Update app.R

app.R

@@ -210,6 +210,822 @@ merge_with_bat_tracking <- function(csv_data, bat_tracking_data) {
   ))
 }
 
+
+##SCRAPER FUNCTIONS
+
+ftp_server <- "ftp.trackmanbaseball.com"
+username <- "FTP_USER"
+password <- "FTP_PASSWORD"
+ftp_base_dir <- "/v3"
+                                            
+
+#Trackman pitch by pitch                                          
+scrape_trackman_pbp <- function(start_date, end_date) {
+  
+  dates <- as.character(seq.Date(as.Date(start_date), as.Date(end_date), by = "day"))
+  all_data <- data.frame()
+  
+  for (date in dates) {
+    year  <- format(as.Date(date), "%Y")
+    month <- format(as.Date(date), "%m")
+    day   <- format(as.Date(date), "%d")
+    
+    ftp_dir <- paste0(ftp_base_dir, "/", year, "/", month, "/", day, "/CSV")
+    ftp_url <- paste0("ftp://", username, ":", password, "@", ftp_server, ftp_dir, "/")
+    
+    # Get list of files for this day
+    file_list <- tryCatch({
+      getURL(ftp_url, ftp.use.epsv = FALSE, dirlistonly = TRUE)
+    }, error = function(e) { NULL })
+    
+    if (is.null(file_list)) next
+    
+    file_list <- unlist(strsplit(file_list, "\n"))
+    
+    # Filter to verified/unverified CSVs (prefer verified)
+    pattern_verified   <- "(\\d{8})-(.+)-\\d+\\.csv$"
+    pattern_unverified <- "(\\d{8})-(.+)-\\d+_unverified\\.csv$"
+    
+    verified   <- grep(pattern_verified, file_list, value = TRUE)
+    unverified <- grep(pattern_unverified, file_list, value = TRUE)
+    
+    verified_stadiums   <- unique(str_match(verified, pattern_verified)[, 2])
+    unverified_stadiums <- str_match(unverified, pattern_unverified)[, 2]
+    
+    files_to_read <- c(
+      verified,
+      unverified[!unverified_stadiums %in% verified_stadiums]
+    )
+    
+    # Read each file directly into memory
+    for (file in files_to_read) {
+      file_url <- paste0(ftp_url, URLencode(trimws(file)))
+      
+      tryCatch({
+        csv_text <- getURL(file_url)
+        rows <- read_csv(I(csv_text), col_types = cols(.default = "c"), show_col_types = FALSE)
+        
+        if ("PlayResult" %in% names(rows)) {
+          all_data <- bind_rows(all_data, rows)
+        }
+      }, error = function(e) {
+        message("Failed: ", file, " - ", e$message)
+      })
+    }
+    
+    message("Done with ", date)
+  }
+  
+  # Deduplicate
+  all_data %>%
+    distinct(PitchUID, .keep_all = TRUE)
+}                                          
+
+
+
+#trackman positional                                           
+scrape_trackman_positional <- function(start_date, end_date) {
+  
+  dates <- as.character(seq.Date(as.Date(start_date), as.Date(end_date), by = "day"))
+  all_data <- data.frame()
+  
+  for (date in dates) {
+    year  <- format(as.Date(date), "%Y")
+    month <- format(as.Date(date), "%m")
+    day   <- format(as.Date(date), "%d")
+    
+    ftp_dir <- paste0(ftp_base_dir, "/", year, "/", month, "/", day, "/CSV")
+    ftp_url <- paste0("ftp://", username, ":", password, "@", ftp_server, ftp_dir, "/")
+    
+    # Get list of files for this day
+    file_list <- tryCatch({
+      getURL(ftp_url, ftp.use.epsv = FALSE, dirlistonly = TRUE)
+    }, error = function(e) { NULL })
+    
+    if (is.null(file_list)) next
+    
+    file_list <- unlist(strsplit(file_list, "\n"))
+    
+    # Filter to verified/unverified CSVs (prefer verified)
+    pattern_verified <- "(\\d{8})-(.+)-\\d+_playerpositioning_FHC\\.csv$"     
+    pattern_unverified <- "(\\d{8})-(.+)-\\d+_unverified_playerpositioning_FHC\\.csv$"  
+    
+    verified   <- grep(pattern_verified, file_list, value = TRUE)
+    unverified <- grep(pattern_unverified, file_list, value = TRUE)
+    
+    verified_stadiums   <- unique(str_match(verified, pattern_verified)[, 2])
+    unverified_stadiums <- str_match(unverified, pattern_unverified)[, 2]
+    
+    files_to_read <- c(
+      verified,
+      unverified[!unverified_stadiums %in% verified_stadiums]
