Fix files only covered by one LCOV report showing 100% coverage (#433)

* Fix files only covered by one LCOV report showing 100% coverage

* Add LCOV reporter tests

diff_cover/violationsreporters/violations_reporter.py

@@ -392,6 +392,8 @@ def _cache_file(self, src_path):
                 src_search_path = src_abs_path
                 if src_search_path not in lcov_document:
                     src_search_path = src_rel_path
+                if src_search_path not in lcov_document:
+                    continue
 
                 # First case, need to define violations initially
                 if violations is None:

tests/test_violations_reporter.py

@@ -5,6 +5,7 @@
 
 import os
 import subprocess
+import tempfile
 import xml.etree.ElementTree as etree
 from io import BytesIO, StringIO
 from subprocess import Popen
@@ -18,6 +19,7 @@
 from diff_cover.violationsreporters.violations_reporter import (
     CppcheckDriver,
     EslintDriver,
+    LcovCoverageReporter,
     PylintDriver,
     Violation,
     XmlCoverageReporter,
@@ -794,6 +796,201 @@ def _coverage_xml(self, file_paths, violations, measured):
         return root
 
 
+class TestLcovCoverageReporterTest:
+    MANY_VIOLATIONS = {
+        Violation(3, None),
+        Violation(7, None),
+        Violation(11, None),
+        Violation(13, None),
+    }
+    FEW_MEASURED = {2, 3, 5, 7, 11, 13}
+
+    FEW_VIOLATIONS = {Violation(3, None), Violation(11, None)}
+    MANY_MEASURED = {2, 3, 5, 7, 11, 13, 17}
+
+    ONE_VIOLATION = {Violation(11, None)}
+    VERY_MANY_MEASURED = {2, 3, 5, 7, 11, 13, 17, 23, 24, 25, 26, 26, 27}
+
+    @pytest.fixture(autouse=True)
+    def patch_git_patch(self, mocker):
+        # Paths generated by git_path are always the given argument
+        _git_path_mock = mocker.patch(
+            "diff_cover.violationsreporters.violations_reporter.GitPathTool"
+        )
+        _git_path_mock.relative_path = lambda path: path
+        _git_path_mock.absolute_path = lambda path: path
+
+    def test_violations(self):
+        # Construct the LCOV report
+        file_paths = ["file1.java", "subdir/file2.java"]
+        violations = self.MANY_VIOLATIONS
+        measured = self.FEW_MEASURED
+        lcov = self._coverage_lcov(file_paths, violations, measured)
+
+        # Parse the report
+        coverage = LcovCoverageReporter([lcov])
+
+        # Expect that the name is set
+        assert coverage.name() == "LCOV"
+
+        # By construction, each file has the same set
+        # of covered/uncovered lines
+        assert violations == coverage.violations("file1.java")
+        assert measured == coverage.measured_lines("file1.java")
+
+        # Try getting a smaller range
+        result = coverage.violations("subdir/file2.java")
+        assert result == violations
+
+        # Once more on the first file (for caching)
+        result = coverage.violations("file1.java")
+        assert result == violations
+
+    def test_two_inputs_first_violate(self):
+        # Construct the LCOV report
+        file_paths = ["file1.java"]
+
+        violations1 = self.MANY_VIOLATIONS
+        violations2 = self.FEW_VIOLATIONS
+
+        measured1 = self.FEW_MEASURED
+        measured2 = self.MANY_MEASURED
+
+        lcov = self._coverage_lcov(file_paths, violations1, measured1)
+        lcov2 = self._coverage_lcov(file_paths, violations2, measured2)
+
+        # Parse the report
+        coverage = LcovCoverageReporter([lcov, lcov2])
+
+        # By construction, each file has the same set
+        # of covered/uncovered lines
+        assert violations1 & violations2 == coverage.violations("file1.java")
+
+        assert measured1 | measured2 == coverage.measured_lines("file1.java")
+
+    def test_two_inputs_second_violate(self):
+        # Construct the LCOV report
+        file_paths = ["file1.java"]
+
+        violations1 = self.MANY_VIOLATIONS
+        violations2 = self.FEW_VIOLATIONS
+
+        measured1 = self.FEW_MEASURED
+        measured2 = self.MANY_MEASURED
+
+        lcov = self._coverage_lcov(file_paths, violations1, measured1)
