Prepare data quality stats feature.

6 files changed, 153 insertions, 43 deletions

diff --git a/src/bm/bmrequest.cpp b/src/bm/bmrequest.cpp
index 9300dc3..5f008b4 100644
--- a/src/bm/bmrequest.cpp
+++ b/src/bm/bmrequest.cpp
@@ -4848,8 +4848,15 @@ QByteArray BMRequest_IndexGetValues::toRequestBuffer(QString *)
     QString request =
         QString(
             "<request type=\"%1\"><args baseTimestamp=\"%2\" medianWinSize=\"%3\" "
-            "cacheKey=\"%4\" />")
-        .arg(name()).arg(baseTimestamp).arg(medianWinSize).arg(cacheKey);
+            "cacheKey=\"%4\" dataQualityStats=\"%5\" dqStatsDiffTol=\"%6\" "
+            "dqStatsStabTol=\"%7\" />")
+        .arg(name())
+        .arg(baseTimestamp)
+        .arg(medianWinSize)
+        .arg(cacheKey)
+        .arg(dataQualityStats ? 1 : 0)
+        .arg(dqStatsDiffTol)
+        .arg(dqStatsStabTol);
 
     for (int i = 0; i < testCaseFilter.size(); ++i)
         request += QString("<testCase name=\"%1\" />").arg(testCaseFilter.at(i));
@@ -4911,6 +4918,14 @@ QByteArray BMRequest_IndexGetValues::toReplyBuffer()
     medianWinSize = argsElem.attributeNode("medianWinSize").value().toInt(&ok);
     Q_ASSERT(ok);
 
+    // Get data quality stats params ...
+    dataQualityStats = argsElem.attributeNode("dataQualityStats").value().toInt(&ok);
+    Q_ASSERT(ok);
+    dqStatsDiffTol = argsElem.attributeNode("dqStatsDiffTol").value().toDouble(&ok);
+    Q_ASSERT(ok);
+    dqStatsStabTol = argsElem.attributeNode("dqStatsStabTol").value().toInt(&ok);
+    Q_ASSERT(ok);
+
     // Get filters ...
     QDomNodeList testCaseNodes = doc.elementsByTagName("testCase");
     for (int i = 0; i < testCaseNodes.size(); ++i)
@@ -5085,7 +5100,7 @@ QByteArray BMRequest_IndexGetValues::toReplyBuffer()
 
         QList<qreal> indexValues;
         QList<int> contrCounts;
-        QList<QList<Index::RankedInfo> > topContr;
+        QList<QList<Index::RankedInfo> > topContr; // Top contributors for each index value
         const int topContrLimit = 10; // ### hard-coded for now!
         QString error_;
         if (!index.computeValues(
diff --git a/src/bm/bmrequest.h b/src/bm/bmrequest.h
index 89c50f3..43accf9 100644
--- a/src/bm/bmrequest.h
+++ b/src/bm/bmrequest.h
@@ -568,10 +568,13 @@ public:
         const int baseTimestamp, const int medianWinSize,
         const QString &cacheKey, const QStringList &testCaseFilter,
         const QStringList &metricFilter, const QStringList &platformFilter,
-        const QStringList &hostFilter, const QStringList &branchFilter)
+        const QStringList &hostFilter, const QStringList &branchFilter,
+        const bool dataQualityStats, const qreal dqStatsDiffTol, const qreal dqStatsStabTol)
         : baseTimestamp(baseTimestamp), medianWinSize(medianWinSize), cacheKey(cacheKey)
         , testCaseFilter(testCaseFilter), metricFilter(metricFilter)
         , platformFilter(platformFilter), hostFilter(hostFilter), branchFilter(branchFilter)
+        , dataQualityStats(dataQualityStats), dqStatsDiffTol(dqStatsDiffTol)
+        , dqStatsStabTol(dqStatsStabTol)
     {}
     BMRequest_IndexGetValues(const QDomDocument &doc) : BMRequest(doc) {}
 
@@ -584,6 +587,9 @@ private:
     mutable QStringList platformFilter;
     mutable QStringList hostFilter;
     mutable QStringList branchFilter;
+    mutable bool dataQualityStats;
+    mutable qreal dqStatsDiffTol;
+    mutable qreal dqStatsStabTol;
 
     QString name() const { return "IndexGetValues"; }
     QByteArray toRequestBuffer(QString *error);
diff --git a/src/bm/index.cpp b/src/bm/index.cpp
index 6bc6c78..be29787 100644
--- a/src/bm/index.cpp
+++ b/src/bm/index.cpp
@@ -280,5 +280,36 @@ bool IndexAlgorithm1::computeValues(
         }
     }
 