+    )
+    
+    # Read each file directly into memory
+    for (file in files_to_read) {
+      file_url <- paste0(ftp_url, URLencode(trimws(file)))
+      
+      tryCatch({
+        csv_text <- getURL(file_url)
+        rows <- read_csv(I(csv_text), col_types = cols(.default = "c"), show_col_types = FALSE)
+        
+        if ("PlayResult" %in% names(rows)) {
+          all_data <- bind_rows(all_data, rows)
+        }
+      }, error = function(e) {
+        message("Failed: ", file, " - ", e$message)
+      })
+    }
+    
+    message("Done with ", date)
+  }
+  
+  # Deduplicate
+  all_data %>%
+    distinct(PitchUID, .keep_all = TRUE)
+}
+
+                                            
+
+
+#Next section is a large section of functions from the pbp parser github to parse ncaa pbp data to get base states                                            
+stripwhite <- function(x) gsub("\\s*$", "", gsub("^\\s*", "", x))
+
+strip_punc <- function(x){ 
+  x=stripwhite(x)
+  x=ifelse(str_sub(x,-1)=='.',gsub("\\.", "", x),x)
+  return(x)}
+
+
+
+##########################################################
+# Functions for parsing 
+
+inn_end = function(top_inn){
+  m=length(top_inn)
+  inn_end=integer(m)
+  for (i in 1:(m-1)){
+    inn_end[i]=ifelse(top_inn[i]!=top_inn[i+1], 1,0)
+  }
+  inn_end[m]=1
+  return(inn_end)
+}
+
+
+game_end = function(game_id){
+  m=length(game_id)
+  game_end=integer(m)
+  for (i in 2:m){
+    if (game_id[i]!=game_id[i-1]){
+      game_end[i-1]=1
+    }
+    game_end[m]=1
+  }
+  return(game_end)
+}
+
+
+
+
+runs_on_play= function(a_txt, h_txt, a_score,h_score){
+  m=length(a_txt)
+  runs_on_play=integer(m)
+  runs_on_play[1]=a_score[1]
+  for (i in 2:m){
+    runs_on_play[i]=case_when(
+      a_txt[i]=='' ~ as.integer(h_score[i]-h_score[i-1]),
+      a_txt[i]!='' ~ as.integer(a_score[i]-a_score[i-1])
+    )
+  }
+  return(runs_on_play)
+}
+
+
+r1_name = function(bat_text, bat_name, r1_text, r1_name, inn_end, game_end, sub_in, sub_out){
+  m=length(bat_text)
+  r1_name= character(m)
+  for (i in 2:m){
+    if (isTRUE(inn_end[i-1]==0 & game_end[i-1]==0)) {
+      r1_name[i]=case_when(
+        sub_out[i-1]!=''&sub_out[i-1]==stripwhite(r1_name[i-1])~sub_in[i-1],
+        (str_detect(bat_text[i-1], '(singled|walked|hit by pitch|reached)') == TRUE) & (str_detect(bat_text[i-1], '(doubled|tripled|homered|advanced|scored|out|stole)') == FALSE) ~ bat_name[i-1],
+        (str_detect(bat_text[i-1], '(reached first)') == TRUE) & (str_detect(bat_text[i-1], '(struck out)') == TRUE) ~ bat_name[i-1],
+        (r1_text[i-1]==''|(str_detect(r1_text[i-1], '(advanced to second|stole second|advanced to third|stole third|scored|out)') == FALSE)) & (str_detect(bat_text[i-1], '(double play|advanced to second|stole second|advanced to third|stole third|scored|caught stealing|picked off|homered)') == FALSE)  ~ r1_name[i-1],
+        (str_detect(bat_text[i-1], '(singled|doubled|tripled|advanced to second|stole second|advanced to third|stole third|scored|homered|out at second c to)') == FALSE) & (str_detect(r1_text[i-1], '(advanced to third|stole third|scored|out at third)') == TRUE) & stripwhite(gsub('((advanced to second|stole second|stole third|advanced to third|scored|out).*$)', '', r1_text[i-1]))!=stripwhite(gsub('((singled|reached).*$)', '', r1_name[i-1])) ~ r1_name[i-1],
+        r1_text[i-1]=='' & stripwhite(gsub('((advanced to second|stole second|stole third|advanced to third|scored|out|failed|Failed|picked off).*$)', '', bat_text[i-1]))!=stripwhite(r1_name[i-1]) ~ r1_name[i-1]
+      )}}
+  return(stripwhite(r1_name))
+}
+
+
+r2_name = function(bat_text, bat_name, r1_text, r1_name, r2_text, r2_name,  inn_end, game_end, sub_in, sub_out){
+  m=length(bat_text)
+  r2_name= character(m)
+  for (i in 2:m){
+    if (isTRUE(inn_end[i-1]==0 & game_end[i-1]==0)) {
+      r2_name[i]=case_when(
+        sub_out[i-1]!=''&sub_out[i-1]==stripwhite(r2_name[i-1])~sub_in[i-1],
