
/// ending: empty (EOF), newline, whitespace before newline, or `#` comment.
fn is_after_yaml_marker(after: &str) -> bool {
    if after.is_empty() {
        return true;
    }
    let b = after.as_bytes()[0];
    // Direct newline or carriage return.
    if b == b'\n' || b == b'\r' {
        return true;
    }
    // Trailing whitespace or comment: skip spaces/tabs, then expect newline/comment/EOF.
    if b == b' ' || b == b'\t' || b == b'#' {
        let rest = after
            .as_bytes()
            .iter()
            .skip_while(|&&c| c == b' ' || c == b'\t')
            .copied()
            .next();
        return rest.is_none() || rest == Some(b'\n') || rest == Some(b'\r') || rest == Some(b'#');
    }
    false
}

/// Skip past the current line (after `---`) to the start of the next line.
fn skip_to_next_line(s: &str) -> &str {
    match s.find('\n') {
        Some(pos) => &s[pos + 1..],
        None => "",
    }
}

/// Parse multi-document YAML into a JSON array, resolving merge keys per doc.
fn parse_multi_document_yaml(content: &str) -> anyhow::Result<serde_json::Value> {
    let mut docs = Vec::new();
    for de in serde_yaml_ng::Deserializer::from_str(content) {
        let mut val: serde_json::Value = serde_json::Value::deserialize(de)?;
        resolve_yaml_merge_keys(&mut val);
        docs.push(val);
    }
    // Precondition: is_multi_document_yaml() returned true, so content has
    // ---  separators and the YAML deserializer always yields at least one doc.
    debug_assert!(!docs.is_empty(), "multi-doc YAML produced zero documents");
    Ok(serde_json::Value::Array(docs))
}

/// Whether a single line (without trailing newline) is a YAML document marker.
fn is_yaml_document_separator_line(line: &str) -> bool {
    line.strip_prefix("---").is_some_and(is_after_yaml_marker)
}

/// Split multi-document YAML into document body strings.
///
/// Returns `(had_leading_marker, bodies)`. Bodies exclude the `---` separator
/// lines. Caller should only invoke when [`is_multi_document_yaml`] is true.
pub(crate) fn split_multi_document_yaml(content: &str) -> (bool, Vec<String>) {
    let mut bodies: Vec<String> = Vec::new();
    let mut current = String::new();
    let mut leading_marker = false;
    let mut first_line = true;
    let mut saw_body = false;

    for line in content.split_inclusive('\n') {
        let without_nl = line.trim_end_matches(['\n', '\r']);
        if is_yaml_document_separator_line(without_nl) {
            if first_line && !saw_body {
                // Leading stream marker (optional preamble), not a boundary.
                leading_marker = true;
                first_line = false;
                continue;
            }
            bodies.push(std::mem::take(&mut current));
            first_line = false;
            continue;
        }
        first_line = false;
        saw_body = true;
        current.push_str(line);
    }
    bodies.push(current);
    (leading_marker, bodies)
}

/// Reassemble multi-document YAML from body strings.
///
/// `eol` is the dominant line ending from the original stream so separators
/// stay CRLF on Windows manifests (mixed EOL confuses diffs and some tools).
fn join_multi_document_yaml(leading_marker: bool, docs: &[String], eol: &str) -> String {
    let mut out = String::new();
    if leading_marker {
        out.push_str("---");
        out.push_str(eol);
    }
    for (i, doc) in docs.iter().enumerate() {
        if i > 0 {
            if !out.ends_with('\n') {
                out.push_str(eol);
            }
            out.push_str("---");
            out.push_str(eol);
        }
        let body = doc.trim_end_matches(['\n', '\r']);
        if !body.is_empty() {
            out.push_str(body);
            out.push_str(eol);
        }
    }
    if out.is_empty() {
        out.push_str(eol);
    }
    out
}

/// Serialize one document body (single-doc preserve path, no multi-doc re-entry).
fn serialize_single_yaml_document(
    original_body: &str,
    old_value: &serde_json::Value,
    new_value: &serde_json::Value,
) -> anyhow::Result<String> {
    if old_value == new_value && !original_body.is_empty() {
        return Ok(original_body.to_string());
    }
    if !original_body.trim().is_empty()
        && let Some(result) = try_preserve_yaml(original_body, old_value, new_value)?
    {
        return Ok(result);
    }
    let body = serialize_value(new_value, &FileFormat::Yaml)?;
    if original_body.is_empty() {
        Ok(body)
    } else {
        Ok(preserve::hoist_comments(original_body, &body))
    }
}

