Add support for SHA256 repositories

beautify_git_hash.py

@@ -95,9 +95,14 @@ def load_git_commit(commit_id):
     return subprocess_check_output(['git', 'cat-file', 'commit', commit_id])
 
 
-def git_commit_hash(commit):
+def git_commit_hash(commit, object_format='sha1'):
     object = 'commit %i\x00%s' % (len(commit), commit)
-    return hashlib.sha1(object).hexdigest()
+    if object_format == 'sha1':
+        return hashlib.sha1(object).hexdigest()
+    elif object_format == 'sha256':
+        return hashlib.sha256(object).hexdigest()
+    else:
+        raise Exception('Unsupported object format: ' + object_format)
 
 
 def commit_line_to_format(line, aggregate_values):
@@ -117,15 +122,24 @@ def commit_to_format(commit):
     aggregate_values = {}
     commit_format = '\n'.join(commit_line_to_format(line, aggregate_values)
                               for line in str(commit).split('\n'))
-    return commit_format, aggregate_values
-
+    tree = commit_format.split('\n', 1)[0].split(' ')[1]
+    object_format = get_hash_type(tree)
+    return commit_format, aggregate_values, object_format
+
+def get_hash_type(hash):
+    if len(hash) == 40:
+        return 'sha1'
+    elif len(hash) == 64:
+        return 'sha256'
+    else:
+        raise Exception('Unsupported hash length: ' + str(len(hash)))
 
 def find_beautiful_git_hash(old_commit, prefix, max_minutes=120):
     ALLOWED_PREFIX_CHARACTERS = '0123456789abcdef'
     if prefix.lstrip(ALLOWED_PREFIX_CHARACTERS) != '':
         raise Exception(
             'Invalid prefix! Only lower case hex digits are allowed: ' + ALLOWED_PREFIX_CHARACTERS)
-    commit_format, old_values = commit_to_format(old_commit)
+    commit_format, old_values, object_format = commit_to_format(old_commit)
     bound = max_minutes*60
     for committer_date_offset in range(bound + 1):
         for author_date_offset in range(committer_date_offset + 1):
@@ -134,7 +148,7 @@ def find_beautiful_git_hash(old_commit, prefix, max_minutes=120):
                 'committer_date_timestamp': old_values['committer_date_timestamp'] + committer_date_offset,
             }
             commit = commit_format % new_values
-            if git_commit_hash(commit).startswith(prefix):
+            if git_commit_hash(commit, object_format=object_format).startswith(prefix):
                 if author_date_offset == committer_date_offset == 0:
                     return None
                 else:

git_commit_chooser_react/src/hash.ts

@@ -1,4 +1,4 @@
-import { SHA1, enc } from 'crypto-js';
+import { SHA1, SHA256 } from 'crypto-js';
 
 interface CommitValues {
   author_date_timestamp: number;
@@ -7,10 +7,16 @@ interface CommitValues {
   committer_date_tz: string;
 }
 
-function git_commit_hash(commit: string): string {
+function git_commit_hash(commit: string, object_format: 'sha1' | 'sha256' = 'sha1'): string {
   const object = `commit ${commit.length}\x00${commit}`;
-  const sha = SHA1(object).toString();
-  return sha;
+  switch (object_format) {
+    case 'sha1':
+      return SHA1(object).toString();
+    case 'sha256':
+      return SHA256(object).toString();
+    default:
+      throw new Error('Unsupported object format: ' + object_format);
+  }
 }
 
 function commit_line_to_format(line: string, aggregate_values: CommitValues): string {
@@ -34,7 +40,7 @@ function commit_line_to_format(line: string, aggregate_values: CommitValues): st
   return format_words.join(' ');
 }
 
-function commit_to_format(commit: string): [string, CommitValues] {
+function commit_to_format(commit: string): [string, CommitValues, 'sha1' | 'sha256'] {
   const aggregate_values: CommitValues = {
     author_date_timestamp: 0,
     author_date_tz: '',
@@ -45,7 +51,9 @@ function commit_to_format(commit: string): [string, CommitValues] {
   const commit_format = commit_lines
     .map((line) => commit_line_to_format(line, aggregate_values))
     .join('\n');
-  return [commit_format, aggregate_values];
+  const tree = commit_lines.find((line) => line.startsWith('tree '))?.split(' ')[1];
+  const object_format: 'sha1' | 'sha256' = tree?.length === 64 ? 'sha256' : 'sha1';
+  return [commit_format, aggregate_values, object_format];
 }
 
 interface CommitTimes {
@@ -63,7 +71,7 @@ export async function find_beautiful_git_hash(
   if (![...prefix].every((c) => allowed_prefix_chars.includes(c))) {
     throw new Error('Invalid prefix! Only lowercase hex digits are allowed');
   }
-  const [commit_format, old_values] = commit_to_format(old_commit);
+  const [commit_format, old_values, object_format] = commit_to_format(old_commit);
 
   const bound = max_minutes * 60;
   const possibilities = (bound + 1) * (bound + 2) / 2;
@@ -104,7 +112,7 @@ export async function find_beautiful_git_hash(
       const commit = commit_format
         .replace('%(author_date_timestamp)i', new_values.author_date_timestamp.toString())
         .replace('%(committer_date_timestamp)i', new_values.committer_date_timestamp.toString());
-      const sha = git_commit_hash(commit);
+      const sha = git_commit_hash(commit, object_format);
       if (sha.startsWith(prefix)) {
         if (author_date_offset === 0 && committer_date_offset === 0) {
           return null;

git_hash_chooser/Cargo.toml

@@ -7,8 +7,10 @@ edition = "2021"
 