+        ((str_detect(bat_text[i-1], '(doubled|advanced to second|stole second)') == TRUE) & (str_detect(bat_text[i-1], '(advanced to third|scored|out|stole third)') == FALSE)) ~ stripwhite(gsub('((doubled|advanced to second|stole second).*$)', '', bat_text[i-1])),
+        ((str_detect(r1_text[i-1], '(advanced to second|stole second)') == TRUE) & (str_detect(r1_text[i-1], '(advanced to third|scored|out|stole third)') == FALSE)) ~ stripwhite(gsub('((advanced to second|stole second).*$)', '', r1_text[i-1])),
+        r2_text[i-1]=='' & stripwhite(gsub('((stole third|advanced to third|scored|out).*$)', '', r1_text[i-1]))!=stripwhite(r2_name[i-1]) & (str_detect(bat_text[i-1], '(advanced to third|stole third|scored|picked off|caught stealing)') == FALSE) ~ r2_name[i-1],
+        r2_text[i-1]=='' & stripwhite(gsub('((out on the play).*$)', '', r1_text[i-1]))!=stripwhite(r2_name[i-1]) & (str_detect(bat_text[i-1], '(double play)') == TRUE) ~ r2_name[i-1],
+        r1_text[i-1]=='' & (str_detect(bat_text[i-1], '(stole third|advanced to third|scored|picked off|homered|caught stealing)') == FALSE) ~ r2_name[i-1],
+        sub_out[i-1]!=''&sub_out[i-1]==stripwhite(r2_name[i-1])~sub_in[i-1]
+      )
+      r2_name[i]=stripwhite(gsub('((singled|reached).*$)', '', r2_name[i]))
+    }
+  }
+  return(stripwhite(r2_name))
+}
+
+
+r3_name = function(bat_text, bat_name, r1_text, r1_name, r2_text, r2_name, r3_text, r3_name, inn_end, game_end, sub_in, sub_out){
+  m=length(bat_text)
+  r3_name= character(m)
+  for (i in 2:m){
+    if (isTRUE(inn_end[i-1]==0 & game_end[i-1]==0)) {
+      r3_name[i]=case_when( 
+        sub_out[i-1]!=''&sub_out[i-1]==stripwhite(r3_name[i-1])~sub_in[i-1],
+        ((str_detect(bat_text[i-1], '(tripled|advanced to third|stole third)') == TRUE) & (str_detect(bat_text[i-1], '(scored|out)') == FALSE)) ~ stripwhite(gsub('((tripled|advanced to third|stole third).*$)', '', bat_text[i-1])),
+        ((str_detect(r1_text[i-1], '(advanced to third|stole third)') == TRUE) & (str_detect(r1_text[i-1], '(scored|out)') == FALSE)) ~ stripwhite(gsub('((advanced to third|stole third).*$)', '', r1_text[i-1])),
+        ((str_detect(r2_text[i-1], '(advanced to third|stole third)') == TRUE) & (str_detect(r2_text[i-1], '(scored|out)') == FALSE)) ~ stripwhite(gsub('((advanced to third|stole third).*$)', '', r2_text[i-1])),
+        r1_text[i-1]=='' & (str_detect(bat_text[i-1], '(scored|stole home|homered)') == FALSE) ~ r3_name[i-1],
+        r2_text[i-1]=='' & stripwhite(gsub('((scored|stole home|out).*$)', '', r1_text[i-1]))!=stripwhite(r3_name[i-1]) & (str_detect(bat_text[i-1], '(scored|stole home)') == FALSE) ~ r3_name[i-1],
+        r3_text[i-1]=='' & (str_detect(r2_text[i-1], '(scored|stole home|out)') == FALSE) & (str_detect(r1_text[i-1], '(scored|stole home|out)') == FALSE) & (str_detect(bat_text[i-1], '(scored|stole home)') == FALSE) ~ r3_name[i-1])
+      r3_name[i]=stripwhite(gsub('((singled|doubled|reached|advanced|stole|failed|Failed|picked off).*$)', '', r3_name[i]))
+    }
+  }
+  return(stripwhite(r3_name))
+}
+
+
+
+
+
+new_game=function(game_end){
+  m = length(game_end)
+  new_game=integer(m)
+  new_game[1]=1
+  for (i in 2:m){
+    new_game[i]=game_end[i-1]
+  }
+  return(new_game)
+}
+
+new_inn=function(inn_end){
+  m = length(inn_end)
+  new_inn=integer(m)
+  new_inn[1]=1
+  for (i in 2:m){
+    new_inn[i]=inn_end[i-1]
+  }
+  return(new_inn)
+}
+
+
+outs_before= function(outs_on_play, new_game, new_inn){
+  m=length(outs_on_play)
+  inn_outs=integer(m)
+  for (i in 2:m){
+    if (isTRUE(new_game[i] == 0 & new_inn[i] == 0)) {
+      inn_outs[i]=((inn_outs[i-1]+outs_on_play[i-1]) %% 3)
+    }
+  }
+  return(inn_outs)
+}
+
+
+
+
+
+
+
+score_before=function(new_game, runs_on_play, top_inning, home_team=1){
+  m=length(new_game)
+  home_score_before=integer(m)
+  away_score_before=integer(m)
+  for (i in 2:m){
