Analyse SVN Logfiles

# Sys.setenv(SVN_IP = NULL, # use saved TortoiseSVN user/password
#            SVN_USER = NULL, # use saved TortoiseSVN user/password 
#            SVN_PW = "svn_password")

logdir <-  file.path(kwb.utils::desktop(), "svn/log")
  
fs::dir_create(logdir)
  ### 1)get infos from "ente"
  ### 1a) size info (warning takes ~ 36 hours!!!!)
  #   kwb.svn::getRepoInfo(tDir = logdir, startRevision = 2, currentRevision = 3,
  #              logs = FALSE)
  ### 1b) log info
  # kwb.svn::getRepoInfo(tDir = logdir, 
  #            logs = FALSE)
  
  svn_repo <- kwb.svn::default_repo()
  kwb.svn::getRepoInfo(tDir = logdir, 
                       startRevision = 2, 
                       currentRevision = 6749, #6749,
                       repo = svn_repo) 
  
  logdata <- kwb.svn::readLogFilesInDirectory(logdir)
  logdata <- logdata[!duplicated(logdata$revision),]
  logdata$author <- as.factor(logdata$author)
  
  ### size info 
  
  sizedir <- file.path(kwb.utils::desktop(), "svn_size")
  
  if(!exists(sizedir)) dir.create(sizedir)
  
  kwb.svn::getRepoInfo(tDir = sizedir, 
                       logs = FALSE,
                       startRevision = 1024, 
                       currentRevision = 1200, #6294,
                       repo = svn_repo) 
  
  enteSize <- kwb.svn::readSizeFiles(fDir = logdir, dbg = TRUE)
  enteSize$size <- enteSize$size[enteSize$size$sumMB < 9000 & enteSize$size$sumMB > 1700,]
  enteSize$size <- enteSize$size[order(enteSize$size$revision),]
  
  enteSize$size$cumsumMB <- cumsum(enteSize$size$sumMB)
  enteSize$size$diffMB <- c(0, diff(enteSize$size$sumMB))
  
  logdat <- merge(logdata, enteSize$size, by.x = "revision", by.y="revision", all.x=TRUE)
  save(logdat,file = "svn.RData")
  plot(revision ~ dateTimeUTC, data=logdat, pch=NA)
  
  legendtext <- levels(logdat$author)
  
  plotcolors <- rainbow(length(legendtext))
  
  legend(x = "topleft", legend = legendtext, fill = plotcolors)
  
  by(
    data = logdat, 
    INDICES = logdat$author, 
    FUN = function(x) {
      points(x$dateTimeUTC, x$revision, col = plotcolors[x$author])
    }
  )  
  
  inGB <- 1024
  inTB <- 1024*1024
  
  plot(sumMB/inGB ~ dateTimeUTC, type = "p", col="blue", data=logdat, 
       ylab="Speicherbedarf Ente (GB)", las=1)
  
  
  max(logdat$size$cumsumMB)/ max(logdat$size$sumMB)
  
  lastcommits <- (nrow(enteSize$size)-10):nrow(enteSize$size)
  plot(sumMB/inGB ~ revision, type = "p", col="blue", data=logdat[lastcommits,], 
       ylab="Speicherbedarf (GB)", las=1)
  
  plot(cumsumMB/inGB ~ revision, type = "l", col="red", 
       data=logdat, las=1,
       xlab="Anzahl Kopien", 
       ylab = "Speicherbedarf (in Gigabyte)", 
       main="Happy Birthsday: 2 Jahre \"Ente\"")
  lines(sumMB/inGB ~ revision, data=logdat, col="blue")
  legend("topleft",
         title = "Strategie:", 
         legend = c("Copy & Paste", "Versionsverwaltung"), 
         col = c("red", "blue"), 
         lty = 1)
  
  res1 <- aggregate(revision ~ author, data = logdat,FUN = length)
  res2 <- aggregate(diffMB ~ author, data = logdat,FUN = sum)
  res <- merge(res1, res2)
  res$MBperCommit <- res$diffMB/res$revision
  myOrder <- order(res$revision)
  par(oma = c(1,4,0,0))
  barplot(100*res$revision[myOrder]/1920, 
          names.arg = res$author[myOrder], 
          horiz = TRUE, 
          las=1,
          xlab= "Prozentualer Anteil an Commits")
  
  teamOW2 <- c("hsonne", "asainz", "meslam", "vphili", "mrustl", "lburho")
  res$team <- "others"
  res$team[res$author %in% teamOW2] <- "OPTIWELLS-2"
  
  myOrder <- order(res$MBperCommit)
  par(oma = c(1,4,0,0))
  barplot(res$MBperCommit[myOrder], 
          names.arg = res$author[myOrder], 
          horiz = TRUE, 
          las=1, 
          xlab= "Speicherplatz (in MB) je Commit")
  
  
  aggregate(revision ~ team, data = res,FUN = sum)

history_dir <-  file.path(kwb.utils::desktop(), "svn/history")
fs::dir_create(history_dir)

## use saved SVN password from Tortoise SVN
repo_rscripts <- kwb.svn::default_repo(username = NULL,
                                       password = NULL, 
                                       root_dir = "svn/kwb/R_Development/trunk/RScripts")

kwb.svn::read_files_history(tdir = history_dir,
                            repo = repo_rscripts)

histories <- kwb.svn::read_histories(history_dir)


histories_file_user <- histories %>% 
  dplyr::group_by(file, username) %>% 
  dplyr::summarise(n_changes = dplyr::n(), 
                   datetime_min = min(datetime), 
                   datetime_max = max(datetime),
                   lines_comments = sum(lines_deleted), 
                   lines_deleted = sum(lines_deleted), 
                   lines_added = sum(lines_added))

histories_per_user <- histories %>% 
  dplyr::group_by(username) %>% 
  dplyr::summarise(n_changes = dplyr::n(),
                   datetime_min = min(datetime), 
                   datetime_max = max(datetime),
                   lines_comments = sum(lines_deleted), 
                   lines_deleted = sum(lines_deleted), 
                   lines_added = sum(lines_added))