Glob Reference

Globs define file patterns for filtering.

Syntax

glob(include, exclude)

Pattern Characters

Pattern	Matches
`*`	Any characters except `/`
`**`	Any characters including `/`
`?`	Single character
`[abc]`	Character class
`[!abc]`	Negated character class

Examples

Include Only

glob(["src/**", "docs/**"])

Include and Exclude

glob(
    include = ["**"],
    exclude = ["**/internal/**"],
)

By Extension

glob(["**/*.py", "**/*.pyi"])

By Directory

glob(["src/**"])
glob(["src/*"])  # One level only

Exclude Multiple

glob(
    include = ["**"],
    exclude = [
        "**/test/**",
        "**/*_test.go",
        "**/testdata/**",
    ],
)

Usage Contexts

origin_files

origin_files = glob(["src/**"])

destination_files

destination_files = glob(
    include = ["**"],
    exclude = ["README.md"],
)

Transformation paths

core.replace(
    before = "old",
    after = "new",
    paths = glob(["**/*.md"]),
)

core.remove

core.remove(glob(["**/*.bak"]))

Advanced Patterns

Match Root Files Only

glob(["*.md"])           # Only root-level .md files
glob(["**/*.md"])        # All .md files recursively

Match Specific Depth

glob(["src/*/*.py"])     # Python files exactly 1 level under src/
glob(["src/**/*.py"])    # Python files at any depth under src/

Multiple Extensions

glob(["**/*.{js,ts,jsx,tsx}"])  # NOT supported - use list instead
glob(["**/*.js", "**/*.ts", "**/*.jsx", "**/*.tsx"])  # Correct

Hidden Files

glob(["**/.*"])          # All hidden files (starting with .)
glob(["**/.github/**"])  # .github directory contents

Common Pitfalls

Pitfall: Missing `**` for Recursive Match

# WRONG: Only matches src/file.py, not src/sub/file.py
glob(["src/*.py"])

# CORRECT: Matches all .py files under src/
glob(["src/**/*.py"])

Pitfall: Exclude Order Matters

Excludes are processed after includes. An excluded path cannot be re-included.

glob(
    include = ["**"],
    exclude = ["internal/**"],
)
# internal/important.txt is EXCLUDED even if you wanted it

Pitfall: Trailing Slashes

Glob patterns should NOT have trailing slashes:

# WRONG
glob(["src/"])

# CORRECT
glob(["src/**"])

Performance Tips

Copybara calculates “roots” from your glob patterns to determine which directories to query from the repository. More specific patterns = fewer files traversed.

Use Specific Paths Over Wildcards

# SLOWER: Scans entire repo, then filters
glob(
    include = ["**"],
    exclude = ["vendor/**", "node_modules/**", "dist/**"],
)

# FASTER: Only queries src/ and docs/ directories
glob(["src/**", "docs/**"])

Understand Root Calculation

Copybara extracts the static prefix before any wildcard to determine query roots:

Pattern	Computed Root
`src/*/.py`	`src`
`foo/bar.txt`	`foo`
`*/.java`	“ (root)
`{foo,bar}/**`	“ (root)

Patterns with wildcards at the start (like **/*.java) force a full repo scan.

Compose Globs with Operators

Use + and - operators to build complex patterns efficiently:

base = glob(["src/**"])
tests = glob(["src/**/*_test.py", "src/**/test_*.py"])

# All source files except tests
origin_files = base - tests

Avoid Deeply Nested Excludes

Each level of glob nesting adds processing overhead:

# Simple and fast
glob(include = ["src/**"], exclude = ["src/internal/**"])

# More complex - avoid if not needed
(glob(["src/**"]) - glob(["src/internal/**"])) + glob(["docs/**"])