+    home_score_before[i]= case_when(
+      new_game[i]==0 & top_inning[i-1]==0 ~ as.numeric(home_score_before[i-1]+runs_on_play[i-1]),
+      new_game[i]==0 & top_inning[i-1]==1 ~ as.numeric(home_score_before[i-1]),
+      TRUE ~ 0)
+    
+    away_score_before[i]= case_when(
+      new_game[i]==0 & top_inning[i-1]==1 ~ as.numeric(away_score_before[i-1]+runs_on_play[i-1]),
+      new_game[i]==0 & top_inning[i-1]==0 ~ as.numeric(away_score_before[i-1]),
+      TRUE ~ 0)
+  }
+  if(home_team==1){
+    return(home_score_before)  
+  }
+  else{return(away_score_before)}
+  
+}
+
+runs_play=function(home_score, away_score, home_score_before, away_score_before, top_inn){
+  n=length(homescore)
+  runs_play=integer(n)
+  for (i in 2:n){
+    case_when(top_inn[i]==0 ~ homescore[i]-homescore_before[i])
+    if (top_inn[i]==0){
+      runs_play[i]=homescore[i]-homescore_before[i]
+    } else{
+      runs_play[i]=roadscore[i]-roadscore_before[i]
+    }
+  }
+  return(runs_play)
+}
+
+
+runs_this_inn=function(end_inn, runs_on_play){
+  m=length(end_inn)
+  runs=integer(m)
+  endinnloc=c(0,grep(1,end_inn))
+  numinns=length(endinnloc)
+  
+  
+  for (j in 2:numinns){
+    for (k in (endinnloc[j-1]+1):endinnloc[j]){
+      runs[k]=sum(runs_on_play[(endinnloc[j-1]+1):endinnloc[j]])
+    }
+  }
+  return(runs)
+}
+
+
+runs_rest_of_inn=function(end_inn, runs_on_play, runs_this_inn){
+  m=length(end_inn)
+  runs=integer(m)
+  
+  endinnloc=c(0,grep(1,end_inn))
+  numinns=length(endinnloc)
+  
+  for (j in 2:numinns){
+    for (k in (endinnloc[j-1]+1):endinnloc[j]){
+      runs[k]=runs_this_inn[k]-sum(runs_on_play[(endinnloc[j-1]+1):(k)])
+    }
+  }
+  runs=runs+runs_on_play
+  return(runs)
+}
+
+
+
+
+
+bat_order_id = function(new_game, top_inn, bat_name){
+  m = length(top_inn)
+  
+  batorder = rep(NA_character_, m)  
+  
+  newgameloc = c(grep(1, new_game), (m+1))
+  numgames = length(newgameloc)
+  
+  for (j in 2:numgames){  
+    kk = 0
+    jj = 0
+    for (i in newgameloc[j-1]:(newgameloc[j]-1)){
+      
+      if (!is.na(top_inn[i]) && !is.na(bat_name[i]) &&
+          top_inn[i] == 1 && bat_name[i] != ''){
+        
+        batorder[i] = (kk %% 9) + 1
+        kk = kk + 1
+        
+      } else if (!is.na(top_inn[i]) && !is.na(bat_name[i]) &&
+                 top_inn[i] == 0 && bat_name[i] != ''){
+        
+        batorder[i] = (jj %% 9) + 1
+        jj = jj + 1
+        
+      } else {
+        batorder[i] = ''   # leave empty if NA or no name
+      }
+    }
+  }
+  return(batorder)
+}
+
+
+
+bat_order_fill=function(bat_order, end_game){
+  m=length(bat_order)
+  for (i in (m):2){
+    if(is.na(bat_order[i-1])==TRUE & end_game[i-1]==0){
+      bat_order[i-1]=bat_order[i]
+    }
+  }
+  
+  for (i in 2:m){
+    if(is.na(bat_order[i])==TRUE){
+      bat_order[i]=bat_order[i-1]
+    }
+  }
+  return(bat_order)
+}
+##########################################################
+
+
+
+
+ncaa_parse=function(pbp_data_frame){
+  
+  pbp_data_frame=pbp_data_frame%>%
+    mutate(
+      tmp_text=paste(away_text, home_text),
+      #  # 
+      sub_fl=case_when(
+        str_detect(tmp_text, '(singled|doubled|tripled|homered|walked|reached|struck out|grounded|flied|lined|popped| hit|infield fly|infield fly|out|double play|triple play)')==TRUE & str_detect(tmp_text, c('pinch hit'))==FALSE ~ 0,
+        str_detect(tmp_text, c('to (p|c|1b|2b|3b|ss|lf|rf|cf|dh)'))==TRUE ~ 1,
+        str_detect(tmp_text, c('pinch hit'))==TRUE ~ 1,
+        str_detect(tmp_text, c('pinch ran'))==TRUE ~ 1,
+        TRUE ~ 0),
+
+      # Split the text up
+      bat_text=gsub('(;|3a|:).*$','', tmp_text),
+      
+      r1_text=case_when(
+        str_detect(tmp_text, '(;|3a|:)')==TRUE ~ stripwhite(gsub('^.*?(;|3a|:)','',tmp_text)),
+        TRUE~''),
+
+      r2_text=case_when(
+        str_detect(r1_text, '(;|3a|:)')==TRUE ~ stripwhite(gsub('^.*?(;|3a|:)','',r1_text)),
+        TRUE~''),
+      