 [dependencies]
 clap = {version = "4.3.0", features = ["derive"]}
+digest = "0.10.7"
 hex = "0.4.3"
 kdam = "0.3.0"
 md5 = "0.7.0"
 rayon = "1.7.0"
 sha1 = "0.10.5"
+sha2 = "0.10.9"

git_hash_chooser/src/main.rs

@@ -1,7 +1,7 @@
 use clap::Parser;
 use kdam::{tqdm, BarExt};
 use rayon::prelude::*;
-use sha1::Digest;
+use digest::Digest;
 use std::error::Error;
 use std::process::Command;
 use std::sync::{Arc, Mutex};
@@ -13,6 +13,13 @@ struct CommitValues {
     committer_date_tz: String,
 }
 
+#[derive(Default, Clone, Copy)]
+enum HashType {
+    #[default]
+    Sha1,
+    Sha256,
+}
+
 fn subprocess_check_output(cmd: &str) -> Result<String, Box<dyn Error>> {
     let output = Command::new("sh").arg("-c").arg(cmd).output()?;
     let stdout = String::from_utf8_lossy(&output.stdout).into_owned();
@@ -28,10 +35,18 @@ fn load_git_commit(commit_id: &str) -> Result<String, Box<dyn Error>> {
     subprocess_check_output(&format!("git cat-file commit {}", commit_id))
 }
 
-fn git_commit_hash(commit: &str) -> String {
+fn git_commit_hash(commit: &str, hash_type: HashType) -> String {
     let object = format!("commit {}\x00{}", commit.len(), commit);
-    let sha = sha1::Sha1::digest(object.as_bytes());
-    format!("{:x}", sha)
+    match hash_type {
+        HashType::Sha1 => {
+            let sha = sha1::Sha1::digest(object.as_bytes());
+            format!("{:x}", sha)
+        }
+        HashType::Sha256 => {
+            let sha = sha2::Sha256::digest(object.as_bytes());
+            format!("{:x}", sha)
+        }
+    }
 }
 
 fn commit_line_to_format(line: &str, aggregate_values: &mut CommitValues) -> String {
@@ -58,7 +73,7 @@ fn commit_line_to_format(line: &str, aggregate_values: &mut CommitValues) -> Str
     format_words.join(" ")
 }
 
-fn commit_to_format(commit: &str) -> Result<(String, CommitValues), Box<dyn Error>> {
+fn commit_to_format(commit: &str) -> Result<(String, CommitValues, HashType), Box<dyn Error>> {
     let mut aggregate_values = CommitValues {
         author_date_timestamp: 0,
         author_date_tz: String::new(),
@@ -71,7 +86,19 @@ fn commit_to_format(commit: &str) -> Result<(String, CommitValues), Box<dyn Erro
         .map(|line| commit_line_to_format(line, &mut aggregate_values))
         .collect::<Vec<String>>()
         .join("\n");
-    Ok((commit_format, aggregate_values))
+    let hash_format = commit
+        .lines()
+        .find(|line| line.starts_with("tree "))
+        .and_then(|line| {
+            let tree_hash = line.split_whitespace().nth(1)?;
+            match tree_hash.len() {
+                40 => Some(HashType::Sha1),
+                64 => Some(HashType::Sha256),
+                _ => None,
+            }
+        })
+        .unwrap_or_default();
+    Ok((commit_format, aggregate_values, hash_format))
 }
 
 fn find_beautiful_git_hash(
@@ -84,7 +111,7 @@ fn find_beautiful_git_hash(
     if !prefix.chars().all(|c| allowed_prefix_chars.contains(c)) {
         return Err("Invalid prefix! Only lower case hex digits are allowed".into());
     }
-    let (commit_format, old_values) = commit_to_format(old_commit)?;
+    let (commit_format, old_values, hash_type) = commit_to_format(old_commit)?;
 
     let lower_bound = min_minutes * 60;
     let upper_bound = max_minutes * 60;
@@ -132,7 +159,7 @@ fn find_beautiful_git_hash(
                         "%(committer_date_timestamp)i",
                         &new_values.committer_date_timestamp.to_string(),
                     );
-                if git_commit_hash(&commit).starts_with(prefix) {
+                if git_commit_hash(&commit, hash_type).starts_with(prefix) {
                     if author_date_offset == 0 && committer_date_offset == 0 {
                         return Some(None);
                     } else {