Files
codeg/src-tauri/src/parsers/mod.rs
2026-03-28 20:15:21 +08:00

793 lines
25 KiB
Rust

pub mod claude;
pub mod cline;
pub mod codex;
pub mod gemini;
pub mod openclaw;
pub mod opencode;
use std::collections::{HashMap, HashSet};
use std::sync::OnceLock;
use regex::Regex;
use crate::models::{
ContentBlock, ConversationDetail, ConversationSummary, MessageTurn, SessionStats, TurnUsage,
};
#[derive(Debug, thiserror::Error)]
pub enum ParseError {
#[error("IO error: {0}")]
Io(#[from] std::io::Error),
#[error("JSON parse error: {0}")]
Json(#[from] serde_json::Error),
#[error("Database error: {0}")]
Db(#[from] sea_orm::DbErr),
#[error("Conversation not found: {0}")]
ConversationNotFound(String),
#[error("Invalid data: {0}")]
InvalidData(String),
}
pub trait AgentParser {
fn list_conversations(&self) -> Result<Vec<ConversationSummary>, ParseError>;
fn get_conversation(&self, conversation_id: &str) -> Result<ConversationDetail, ParseError>;
}
/// Truncate a string to `max_len` characters, appending "..." if truncated.
pub fn truncate_str(s: &str, max_len: usize) -> String {
if s.chars().count() <= max_len {
s.to_string()
} else {
let truncated: String = s.chars().take(max_len).collect();
format!("{}...", truncated)
}
}
/// Aggregate turn-level usage and duration into a single `SessionStats`.
pub fn compute_session_stats(turns: &[MessageTurn]) -> Option<SessionStats> {
let mut total_in = 0u64;
let mut total_out = 0u64;
let mut total_cache_create = 0u64;
let mut total_cache_read = 0u64;
let mut total_duration = 0u64;
let mut has_data = false;
for turn in turns {
if let Some(ref u) = turn.usage {
total_in += u.input_tokens;
total_out += u.output_tokens;
total_cache_create += u.cache_creation_input_tokens;
total_cache_read += u.cache_read_input_tokens;
has_data = true;
}
if let Some(d) = turn.duration_ms {
total_duration += d;
}
}
if !has_data {
return None;
}
Some(SessionStats {
total_usage: Some(TurnUsage {
input_tokens: total_in,
output_tokens: total_out,
cache_creation_input_tokens: total_cache_create,
cache_read_input_tokens: total_cache_read,
}),
total_tokens: Some(total_in + total_out + total_cache_create + total_cache_read),
total_duration_ms: total_duration,
context_window_used_tokens: None,
context_window_max_tokens: None,
context_window_usage_percent: None,
})
}
fn model_capacity_suffix_regex() -> &'static Regex {
static RE: OnceLock<Regex> = OnceLock::new();
RE.get_or_init(|| {
Regex::new(r"(?i)\[\s*([0-9]+(?:\.[0-9]+)?)\s*([km])\s*\]\s*$")
.expect("valid model capacity regex")
})
}
fn parse_model_capacity_suffix(model: &str) -> Option<u64> {
let captures = model_capacity_suffix_regex().captures(model.trim())?;
let value = captures.get(1)?.as_str().parse::<f64>().ok()?;
if !value.is_finite() || value <= 0.0 {
return None;
}
let unit = captures
.get(2)
.map(|m| m.as_str().to_ascii_lowercase())
.unwrap_or_default();
let multiplier = match unit.as_str() {
"m" => 1_000_000.0,
"k" => 1_000.0,
_ => return None,
};
Some((value * multiplier) as u64)
}
pub fn infer_context_window_max_tokens(model: Option<&str>) -> Option<u64> {
let raw = model?.trim();
if raw.is_empty() {
return None;
}
if let Some(suffixed_limit) = parse_model_capacity_suffix(raw) {
return Some(suffixed_limit);
}
let normalized = raw
.rsplit('/')
.next()
.unwrap_or(raw)
.split(':')
.next()
.unwrap_or(raw)
.trim()
.to_ascii_lowercase();
if normalized.starts_with("claude") {
return Some(200_000);
}
if normalized.starts_with("gemini") {
return Some(1_000_000);
}
match normalized.as_str() {
"gpt-5.2-codex" | "gpt-5.1-codex-max" | "gpt-5.1-codex-mini" | "gpt-5.2" => Some(258_000),
"gpt-5.1" | "gpt-5.1-codex" | "gpt-4o" | "gpt-4o-mini" | "gpt-4-turbo" | "o1-mini"
| "o1-preview" => Some(128_000),
"gpt-4" => Some(8_192),
"o3" | "o3-mini" | "o1" => Some(200_000),
_ => {
if normalized.starts_with("gpt-5") {
Some(258_000)
} else if normalized.starts_with("gpt-4o")