+
+    const bool computeDataQualityStatistics = false; // 4 NOW
+    if (computeDataQualityStatistics) {
+        // Step 1: Compute the MaxESSTotalCount and MaxESSStableCount for each
+        //         contributing RH (compute for the exact median-smoothed
+        //         values that formed the basis for computing the index)
+        //
+        // Step 2: Compute the complete distribution of MaxESSTotalCount
+        //         (note that the number of distinct counts are likely to be only a small
+        //         fraction of the number of RHs):
+        //
+        //           TC(c1) = <# of RHs with a MaxESSTotalCount of c1>
+        //           TC(c2) = <# of RHs with a MaxESSTotalCount of c2>
+        //           ...
+        //           TC(cN) = <# of RHs with a MaxESSTotalCount of cN>
+        //
+        //         (TC = Total Count, and N is number of distinct counts)
+        //          
+        // Step 3: Compute the percentile distribution (for the 10 levels 10%, 20%, ..., 100%)
+        //         of the SF (stability fraction) values (where the SF for a given
+        //         RH is MaxESSStableCount / MaxESSTotalCount):
+        //
+        //           SFP(100) = <the max SF value for the worst 100% of the RHs (i.e. all RHs!)>
+        //           SFP(90)  = <the max SF value for the worst  90% of the RHs>
+        //           SFP(80)  = <the max SF value for the worst  80% of the RHs>
+        //           ...
+        //           SFP(10)  = <the max SF value for the worst  10% of the RHs>
+        //
+        //           (SFP = Stability Fraction Percentile)
+    }
+
     return true;
 }
diff --git a/src/bmclient/main.cpp b/src/bmclient/main.cpp
index 738e734..da99f3c 100644
--- a/src/bmclient/main.cpp
+++ b/src/bmclient/main.cpp
@@ -1036,9 +1036,49 @@ BMRequest * Executor::createIndexGetValuesRequest(
     if (!BMMisc::getMultiOption(args, "-branch", &branchFilter, error))
         return 0;
 
+    // Get data quality stats params ...
+    bool dataQualityStats = false;
+    qreal dqStatsDiffTol = -1;
+    int dqStatsStabTol = -1;
+    if (BMMisc::hasOption(args, "-dataqualitystats")) {
+        dataQualityStats = true;
+
+        // Difference tolerance ...
+        if (BMMisc::getOption(args, "-dqstatsdifftol", &values, 1, 0, error)) {
+            bool ok;
+            dqStatsDiffTol = values.at(0).toDouble(&ok);
+            if (!ok) {
+                *error =
+                    "failed to extract difference tolerance for data quality stats as a double";
+                return 0;
+            }
+            dqStatsDiffTol = qMax(dqStatsDiffTol, 0.0);
+        } else {
+            if (error->isEmpty())
+                *error = QString("-dqstatsdifftol option not found");
+            return 0;
+        }
+
+        // Stability tolerance ...
+        if (BMMisc::getOption(args, "-dqstatsstabtol", &values, 1, 0, error)) {
+            bool ok;
+            dqStatsStabTol = values.at(0).toInt(&ok);
+            if ((!ok) || (dqStatsStabTol < 2)) {
+                *error =
+                    "failed to extract stabiliity tolerance for data quality stats as an "
+                    "integer >= 2";
+                return 0;
+            }
+        } else {
+            if (error->isEmpty())
+                *error = QString("-dqstatsstabtol option not found");
+            return 0;
+        }
+    }
+
     return new BMRequest_IndexGetValues(
         baseTimestamp, medianWinSize, cacheKey, testCaseFilter, metricFilter, platformFilter,
-        hostFilter, branchFilter);
+        hostFilter, branchFilter, dataQualityStats, dqStatsDiffTol, dqStatsStabTol);
 }
 
 BMRequest * Executor::createIndexPutConfigRequest(const QStringList &args, QString *error) const
@@ -1582,7 +1622,8 @@ class DirectExecutor : public Executor
             "      -basetimestamp <...> [-medianwinsize <...> (default = 8)] \\\n"
             "      [-testcase <...> -testcase <...> ...] [-metric <...> -metric <...> ...] \\\n"
             "      [-platform <...> -platform <...> ...] [-host <...> -host <...> ...] \\\n"
-            "      [-branch <...> -branch <...> ...]\n"
+            "      [-branch <...> -branch <...> ...] \\\n"
+            "      [-dataqualitystats -dqstatsdifftol <...>] \n"
 
             <<
             "index get plot  <SAME AS 'index get values' except that the optional \\\n"
diff --git a/src/bmweb/indexsection.js b/src/bmweb/indexsection.js
index c3fac25..05d15d3 100644
--- a/src/bmweb/indexsection.js
+++ b/src/bmweb/indexsection.js
@@ -69,6 +69,15 @@ function updateComputeIndexLink(defaultBaseTime, defaultMedianWinSize)
     url += createFilterOptionsString("-host",     getCheckedFilterValues("ixHostFilter"));
     url += createFilterOptionsString("-branch",   getCheckedFilterValues("ixBranchFilter"));
 
