Skip to content

Commit 24c7149

Browse files
committed
Add graph visualizations to anomaly detection
1 parent 5d1a5e4 commit 24c7149

File tree

4 files changed

+222
-1
lines changed

4 files changed

+222
-1
lines changed
Lines changed: 65 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,65 @@
1+
// Anomaly Detection Graphs: Find top nodes marked as "hub" including their incoming dependencies and output them in Graphviz format.
2+
3+
RETURN "graph [label=\"" + $projection_language + " " + $projection_node_label + " - Top Rank " + $projection_node_rank + " Hub\"];" AS graphVizDotNotationLine
4+
UNION ALL
5+
// Step 1: Query overall statistics, e.g. min/max weight for later normalization
6+
MATCH (sourceForStatistics)-[dependencyForStatistics:DEPENDS_ON]->(targetForStatistics)
7+
WHERE $projection_node_label IN labels(sourceForStatistics)
8+
AND $projection_node_label IN labels(targetForStatistics)
9+
WITH min(coalesce(dependencyForStatistics.weight25PercentInterfaces, dependencyForStatistics.weight)) AS minWeight
10+
,max(coalesce(dependencyForStatistics.weight25PercentInterfaces, dependencyForStatistics.weight)) AS maxWeight
11+
// Step 2: Query direct dependencies to the target
12+
MATCH (target)
13+
WHERE $projection_node_label IN labels(target)
14+
AND target.anomalyScore > 0
15+
AND target.anomalyHubRank = toInteger($projection_node_rank)
16+
MATCH (source)-[directDependency:DEPENDS_ON]->(target)
17+
WHERE $projection_node_label IN labels(source)
18+
ORDER BY directDependency.weight DESC
19+
WITH minWeight
20+
,maxWeight
21+
,target
22+
,collect(source)[0..60] AS sources
23+
,collect(directDependency)[0..60] AS directDependencies
24+
// Step 3: Query dependencies among sources
25+
UNWIND sources AS source1
26+
UNWIND sources AS source2
27+
MATCH (source1)-[indirectDependency:DEPENDS_ON]->(source2)
28+
WITH minWeight
29+
,maxWeight
30+
,target
31+
,directDependencies
32+
,collect(indirectDependency) AS indirectDependencies
33+
WITH *, directDependencies + indirectDependencies AS allDependencies
34+
// Step 4: Prepare results in GraphViz format for all dependencies
35+
UNWIND allDependencies AS dependency
36+
WITH *, (endNode(dependency) = target) AS isTargetEndNode
37+
WITH *, CASE WHEN isTargetEndNode THEN endNode(dependency) ELSE null END AS targetEndNodeOrNull
38+
WITH *, CASE WHEN isTargetEndNode THEN null ELSE endNode(dependency) END AS nonTargetEndNodeOrNull
39+
WITH *, coalesce(dependency.weight25PercentInterfaces, dependency.weight) AS weight
40+
WITH *, toFloat(weight - minWeight) / toFloat(maxWeight - minWeight) AS normalizedWeight
41+
WITH *, round((normalizedWeight * 2) + 0.4, 1) AS penWidth
42+
WITH *, coalesce(target.fqn, target.globalFqn, target.fileName, target.signature, target.name) AS targetName
43+
WITH *, replace(replace(targetName, '.', '.\\n'), '/', '/\\n') AS targetNameSplit
44+
WITH *, "\\n(hub #" + targetEndNodeOrNull.anomalyHubRank + ")" AS hubSubLabel
45+
WITH *, "\"" + targetNameSplit + hubSubLabel + "\"" AS hubLabel
46+
WITH *, coalesce("hub [label=" + hubLabel+ ";]; ", "") AS hubNode
47+
WITH *, "\"" + startNode(dependency).name + "\"" AS sourceNode
48+
WITH *, coalesce("\"" + nonTargetEndNodeOrNull.name + "\"", "\"hub\"") AS targetNode
49+
WITH *, " -> " + targetNode
50+
+ " [label = " + weight + ";"
51+
+ " penwidth = " + penWidth + ";"
52+
+ " ];" AS graphVizDotNotationEdge
53+
WITH *, hubNode + sourceNode + coalesce(graphVizDotNotationEdge, " [];") AS graphVizDotNotationLine
54+
ORDER BY target.anomalyHubRank DESC, target.name ASC
55+
RETURN DISTINCT graphVizDotNotationLine
56+
//Debugging
57+
//,startNode(dependency).name AS sourceName
58+
//,endNode(dependency).name AS targetName
59+
//,hubNode
60+
//,penWidth
61+
//,normalizedWeight
62+
//,dependency.weight AS weight
63+
//,minWeight
64+
//,maxWeight
65+
LIMIT 100
Lines changed: 26 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,26 @@
1+
// This is a GraphViz dot template file for the visualization of a anomaly archetype graph.
2+
// The main part of the template is marked by the comments "Begin-Template" and "End-Template".
3+
// It also contains a simple example graph.
4+
//
5+
strict digraph top_hub_template {
6+