|| normalized.starts_with("gpt-4.1")
|| normalized.starts_with("gpt-4-turbo")
{
Some(128_000)
} else if normalized.starts_with("o3") || normalized == "o1" {
Some(200_000)
} else if normalized.starts_with("o1-mini") || normalized.starts_with("o1-preview") {
Some(128_000)
} else {
None
}
}
}
}
pub fn latest_turn_total_usage_tokens(turns: &[MessageTurn]) -> Option<u64> {
turns.iter().rev().find_map(|turn| {
turn.usage.as_ref().map(|usage| {
usage
.input_tokens
.saturating_add(usage.output_tokens)
.saturating_add(usage.cache_creation_input_tokens)
.saturating_add(usage.cache_read_input_tokens)
})
})
}
pub fn merge_context_window_stats(
stats: Option<SessionStats>,
used_tokens: Option<u64>,
max_tokens: Option<u64>,
) -> Option<SessionStats> {
if used_tokens.is_none() && max_tokens.is_none() {
return stats;
}
let usage_percent = match (used_tokens, max_tokens) {
(Some(used), Some(max)) if max > 0 => Some((used as f64 / max as f64) * 100.0),
_ => None,
};
match stats {
Some(mut s) => {
s.context_window_used_tokens = used_tokens;
s.context_window_max_tokens = max_tokens;
s.context_window_usage_percent = usage_percent;
Some(s)
}
None => Some(SessionStats {
total_usage: None,
total_tokens: None,
total_duration_ms: 0,
context_window_used_tokens: used_tokens,
context_window_max_tokens: max_tokens,
context_window_usage_percent: usage_percent,
}),
}
}
/// Relocate orphaned tool_result blocks to the turn that contains their matching tool_use.
///
/// After `group_into_turns` splits assistant rounds, async tool execution can cause
/// a tool_result to land in a later turn than its corresponding tool_use.
/// This post-processing step moves such orphaned results back.
pub fn relocate_orphaned_tool_results(turns: &mut Vec<MessageTurn>) {
// Build map: tool_use_id → turn index
let mut tool_use_turn: HashMap<String, usize> = HashMap::new();
for (idx, turn) in turns.iter().enumerate() {
for block in &turn.blocks {
if let ContentBlock::ToolUse {
tool_use_id: Some(ref id),
..
} = block
{
tool_use_turn.insert(id.clone(), idx);
}
}
}
if tool_use_turn.is_empty() {
return;
}
// Collect (source_turn, target_turn, block) for orphaned results
let mut relocations: Vec<(usize, usize, ContentBlock)> = Vec::new();
for (turn_idx, turn) in turns.iter().enumerate() {
for block in &turn.blocks {
if let ContentBlock::ToolResult {
tool_use_id: Some(ref id),
..
} = block
{
if let Some(&target) = tool_use_turn.get(id) {
if target != turn_idx {
relocations.push((turn_idx, target, block.clone()));
}
}
}
}
}
if relocations.is_empty() {
return;
}
// Build set of (turn_idx, tool_use_id) to remove
let remove_set: HashMap<usize, Vec<String>> = {
let mut map: HashMap<usize, Vec<String>> = HashMap::new();
for (from, _, block) in &relocations {
if let ContentBlock::ToolResult {
tool_use_id: Some(ref id),
..
} = block
{
map.entry(*from).or_default().push(id.clone());
}
}
map
};
// Remove from source turns
for (&turn_idx, ids) in &remove_set {
turns[turn_idx].blocks.retain(|block| {
if let ContentBlock::ToolResult {
tool_use_id: Some(ref id),
..
} = block
{
!ids.contains(id)
} else {
true
}
});
}
// Append to target turns
for (_, target, block) in relocations {
turns[target].blocks.push(block);
}
// Remove turns that became empty after relocation
turns.retain(|turn| !turn.blocks.is_empty());
}
/// Convert Read tool output from numbered-line format to `{"start_line":N,"content":"..."}`.
///
/// Claude Code embeds line numbers in Read output like ` 115→content`.
/// This splits on the `→` delimiter (or tab for older `cat -n` format),
/// extracts the starting line number, and returns clean content.
pub fn structurize_read_tool_output(turns: &mut [MessageTurn]) {
let mut read_tool_ids: HashSet<String> = HashSet::new();
for turn in turns.iter() {
for block in &turn.blocks {
if let ContentBlock::ToolUse {
tool_use_id: Some(ref id),
ref tool_name,
..