/// Write multi-document YAML while keeping `---` document separators.
///
/// Parse models multi-doc streams as a JSON array. Naively serializing that
/// array emits a single sequence document (`- kind: ...`), which is not a
/// multi-document stream and breaks tools that expect `---` separators
/// (e.g. `kubectl apply -f`).
fn serialize_multi_document_yaml(
    original_content: &str,
    old_value: &serde_json::Value,
    new_value: &serde_json::Value,
) -> anyhow::Result<String> {
    if old_value == new_value {
        return Ok(original_content.to_string());
    }

    let (leading_marker, bodies) = split_multi_document_yaml(original_content);

    let Some(new_docs) = new_value.as_array() else {
        // Unexpected: multi-doc parse always yields an array. Fall back.
        let body = serialize_value(new_value, &FileFormat::Yaml)?;
        return Ok(preserve::hoist_comments(original_content, &body));
    };

    let old_docs = old_value.as_array().map(|a| a.as_slice()).unwrap_or(&[]);
    let mut out_docs: Vec<String> = Vec::with_capacity(new_docs.len());

    // When document count is unchanged, pair by index (in-place field edits).
    // When count changes (whole-doc delete/append), pair by value identity so
    // surviving docs keep their original bodies/comments. Index pairing after
    // `doc.delete 0` wrongly maps body[0] onto the former doc 1.
    if old_docs.len() == new_docs.len() {
        for (i, new_doc) in new_docs.iter().enumerate() {
            let orig_body = bodies.get(i).map(String::as_str).unwrap_or("");
            let old_doc = old_docs.get(i);
            match old_doc {
                Some(old_doc) if old_doc == new_doc && !orig_body.is_empty() => {
                    out_docs.push(orig_body.to_string());
                }
                Some(old_doc) => {
                    out_docs.push(serialize_single_yaml_document(orig_body, old_doc, new_doc)?);
                }
                None => {
                    out_docs.push(serialize_value(new_doc, &FileFormat::Yaml)?);
                }
            }
        }
    } else {
        let mut used = vec![false; old_docs.len()];
        for new_doc in new_docs {
            let match_j = old_docs.iter().enumerate().find_map(|(j, old_doc)| {
                if !used[j] && old_doc == new_doc {
                    Some(j)
                } else {
                    None
                }
            });
            match match_j {
                Some(j) => {
                    used[j] = true;
                    let orig_body = bodies.get(j).map(String::as_str).unwrap_or("");
                    if !orig_body.is_empty() {
                        out_docs.push(orig_body.to_string());
                    } else {
                        out_docs.push(serialize_value(new_doc, &FileFormat::Yaml)?);
                    }
                }
                None => {
                    // New or edited without exact old identity: full serialize.
                    out_docs.push(serialize_value(new_doc, &FileFormat::Yaml)?);
                }
            }
        }
    }

    let eol = crate::write::detect_eol(original_content);
    Ok(join_multi_document_yaml(leading_marker, &out_docs, eol))
}

// ---------------------------------------------------------------------------
// Pure data helpers (value parsing, flattening, diffing)
// ---------------------------------------------------------------------------

/// Parse a CLI value string into a [`serde_json::Value`].
///
/// Recognition order: JSON-quoted string, JSON object/array, boolean, null,
/// i64, f64, then fallback to bare string.
///
/// Bare floats (`2.0`, `1.5`) become JSON numbers by design (batch/CLI
/// contract). For a string version field, quote the value:
/// `doc set package.json version '"2.0"'` or plan/MCP typed string.
pub fn parse_value(s: &str) -> serde_json::Value {
    // JSON-quoted string
    if s.starts_with('"')
        && s.ends_with('"')
        && let Ok(v) = serde_json::from_str::<serde_json::Value>(s)
    {
        return v;
    }
    // JSON object or array
    if (s.starts_with('{') || s.starts_with('['))
        && let Ok(v) = serde_json::from_str::<serde_json::Value>(s)
    {
        return v;
    }
    // Booleans
    if s == "true" {
        return serde_json::Value::Bool(true);
    }
    if s == "false" {
        return serde_json::Value::Bool(false);
    }
    // Null
    if s == "null" {
        return serde_json::Value::Null;
    }
    // Integer
    if let Ok(n) = s.parse::<i64>() {
        return serde_json::Value::Number(n.into());
    }
    // Float
    if let Ok(n) = s.parse::<f64>()
        && let Some(num) = serde_json::Number::from_f64(n)
    {
        return serde_json::Value::Number(num);
    }
    // Bare string
    serde_json::Value::String(s.to_string())
}

