Table of Contents

1. Overview
2. Local Search for Commits with Specific Content
   2.1. Using git log with -S and -G Options
   2.2. Searching Across All Branches
   2.3. Combining with Other Options
3. Remote Search for Commits with Specific Content
   3.1. Fetching All Branches Locally
   3.2. Using Remote Repository Features
   3.3. Utilizing GitHub's API
4. Practical Examples
   4.1. Finding Commits Adding a Specific Function
   4.2. Locating Changes Containing a Configuration Parameter
   4.3. Searching Remotely on GitHub
5. Important Considerations
6. Summary

Overview

When you need to track down commits that introduced or modified specific code (instead of just searching by commit message), Git’s built-in commands and remote repository features can help. Below are methods for searching locally across all branches, as well as options for searching remote repositories (like GitHub).

Local Search for Commits with Specific Content

Using git log with -S and -G Options

• -S<string>: Searches commits that add or remove an exact string.
• -G<regex>: Searches commits where changes match a given regular expression.

Basic syntax:

git log [options] -- <file_or_path>

Examples:

• Search for commits that added/removed the string initialize:

  git log -S'initialize' -- path/to/file
  

• Search for commits matching the regex init.*:

  git log -G'init.*' -- path/to/file
  

Searching Across All Branches

To extend the search to every branch, add --all:

git log --all -S'your_search_string'

You can omit -- path/to/file if you want to search in all files within the repo.

Combining with Other Options

• --oneline: Concise output (one commit per line).
• --pretty=format:"...": Customizable output format.
• -p: Show the diff of each commit.
• --stat: Show a summary of changes.

Example:

git log --all -G'ENABLE_FEATURE_X' -p

This searches across all branches (--all) for any changes matching the regex ENABLE_FEATURE_X and shows the actual diffs introduced in each commit (-p).

Remote Search for Commits with Specific Content

Fetching All Branches Locally

First, ensure your local repository is up to date with all branches from the remote:

git fetch --all --prune

Then, use the same local search methods described above, but now you have every branch available.

Using Remote Repository Features

On GitHub or similar platforms, you can search directly via their web interfaces.

GitHub Code Search:

1. Go to https://github.com/search.
2. Use advanced syntax, for example:

   repo:username/repository branch:* in:commit "search_string"
   

This looks for commits containing search_string in their content for all branches in the specified repo.

Utilizing GitHub's API

For automated or more advanced searches, you can use the GitHub REST or GraphQL APIs. For instance, the GraphQL API can be queried to search commit diffs.

Practical Examples

Finding Commits Adding a Specific Function

git log --all -S'initializeApp' -- '*.js'

This finds commits across all branches (--all) that introduced or removed the string initializeApp in any JavaScript file.

Locating Changes Containing a Configuration Parameter

git log --all -G'ENABLE_FEATURE_X' -p

Shows any commits that modified lines containing ENABLE_FEATURE_X, and -p will include the actual patches so you can see the differences.

Searching Remotely on GitHub

You can combine filters (user/repo, branch, content, etc.) in the search box. A typical search query might be:

repo:myusername/myrepo branch:* in:commit "refactor"

GitHub will then list the commits that have refactor in their diffs.

Important Considerations

1. Performance: Searching across all branches and all commits can be time-consuming in large repos. Use options like --since, --until, or path filtering for faster results.
2. Rename Detection: If files were renamed, use -M to adjust similarity thresholds:

   git log --all -S'search_string' -M50%
   

3. Binary Files: Searching diffs in binary files often yields limited results.
4. Commit Range: Restrict the search to a specific range:

   git log commit_hash1..commit_hash2 -S'search_string'
   

5. Permissions: Make sure you have rights to access private branches or repositories.
6. Updated Local Repository: Always fetch the latest remote changes to ensure your local search is up to date:

   git fetch --all
   

Summary

To find commits (across any or all branches) that contain a certain string in the actual changes, you can:

1. Locally: Use git log --all -S'search_string' or git log --all -G'pattern' with additional options (-p, --stat, etc.) after fetching all remote branches.
2. Remotely: Use platforms like GitHub’s search or API to query commit diffs without fully cloning or fetching everything locally (especially helpful for large repos or partial queries).

These techniques help you trace the origin of specific code snippets, debug issues, or gather historical context on how your codebase has evolved.