} = block
{
let name = tool_name.to_lowercase();
if matches!(
name.as_str(),
"read" | "read_file" | "readfile" | "read file" | "cat" | "view"
) {
read_tool_ids.insert(id.clone());
}
}
}
}
for turn in turns.iter_mut() {
for block in turn.blocks.iter_mut() {
let is_read_result = matches!(
block,
ContentBlock::ToolResult { tool_use_id: Some(ref id), .. }
if read_tool_ids.contains(id)
);
if !is_read_result {
continue;
}
if let ContentBlock::ToolResult {
ref mut output_preview,
..
} = block
{
if let Some(ref text) = output_preview {
if let Some(json) = strip_numbered_lines(text) {
*output_preview = Some(json);
}
}
}
}
}
}
/// Known delimiters between line number and content.
const LINE_NUM_DELIMITERS: &[&str] = &["", "\t"];
/// Try to split a line at a known delimiter, returning (line_number, content).
fn split_line_number(line: &str) -> Option<(u64, &str)> {
for delim in LINE_NUM_DELIMITERS {
if let Some(pos) = line.find(delim) {
let prefix = line[..pos].trim();
if let Ok(num) = prefix.parse::<u64>() {
let content_start = pos + delim.len();
return Some((num, &line[content_start..]));
}
}
}
None
}
/// If most lines have a recognized line-number prefix, strip them all
/// and return `{"start_line":N,"content":"clean text"}`.
pub fn strip_numbered_lines(text: &str) -> Option<String> {
let raw_lines: Vec<&str> = text.lines().collect();
if raw_lines.len() < 2 {
return None;
}
let matched = raw_lines
.iter()
.filter(|l| l.is_empty() || split_line_number(l).is_some())
.count();
if matched < raw_lines.len() * 4 / 5 {
return None;
}
let mut start_line: u64 = 1;
let mut first = true;
let stripped: Vec<&str> = raw_lines
.iter()
.map(|line| {
if let Some((num, content)) = split_line_number(line) {
if first {
start_line = num;
first = false;
}
content
} else {
first = false;
*line
}
})
.collect();
Some(
serde_json::json!({
"start_line": start_line,
"content": stripped.join("\n")
})
.to_string(),
)
}
/// Resolve line numbers for `*** Update File` / `*** Add File` style patches.
///
/// When a hunk header is just `@@` without `-N,M +N,M`, this reads the actual
/// file from disk and matches the context lines to calculate real line numbers.
/// Falls back gracefully if the file doesn't exist or context doesn't match.
pub fn resolve_patch_line_numbers(turns: &mut [MessageTurn], cwd: Option<&str>) {
for turn in turns.iter_mut() {
for block in turn.blocks.iter_mut() {
if let ContentBlock::ToolUse {
ref tool_name,
ref mut input_preview,
..
} = block
{
let name = tool_name.to_lowercase();
if !matches!(
name.as_str(),
"apply_patch" | "edit" | "patch" | "applypatch"
) {
continue;
}
if let Some(ref text) = input_preview {
if text.contains("@@\n") || text.contains("@@\r\n") {
if let Some(resolved) = resolve_patch_text(text, cwd) {
*input_preview = Some(resolved);
}
}
}
}
}
}
}
/// Resolve a single patch text, replacing bare `@@` with `@@ -N,M +N,M @@`.
pub fn resolve_patch_text(patch: &str, cwd: Option<&str>) -> Option<String> {
let mut output = String::with_capacity(patch.len() + 256);
let mut current_file_path: Option<String> = None;
let mut file_lines: Option<Vec<String>> = None;
let mut any_resolved = false;
let lines: Vec<&str> = patch.lines().collect();
let mut i = 0;
while i < lines.len() {
let line = lines[i];
// Detect file markers
if line.starts_with("*** Update File: ") || line.starts_with("*** Add File: ") {
let marker_end = if line.starts_with("*** Update File: ") {
17
} else {
14
};
let path = line[marker_end..].trim();
current_file_path = Some(path.to_string());
file_lines = load_file_lines(path, cwd);
output.push_str(line);
output.push('\n');
i += 1;
continue;
}
// Detect bare @@ hunk header (no line numbers)
if line == "@@" {
if let (Some(ref fl), true) = (&file_lines, current_file_path.is_some()) {
// Collect context lines from this hunk to find match position
let hunk_lines = collect_hunk_lines(&lines, i + 1);
if let Some((old_start, old_count, new_count)) =
find_hunk_position(fl, &hunk_lines)
{
let new_start = old_start; // same start for context-based patches
output.push_str(&format!(
"@@ -{},{} +{},{} @@\n",
old_start, old_count, new_start, new_count
));
any_resolved = true;
i += 1;
continue;
}
}
// Fallback: keep bare @@
output.push_str(line);
output.push('\n');
i += 1;
continue;
}
output.push_str(line);
output.push('\n');
i += 1;
}
if any_resolved { Some(output) } else { None }
}
/// Load file lines from disk, trying both absolute path and cwd-relative.