/// Recursively enumerate all leaf selector paths in a JSON value.
///
/// Uses a mutable `String` buffer for the path prefix to avoid
/// allocating a new `String` via `format!()` at every recursion level.
/// The buffer is extended and truncated as the recursion descends and
/// ascends, so only leaf paths produce a final `String::clone()`.
/// Append a key to the path buffer, quoting it if it contains separator
/// characters (`.`, `[`, `]`) to prevent ambiguity in flattened paths.
fn push_key_quoted(buf: &mut String, k: &str) {
    if k.contains('.') || k.contains('[') || k.contains(']') || k.contains('"') {
        buf.push('"');
        buf.push_str(&k.replace('"', "\\\""));
        buf.push('"');
    } else {
        buf.push_str(k);
    }
}

pub fn flatten_value<'a>(
    value: &'a serde_json::Value,
    buf: &mut String,
    out: &mut Vec<(String, &'a serde_json::Value)>,
) {
    match value {
        serde_json::Value::Object(map) if !map.is_empty() => {
            for (k, v) in map {
                let restore = buf.len();
                if !buf.is_empty() {
                    buf.push('.');
                }
                push_key_quoted(buf, k);
                flatten_value(v, buf, out);
                buf.truncate(restore);
            }
        }
        serde_json::Value::Array(arr) if !arr.is_empty() => {
            for (i, v) in arr.iter().enumerate() {
                let restore = buf.len();
                buf.push('[');
                let _ = std::fmt::Write::write_fmt(buf, format_args!("{i}"));
                buf.push(']');
                flatten_value(v, buf, out);
                buf.truncate(restore);
            }
        }
        _ => {
            out.push((buf.clone(), value));
        }
    }
}

/// Entry in a structured diff.
#[derive(Debug, Clone, serde::Serialize)]
pub struct DiffEntry {
    pub path: String,
    pub kind: &'static str,
    #[serde(skip_serializing_if = "Option::is_none")]
    pub old_value: Option<serde_json::Value>,
    #[serde(skip_serializing_if = "Option::is_none")]
    pub new_value: Option<serde_json::Value>,
}

/// Compute structural differences between two JSON values.
///
/// Uses a mutable `String` buffer for the path prefix to avoid
/// allocating a new `String` via `format!()` at every recursion level.
pub fn diff_values(
    a: &serde_json::Value,
    b: &serde_json::Value,
    buf: &mut String,
    out: &mut Vec<DiffEntry>,
) {
    match (a, b) {
        (serde_json::Value::Object(ma), serde_json::Value::Object(mb)) => {
            for (k, va) in ma {
                let restore = buf.len();
                if !buf.is_empty() {
                    buf.push('.');
                }
                push_key_quoted(buf, k);
                if let Some(vb) = mb.get(k) {
                    diff_values(va, vb, buf, out);
                } else {
                    out.push(DiffEntry {
                        path: buf.clone(),
                        kind: "removed",
                        old_value: Some(va.clone()),
                        new_value: None,
                    });
                }
                buf.truncate(restore);
            }
            for (k, vb) in mb {
                if !ma.contains_key(k) {
                    let restore = buf.len();
                    if !buf.is_empty() {
                        buf.push('.');
                    }
                    push_key_quoted(buf, k);
                    out.push(DiffEntry {
                        path: buf.clone(),
                        kind: "added",
                        old_value: None,
                        new_value: Some(vb.clone()),
                    });
                    buf.truncate(restore);
                }
            }
        }
        (serde_json::Value::Array(aa), serde_json::Value::Array(ab)) => {
            let max_len = aa.len().max(ab.len());
            for i in 0..max_len {
                let restore = buf.len();
                buf.push('[');
                let _ = std::fmt::Write::write_fmt(buf, format_args!("{i}"));
                buf.push(']');
                match (aa.get(i), ab.get(i)) {
                    (Some(va), Some(vb)) => diff_values(va, vb, buf, out),
                    (Some(va), None) => out.push(DiffEntry {
                        path: buf.clone(),
                        kind: "removed",
                        old_value: Some(va.clone()),
                        new_value: None,
                    }),
                    (None, Some(vb)) => out.push(DiffEntry {
                        path: buf.clone(),
                        kind: "added",
                        old_value: None,
                        new_value: Some(vb.clone()),
                    }),
                    (None, None) => {}
                }
                buf.truncate(restore);
            }
        }
        _ => {
            if a != b {
                out.push(DiffEntry {
                    path: buf.clone(),
                    kind: "changed",
                    old_value: Some(a.clone()),
                    new_value: Some(b.clone()),
                });
            }
        }
    }
}

