streaming parser; reporting on last visited tag; maybe fix through no…

…t re-enconding file's contents

simplified, rearranged code
allowing garbage collection
capturing error context and providing it to user

docker-qgis/process_projectfile.py

@@ -1,6 +1,7 @@
+import io
 import logging
 from pathlib import Path
-from typing import Dict
+from typing import NamedTuple
 from xml.etree import ElementTree
 
 from qfieldcloud.qgis.utils import BaseException, get_layers_data, layers_data_to_string
@@ -37,20 +38,89 @@ class FailedThumbnailGenerationException(BaseException):
     message = "Failed to generate project thumbnail:\n%(reason)s"
 
 
+class LocationError(NamedTuple):
+    line: int
+    column: int
+
+    @classmethod
+    def from_exception_msg(
+        cls, maybe_invalid_token_error_msg: str
+    ) -> "LocationError" | None:
+        """Get column and line numbers from the provided error message."""
+        if "invalid token" not in maybe_invalid_token_error_msg.lower():
+            logger.error("Unable to find 'invalid token' details in the given message")
+            return None
+
+        _, details = maybe_invalid_token_error_msg.split(":")
+        line, column = details.split(",")
+        _, line_number = line.strip().split(" ")
+        _, column_number = column.strip().split(" ")
+        return cls(int(line_number), int(column_number))
+
+
+class ErrorReport(NamedTuple):
+    above: list[str]
+    below: list[str]
+    erroneous_line: str
+    culprit: bytes
+
+    @classmethod
+    def wrap_context(
+        cls, fh: io.TextIOWrapper, location: LocationError
+    ) -> "ErrorReport":
+        """Expand lines above and below error location to provide some context as feedback."""
+        above = []
+        below = []
+        erroneous_line = ""
+        culprit = b""
+
+        fh.seek(0)
+        for cursor_pos, line in enumerate(fh, start=1):
+            if location.line - cursor_pos in range(1, 3):
+                above.append(line)
+            elif cursor_pos - location.line in range(1, 3):
+                below.append(line)
+            elif location.line == cursor_pos:
+                erroneous_line = line
+                culprit = bytes(line[location.column], encoding="utf-8")
+
+        return cls(
+            above=above, below=below, erroneous_line=erroneous_line, culprit=culprit
+        )
+
+    def render(self) -> str:
+        above = "\n".join(self.above)
+        below = "\n".join(self.below)
+        return f"""
+            Lines above: ***\n{above}\n***
+            Erroneous line: ***\n{self.erroneous_line}\n***
+            Lines below: ***\n{below}
+        """
+
+
 def check_valid_project_file(project_filename: Path) -> None:
     logger.info("Check QGIS project file validity…")
 
     if not project_filename.exists():
         raise ProjectFileNotFoundException(project_filename=project_filename)
 
     if project_filename.suffix == ".qgs":
-        try:
-            with open(project_filename) as f:
-                ElementTree.fromstring(f.read())
-        except ElementTree.ParseError as err:
-            raise InvalidXmlFileException(
-                project_filename=project_filename, xml_error=err
-            )
+        with open(project_filename, encoding="utf-8") as fh:
+            try:
+                for event, elem in ElementTree.iterparse(fh):
+                    continue
+            except ElementTree.ParseError as error:
+                error_msg = str(error)
+                location = LocationError.from_exception_msg(error_msg)
+                if location:
+                    contextualized_error = ErrorReport.wrap_context(fh, location)
+                    xml_error = contextualized_error.render()
+                else:
+                    xml_error = error_msg
+                raise InvalidXmlFileException(
+                    project_filename=project_filename,
+                    xml_error=xml_error,
+                )
     elif project_filename.suffix != ".qgz":
         raise InvalidFileExtensionException(
             project_filename=project_filename, extension=project_filename.suffix
@@ -71,7 +141,7 @@ def load_project_file(project_filename: Path) -> QgsProject:
     return project
 
 
-def extract_project_details(project: QgsProject) -> Dict[str, str]:
+def extract_project_details(project: QgsProject) -> dict[str, str]:
     """Extract project details"""
     logger.info("Extract project details…")