+      r3_text=case_when(
+        str_detect(r2_text, '(;|3a|:)')==TRUE ~ stripwhite(gsub('^.*?(;|3a|:)','',r2_text)),
+        TRUE~''),
+      
+      r2_text=stripwhite(gsub('(;|3a|:).*$','',r2_text)),
+      
+      r1_text=stripwhite(gsub('(;|3a|:).*$','',r1_text)),
+       
+      # Event code: same as retrosheet
+      event_cd=case_when(
+        sub_fl==1 ~ 1,
+        str_sub(stripwhite(tmp_text),1,1)=='(' ~ 1,
+        str_detect(tmp_text, '(hitting out of turn| for |No play|halted|delay|postponed|ejected|suspended|coach|sunny|review|challenged|HC|\\*\\*)') == TRUE ~ 1,
+        str_detect(tmp_text,'struck out') == TRUE ~ 3,
+        str_detect(tmp_text,'stole') == TRUE ~ 4,
+        (str_detect(tmp_text,'(caught stealing|out at second c to|out at third c to)') == TRUE) & (str_detect(tmp_text,'(bunt|grounded)') == FALSE) ~ 6,
+        str_detect(tmp_text,'picked off') == TRUE ~ 8,
+        str_detect(tmp_text,'wild pitch') == TRUE ~ 9,
+        str_detect(tmp_text,'passed ball') == TRUE ~ 10,
+        str_detect(tmp_text,'balk') == TRUE ~ 11,
+        str_detect(tmp_text,'Dropped foul') == TRUE ~ 13,
+        str_detect(tmp_text,'walked') == TRUE ~ 14,
+        str_detect(tmp_text,'hit by pitch') == TRUE ~ 16,
+        str_detect(tmp_text,'interference') == TRUE ~ 17,
+        str_detect(tmp_text,'error') == TRUE ~ 18,
+        str_detect(tmp_text,'muffed') == TRUE ~ 18,
+        str_detect(tmp_text,'dropped') == TRUE ~ 18,
+        str_detect(tmp_text,'fielder\'s choice') == TRUE ~ 19,
+        str_detect(tmp_text,'singled') == TRUE ~ 20,
+        str_detect(tmp_text,'doubled') == TRUE ~ 21,
+        str_detect(tmp_text,'tripled') == TRUE ~ 22,
+        str_detect(tmp_text,'homered') == TRUE ~ 23,
+        str_detect(tmp_text, '(flied out|grounded out|popped|fouled out|lined out| infield fly|double play|triple play|out at (first|second|third|home))') == TRUE ~ 2,
+        str_detect(tmp_text, 'advanced') == TRUE ~ 12,
+        TRUE ~ 0),
+      
+      
+      # Bat name
+      bat_name= case_when(
+        event_cd %in% c(0,1)~'',
+        str_detect(bat_text, '(Batter|Runner\'s interference)')==TRUE ~'',
+        str_detect(bat_text, '(walked|singled|doubled|tripled|reached|struck out|grounded out)')==FALSE & str_detect(bat_text, '(advanced|caught stealing|stole|picked off|out at (first|second|third|home)|tagged out)')==TRUE ~ '',
+        str_detect(bat_text, '(singled|doubled|tripled|homered|walked|reached|struck out|grounded|flied|lined|popped|hit | out |fouled out|pinch hit|infield fly|intentionally walked|was intentionally walked|fouled into double play)')==TRUE ~ gsub('((singled|doubled|tripled|homered|walked|reached|struck out|grounded|flied|lined|popped|hit | out |fouled out|pinch hit|infield fly|intentionally walked|was intentionally walked|fouled into double play).*$)', '', bat_text),
+        str_detect(stripwhite(r1_text), 'caught stealing  c to (2b|3b), double play.')==TRUE ~ bat_text,
+        TRUE ~ ''),
+      
+      #  Sub in
+      sub_in= case_when(
+        sub_fl==1&str_detect(bat_text, 'to (p|c|1b|2b|3b|ss|lf|rf|cf|dh)')==TRUE ~ stripwhite(gsub('(to (p|c|1b|2b|3b|ss|lf|rf|cf|dh).*$)', '', bat_text)),
+        sub_fl==1&str_detect(bat_text, 'pinch ran for')==TRUE ~ stripwhite(gsub('pinch ran for.*$', '', bat_text)),
+        sub_fl==1&str_detect(bat_text, 'pinch hit for')==TRUE ~ stripwhite(gsub('pinch hit for.*$', '', bat_text)),
+        TRUE ~ ''),
+      
+      # Sub out
+      sub_out= case_when(
+        sub_fl==1&str_detect(bat_text, 'to (p|c|1b|2b|3b|ss|lf|rf|cf|dh) for')==TRUE ~ gsub('^.*to (p|c|1b|2b|3b|ss|lf|rf|cf|dh) for', '', bat_text),
+        sub_fl==1&str_detect(bat_text, 'pinch ran for')==TRUE ~ gsub('^.*pinch ran for', '', bat_text),
+        sub_fl==1&str_detect(bat_text, 'pinch hit')==TRUE ~ gsub('^.*pinch hit for', '', bat_text),
+        TRUE ~ ''),
+      # Clean sub out
+      sub_out=strip_punc(sub_out),
