From 76da280dab6ac242dcaf9a6c04223d38474b4f7e Mon Sep 17 00:00:00 2001
From: Sander Schaminee <sander.schaminee@geocat.net>
Date: Tue, 8 Oct 2024 11:16:43 +0200
Subject: [PATCH] Fix sample feature expression
 [GeoCat/qgis-bridge-plugin#189], fix regex char escape warning

---
 bridgestyle/qgis/expressions.py  | 34 +++++++++++++++++++++++---------
 bridgestyle/sld/fromgeostyler.py |  6 ++++--
 2 files changed, 29 insertions(+), 11 deletions(-)

diff --git a/bridgestyle/qgis/expressions.py b/bridgestyle/qgis/expressions.py
index 3fd2825..ccc698e 100644
--- a/bridgestyle/qgis/expressions.py
+++ b/bridgestyle/qgis/expressions.py
@@ -4,7 +4,7 @@
     from qgis.core import (
         QgsExpressionNode, QgsExpression, QgsExpressionNodeBinaryOperator,
         QgsMapLayer, QgsVectorLayer, QgsFeatureRequest, QgsExpressionContext,
-        QgsExpressionContextUtils, QgsFields
+        QgsExpressionContextUtils, QgsFields, QgsFeature
     )
 except (ImportError, ModuleNotFoundError):
     QgsExpressionNodeBinaryOperator = None
@@ -150,18 +150,13 @@ def __init__(self, layer: QgsMapLayer):
 
         self.layer = layer
 
-        if not isinstance(layer, QgsVectorLayer):
-            # Can't sample features from non-vector layers
-            return
-
-        feature = next(layer.getFeatures(QgsFeatureRequest().setFilterFid(0)))
+        feature = self._get_feature(layer)
         if feature is None:
-            # Empty layer: can't get a feature
-            self.warnings.add(f"Layer '{layer.name()}' contains no features")
+            # Not a vector layer or no features found
             return
 
         # Get fields
-        self.fields = layer.fields()
+        self.fields = feature.fields()
 
         # Set context to the first feature
         try:
@@ -172,6 +167,27 @@ def __init__(self, layer: QgsMapLayer):
         except Exception as e:
             self.warnings.add(f"Can't get expression context for layer '{layer.name()}': {str(e)}")
 
+    def _get_feature(self, layer: QgsMapLayer) -> Optional[QgsFeature]:
+        """ Returns the first feature of the given vector layer, or None if there aren't any. """
+        if not isinstance(layer, QgsVectorLayer):
+            # Can't sample features from non-vector layers
+            return None
+
+        try:
+            feature = None
+            for ft in layer.getFeatures(QgsFeatureRequest().setLimit(10)):
+                # Sample 10 features and use the first valid one
+                if ft and ft.isValid():
+                    feature = ft
+                    break
+            if not feature:
+                raise ValueError("no valid feature found")
+        except Exception as e:
+            self.warnings.add(f"Can't get sample feature for layer '{layer.name()}': {str(e)}")
+            return None
+
+        return feature
+
     def __del__(self):
         """ Cleans up the expression converter instance. """
         self.layer = None
diff --git a/bridgestyle/sld/fromgeostyler.py b/bridgestyle/sld/fromgeostyler.py
index 3216f10..e0e0333 100644
--- a/bridgestyle/sld/fromgeostyler.py
+++ b/bridgestyle/sld/fromgeostyler.py
@@ -1,5 +1,5 @@
 import os
-import re
+from re import compile
 from xml.dom import minidom
 from xml.etree import ElementTree
 from xml.etree.ElementTree import Element, SubElement
@@ -14,6 +14,8 @@
 from ..version import __version__
 from ..geostyler.custom_properties import WellKnownText
 
+REGEX_NONWORDCHARS = compile(r'\W')
+
 _warnings = []
 
 
@@ -87,7 +89,7 @@ def _replaceSpecialCharacters(replacement, text=""):
     """
     Replace all characters that are not matching one of [a-zA-Z0-9_].
     """
-    return re.sub('[^\w]', replacement, text)
+    return REGEX_NONWORDCHARS.sub(replacement, text)
 
 
 def _createSymbolizers(symbolizers) -> list: