Get list of Github repositories in build scans

process-scans.sh

#!/bin/bash

# init parameters
readonly geUrl=$1
readonly bearerToken=$2
readonly daysAgo=$3

# init global vars
readonly maxBuildsPerBatch=100
readonly recoveryFile="$0.out.build-scan-id"
readonly outputFile="$0.out.txt"
readonly workFile="$0.out.tmp.txt"
buildScans=""
buildCount=1

#############
# functions #
#############
function getBuildScansInTimeWindowWithGitRepositoryAsCustomValue() {
    local filter=$1
    local encodedQuery=$(echo "value:\"Git repository=*\" and buildStartTime > -${daysAgo}d" | jq -sRr @uri)
    buildScans=$(curl --silent "${geUrl}/api/builds?${filter}&maxBuilds=${maxBuildsPerBatch}&reverse=false&query=${encodedQuery}" --header "authorization: Bearer ${bearerToken}" | jq -r ".[] | [.id,.availableAt,.buildToolType] | @csv")
}

function getBuildScans() {
    getBuildScansInTimeWindowWithGitRepositoryAsCustomValue "fromInstant=0"
}

function getBuildScansFromBuildId() {
    local buildId=$1
    getBuildScansInTimeWindowWithGitRepositoryAsCustomValue "fromBuild=${buildId}"
}

########
# main #
########

# check if recovery data are present
recoveryBuildScanId=$(cat ${recoveryFile} 2>/dev/null)

if [[ -z "${recoveryBuildScanId}" ]]
then
    # Clean output files
    rm -f ${outputFile}
    rm -f ${workFile}

    # Get first batch of builds
    getBuildScans
else
    echo "Recovering data from build scan id ${recoveryBuildScanId} (you can trigger a fresh run by deleting ${recoveryFile})"

    # Get first batch of builds
    getBuildScansFromBuildId ${recoveryBuildScanId}
fi

# Iterate over build scan list
while [ ! -z "${buildScans}" ]
do 
    buildScanId=""

    for buildScan in ${buildScans}; do
        # Remove "
        buildScan=$(echo "${buildScan}" | sed -e 's/"//g')

        # Parse fields
        buildScanId=$(echo "${buildScan}" | cut -d, -f 1)
        buildScanDate=$(echo "${buildScan}" | cut -d, -f 2)
        buildTool=$(echo "${buildScan}" | cut -d, -f 3)
        echo "Build ${buildCount}: Processing ${buildScanId} (${buildTool})"

        # save build scan id for recovery mode
        echo "${buildScanId}" > ${recoveryFile}

        # Get Git repository
        buildData=$(curl --silent ${geUrl}/api/builds/${buildScanId}/${buildTool}-attributes --header "authorization: Bearer ${bearerToken}" | jq -r '.values[] | select(.name | contains("Git repository")) | .value')

        # remove .git extension
        buildData=$(echo ${buildData} | sed -e 's/\.git$//')

        # append to file if not already present
        grep -q "${buildData}" $0.out.tmp.txt 2>/dev/null || echo ${buildData} >> $0.out.tmp.txt
        buildCount=$((buildCount+1))
    done

    getBuildScansFromBuildId ${buildScanId}
done 

# Remove recovery data
rm -f ${recoveryFile}

# Sort results
sort ${workFile} > ${outputFile}

# Display results
echo "List of repositories found:"
echo ""
cat ${outputFile}

Prerequisites:

• Develocity 2023.3 or above
• Access token with API Read permissions

Call me with:
./process-scans.sh <GE_URL> <GE_ACCESS_TOKEN> <NB_DAYS_FROM_NOW>

This will collect the Git repositories added as custom values (key Git repository) to build scans belonging to the date range [NOW, NOW - <NB_DAYS_FROM_NOW> days]