+      
+      
+      # Game end
+      game_end = game_end(game_id),
+      
+      # New game
+      new_game=new_game(game_end),
+      
+      # Top inning
+      top_inning=ifelse(away_text=='', 0,1),
+      # End of inning
+      inn_end = inn_end(top_inning),
+      # Runner names
+      r1_name=r1_name(bat_text, bat_name, r1_text, r1_name, inn_end, game_end, sub_in, sub_out),
+      r2_name =r2_name(bat_text, bat_name, r1_text, r1_name, r2_text, r2_name, inn_end, game_end, sub_in, sub_out),
+      r3_name =r3_name(bat_text, bat_name, r1_text, r1_name, r2_text, r2_name, r3_text, r3_name, inn_end, game_end, sub_in, sub_out),
+      # Clean runner names
+      r1_name=replace(r1_name,is.na(r1_name),''),
+      r2_name=replace(r2_name,is.na(r2_name),''),
+      r3_name=replace(r3_name,is.na(r3_name),''),
+      
+      
+      
+      
+      
+      
+      # Fix repeat bat names
+      bat_name=case_when(
+        bat_name!='' & stripwhite(bat_name)==stripwhite(r1_name)~ '',
+        bat_name!='' & stripwhite(bat_name)==stripwhite(r2_name)~ '',
+        bat_name!='' & stripwhite(bat_name)==stripwhite(r3_name)~ '',
+        TRUE ~ bat_name),
+      
+      # 
+      outs_on_play=case_when(
+        event_cd %in% c(0,1) ~ 0,
+        str_count(bat_text, 'triple play') == 1 ~ 3,
+        str_count(bat_text, 'double play') == 1 ~ 2,
+        (str_detect(bat_text, '( out|popped)') == TRUE) &  (str_detect(bat_text, '(reached)') == TRUE) ~ 0,
+        # 1 out
+        ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==TRUE)) ~ 1,
+        #  2 outs
+        ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==TRUE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==TRUE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == FALSE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==TRUE)) ~ 2,
+        # 3 outs 
+        ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==FALSE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==FALSE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==TRUE)) |
+          ((str_detect(bat_text, '( out |popped|infield fly)') == TRUE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==FALSE) &(str_detect(r3_text, '( out |popped)')==TRUE)) |
+          ((str_detect(bat_text, '( out |popped)') == FALSE) & (str_detect(r1_text, '( out |popped)')==TRUE) & (str_detect(r2_text, '( out |popped)')==TRUE) &(str_detect(r3_text, '( out |popped)')==TRUE)) ~ 3,
+        TRUE ~ 0),
+      
+      # New inning
+      new_inn=new_inn(inn_end),
+      # Outs before
+      outs_before=outs_before(outs_on_play, new_game, new_inn),
+      # Outs after
+      outs_after=outs_before+outs_on_play,
+      
+      # Base code
+      base_cd_before=case_when(
+        stripwhite(r1_name)!='' & r2_name=='' & r3_name=='' ~ 1,
+        r1_name=='' & r2_name!='' & r3_name=='' ~ 2,
+        r1_name!='' & r2_name!='' & r3_name=='' ~ 3,
+        r1_name=='' & r2_name=='' & r3_name!='' ~ 4,
+        r1_name!='' & r2_name=='' & r3_name!='' ~ 5,
+        r1_name=='' & r2_name!='' & r3_name!='' ~ 6,
+        r1_name!='' & r2_name!='' & r3_name!='' ~ 7,
+        TRUE~0),
+      
+      # Batting order
+      bat_order=bat_order_id(new_game, top_inning, bat_name),
+      
+      # Hit type
+      hit_type=case_when(
+        event_cd==3 ~ 'K',
+        str_detect(bat_text,'(bunt)')==TRUE ~ 'B',
+        str_detect(bat_text, '(bunt)')==FALSE & str_detect(bat_text, '(SAC)')==TRUE & str_detect(bat_text, '(flied|popped)')==FALSE ~ 'B',
+        str_detect(bat_text,'(grounded out|(p|3b|2b|ss|1b) to (p|3b|2b|ss|1b|c))')==TRUE ~ 'GO',
+        str_detect(bat_text,'(flied|fouled out to (lf|rf))')==TRUE ~ 'FO',
+        str_detect(bat_text,'(lined)')==TRUE ~ 'LO',
+        str_detect(bat_text,'(popped|infield fly|fouled out to (p|3b|2b|ss|1b|c))')==TRUE ~ 'PO',
+        TRUE ~ ''   ),
+      
+      # Runs on play