pub fn load_file_lines(path: &str, cwd: Option<&str>) -> Option<Vec<String>> {
use std::fs;
use std::path::Path;
let p = Path::new(path);
if p.is_absolute() {
if let Ok(content) = fs::read_to_string(p) {
return Some(content.lines().map(|l| l.to_string()).collect());
}
}
if let Some(base) = cwd {
let full = Path::new(base).join(path);
if let Ok(content) = fs::read_to_string(&full) {
return Some(content.lines().map(|l| l.to_string()).collect());
}
}
None
}
/// Collect lines belonging to a hunk (until next `@@` or `*** ` marker or end).
fn collect_hunk_lines<'a>(lines: &'a [&'a str], start: usize) -> Vec<&'a str> {
let mut result = Vec::new();
for &line in &lines[start..] {
if line == "@@"
|| line.starts_with("*** ")
{
break;
}
result.push(line);
}
result
}
/// Find where a hunk's context lines match in the file, returning (start_line, old_count, new_count).
/// `start_line` is 1-based.
///
/// The file on disk may be in either pre-patch or post-patch state, and may
/// have been further modified. We try three strategies in order:
/// 1. Contiguous match of context+added lines (post-patch file, no further edits)
/// 2. Contiguous match of context+deleted lines (pre-patch file)
/// 3. Subsequence match of context-only lines (file has been further modified)
fn find_hunk_position(
file_lines: &[String],
hunk_lines: &[&str],
) -> Option<(usize, usize, usize)> {
let mut old_count = 0usize;
let mut new_count = 0usize;
for hl in hunk_lines {
if hl.starts_with(' ') {
old_count += 1;
new_count += 1;
} else if hl.starts_with('-') {
old_count += 1;
} else if hl.starts_with('+') {
new_count += 1;
}
}
// Strategy 1: contiguous match of context+added (post-patch)
let new_view: Vec<&str> = hunk_lines
.iter()
.filter(|l| l.starts_with(' ') || l.starts_with('+'))
.map(|l| &l[1..])
.collect();
if let Some(pos) = find_contiguous(file_lines, &new_view) {
return Some((pos + 1, old_count, new_count));
}
// Strategy 2: contiguous match of context+deleted (pre-patch)
let old_view: Vec<&str> = hunk_lines
.iter()
.filter(|l| l.starts_with(' ') || l.starts_with('-'))
.map(|l| &l[1..])
.collect();
if let Some(pos) = find_contiguous(file_lines, &old_view) {
return Some((pos + 1, old_count, new_count));
}
// Strategy 3: subsequence match of context-only lines (file further modified)
let ctx_only: Vec<&str> = hunk_lines
.iter()
.filter(|l| l.starts_with(' '))
.map(|l| &l[1..])
.collect();
if let Some(pos) = find_subsequence(file_lines, &ctx_only) {
return Some((pos + 1, old_count, new_count));
}
None
}
/// Find contiguous `view` lines in `file_lines`. Returns 0-based start index.
fn find_contiguous(file_lines: &[String], view: &[&str]) -> Option<usize> {
if view.is_empty() || view.len() > file_lines.len() {
return None;
}
let first = view[0];
for i in 0..=(file_lines.len() - view.len()) {
if file_lines[i].as_str() != first {
continue;
}
if view.iter().enumerate().all(|(j, v)| file_lines[i + j].as_str() == *v) {
return Some(i);
}
}
None
}
/// Find `needles` as an ordered subsequence in `file_lines` within a small window.
/// Returns 0-based index of the first needle's position.
fn find_subsequence(file_lines: &[String], needles: &[&str]) -> Option<usize> {
if needles.is_empty() {
return None;
}
let first = needles[0];
for start in 0..file_lines.len() {
if file_lines[start].as_str() != first {
continue;
}
let mut cursor = start + 1;
let mut all_found = true;
for &needle in &needles[1..] {
// Allow up to 10 lines gap between consecutive context lines
let limit = std::cmp::min(cursor + 10, file_lines.len());
match file_lines[cursor..limit]
.iter()
.position(|fl| fl.as_str() == needle)
{
Some(offset) => cursor = cursor + offset + 1,
None => {
all_found = false;
break;
}
}
}
if all_found {
return Some(start);
}
}
None
}
/// Extract the last path component as the folder name.
pub fn folder_name_from_path(path: &str) -> String {
path.rsplit(['/', '\\']).next().unwrap_or(path).to_string()
}
/// Normalize a filesystem path string for tolerant cross-platform comparison.
