ensure file integrity for tex formatters

build/utils/__init__.py

@@ -1,11 +1,9 @@
 from .build_information import BuildInfo
 from .pytex_file import TexFileToFormat
 from .pytex_msg import pytex_msg
-from .checksum import md5
 
 __all__ = [
     'BuildInfo',
     'TexFileToFormat',
-    'pytex_msg',
+    'pytex_msg'
-    'md5'
 ]

build/utils/pytex_file.py

@@ -5,7 +5,7 @@ from PyTeX.build.git_hook import is_recent, get_latest_commit
 from PyTeX import PackageFormatter, ClassFormatter, DictionaryFormatter
 from PyTeX.errors import *
 from .pytex_msg import pytex_msg
-from .checksum import md5
+from PyTeX.utils import md5
 
 from .build_information import BuildInfo
 
@@ -129,7 +129,11 @@ class TexFileToFormat:
         else:
             raise ProgrammingError
         formatter.make_default_macros()
-        written_files = formatter.format_file(self.src_path, self.build_path)
+        written_files = formatter.format_file(
+            input_path=self.src_path,
+            output_dir=self.build_path,
+            relative_name=str(self.src_path.relative_to(self.src_root)),
+            last_build_info=self.last_build_info_all)
         build_infos = []
         for written_file in written_files:
             info = {

default_formatters/dictionary_formatter.py

@@ -29,7 +29,13 @@ class DictionaryFormatter(Formatter):
     def expected_file_name(self):
         return self.file_name
 
-    def format_file(self, input_path: Path, output_dir: Path = None) -> List[str]:
+    def format_file(
+            self,
+            input_path: Path,
+            output_dir: Path = None,
+            relative_name: Optional[str] = None,
+            last_build_info: Optional[List[Dict]] = None
+    ) -> List[str]:
         self.source_file_name = str(input_path.name)
         written_files = []
 

errors/__init__.py

@@ -1,10 +1,12 @@
 from .errors import PyTexError, SubmoduleDirtyForbiddenError, ProgrammingError, ExtraHeaderFileNotFoundError, \
-    UnknownTexVersionError
+    UnknownTexVersionError, ModifiedFileInBuildDirectoryError, UnknownFileInBuildDirectoryError
 
 __all__ = [
     'PyTexError',
     'SubmoduleDirtyForbiddenError',
     'ProgrammingError',
     'ExtraHeaderFileNotFoundError',
-    'UnknownTexVersionError'
+    'UnknownTexVersionError',
+    'ModifiedFileInBuildDirectoryError',
+    'UnknownFileInBuildDirectoryError'
 ]

errors/errors.py

@@ -33,3 +33,24 @@ class UnknownTexVersionError(PyTexError):
             f"Unknown TeX version {tex_version}given. Only 'LaTeX2e' and 'LaTeX3' "
             f"are currently supported"
         )
+
+
+class ModifiedFileInBuildDirectoryError(PyTexError):
+    def __init__(self, filename: str):
+        super().__init__(
+            f"File '{filename}' in the build directory has been modified since the last build. "
+            f"Refusing to overwrite a modified file, since you could lose your manual changes. "
+            f"If you are sure you do not need this anymore, delete it manually and build again. "
+            f"Note that for exactly this reason, it is strongly discouraged to edit built files directly."
+        )
+
+
+class UnknownFileInBuildDirectoryError(PyTexError):
+    def __init__(self, filename: str):
+        super().__init__(
+            f"Unknown file {filename} in build directory found. "
+            f"PyTeX has no knowledge whether this file has been built by PyTeX. "
+            f"Refusing to overwrite this file, since you could lose your data. "
+            f"If you are sure, this can be got rid of, delete the file manually, "
+            f"and run the build again."
+        )

formatter/formatter.py

@@ -1,5 +1,5 @@
 from pathlib import Path
-from typing import List
+from typing import List, Dict, Optional
 
 
 class Formatter:
@@ -10,7 +10,7 @@ class Formatter:
     def make_default_macros(self) -> None:
         pass
 
-    def format_file(self, input_path: Path, output_dir: Path) -> List[str]:
+    def format_file(self, input_path: Path, output_dir: Path, last_build_info: Optional[List[Dict]] = None) -> List[str]:
         pass
 
     def expected_file_name(self) -> str:

formatter/tex_formatter.py

@@ -5,7 +5,8 @@ from typing import Dict, Optional, List
 from datetime import *
 
 from PyTeX.base import Attributes, Args
-from PyTeX.errors import UnknownTexVersionError
+from PyTeX.errors import *
+from PyTeX.utils import md5
 
 from .formatter import Formatter
 
@@ -112,7 +113,12 @@ class TexFormatter(Formatter):
             'format_kwargs': kwargs
         }
 
-    def format_file(self, input_path: Path, output_dir: Path = None) -> List[str]:
+    def format_file(
+            self,
+            input_path: Path,
+            output_dir: Path = None,
+            relative_name: Optional[str] = None,
+            last_build_info: Optional[List[Dict]] = None) -> List[str]:
         self.source_file_name = str(input_path.name)
         input_file = input_path.open()
         lines = input_file.readlines()
@@ -127,5 +133,15 @@ class TexFormatter(Formatter):
         if output_dir is None:
             output_dir = input_path.parent
         output_dir.mkdir(parents=True, exist_ok=True)
+        if (output_dir / self.file_name).exists():
+            found = False
+            for info in last_build_info:
+                if info['source file'] == relative_name:
+                    if not md5(output_dir / self.file_name) == info['md5sum']:
+                        raise ModifiedFileInBuildDirectoryError(str(output_dir / self.file_name))
+                    found = True
+            if not found:
+                raise UnknownFileInBuildDirectoryError(str(output_dir / self.file_name))
+
         (output_dir / self.file_name).write_text(''.join(newlines))
         return [str(self.file_name)]

utils/__init__.py

@@ -0,0 +1,5 @@
+from. checksum import md5
+
+__all__ = [
+    'md5'
+]