//Begin-Template
7+
graph [layout = "fdp"; start = "7", fontname = "Helvetica,Arial,sans-serif"; labelloc = "t"];
8+
node [fontsize = 8;];
9+
edge [fontsize = 4;];
10+
node [style = "filled"; color = "0.58 0.75 0.75"; fillcolor = "0.58 0.15 0.99"; margin = "0.00001,0.00001";];
11+
edge [color = "0.58 0.75 0.85"; arrowsize = "0.4";];
12+
13+
hub [shape = "doublecircle";];
14+
hub [fontsize = 10;];
15+
hub [color = "0.52 0.7 0.7"; fillcolor = "0.52 0.4 0.9"; penwidth = 3;];
16+
17+
limit_hint [color = "0.52 0.7 0.7"; fillcolor = "0.52 0.4 0.9";]
18+
limit_hint [shape = "note"; penwidth = 2; fontsize = 10]
19+
limit_hint [label = "limited to\n50 nodes...";]
20+
limit_hint -> hub // Signals that the number of edges might have been limited
21+
22+
//End-Template
23+
"A" -> "hub" [penwidth = 1.0; label = 1;];
24+
"A" -> "B" [penwidth = 3.0; label = 4;];
25+
"B" -> "hub" [penwidth = 2.0; label = 2;];
26+
}
Lines changed: 130 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,130 @@
1+
#!/usr/bin/env bash
2+
3+
# Executes selected anomaly detection Cypher queries for GraphViz visualization.
4+
# Visualizes top ranked anomaly archetypes.
5+
# Requires an already running Neo4j graph database with already scanned and analyzed artifacts.
6+
# The reports (csv, dot and svg files) will be written into the sub directory reports/anomaly-detection/{language}_{codeUnit}.
7+
8+
# Requires executeQueryFunctions.sh, visualizeQueryResults.sh, cleanupAfterReportGeneration.sh
9+
10+
# Fail on any error ("-e" = exit on first error, "-o pipefail" exist on errors within piped commands)
11+
set -o errexit -o pipefail
12+
13+
# Overrideable Constants (defaults also defined in sub scripts)
14+
REPORTS_DIRECTORY=${REPORTS_DIRECTORY:-"reports"}
15+
16+
## Get this "scripts/reports" directory if not already set
17+
# Even if $BASH_SOURCE is made for Bourne-like shells it is also supported by others and therefore here the preferred solution.
18+
# CDPATH reduces the scope of the cd command to potentially prevent unintended directory changes.
19+
# This way non-standard tools like readlink aren't needed.
20+
ANOMALY_DETECTION_GRAPHS_DIR=${REPORTS_SCRIPT_DIR:-$( CDPATH=. cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd -P )}
21+
#echo "anomalyDetectionGraphVisualization: ANOMALY_DETECTION_GRAPHS_DIR=${ANOMALY_DETECTION_GRAPHS_DIR}"
22+
23+
# Get the "scripts" directory by taking the path of this script and going one directory up.
24+
SCRIPTS_DIR=${SCRIPTS_DIR:-"${ANOMALY_DETECTION_GRAPHS_DIR}/../../../scripts"} # Repository directory containing the shell scripts
25+
# echo "anomalyDetectionGraphVisualization: SCRIPTS_DIR=${SCRIPTS_DIR}"
26+
27+
# Get the "scripts/visualization" directory.
28+
VISUALIZATION_SCRIPTS_DIR=${VISUALIZATION_SCRIPTS_DIR:-"${SCRIPTS_DIR}/visualization"} # Repository directory containing the shell scripts for visualization
29+
# echo "anomalyDetectionGraphVisualization: VISUALIZATION_SCRIPTS_DIR=${VISUALIZATION_SCRIPTS_DIR}"
30+
31+
# Define functions to execute cypher queries from within a given file
32+
source "${SCRIPTS_DIR}/executeQueryFunctions.sh"
33+
34+
# Runs a parametrized query, converts their results in GraphViz format and creates a Graph visualization.
35+
# Outputs (at most) 10 indexed files (for report_name="TopHub" then TopHub1, TopHub2,...) with a focused visualization of one selected node and its surroundings.
36+
#
37+
# Required Parameters:
38+
# - report_name=...
39+
# Name of the query, the template and then also the resulting visualization file.
40+
# - projection_language=...
41+
# Name of the associated programming language. Examples: "Java", "Typescript"
42+
# - projection_node_label=...
43+
# Label of the nodes that will be used for the projection. Example: "Package"
44+
create_graph_visualization() {
45+
local nodeLabel
46+
nodeLabel=$( extractQueryParameter "projection_node_label" "${@}" )
47+
48+
local language
49+
language=$( extractQueryParameter "projection_language" "${@}" )
50+
51+
local report_name
52+
report_name=$( extractQueryParameter "report_name" "${@}" )
53+
54+
echo "anomalyDetectionGraphVisualization: $(date +'%Y-%m-%dT%H:%M:%S%z') Creating ${language} ${nodeLabel} ${report_name} visualizations..."
