FIX: Filter LOG statements from coverage data, not just HTML report

The previous implementation only filtered LOG statements in the HTML report
using genhtml --omit-lines, but the coverage.xml sent to Azure DevOps still
included them, causing incorrect coverage percentage (25%).

Changes:
- Add Python filter to remove DA entries for LOG macro lines from total.info
- Filter uses regex pattern '\bLOG[A-Z_]*\s*\(' to match all LOG variants
- Properly recalculate LH/LF counters after filtering
- Both genhtml and lcov_cobertura now use the filtered data

This ensures consistent coverage metrics across HTML reports and Azure DevOps.

Author: gargsaumya

generate_codecov.sh

@@ -88,20 +88,113 @@ echo "[STEP 4] Merging Python + C++ coverage"
 echo "==================================="
 
 # Merge LCOV reports (ignore inconsistencies in Python LCOV export)
-# Use --omit-lines to exclude LOG macro calls from coverage
-# The regex matches LOG, LOG_ERROR, LOG_WARNING, etc.
-echo "[ACTION] Merging Python and C++ coverage with LOG exclusion"
-lcov -a python-coverage.info -a cpp-coverage.info -o total.info \
-  --ignore-errors inconsistent,corrupt \
-  --omit-lines '\bLOG[A-Z_]*\('
+echo "[ACTION] Merging Python and C++ coverage"
+lcov -a python-coverage.info -a cpp-coverage.info -o total-unfiltered.info \
+  --ignore-errors inconsistent,corrupt
 
-echo "[INFO] LOG statements excluded from coverage using --omit-lines"
+echo "[INFO] Coverage merged successfully"
 
 # Normalize paths so everything starts from mssql_python/
 echo "[ACTION] Normalizing paths in LCOV report"
-sed -i "s|$(pwd)/||g" total.info
+sed -i "s|$(pwd)/||g" total-unfiltered.info
+
+# Filter out LOG statements from coverage data
+# Since we joined multi-line LOGs during build, they're now on single lines
+# We can filter them using a regex pattern
+echo "[ACTION] Filtering LOG statements from coverage data"
+python3 - <<'PYTHON_FILTER'
+import re
+import os
+
+with open('total-unfiltered.info', 'r') as f:
+    content = f.read()
+
+# Process file section by section
+sections = content.split('SF:')
+output_sections = []
+
+if sections[0].strip():
+    output_sections.append(sections[0])  # Keep any header
+
+total_excluded = 0
+LOG_PATTERN = re.compile(r'\bLOG[A-Z_]*\s*\(')
+
+for section in sections[1:]:
+    if not section.strip():
+        continue
+    
+    lines = section.split('\n')
+    source_file = lines[0].strip()
+    
+    # Read source file to find lines with LOG macros
+    excluded_lines = set()
+    if os.path.exists(source_file):
+        try:
+            with open(source_file, 'r', encoding='utf-8', errors='ignore') as src:
+                for line_num, src_line in enumerate(src, 1):
+                    if LOG_PATTERN.search(src_line):
+                        excluded_lines.add(line_num)
+        except Exception as e:
+            print(f"[WARNING] Could not read {source_file}: {e}")
+    
+    # Process section lines - filter DA entries and recalculate LH/LF
+    new_section = [f'SF:{source_file}']
+    da_entries = []
+    other_lines = []
+    lines_hit = 0
+    lines_found = 0
+    
+    for line in lines[1:]:
+        line = line.rstrip()
+        if not line:
+            continue
+            
+        if line.startswith('DA:'):
+            # DA:line_number,hit_count
+            match = re.match(r'DA:(\d+),(\d+)', line)
+            if match:
+                line_num = int(match.group(1))
+                hit_count = int(match.group(2))
+                
+                if line_num not in excluded_lines:
+                    da_entries.append(line)
+                    lines_found += 1
+                    if hit_count > 0:
+                        lines_hit += 1
+                else:
+                    total_excluded += 1
+        elif line.startswith('LH:') or line.startswith('LF:'):
+            # Skip old counters - we'll add new ones
+            continue
+        else:
+            other_lines.append(line)
+    
+    # Rebuild section with correct order: SF, DA entries, other, LH/LF, end_of_record
+    new_section.extend(da_entries)
+    
+    # Add other lines (FN, FNDA, etc.) but keep end_of_record for last
+    for line in other_lines:
+        if line != 'end_of_record':
+            new_section.append(line)
+    
+    # Add recalculated counters
+    new_section.append(f'LH:{lines_hit}')
+    new_section.append(f'LF:{lines_found}')
+    new_section.append('end_of_record')
+    
+    output_sections.append('\n'.join(new_section))
+
+# Write filtered coverage with recalculated statistics
+with open('total.info', 'w') as f:
+    f.write('\n'.join(output_sections))
+
+print(f"[INFO] Filtered {total_excluded} LOG statement lines from coverage")
+PYTHON_FILTER
+
+echo "[INFO] LOG statements filtered from coverage data"
 
 # Generate full HTML report
+echo "[ACTION] Generating HTML coverage report"
 genhtml total.info \
   --output-directory unified-coverage \
   --quiet \