Addresses #268 : local builds should respect .gitignore

repo2docker/buildpacks/base.py

@@ -4,6 +4,7 @@
 import io
 import os
 import re
+import pathspec
 import logging
 import docker
 
@@ -170,7 +171,7 @@ def get_base_packages(self):
             # FIXME: Use npm from nodesource!
             # Everything seems to depend on npm these days, unfortunately.
             "npm",
-            
+
             "unzip",
         }
 
@@ -349,7 +350,23 @@ def _filter_tar(tar):
             src_path = os.path.join(os.path.dirname(__file__), *src_parts)
             tar.add(src_path, src, filter=_filter_tar)
 
-        tar.add('.', 'src/', filter=_filter_tar)
+        _exclude_tar = None
+        if os.path.exists(".gitignore"):
+            with open(".gitignore") as gitignore_fh:
+                ignorespec = pathspec.PathSpec.from_lines('gitignore',
+                                                          gitignore_fh)
+                def _exclude_tar(filepath):
+                    # Conditionally exclude files based on the pathspecs
+                    # mentioned in the `.gitignore` file.
+                    # Note that, the behaviour of this function is
+                    # not **exactly** same as the way `git` excludes files
+                    # based on `.gitignore`.
+                    #
+                    # https://github.com/cpburnz/python-path-specification/issues/19
+                    filepath = os.path.relpath(filepath, "./")
+                    return ignorespec.match_file(filepath)
+
+        tar.add('.', 'src/', exclude=_exclude_tar, filter=_filter_tar)
 
         tar.close()
         tarf.seek(0)

setup.py

@@ -10,6 +10,7 @@
         'escapism',
         'jinja2',
         'ruamel.yaml>=0.15',
+        'pathspec>=0.5.5',
     ],
     python_requires='>=3.4',
     author='Yuvi Panda',

tests/gitignore.py

@@ -0,0 +1,93 @@
+"""
+Local builds respect .gitignore
+Tests that files excluded in gitignore are not packaged inside the image
+"""
+import os
+import subprocess
+import tempfile
+import time
+
+GITIGNORE_CONTENT="""
+*.csv
+!A/0.csv
+A/1.tsv
+!/B/2.csv
+B/1.tsv
+C/
+"""
+
+VERIFY_SCRIPT="""
+#!/usr/bin/env python
+import os
+
+assert os.path.exists("A/0.csv")
+assert not os.path.exists("A/1.csv")
+assert not os.path.exists("A/2.csv")
+
+assert os.path.exists("A/0.tsv")
+assert not os.path.exists("A/1.tsv")
+assert os.path.exists("A/2.tsv")
+
+assert not os.path.exists("B/0.csv")
+assert not os.path.exists("B/1.csv")
+assert os.path.exists("B/2.csv")
+
+assert os.path.exists("B/0.tsv")
+assert not os.path.exists("B/1.tsv")
+assert os.path.exists("B/2.tsv")
+
+assert not os.path.exists("C/0.csv")
+assert not os.path.exists("C/1.csv")
+assert not os.path.exists("C/2.csv")
+
+assert not os.path.exists("C/0.tsv")
+assert not os.path.exists("C/1.tsv")
+assert not os.path.exists("C/2.tsv")
+"""
+
+def test_gitignore():
+    """
+    Local builds respect .gitignore
+    """
+    def create_directory_structure():
+        # Create Directories and files
+        #
+        # It is important to create the directory structure programmatically
+        # because if we include a folder with a .gitignore in the repository,
+        # then git uses it to exclude files from being checked into
+        # the repository
+        for dirName in ['A', 'B', 'C']:
+            os.mkdir(dirName)
+            for file_ids in range(3):
+                # Write csv
+                fp = open(os.path.join(dirName, str(file_ids)+".csv"), "w")
+                fp.write("test")
+                fp.close()
+                # Write .tsv
+                fp = open(os.path.join(dirName, str(file_ids)+".tsv"), "w")
+                fp.write("test")
+                fp.close()
+        # Create .gitignore
+        with open(".gitignore", "w") as fp:
+            fp.write(GITIGNORE_CONTENT)
+        # Create a verify script
+        with open("verify.py", "w") as fp:
+            fp.write(VERIFY_SCRIPT)
+            # Make the file executable
+            mode = os.fstat(fp.fileno()).st_mode
+            mode |= 0o111
+            os.fchmod(fp.fileno(), mode & 0o7777)
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        os.chdir(tmpdir)
+        create_directory_structure()
+        username = os.getlogin()
+        euid = os.getegid()
+        subprocess.check_output([
+            'repo2docker',
+            '--user-id', str(euid),
+            '--user-name', username,
+            tmpdir,
+            'python',
+            '/home/{}/verify.py'.format(username)
+        ], stderr=subprocess.STDOUT)