55+
56+
local detail_report_directory_name="${language}_${nodeLabel}"
57+
local detail_report_directory="${FULL_REPORT_DIRECTORY}/${detail_report_directory_name}/GraphVisualizations"
58+
mkdir -p "${detail_report_directory}"
59+
60+
#TODO output all 10 indices
61+
for index in {1..2}; do
62+
# Query Graph data
63+
local resultFileName="${detail_report_directory}/${report_name}${index}"
64+
local queryResultFile="${resultFileName}.csv"
65+
execute_cypher "${ANOMALY_DETECTION_GRAPHS_DIR}/${report_name}.cypher" "${@}" "projection_node_rank=${index}" > "${queryResultFile}" || true
66+
67+
# Remove empty files
68+
# Note: After that, detail_report_directory might be deleted as well.
69+
# In that case the image generation is finished and the loop needs to be terminated.
70+
source "${SCRIPTS_DIR}/cleanupAfterReportGeneration.sh" "${detail_report_directory}"
71+
# Stop generation as soon as the first query result is empty or the directory is deleted.
72+
if [ ! -f "${queryResultFile}" ] ; then
73+
break;
74+
fi
75+
76+
# Generate svg image using GraphViz
77+
source "${VISUALIZATION_SCRIPTS_DIR}/visualizeQueryResults.sh" "${queryResultFile}" --template "${ANOMALY_DETECTION_GRAPHS_DIR}/TopHub.template.gv"
78+
79+
# Clean up after graph visualization image generation:
80+
rm -rf "${queryResultFile}" # Remove query result # TODO reactive cleanup
81+
# Collect graphviz files in a "graphviz" sub directory
82+
mkdir -p "${detail_report_directory}/graphviz"
83+
mv -f "${resultFileName}.gv" "${detail_report_directory}/graphviz"
84+
85+
# Create visualization reference Markdown file to be embeddable in main Markdown report
86+
if [ "${index}" == "1" ]; then
87+
{
88+
echo ""
89+
echo "##### ${language} ${nodeLabel} - ${report_name} Graph Visualizations"
90+
echo ""
91+
} > "${detail_report_directory}/VisualizationsReference.md"
92+
fi
93+
echo "![${report_name} ${index}](./${detail_report_directory_name}/GraphVisualizations/${report_name}${index}.svg)" >> "${detail_report_directory}/VisualizationsReference.md"
94+
done
95+
}
96+
97+
# Run queries, outputs their results in GraphViz format and create Graph visualizations.
98+
#
99+
# Required Parameters:
100+
# - projection_language=...
101+
# Name of the associated programming language. Examples: "Java", "Typescript"
102+
# - projection_node_label=...
103+
# Label of the nodes that will be used for the projection. Example: "Package"
104+
anomaly_detection_graph_visualization() {
105+
106+
create_graph_visualization "report_name=TopHub" "${@}"
107+
108+
}
109+
110+
111+
# Create report directory
112+
REPORT_NAME="anomaly-detection"
113+
FULL_REPORT_DIRECTORY="${REPORTS_DIRECTORY}/${REPORT_NAME}"
114+
mkdir -p "${FULL_REPORT_DIRECTORY}"
115+
116+
# Query Parameter key pairs for projection and algorithm side
117+
QUERY_NODE="projection_node_label"
118+
QUERY_LANGUAGE="projection_language"
119+
120+
# -- Detail Reports for each code type -------------------------------
121+
122+
#TODO output all types
123+
#anomaly_detection_graph_visualization "${QUERY_NODE}=Artifact" "${QUERY_LANGUAGE}=Java"
124+
anomaly_detection_graph_visualization "${QUERY_NODE}=Package" "${QUERY_LANGUAGE}=Java"
125+
#anomaly_detection_graph_visualization "${QUERY_NODE}=Type" "${QUERY_LANGUAGE}=Java"
126+
#anomaly_detection_graph_visualization "${QUERY_NODE}=Module" "${QUERY_LANGUAGE}=Typescript"
127+
128+
# ---------------------------------------------------------------
129+
130+
echo "anomalyDetectionGraphVisualization: $(date +'%Y-%m-%dT%H:%M:%S%z') Successfully finished."

scripts/visualization/visualizeQueryResults.sh

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -17,7 +17,7 @@ echo "visualizeQueryResults: VISUALIZATION_SCRIPTS_DIR=${VISUALIZATION_SCRIPTS_D
1717
# Read the first unnamed input argument containing the version of the project
1818
inputCsvFileName=""
1919
case "${1}" in
20-
"--"*) ;; # Skipping named command line options to forward them later to the "analyze" command
20+
"--"*) ;; # Skipping named command line options to forward them later to the "convertQueryResultCsvToGraphVizDotFile" command
2121
*)
2222
inputCsvFileName="${1}"
2323
shift || true

0 commit comments

Comments
 (0)