+      runs_on_play=(as.numeric(str_count(tmp_text, '(advanced to home)'))+as.numeric(str_count(tmp_text, '(scored)')) + as.numeric(str_count(tmp_text, '(homered)')) + as.numeric(str_count(tmp_text, '(stole home)'))-as.numeric(str_count(tmp_text, '(scored, scored)'))),
+      
+      # Away score
+      away_score_before=score_before(new_game, runs_on_play, top_inning, home_team=0),
+      
+      # Home score
+      home_score_before=score_before(new_game, runs_on_play, top_inning, home_team=1),
+      
+      # #  Away score after
+      away_score_after=case_when(
+        top_inning==1 ~away_score_before+ runs_on_play,
+        TRUE ~ away_score_before),
+      
+      # # Home score after
+      home_score_after=case_when(
+        top_inning==0 ~home_score_before+ runs_on_play,
+        TRUE ~ home_score_before),
+      
+      # Runs this inning
+      runs_this_inn=runs_this_inn(inn_end, runs_on_play),
+      
+      # Runs rest of inning
+      runs_roi=runs_rest_of_inn(inn_end,runs_on_play, runs_this_inn),
+      
+      # Intentional walk 
+      int_bb_fl=case_when(
+        str_detect(tmp_text,'intentionally ') == TRUE ~ 1,
+        TRUE ~ 0
+      ),
+      
+      # Sac bunts
+      sh_fl=case_when(
+        str_detect(bat_text, '(SAC)')==TRUE & str_detect(bat_text, '(flied|popped)')==FALSE ~ 1,
+        TRUE~0),
+      
+      # Sac flys
+      sf_fl=case_when(
+        str_detect(bat_text, '(SAC)')==TRUE & str_detect(bat_text, '(flied|popped)')==TRUE ~ 1,
+        str_detect(bat_text, '(SAC)')==FALSE & str_detect(bat_text, '(flied|popped)')==TRUE & str_detect(bat_text, '(RBI)')==TRUE~1,
+        TRUE~0 )
+    )
+  
+    
+  pbp_data_frame=pbp_data_frame%>%
+    mutate(bat_order=bat_order_fill(bat_order, game_end))
+
+  return(pbp_data_frame)
+  
+}
+
+prefixes <- c("St\\.", "Mc", "De", "Di", "Van", "Von")
+
+#Get NCAA Game IDs
+get_ncaa_schedule <- function(date) {
+  
+  date <- as.Date(date)
+  api_url <- sprintf(
+    "https://ncaa-api.henrygd.me/scoreboard/baseball/d1/%04d/%02d/%02d/all-conf",
+    as.integer(format(date, "%Y")),
+    as.integer(format(date, "%m")),
+    as.integer(format(date, "%d"))
+  )
+  
+  
+  res <- GET(api_url, user_agent("Mozilla/5.0"))
+  stop_for_status(res)
+  
+  dat <- fromJSON(content(res, "text", encoding = "UTF-8"), flatten = TRUE)
+  
+  
+  
+  dat$games <- dat$games %>%
+    tidyr::unnest(game.away.conferences) %>%
+    rename(away_conference = conferenceName) %>% 
+    dplyr::select(-conferenceSeo) %>% 
+    tidyr::unnest(game.home.conferences) %>% 
+    rename(home_conference = conferenceName)
+  
+  tibble(
+    Date = dat$games$game.startDate,
+    GameID = dat$games$game.url %>% 
+      str_extract("(?<=/game/)\\d+"), 
+    HomeTeam = dat$games$game.home.names.short, 
+    AwayTeam = dat$games$game.away.names.short, 
+   StartTime = dat$games$game.startTime,
+    HomeScore = dat$games$game.home.score,
+    AwayScore = dat$games$game.away.score, 
+    HomeRecord = dat$games$game.home.description,
+    AwayRecord = dat$games$game.away.description, 
+    HomeConference = dat$games$home_conference,
+    AwayConference = dat$games$away_conference
+  )
+}
+                                            
+
+#Scrapes NCAA PBP based on game IDs                                          
+get_ncaa_pbp <- function(game_id){
+  
+  url <- sprintf("https://ncaa-api.henrygd.me/game/%s/play-by-play", game_id)
+  res <- GET(url)
+  stop_for_status(res)
+  
+  dat <- fromJSON(content(res, "text", encoding = "UTF-8"), flatten = TRUE)
+  
+  teams_df <- dat$teams
+  
+  team_home <- teams_df %>% filter(isHome == TRUE) %>% pull(nameShort)
+  team_visitor <- teams_df %>% filter(isHome == FALSE) %>% pull(nameShort)
+  
+  home_id <- teams_df %>% filter(isHome == TRUE) %>% pull(teamId)
+  visitor_id <- teams_df %>% filter(isHome == FALSE) %>% pull(teamId)
+  
+  home_name_id <- teams_df %>% 
+    filter(isHome == TRUE) %>%
