Duplicate/conflicting fix locations are produced by analysis

[ChrisTimperley]

@rshariffdeen Heads up that some fix locations are being reported twice, leading to duplicate candidate patches. For example, libxml2/CVE-2016-1838. Below is a snippet from the report.json for that bug scenario:

    "fix-locations": [
      {
        "constraint": "((@var(integer, tlen) < 16388) && (0 <= @var(integer, tlen)))",
        "location": "/data/vulnloc/libxml2/CVE-2016-1838/src/parser.c:9836:5",
        "values-file": "/data/vulnloc/libxml2/CVE-2016-1838/analysis/values/#data#vulnloc#libxml2#CVE-2016-1838#src#parser.c:9836:5.csv",
        "variables": [
          {
            "column": 13,
            "line": 9837,
            "name": "ctxt->input->cur[tlen]",
            "type": "integer"
          },
          {
            "column": 13,
            "line": 9837,
            "name": "ctxt",
            "type": "integer"
          },
          {
            "column": 13,
            "line": 9837,
            "name": "ctxt->input->cur[tlen]",
            "type": "pointer"
          },
          {
            "column": 13,
            "line": 9837,
            "name": "ctxt",
            "type": "pointer"
          },
          {
            "column": 30,
            "line": 9837,
            "name": "tlen",
            "type": "integer"
          },
          {
            "column": 26,
            "line": 9837,
            "name": "ctxt->input->cur",
            "type": "pointer"
          },
          {
            "column": 19,
            "line": 9837,
            "name": "ctxt->input",
            "type": "pointer"
          }
        ]
      },
      {
        "constraint": "((@var(integer, tlen) < 16388) && (0 <= @var(integer, tlen)))",
        "location": "/data/vulnloc/libxml2/CVE-2016-1838/src/parser.c:9836:5",
        "values-file": "/data/vulnloc/libxml2/CVE-2016-1838/analysis/values/#data#vulnloc#libxml2#CVE-2016-1838#src#parser.c:9836:24.csv",
        "variables": []
      }
    ]

[ChrisTimperley]

@rshariffdeen It looks like this is still an issue. Below is an except from the localization.json for tests/program-features/switch-case:

[
  {
    "constraint": "((@var(integer, b) - 6) != 0)",
    "distance": 1,
    "location": "/CrashRepair/tests/program-features/switch-case/src/test.c:19:5",
    "values-file": "/CrashRepair/tests/program-features/switch-case/analysis/values/#CrashRepair#tests#program-features#switch-case#src#test.c:21:13.csv",
    "variables": []
  },
  {
    "constraint": "((@var(integer, b) - 6) != 0)",
    "distance": 5,
    "location": "/CrashRepair/tests/program-features/switch-case/src/test.c:19:5",
    "values-file": "/CrashRepair/tests/program-features/switch-case/analysis/values/#CrashRepair#tests#program-features#switch-case#src#test.c:20:9.csv",
    "variables": [
      {
        "column": 13,
        "line": 20,
        "name": "a",
        "type": "integer"
      }
    ]
  }
]

Notice that we have two entries for the same location but those entries have conflicting distances and variable information. The value files for each of them also look wrong.

[rshariffdeen]

I am not able to reproduce this, which commit was used for this @ChrisTimperley ? Actually I don't see any location in line 19 is localized after commit 9b2d69f