feat(engines): add basic excel validation

document_merge_service/api/data/xlsx-not-valid.xlsx

@@ -0,0 +1 @@
+asdf

document_merge_service/api/engines.py

@@ -186,9 +186,13 @@ def merge(self, data, buf):
             return buf
 
 
+_placeholder_match = re.compile(r"^\s*{{\s*([^{}]+)\s*}}\s*$")
+
+
 class XlsxTemplateEngine:
     def __init__(self, template):
         self.template = template
+        self.writer = None
 
     def validate_is_xlsx(self):
         try:
@@ -201,10 +205,39 @@ def validate(self, available_placeholders=None, sample_data=None):
         self.validate_template_syntax(available_placeholders, sample_data)
 
     def validate_template_syntax(self, available_placeholders=None, sample_data=None):
-        pass
+        # We cannot use jinja to validate because xltpl uses jinja's lexer directly
+        if not sample_data:
+            sample_data = {}
+        buf = io.BytesIO()
+        try:
+            self.merge(sample_data, buf)
+        except TemplateSyntaxError as exc:
+            arg_str = ";".join(exc.args)
+            raise exceptions.ValidationError(f"Syntax error in template: {arg_str}")
+        if available_placeholders:
+            placeholders = []
+            for sheet in self.writer.sheet_resource_map.sheet_state_list:
+                if not sheet.sheet_resource:
+                    continue
+                tree = sheet.sheet_resource.sheet_tree
+                self.collect_placeholders(tree._children, placeholders)
+            missing = set(available_placeholders) - set(placeholders)
+            if missing:
+                raise exceptions.ValidationError(
+                    f"Template uses unavailable placeholders: {str(missing)}"
+                )
+
+    def collect_placeholders(self, children, placeholders):
+        for child in children:
+            if hasattr(child, "value"):
+                value = str(child.value)
+                re_match = _placeholder_match.match(value)
+                if re_match:
+                    placeholders.append(re_match.group(1))
+            self.collect_placeholders(child._children, placeholders)
 
     def merge(self, data, buf):
-        writer = BookWriter(self.template)
+        self.writer = writer = BookWriter(self.template)
 
         writer.jinja_env.filters.update(get_jinja_filters())
         writer.jinja_env.globals.update(dir=dir, getattr=getattr)

document_merge_service/api/tests/test_structure.py → document_merge_service/api/tests/test_excel.py

@@ -1,6 +1,8 @@
 import io
 
 import openpyxl
+import pytest
+from rest_framework import exceptions
 
 from ..data import django_file
 from ..engines import XlsxTemplateEngine
@@ -17,6 +19,8 @@
     ],
 }
 
+_available = ["key0", "key1.subkey1"]
+
 
 def test_structure():
     tmpl = django_file("xlsx-structure.xlsx")
@@ -31,3 +35,21 @@ def test_structure():
         assert ws["A5"].value == "Item: mixed"
         assert ws["A6"].value == "Item: list"
         assert ws["A7"].value == "Subitem: xdata2"
+    engine.validate(_available, _structure)
+    _available.append("huhu")
+    with pytest.raises(exceptions.ValidationError):
+        engine.validate(_available, _structure)
+
+
+def test_syntax_error():
+    tmpl = django_file("xlsx-syntax.xlsx")
+    engine = XlsxTemplateEngine(tmpl)
+    with pytest.raises(exceptions.ValidationError):
+        engine.validate(_available, _structure)
+
+
+def test_valid_error():
+    tmpl = django_file("xlsx-not-valid.xlsx")
+    engine = XlsxTemplateEngine(tmpl)
+    with pytest.raises(exceptions.ParseError):
+        engine.validate(_available, _structure)

document_merge_service/settings.py

@@ -10,7 +10,7 @@
 django_root = environ.Path(__file__) - 2
 
 ENV_FILE = env.str("ENV_FILE", default=django_root(".env"))
-if os.path.exists(ENV_FILE):
+if os.path.exists(ENV_FILE):  # pragma: no cover
     environ.Env.read_env(ENV_FILE)
 
 # per default production is enabled for security reasons