+
+    if (document.getElementById("ixDQStatsEnabled").checked) {
+        // Add parameters for data quality statistics ...
+        url += " -dataqualitystats"
+            + " -dqstatsdifftol " + document.getElementById("ixDQStatsDiffTol").value
+            + " -dqstatsstabtol " + selectedOptions("ixDQStatsStabTol")[0];
+    }
+
+
     // Update link ...
     link = document.getElementById("ixComputeIndexLink");
     link.setAttribute("href", url);
@@ -537,14 +546,6 @@ function IndexSection()
         table = section.appendChild(document.createElement("table"));
         table.setAttribute("style", "border:0px; padding:0px");
         tr = table.insertRow(0);
-        td = tr.insertCell(0);
-        td.setAttribute("style", "border:0px; padding:0px");
-        linksFieldset = td.appendChild(document.createElement("fieldset"));
-        linksFieldset.setAttribute("style", "border:0px; padding:2px");
-
-        table = linksFieldset.appendChild(document.createElement("table"));
-        table.setAttribute("style", "border:0px; padding:0px");
-        tr = table.insertRow(0);
 
         // ... 'compute index' link ...
         td = tr.insertCell(tr.cells.length);
@@ -562,22 +563,51 @@ function IndexSection()
                 + ")"
         );
 
-        // ... 'compute stats' link ...
+        // ... data quality stats fieldset ...
         td = tr.insertCell(tr.cells.length);
-        td.setAttribute("style", "border:0px; padding:2px");
-        link = td.appendChild(document.createElement("a"));
-        link.id = "ixComputeStatsLink";
-        link.appendChild(document.createTextNode("Compute Stats"));
-        link.setAttribute("class", "ixLink_stats");
-        link.setAttribute("href", "dummy"); // for correct initial rendering
-        link.setAttribute(
-            "onmouseover",
-            "updateComputeStatsLink("
-                + defaultBaseTime
-                + ", " + defaultMedianWinSize
-                + ")"
-        );
+        td.setAttribute("style", "border:0px; padding:0px");
+        fieldset = td.appendChild(document.createElement("fieldset"));
+        legend = fieldset.appendChild(document.createElement("legend"));
+        legend.appendChild(document.createTextNode("Data Quality Statistics"));
+        legend.setAttribute("style", "font-size:12");
+        table = fieldset.appendChild(document.createElement("table"));
+        table.setAttribute("style", "border:0px; padding:0px");
 
+        tr = table.insertRow(table.rows.length);
+        td = tr.insertCell(0);
+        td.innerHTML = "Enabled: ";
+        td.setAttribute("style", "text-align:right");
+        td = tr.insertCell(1);
+        td.setAttribute("style", "text-align:left");
+        input = td.appendChild(document.createElement("input"));
+        input.setAttribute("type", "checkbox");
+        input.setAttribute("id", "ixDQStatsEnabled");
+
+        tr = table.insertRow(table.rows.length);
+        td = tr.insertCell(0);
+        td.setAttribute("style", "text-align:right");
+        td.innerHTML = "Difference tolerance (%): ";
+        td = tr.insertCell(1);
+        td.setAttribute("style", "text-align:left");
+        input = td.appendChild(document.createElement("input"));
+        input.setAttribute("type", "text");
+        input.setAttribute("id", "ixDQStatsDiffTol");
+        input.setAttribute("Value", "10");
+
+        tr = table.insertRow(table.rows.length);
+        td = tr.insertCell(0);
+        td.setAttribute("style", "text-align:right");
+        td.innerHTML = "Stability tolerance: ";
+        td = tr.insertCell(1);
+        td.setAttribute("style", "text-align:left");
+        select = td.appendChild(document.createElement("select"));
+        select.setAttribute("id", "ixDQStatsStabTol");
+        select.options[0] = new Option("2", "2");
+        select.options[1] = new Option("3", "3");
+        select.options[2] = new Option("4", "4");
+        select.options[3] = new Option("5", "5");
+        select.options[4] = new Option("6", "6");
+        select.options[2].selected = true;
 
         // Create 'Configuration' section ...
         section.appendChild(document.createElement("br"));
@@ -623,6 +653,7 @@ function IndexSection()
         tr = table.insertRow(table.rows.length);
         td = tr.insertCell(0);
         td.innerHTML = "Base timestamp (secs since 1970): ";
+        td.setAttribute("style", "text-align:right");
         td = tr.insertCell(1);
         input = td.appendChild(document.createElement("input"));
         input.setAttribute("type", "text");
@@ -633,6 +664,7 @@ function IndexSection()
         tr = table.insertRow(table.rows.length);
         td = tr.insertCell(0);
         td.innerHTML = "Median window size: ";
+        td.setAttribute("style", "text-align:right");
         td = tr.insertCell(1);
         input = td.appendChild(document.createElement("input"));
         input.setAttribute("type", "text");
diff --git a/src/bmweb/style.css b/src/bmweb/style.css
index cda1891..83e010b 100644
--- a/src/bmweb/style.css
+++ b/src/bmweb/style.css
@@ -101,21 +101,6 @@ a.ixLink_index:hover, a.ixLink_index:active {
     background-color:#696969;
 }
 
-a.ixLink_stats:link, a.ixLink_stats:visited {
-    display:block;
-    font-weight:bold;
-    font-size:12px;
-    color:#ffffff;
-    background-color:#c97f47;
-    width:120px;
-    text-align:center;
-    padding:4px;
-    text-decoration:none;
-}
-a.ixLink_stats:hover, a.ixLink_stats:active {
-    background-color:#a26739;
-}
-
 a.bcLink:link, a.bcLink:visited {
     display:block;
     font-weight:bold;