incremental building: skip packages that were not modified and are already built

build.py

@@ -1,8 +1,15 @@
-from build_scripts.build import build
+import sys
 from pathlib import Path
 
+from build_scripts.build import build
+
 if __name__ == "__main__":
+    check_existence = True
+    if len(sys.argv) == 2:
+        if sys.argv[1] == '--only-new':
+            check_existence = False
     build(
         src_dir=Path('./src').resolve(),
-        build_dir=Path('./build').resolve()
+        build_dir=Path('./build').resolve(),
+        check_existence=check_existence
     )

build_scripts/build/build.py

@@ -3,12 +3,12 @@ import git
 
 import PyTeX
 
-from build_scripts.git_hook.git_version import get_latest_commit
+from build_scripts.git_hook import get_latest_commit, is_recent
 
 from .build_information import build_information
 
 
-def build(src_dir: Path, build_dir: Path):
+def build(src_dir: Path, build_dir: Path, check_existence: bool = True):
     print('[PyTeX] Getting git repository information...')
     extra_header, repo_description = build_information()
     print('[PyTeX] Building version {version} of LatexPackages'.format(version=repo_description))
@@ -17,22 +17,51 @@ def build(src_dir: Path, build_dir: Path):
         extra_header += ['WARNING: Local changes to git repository detected.',
                          '         The build will not be reproducible (!)']
     num_packages = 0
+    num_skipped_packages = 0
     num_classes = 0
+    num_skipped_classes = 0
 
     for file in src_dir.rglob('*.pysty'):
+        output_dir = build_dir / str(file.parent.relative_to(src_dir))
+        if not is_recent(file, git.Repo()):
+            if not check_existence:
+                print('[PyTex] Skipping file {file} since it was not modified since last build'.format(file=file.name))
+                num_skipped_packages += 1
+                continue
+            else:
+                if (output_dir / ('mkessler-' + str(file.with_suffix('.sty').name))).exists():
+                    print('[PyTex] Skipping file {file} since it was not modified since '
+                          'last build and already exists'.format(file=file.name))
+                    num_skipped_packages += 1
+                    continue
+
         num_packages += 1
         formatter = PyTeX.PackageFormatter(
             package_name=file.with_suffix('').name,
             extra_header=extra_header)
         print('[PyTeX] Writing file {}'.format(formatter.file_name))
         formatter.make_default_macros()
-        formatter.format_file(file, build_dir / str(file.parent.relative_to(src_dir)))
+        formatter.format_file(file, output_dir)
     for file in src_dir.rglob('*.pycls'):
+        output_dir = build_dir / str(file.parent.relative_to(src_dir))
+        if not is_recent(file, git.Repo()):
+            if not check_existence:
+                print('[PyTex] Skipping file {file} since it was not modified since last build'.format(file=file.name))
+                num_skipped_classes += 1
+                continue
+            else:
+                if (output_dir / ('mkessler-' + str(file.with_suffix('.cls').name))).exists():
+                    print('[PyTex] Skipping file {file} since it was not modified since '
+                          'last build and already exists'.format(file=file.name))
+                    num_skipped_classes += 1
+                    continue
+        output_dir = build_dir / str(file.parent.relative_to(src_dir))
         num_classes += 1
         formatter = PyTeX.ClassFormatter(
             class_name=file.with_suffix('').name,
             extra_header=extra_header)
         print('[PyTeX] Writing class file {}'.format(formatter.file_name))
         formatter.make_default_macros()
-        formatter.format_file(file, build_dir / str(file.parent.relative_to(src_dir)))
-    print(f'[PyTeX] Successfully built {num_packages} packages and {num_classes} classes in {build_dir}/')
+        formatter.format_file(input_path=file, output_dir=output_dir)
+    print(f'[PyTeX] Successfully built {num_packages} packages (skipped {num_skipped_packages}) '
+          f'and {num_classes} classes (skipped {num_skipped_classes}) in {build_dir}/')

build_scripts/git_hook/__init__.py

@@ -1,7 +1,9 @@
 from .git_version import git_describe, get_history, get_latest_commit
+from .recent import is_recent
 
 __all__ = [
     'git_describe',
     'get_history',
-    'get_latest_commit'
+    'get_latest_commit',
+    'is_recent'
 ]

build_scripts/git_hook/recent.py

@@ -0,0 +1,14 @@
+from .git_version import get_latest_commit
+
+
+def is_recent(file, repo):
+    modified_files = [item.a_path for item in repo.index.diff(None)]
+    if file in modified_files:
+        return True
+
+    for parent in get_latest_commit(repo).parents:
+        newly_committed_files = [item.a_path for item in repo.index.diff(parent)]
+        if file in newly_committed_files:
+            return True
+    return False
+