blob: 33664b0eb5524cbc08334a5d4857b496aa93989d [file] [log] [blame]
# Copyright 2017 Open Networking Foundation (ONF)
#
# Please refer questions to either the onos test mailing list at <onos-test@onosproject.org>,
# the System Testing Plans and Results wiki page at <https://wiki.onosproject.org/x/voMg>,
# or the System Testing Guide page at <https://wiki.onosproject.org/x/WYQg>
#
# TestON is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# TestON is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with TestON. If not, see <http://www.gnu.org/licenses/>.
#
# If you have any questions, or if you don't understand R,
# please contact Jeremy Ronquillo: j_ronquillo@u.pacific.edu
pipelineMinValue = 1000
# **********************************************************
# STEP 1: Data management.
# **********************************************************
print( "**********************************************************" )
print( "STEP 1: Data management." )
print( "**********************************************************" )
# Command line arguments are read. Args include the database credentials, test name, branch name, and the directory to output files.
print( "Reading commmand-line args." )
args <- commandArgs( trailingOnly=TRUE )
databaseHost <- 1
databasePort <- 2
databaseUserID <- 3
databasePassword <- 4
testSuiteName <- 5
branchName <- 6
testsToInclude <- 7
buildsToShow <- 8
saveDirectory <- 9
# ----------------
# Import Libraries
# ----------------
print( "Importing libraries." )
library( ggplot2 )
library( reshape2 )
library( RPostgreSQL )
# -------------------
# Check CLI Arguments
# -------------------
print( "Verifying CLI args." )
if ( is.na( args[ saveDirectory ] ) ){
print( paste( "Usage: Rscript testCategoryTrend.R",
"<database-host>",
"<database-port>",
"<database-user-id>",
"<database-password>",
"<test-suite-name>",
"<branch-name>",
"<tests-to-include-(as-one-string)>",
"<builds-to-show>",
"<directory-to-save-graphs>",
sep=" " ) )
quit( status = 1 ) # basically exit(), but in R
}
# -------------------------------
# Create Title and Graph Filename
# -------------------------------
print( "Creating title of graph." )
title <- paste( args[ testSuiteName ],
" Test Results Trend - ",
args[ branchName ],
" \n Results of Last ",
args[ buildsToShow ],
" Nightly Builds",
sep="" )
print( "Creating graph filename." )
outputFile <- paste( args[ saveDirectory ],
args[ testSuiteName ],
"_",
args[ branchName ],
"_overview.jpg",
sep="" )
# ------------------
# SQL Initialization
# ------------------
print( "Initializing SQL" )
con <- dbConnect( dbDriver( "PostgreSQL" ),
dbname = "onostest",
host = args[ databaseHost ],
port = strtoi( args[ databasePort ] ),
user = args[ databaseUserID ],
password = args[ databasePassword ] )
# ---------------------
# Test Case SQL Command
# ---------------------
print( "Generating Test Case SQL command." )
tests <- "'"
for ( test in as.list( strsplit( args[ testsToInclude ], "," )[[1]] ) ){
tests <- paste( tests, test, "','", sep="" )
}
tests <- substr( tests, 0, nchar( tests ) - 2 )
command <- paste( "SELECT * ",
"FROM executed_test_tests a ",
"WHERE ( SELECT COUNT( * ) FROM executed_test_tests b ",
"WHERE b.branch='",
args[ branchName ],
"' AND b.actual_test_name IN (",
tests,
") AND a.actual_test_name = b.actual_test_name AND a.date <= b.date AND b.build >= ",
pipelineMinValue,
" ) <= ",
args[ buildsToShow ],
" AND a.branch='",
args[ branchName ],
"' AND a.actual_test_name IN (",
tests,
") AND a.build >= ",
pipelineMinValue,
" ORDER BY a.actual_test_name DESC, a.date DESC",
sep="")
print( "Sending SQL command:" )
print( command )
dbResult <- dbGetQuery( con, command )
maxBuild <- max( dbResult[ 'build' ] ) - strtoi( args[ buildsToShow ] )
dbResult <- dbResult[ which( dbResult[,4]>maxBuild ), ]
print( dbResult )