/// Recursively resolve YAML merge keys (`<<`) in a parsed JSON value.
///
/// When `serde_yaml_ng` deserializes `<<: *anchor`, it produces a literal
/// `"<<"` key whose value is the referenced mapping.  This function walks
/// the tree and flattens those entries into the parent object, matching
/// YAML merge-key semantics (existing keys take precedence).
///
/// Recursion stops at `MAX_MERGE_DEPTH` (128, same cap as `deep_merge`).
fn resolve_yaml_merge_keys(value: &mut serde_json::Value) {
    resolve_yaml_merge_keys_inner(value, 0);
}

fn resolve_yaml_merge_keys_inner(value: &mut serde_json::Value, depth: usize) {
    if depth >= navigate::MAX_MERGE_DEPTH {
        return;
    }
    match value {
        serde_json::Value::Object(map) => {
            // First, recurse into all child values (including the merge value itself).
            for v in map.values_mut() {
                resolve_yaml_merge_keys_inner(v, depth + 1);
            }

            // Then resolve `<<` if present.
            if let Some(merge_val) = map.remove("<<") {
                match merge_val {
                    serde_json::Value::Object(merged) => {
                        for (k, v) in merged {
                            map.entry(k).or_insert(v);
                        }
                    }
                    serde_json::Value::Array(arr) => {
                        // Multiple merges: `<<: [*a, *b]` — first wins.
                        for item in arr {
                            if let serde_json::Value::Object(merged) = item {
                                for (k, v) in merged {
                                    map.entry(k).or_insert(v);
                                }
                            }
                        }
                    }
                    _ => {
                        // Non-object merge value — put it back as-is.
                        map.insert("<<".to_string(), merge_val);
                    }
                }
            }
        }
        serde_json::Value::Array(arr) => {
            for v in arr {
                resolve_yaml_merge_keys_inner(v, depth + 1);
            }
        }
        _ => {}
    }
}

// ---------------------------------------------------------------------------
// Unified doc mutation dispatch
// ---------------------------------------------------------------------------

/// Describes a single mutation to apply to a parsed document root.
///
/// This enum captures the 9 doc write operations so that both the CLI
/// (`cmd/doc.rs`) and the transaction engine (`tx.rs`) share a single
/// dispatch path instead of duplicating the match logic.
#[derive(Debug)]
pub enum DocMutation {
    Set {
        selector: String,
        value: serde_json::Value,
    },
    Delete {
        selector: String,
    },
    Merge {
        /// When set, deep-merge into this selector (e.g. multi-doc `0`).
        selector: Option<String>,
        value: serde_json::Value,
    },
    Append {
        selector: String,
        value: serde_json::Value,
    },
    Prepend {
        selector: String,
        value: serde_json::Value,
    },
    Update {
        selector: String,
        value: serde_json::Value,
    },
    Move {
        from: String,
        to: String,
    },
    Ensure {
        selector: String,
        value: serde_json::Value,
    },
    DeleteWhere {
        selector: String,
        predicate: String,
    },
}

/// Result of applying a [`DocMutation`] to a document root.
#[derive(Debug)]
pub enum MutationResult {
    /// The mutation was applied and the document was modified.
    Applied,
    /// A delete / delete-where removed this many items (always >= 1).
    ///
    /// Callers that only care about "did anything change" can treat this like
    /// [`Applied`](Self::Applied). Callers that need counts (CLI/MCP JSON)
    /// use the payload for `removed`.
    Removed(usize),
    /// The selector matched nothing (e.g. delete on a missing key).
    NoMatch,
    /// The path already exists (used by `Ensure` when no write is needed).
    AlreadyExists,
    /// A type error occurred (e.g. append to a non-array). The string
    /// includes the operation name prefix for backward-compatible error
    /// messages (e.g. "doc append: target at 'x' is not an array").
    TypeError(String),
}