+        lcov2 = self._coverage_lcov(file_paths, violations2, measured2)
+
+        # Parse the report
+        coverage = LcovCoverageReporter([lcov2, lcov])
+
+        # By construction, each file has the same set
+        # of covered/uncovered lines
+        assert violations1 & violations2 == coverage.violations("file1.java")
+
+        assert measured1 | measured2 == coverage.measured_lines("file1.java")
+
+    def test_three_inputs(self):
+        # Construct the LCOV report
+        file_paths = ["file1.java"]
+
+        violations1 = self.MANY_VIOLATIONS
+        violations2 = self.FEW_VIOLATIONS
+        violations3 = self.ONE_VIOLATION
+
+        measured1 = self.FEW_MEASURED
+        measured2 = self.MANY_MEASURED
+        measured3 = self.VERY_MANY_MEASURED
+
+        lcov = self._coverage_lcov(file_paths, violations1, measured1)
+        lcov2 = self._coverage_lcov(file_paths, violations2, measured2)
+        lcov3 = self._coverage_lcov(file_paths, violations3, measured3)
+
+        # Parse the report
+        coverage = LcovCoverageReporter([lcov2, lcov, lcov3])
+
+        # By construction, each file has the same set
+        # of covered/uncovered lines
+        assert violations1 & violations2 & violations3 == coverage.violations(
+            "file1.java"
+        )
+
+        assert measured1 | measured2 | measured3 == coverage.measured_lines(
+            "file1.java"
+        )
+
+    def test_different_files_in_inputs(self):
+        # Construct the LCOV report
+        lcov_repots = [
+            self._coverage_lcov(["file.java"], self.MANY_VIOLATIONS, self.FEW_MEASURED),
+            self._coverage_lcov(
+                ["other_file.java"], self.FEW_VIOLATIONS, self.MANY_MEASURED
+            ),
+        ]
+
+        # Parse the report
+        coverage = LcovCoverageReporter(lcov_repots)
+
+        assert self.MANY_VIOLATIONS == coverage.violations("file.java")
+        assert self.FEW_VIOLATIONS == coverage.violations("other_file.java")
+
+    def test_empty_violations(self):
+        """
+        Test that an empty violations report is handled properly
+        """
+        # Construct the LCOV report
+        file_paths = ["file1.java"]
+
+        violations1 = self.MANY_VIOLATIONS
+        violations2 = set()
+
+        measured1 = self.FEW_MEASURED
+        measured2 = self.MANY_MEASURED
+
+        lcov = self._coverage_lcov(file_paths, violations1, measured1)
+        lcov2 = self._coverage_lcov(file_paths, violations2, measured2)
+
+        # Parse the report
+        coverage = LcovCoverageReporter([lcov2, lcov])
+
+        # By construction, each file has the same set
+        # of covered/uncovered lines
+        assert violations1 & violations2 == coverage.violations("file1.java")
+
+        assert measured1 | measured2 == coverage.measured_lines("file1.java")
+
+    def test_no_such_file(self):
+        # Construct the LCOV report with no source files
+        lcov = self._coverage_lcov([], [], [])
+
+        # Parse the report
+        coverage = LcovCoverageReporter(lcov)
+
+        # Expect that we get no results
+        result = coverage.violations("file.java")
+        assert result == set()
+
+    def _coverage_lcov(self, file_paths, violations, measured):
+        """
+        Build an LCOV document based on the provided arguments.
+        """
+
+        violation_lines = {violation.line for violation in violations}
+
+        with tempfile.NamedTemporaryFile("w", delete=False) as f:
+            for file_path in file_paths:
+                f.write(f"SF:{file_path}\n")
+                for line_num in measured:
+                    f.write(
+                        f"DA:{line_num},{0 if line_num in violation_lines else 1}\n"
+                    )
+                f.write("end_of_record\n")
+        try:
+            return LcovCoverageReporter.parse(f.name)
+        finally:
+            os.unlink(f.name)
+
+
 class TestPycodestyleQualityReporterTest:
     def test_quality(self, mocker, process_patcher):
         # Patch the output of `pycodestyle`