+    mutate(name_id = paste0(toupper(str_sub(nameShort, 1, 3)), "_", toupper(str_sub(teamName, 1, 3)))) %>%
+    pull(name_id)
+  
+  away_name_id <- teams_df %>% 
+    filter(isHome == FALSE) %>%
+    mutate(name_id = paste0(toupper(str_sub(nameShort, 1, 3)), "_", toupper(str_sub(teamName, 1, 3)))) %>%
+    pull(name_id)
+  
+  pbp <- dat$periods %>%
+    unnest(playbyplayStats, names_sep = "_") %>%
+    unnest(playbyplayStats_plays, names_sep = "_") %>%
+    mutate(
+      game_id = game_id,
+      away_team = team_visitor, 
+      home_team = team_home,
+      PlayNumber = row_number(),
+      BatterTeam = ifelse(playbyplayStats_teamId == home_id, team_home, team_visitor),
+      PitcherTeam = ifelse(playbyplayStats_teamId == home_id, team_visitor, team_home),
+      away_text = ifelse(playbyplayStats_teamId == visitor_id, playbyplayStats_plays_playText, ""),
+      home_text = ifelse(playbyplayStats_teamId == home_id, playbyplayStats_plays_playText, ""),
+      Score = ifelse(is.na(playbyplayStats_plays_homeScore) & is.na(playbyplayStats_plays_visitorScore), 
+                     "", 
+                     paste0(playbyplayStats_plays_visitorScore, "-", playbyplayStats_plays_homeScore)),
+      home_name_id = home_name_id,
+      away_name_id = away_name_id
+    ) %>%
+    dplyr::select(game_id, Inning = periodNumber, PlayNumber, BatterTeam, PitcherTeam,
+                  away_team, home_team, away_text, home_text, Score, home_name_id, away_name_id)
+  
+  return(pbp)
+}
+
+
+scrape_clean_ncaa_pbp <- function(start_date, end_date) {
+
+  games_list <- tibble()
+  counter <- 0
+  
+  dates_list <- seq(as.Date(start_date), as.Date(end_date), by = "day")
+  
+  
+  #Get Schedule Data for Game IDs
+  for (date in dates_list) {
+    
+    counter <- counter + 1
+    message(sprintf(
+      "%5.1f%% | %d / %d | %s",
+      100 * counter / length(dates_list), counter, length(dates_list), as.Date(date)
+    ))
+    
+    new_game_ids <- tryCatch(
+      get_ncaa_schedule(date),
+      error = function(e) NULL
+    )
+    
+    if (is.null(new_game_ids) || nrow(new_game_ids) == 0) {
+      message(" -- no games, skipping")
+      next
+    }
+    
+    new_game_ids <- new_game_ids %>%
+      distinct(GameID, .keep_all = TRUE)
+    games_list <- rbind(games_list, new_game_ids)
+    
+    Sys.sleep(0.2)
+  }
+  
+  df <- tibble()
+  counter <- 0
+  
+  #Get PBP Data
+  for (i in seq_len(nrow(games_list))){
+    
+    counter <- counter + 1
+    
+    new_data <- tryCatch(
+      get_ncaa_pbp(games_list$GameID[i]),
+      error = function(e) NULL
+    )
+    
+    if (is.null(new_data) || nrow(new_data) == 0) {
+      message(" -- no data, skipping")
+      next
+    }
+    
+    new_data <- new_data %>% 
+      mutate(Date = games_list$Date[i])
+  
+    df <- rbind(new_data, df) 
+    
+  
+    
+    message(sprintf(
+      "%5.1f%% | %d / %d | %s",
+      100 * counter / length(unique(games_list$GameID)), counter, length(unique(games_list$GameID)), games_list$GameID[i]
+    ))
+    
+    Sys.sleep(0.2)
+
+  }
+
+  pbp <- ncaa_parse(df2)
+
+}
+
+
+                                            
+                                            
 # UI
 ui <- fluidPage(
   tags$head(
@@ -1561,13 +2377,11 @@ removeModal()
     
     result <- tryCatch({
       if (input$scrape_source == "pbp") {
-        scrape_pbp(input$start_date, input$end_date)
+        scrape_trackman_pbp(input$start_date, input$end_date)
       } else if (input$scrape_source == "pos") {
-        scrape_positional(input$start_date, input$end_date)
+        scrape_trackman_positional(input$start_date, input$end_date)
       } else if (input$scrape_source == "ncaa") {
-        # placeholder — put your NCAA scraper here
-        scrape_status_msg("NCAA scraper not yet implemented.")
-        NULL
+        scrape_clean_ncaa_pbp(input$start_date, input$end_date)
       }
     }, error = function(e) {
       scrape_status_msg(paste("Error:", e$message))