fixup! Add git history commit and author statistics

Author: JohT

cypher/GitLog/List_git_file_directories_with_commit_statistics.cypher

@@ -16,12 +16,9 @@
       ,fileName
       ,filePath                                         AS fileRelativePath
       ,split(git_commit.author, ' <')[0]                AS author
-      ,split(split(git_commit.author, ' <')[1], '>')[0] AS authorEmail
-      ,git_commit.author                                AS authorFull
       ,max(git_commit.sha)                              AS maxCommitSha
-      ,COUNT(DISTINCT git_commit.sha)                   AS commitCount
+      ,collect(DISTINCT git_commit.sha)                 AS commitHashes
       ,date(max(git_commit.date))                       AS lastCommitDate
-ORDER BY filePath ASCENDING, commitCount DESCENDING
 UNWIND pathElements AS pathElement
   WITH *
       ,coalesce(nullif(split(fileRelativePath, '/' + pathElement)[0], fileRelativePath), '')  AS parent
@@ -34,37 +31,29 @@ UNWIND pathElements AS pathElement
       ,split(parent, '/')[-1]                     AS directoryParentName
       ,size(split(directory, '/'))                AS directoryPathLength
       ,author
-      ,authorEmail
-      ,authorFull
-      ,count(DISTINCT fileRelativePath)           AS fileCount
+      ,collect(DISTINCT fileRelativePath)         AS files
       ,max(date(fileCreatedAtTimestamp) )         AS lastCreationDate
       ,max(date(fileLastModificationAtTimestamp)) AS lastModificationDate
-      ,sum(commitCount)                           AS commitCount
+      ,apoc.coll.toSet(apoc.coll.flatten(collect(commitHashes))) AS commitHashes
       ,max(maxCommitSha)                          AS maxCommitSha
       ,max(lastCommitDate)                        AS lastCommitDate
       ,max(fileRelativePath)                      AS maxFileRelativePath
       ,duration.inDays(max(lastCommitDate), date()).days                      AS daysSinceLastCommit
       ,duration.inDays(max(fileCreatedAtTimestamp), datetime()).days          AS daysSinceLastCreation
       ,duration.inDays(max(fileLastModificationAtTimestamp), datetime()).days AS daysSinceLastModification
 // Assure that the authors are ordered by their commit count descending per directory
-ORDER BY directoryPath ASCENDING, commitCount DESCENDING
+ORDER BY directoryPath ASCENDING, size(commitHashes) DESCENDING
   WITH directoryPath
       ,directoryName
       ,directoryParentPath
       ,directoryParentName
       ,directoryPathLength
       ,collect(author)[0]                         AS mainAuthor
-      ,collect(authorEmail)[0]                    AS mainAuthorEmail
-      ,collect(authorFull)[0]                     AS mainAuthorFull
       ,collect(author)[1]                         AS secondAuthor
-      ,collect(authorEmail)[1]                    AS secondAuthorEmail
-      ,collect(authorFull)[1]                     AS secondAuthorFull
       ,collect(author)[2]                         AS thirdAuthor
-      ,collect(authorEmail)[2]                    AS thirdAuthorEmail
-      ,collect(authorFull)[2]                     AS thirdAuthorFull
-      ,count(DISTINCT authorFull)                 AS authorCount
-      ,sum(fileCount)                             AS fileCount
-      ,sum(commitCount)                           AS commitCount
+      ,count(DISTINCT author)                     AS authorCount
+      ,size(apoc.coll.toSet(apoc.coll.flatten(collect(files))))        AS fileCount
+      ,size(apoc.coll.toSet(apoc.coll.flatten(collect(commitHashes)))) AS commitCount
       ,max(lastCreationDate)                      AS lastCreationDate
       ,max(lastModificationDate)                  AS lastModificationDate
       ,max(maxCommitSha)                          AS maxCommitSha
@@ -79,14 +68,8 @@ RETURN collect(directoryPath)[-1]                  AS directoryPath
       ,collect(directoryParentPath)[0]             AS directoryParentPath
       ,collect(directoryParentName)[0]             AS directoryParentName
       ,mainAuthor
-      ,mainAuthorEmail
-      ,mainAuthorFull
       ,secondAuthor
-      ,secondAuthorEmail
-      ,secondAuthorFull
       ,thirdAuthor
-      ,thirdAuthorEmail
-      ,thirdAuthorFull
       ,authorCount
       ,fileCount
       ,commitCount

cypher/GitLog/List_git_file_directories_with_commit_statistics_by_author.cypher

@@ -42,7 +42,6 @@ echo "GitHistoryCsv: $(date +'%Y-%m-%dT%H:%M:%S%z') Processing git history..."
 # Detailed git file statistics
 execute_cypher "${GIT_LOG_CYPHER_DIR}/List_git_files_with_commit_statistics_by_author.cypher" > "${FULL_REPORT_DIRECTORY}/List_git_files_with_commit_statistics_by_author.csv"
 execute_cypher "${GIT_LOG_CYPHER_DIR}/List_git_files_that_were_changed_together_with_another_file.cypher" > "${FULL_REPORT_DIRECTORY}/List_git_files_that_were_changed_together_with_another_file.csv"
-execute_cypher "${GIT_LOG_CYPHER_DIR}/List_git_file_directories_with_commit_statistics_by_author.cypher" > "${FULL_REPORT_DIRECTORY}/List_git_file_directories_with_commit_statistics_by_author.csv"
 execute_cypher "${GIT_LOG_CYPHER_DIR}/List_git_file_directories_with_commit_statistics.cypher" > "${FULL_REPORT_DIRECTORY}/List_git_file_directories_with_commit_statistics.csv"
 
 # Overall distribution of how many files were changed with one git commit, how many were changed with two, etc.