/// This intentionally does not hit the filesystem (no canonicalize), and only
/// normalizes separators/casing differences that commonly break exact matching.
pub fn normalize_path_for_matching(path: &str) -> String {
let mut normalized = path.trim().replace('\\', "/");
#[cfg(target_os = "windows")]
{
if let Some(stripped) = normalized.strip_prefix("//?/") {
normalized = stripped.to_string();
}
normalized = normalized.to_ascii_lowercase();
}
while normalized.ends_with('/') {
if normalized == "/" {
break;
}
// Keep Windows drive root such as "c:/" intact.
if normalized.len() == 3
&& normalized.as_bytes().get(1) == Some(&b':')
&& normalized.as_bytes().get(2) == Some(&b'/')
{
break;
}
normalized.pop();
}
normalized
}
pub fn path_eq_for_matching(left: &str, right: &str) -> bool {
normalize_path_for_matching(left) == normalize_path_for_matching(right)
}
#[cfg(test)]
mod tests {
use chrono::Utc;
use super::{
infer_context_window_max_tokens, latest_turn_total_usage_tokens,
merge_context_window_stats, path_eq_for_matching,
};
use crate::models::{MessageTurn, SessionStats, TurnRole, TurnUsage};
#[test]
fn infers_model_context_limits() {
assert_eq!(
infer_context_window_max_tokens(Some("claude-sonnet-4-6")),
Some(200_000)
);
assert_eq!(
infer_context_window_max_tokens(Some("gemini-2.5-pro")),
Some(1_000_000)
);
assert_eq!(
infer_context_window_max_tokens(Some("claude-sonnet-4-6 [1.5M]")),
Some(1_500_000)
);
assert_eq!(infer_context_window_max_tokens(Some("unknown-model")), None);
}
#[test]
fn picks_latest_turn_usage_total_tokens() {
let timestamp = Utc::now();
let turns = vec![
MessageTurn {
id: "turn-0".to_string(),
role: TurnRole::Assistant,
blocks: vec![],
timestamp,
usage: Some(TurnUsage {
input_tokens: 10,
output_tokens: 20,
cache_creation_input_tokens: 30,
cache_read_input_tokens: 40,
}),
duration_ms: None,
model: None,
},
MessageTurn {
id: "turn-1".to_string(),
role: TurnRole::Assistant,
blocks: vec![],
timestamp,
usage: Some(TurnUsage {
input_tokens: 11,
output_tokens: 21,
cache_creation_input_tokens: 31,
cache_read_input_tokens: 41,
}),
duration_ms: None,
model: None,
},
];
assert_eq!(latest_turn_total_usage_tokens(&turns), Some(104));
}
#[test]
fn merges_context_window_stats() {
let merged = merge_context_window_stats(None, Some(1500), Some(3000))
.expect("context stats should exist");
assert_eq!(merged.context_window_used_tokens, Some(1500));
assert_eq!(merged.context_window_max_tokens, Some(3000));
assert!(merged.total_usage.is_none());
let percent = merged
.context_window_usage_percent
.expect("usage percent should exist");
assert!((percent - 50.0).abs() < f64::EPSILON);
let existing = Some(SessionStats {
total_usage: Some(TurnUsage {
input_tokens: 1,
output_tokens: 2,
cache_creation_input_tokens: 3,
cache_read_input_tokens: 4,
}),
total_tokens: Some(10),
total_duration_ms: 100,
context_window_used_tokens: None,
context_window_max_tokens: None,
context_window_usage_percent: None,
});
let merged_existing =
merge_context_window_stats(existing, Some(200), Some(1000)).expect("merged");
assert_eq!(merged_existing.total_tokens, Some(10));
assert_eq!(merged_existing.context_window_used_tokens, Some(200));
assert_eq!(merged_existing.context_window_max_tokens, Some(1000));
}
#[test]
fn path_matching_handles_separator_differences() {
assert!(path_eq_for_matching(
"/Users/demo/workspace/codeg",
"/Users/demo/workspace/codeg/"
));
assert!(path_eq_for_matching(
"C:\\Users\\demo\\workspace\\codeg",
"C:/Users/demo/workspace/codeg"
));
}
}