/// Apply a [`DocMutation`] to an in-memory document root.
///
/// Callers are responsible for:
/// - Parsing the file and providing the root `Value`
/// - Serializing the modified root back to disk
/// - Mapping [`MutationResult`] to the appropriate exit code or error
pub fn apply_doc_mutation(
    root: &mut serde_json::Value,
    mutation: DocMutation,
) -> anyhow::Result<MutationResult> {
    match mutation {
        DocMutation::Set { selector, value } => {
            let sel = selector::parse_anyhow(&selector)?;
            set_at_path(root, &sel, value)?;
            Ok(MutationResult::Applied)
        }
        DocMutation::Delete { selector } => {
            let sel = selector::parse_anyhow(&selector)?;
            // delete_at_selector returns TypeError for bare keys on array parents.
            if delete_at_selector(root, &sel)? {
                Ok(MutationResult::Removed(1))
            } else {
                Ok(MutationResult::NoMatch)
            }
        }
        DocMutation::Merge { selector, value } => {
            // Multi-document YAML / top-level JSON arrays: deep_merge replaces
            // a non-object base with the overlay wholesale (object *or* array
            // overlay). Fail closed for any overlay on an array root so agents
            // cannot wipe a multi-doc stream (fixrealloop / MPI; #1872 incomplete).
            // Prefer `selector: "0"` (or CLI `--selector 0`) to merge into one doc.
            let target = if let Some(sel) = selector.as_deref().filter(|s| !s.is_empty()) {
                let parsed = selector::parse_anyhow(sel)?;
                navigate_mut(root, &parsed, false, "doc.merge")?
            } else {
                root
            };
            if target.is_array() {
                return Ok(MutationResult::TypeError(
                    "doc merge: target is a top-level array (multi-document YAML or JSON \
                     array); deep-merge would replace the whole stream with the overlay. \
                     Pass a selector to an object document (e.g. `--selector 0` / plan \
                     `\"selector\": \"0\"`) or use doc.set under `0.` / `[0].`"
                        .into(),
                ));
            }
            if !target.is_object() && !value.is_object() {
                // deep_merge replaces non-object bases; still allow object overlay
                // onto object or empty. Non-object target with object value is ok.
            }
            reject_blank_merge_overlay(&value)?;
            deep_merge(target, &value);
            Ok(MutationResult::Applied)
        }
        DocMutation::Append { selector, value } => {
            let sel = selector::parse_anyhow(&selector)?;
            let target = navigate_mut(root, &sel, false, "doc.append")?;
            match target.as_array_mut() {
                Some(arr) => {
                    arr.push(value);
                    Ok(MutationResult::Applied)
                }
                None => Ok(MutationResult::TypeError(format!(
                    "doc append: target at '{selector}' is not an array"
                ))),
            }
        }
        DocMutation::Prepend { selector, value } => {
            let sel = selector::parse_anyhow(&selector)?;
            let target = navigate_mut(root, &sel, false, "doc.prepend")?;
            match target.as_array_mut() {
                Some(arr) => {
                    arr.insert(0, value);
                    Ok(MutationResult::Applied)
                }
                None => Ok(MutationResult::TypeError(format!(
                    "doc prepend: target at '{selector}' is not an array"
                ))),
            }
        }
        DocMutation::Update { selector, value } => {
            let sel = selector::parse_anyhow(&selector)?;
            if update_matching(root, &sel, &value)? == 0 {
                // Bare key on multi-doc / array root: soft no_matches hides shape errors.
                if let Some(hint) = query::array_root_bare_key_hint(root, &sel) {
                    Ok(MutationResult::TypeError(hint))
                } else {
                    Ok(MutationResult::NoMatch)
                }
            } else {
                Ok(MutationResult::Applied)
            }
        }
        DocMutation::Move { from, to } => {
            let from_sel = selector::parse_anyhow(&from)?;
            let to_sel = selector::parse_anyhow(&to)?;
            move_at_path(root, &from_sel, &to_sel)?;
            Ok(MutationResult::Applied)
        }
        DocMutation::Ensure { selector, value } => {
            let sel = selector::parse_anyhow(&selector)?;
            if !selector::eval_result(root, &sel)?.is_empty() {
                Ok(MutationResult::AlreadyExists)
            } else {
                set_at_path(root, &sel, value)?;
                Ok(MutationResult::Applied)
            }
        }
        DocMutation::DeleteWhere {
            selector,
            predicate,
        } => {
            let sel = selector::parse_anyhow(&selector)?;
            let removed = delete_where(root, &sel, &predicate)?;
            if removed == 0 {
                Ok(MutationResult::NoMatch)
            } else {
                Ok(MutationResult::Removed(removed))
            }
        }
    }
}

#[cfg(test)]
mod tests;