# **********************************************************
# STEP 2: Organize data.
# **********************************************************
print( "**********************************************************" )
print( "STEP 2: Organize Data." )
print( "**********************************************************" )
t <- subset( dbResult, select=c( "actual_test_name", "build", "num_failed" ) )
t$num_failed <- ceiling( t$num_failed / ( t$num_failed + 1 ) )
t$num_planned <- 1
fileData <- aggregate( t$num_failed, by=list( Category=t$build ), FUN=sum )
colnames( fileData ) <- c( "build", "num_failed" )
fileData$num_planned <- ( aggregate( t$num_planned, by=list( Category=t$build ), FUN=sum ) )$x
fileData$num_passed <- fileData$num_planned - fileData$num_failed
print(fileData)
# --------------------
# Construct Data Frame
# --------------------
#
dataFrame <- melt( subset( fileData, select=c( "num_failed", "num_passed", "num_planned" ) ) )
dataFrame$build <- fileData$build
colnames( dataFrame ) <- c( "status", "results", "build" )
dataFrame$num_failed <- fileData$num_failed
dataFrame$num_passed <- fileData$num_passed
dataFrame$num_planned <- fileData$num_planned
dataFrame$iterative <- seq( 1, nrow( fileData ), by = 1 )
print( "Data Frame Results:" )
print( dataFrame )
# **********************************************************
# STEP 3: Generate graphs.
# **********************************************************
print( "**********************************************************" )
print( "STEP 3: Generate Graph." )
print( "**********************************************************" )
# -------------------
# Main Plot Generated
# -------------------
print( "Creating main plot." )
# Create the primary plot here.
# ggplot contains the following arguments:
# - data: the data frame that the graph will be based off of
# - aes: the asthetics of the graph which require:
# - x: x-axis values (usually iterative, but it will become build # later)
# - y: y-axis values (usually tests)
# - color: the category of the colored lines (usually status of test)
mainPlot <- ggplot( data = dataFrame, aes( x = iterative,
y = results,
color = status ) )
# -------------------
# Main Plot Formatted
# -------------------
print( "Formatting main plot." )
# geom_ribbon is used so that there is a colored fill below the lines. These values shouldn't be changed.
failedColor <- geom_ribbon( aes( ymin = 0,
ymax = dataFrame$num_failed ),
fill = "#ff0000",
linetype = 0,
alpha = 0.07 )
passedColor <- geom_ribbon( aes( ymin = 0,
ymax = dataFrame$num_passed ),
fill = "#0083ff",
linetype = 0,
alpha = 0.05 )
plannedColor <- geom_ribbon( aes( ymin = 0,
ymax = dataFrame$num_planned ),
fill = "#000000",
linetype = 0,
alpha = 0.01 )
# Colors for the lines
lineColors <- scale_color_manual( values=c( "#ff0000", # fail
"#0083ff", # pass
"#000000"),
labels = c( "Containing Failures",
"No Failures",
"Total Built" ) ) # planned
# ------------------------------
# Fundamental Variables Assigned
# ------------------------------
print( "Generating fundamental graph data." )
theme_set( theme_grey( base_size = 26 ) ) # set the default text size of the graph.
xScaleConfig <- scale_x_continuous( breaks = dataFrame$iterative,
label = dataFrame$build )
yScaleConfig <- scale_y_continuous( breaks = seq( 0, max( dataFrame$results ),
by = ceiling( max( dataFrame$results ) / 10 ) ) )
xLabel <- xlab( "Build Number" )
yLabel <- ylab( "Tests" )
imageWidth <- 15
imageHeight <- 10
imageDPI <- 200
# Set other graph configurations here.
theme <- theme( plot.title = element_text( hjust = 0.5, size = 32, face ='bold' ),
axis.text.x = element_text( angle = 0, size = 14 ),
legend.position = "bottom",
legend.text = element_text( size = 22 ),
legend.title = element_blank(),
legend.key.size = unit( 1.5, 'lines' ),
plot.subtitle = element_text( size=16, hjust=1.0 ) )
subtitle <- paste( "Last Updated: ", format( Sys.time(), format = "%b %d, %Y at %I:%M %p %Z" ), sep="" )
title <- labs( title = title, subtitle = subtitle )
# Store plot configurations as 1 variable
fundamentalGraphData <- mainPlot +
plannedColor +
passedColor +
failedColor +
xScaleConfig +
yScaleConfig +
xLabel +
yLabel +
theme +
title +
lineColors
# ----------------------------
# Generating Line Graph Format
# ----------------------------
print( "Generating line graph." )
lineGraphFormat <- geom_line( size = 1.1 )
pointFormat <- geom_point( size = 3 )
result <- fundamentalGraphData +
lineGraphFormat +
pointFormat
# -----------------------
# Exporting Graph to File
# -----------------------
print( paste( "Saving result graph to", outputFile ) )
tryCatch( ggsave( outputFile,
width = imageWidth,
height = imageHeight,
dpi = imageDPI ),
error = function( e ){
print( "[ERROR] There was a problem saving the graph due to a graph formatting exception. Error dump:" )
print( e )
quit( status = 1 )
}
)
print( paste( "[SUCCESS] Successfully wrote result graph out to", outputFile ) )
quit( status = 0 )