How to ignore some composers

[simoncn]

Can you describe in a bit more detail the mismatch between your tags and the browsing structure you would like to have?

[ac16161]

I have been experimenting with writing some code to build the tagupdate.txt file based on input parameters, so that I can have a collection of favourite genres bundled together - Classical on its own and then a set of others. Also, the code ought to be easier to maintain. I have written the following in R, the output file looks like a perfectly formatted tagupdate.txt file, but Minimserver is ignoring it. My manually created file adding a main library view of Classical music works fine. Even if you don't know R, you are probably way more advanced than me and can see what this is doing. Any ideas why the code-generated version of the tag file is being ignored?

Code:

# filename of tag update file
fn <- "M:/tagupdate.txt"

# genre tags, group using * character
# first entry is display genre in Minimserver, subsequent entries are synonyms
# R requires substitution of \ with \\ otherwise it is treated as an escape character

ingenre <- c("Electronica*Electronic*Electro",
             "Hip-Hop/Rap*Hip Hop/Rap*Hip-Hop*Rap*Alternative Rap",
             "Pop/Rock*Pop\\Rock*Pop Rock",
             "Pop*Dance*New Wave",
             "World*International",
             "Soundtrack*Original Score",
             "Alternative*Psychadelic*Alternative & Punk",
             "New Age*Newage",
             "Soul*R&B/Soul*Soul / Funk / R&B",
             "R&B*Blues",
             "Easy Listening*Swing*Adult Contemporary",
             "Rock*Metal",
             "Traditional*Traditional, Christmas")

# favourite non-classical genres
favgenre <- c("Alternative","Easy Listening","Electronica","Jazz","Latin","Pop","Pop/Rock","R&B","Reggae","Rock","Singer/Songwriter","Soul","Soundtrack","World")

con <- file(fn,open="w",encoding="UTF-8")
sink(con)
for (i in 1:length(ingenre)){
  graw <- unlist(strsplit(ingenre[i],split="*",fixed=TRUE))
  for (j in 2:length(graw)){
    cat(paste("@GENRE=",graw[j],sep=""),"\n")
    cat(paste("=GENRE=",graw[1],sep=""),"\n","\n")
  }
}

# create additional main library genres
cat("@GENRE=Classical","\n")
cat("+CLASSICALGENRE=Classical","\n","\n")

for (i in 1:length(favgenre)){
  cat(paste("@GENRE=",favgenre[i],sep=""),"\n")
  cat("+FAVGENRE=Favourites","\n","\n")
}

closeAllConnections()

[simoncn]

Are there any messages in the MinimServer log about this?

If there are no messages, please post the output from this program as a file attachment (preferably zipped). If the output is very large, please post a section of it. Many thanks.

[ac16161]

  tagupdate.zip (Size: 434 bytes / Downloads: 0)
  tagupdate_r.zip (Size: 504 bytes / Downloads: 0)

First of these tage files works fine, second one causes Minimserver to stall on starting. I don't see any issues in the log.

[ac16161]

Found the problem: my R code was inserting a trailing blank before the carriage return character, this caused the tag update to fail. I have tweaked the code to suppress these. The following now builds the tagupdate.txt file with the synonyms for genres in one place, and allows for favourite genres to be bundled.

Code:

# filename of tag update file
fn <- "M:/tagupdate.txt"

# genre tags, group using * character
# first entry is display genre in Minimserver, subsequent entries are synonyms
# R requires substitution of \ with \\ otherwise it is treated as an escape character

ingenre <- c("Electronica*Electronic*Electro",
             "Hip-Hop/Rap*Hip Hop/Rap*Hip-Hop*Rap*Alternative Rap",
             "Pop/Rock*Pop\\Rock*Pop Rock",
             "Pop*Dance*New Wave",
             "World*International",
             "Soundtrack*Original Score",
             "Alternative*Psychadelic*Alternative & Punk",
             "New Age*Newage",
             "Soul*R&B/Soul*Soul / Funk / R&B",
             "R&B*Blues",
             "Easy Listening*Swing*Adult Contemporary",
             "Rock*Metal",
             "Traditional*Traditional, Christmas")

# favourite non-classical genres
favgenre <- c("Alternative","Easy Listening","Electronica","Jazz","Latin","Pop","Pop/Rock","R&B","Reggae","Rock","Singer/Songwriter","Soul","Soundtrack","World")

con <- file(fn,open="w",encoding="UTF-8")
sink(con)
for (i in 1:length(ingenre)){
  graw <- unlist(strsplit(ingenre[i],split="*",fixed=TRUE))
  for (j in 2:length(graw)){
    cat(paste("@GENRE=",graw[j],"\n",sep=""))
    cat(paste("=GENRE=",graw[1],"\n",sep=""))
  }
}

# create additional main library genres
cat(paste("@GENRE=Classical","\n",sep=""))
cat(paste("+CLASSICALGENRE=Classical","\n",sep=""))

for (i in 1:length(favgenre)){
  cat(paste("@GENRE=",favgenre[i],"\n",sep=""))
  cat(paste("+FAVGENRE=Favourites","\n",sep=""))
}

closeAllConnections()

[simoncn]

Thanks for letting me know. Trailing blanks are automatically stripped from tag values, so I think MinimServer should ignore trailing blanks in the tag update file if present. I will make this change in the next update.

[ac16161]

Thanks for that Simon. And for anyone who is interested in building a tagupdate.txt file via R, here is an extension based on processing the all-tags txt file that Minimserver can create - this code reads the txt file from Minimserver, gets the lines in the file for genre tags, then creates a vector of (in this case) non-classical tags which can be processed using code in the previous post, so I can now select from classical/non-classical via the main page in my Kazoo control point.

Code:

intags <- "L:/tag.txt"
tagdat <- readLines(intags)
genres <- gsub("GENRE=","",tagdat[grep("GENRE=",tagdat)])
# non-classical genres
ncgenres <- genres[!grepl("Classical",genres)]

[ac16161]

Can I confirm my understanding of how the tag updating works: essentially, all actions associated with a tag @xxx= statement must be blocked together?

E.g. if at one point in the file we have

@GENRE=Pop Rock
=GENRE=Pop/Rock

but then later in the file we have

@GENRE=Pop Rock
+MAINGENRE=Main

then only the second update gets made. For both updates to apply, we need

@GENRE=Pop Rock
=GENRE=Pop/Rock
+MAINGENRE=Main

I ask this since the behaviour based on a tag update file that firstly processes all the synonyms and then processes the addition of main library tags did not quite work - I had to merge these steps together, which made the R code slightly more fiddly but the end result is now pretty solid.

[simoncn]

Yes, this is how it works at present. Supporting multiple actions for the same tag value is tricky because the first action might change or remove the tag value which would cause a problem for matching the @ instruction in the second action. There are ways round this but they would add significant complexity and overhead.

[simoncn]

(01-09-2018 10:48)simoncn Wrote:  Thanks for letting me know. Trailing blanks are automatically stripped from tag values, so I think MinimServer should ignore trailing blanks in the tag update file if present. I will make this change in the next update.

This change is now available in MinimServer update 125.

Edit 1: Actually, this change didn't make it into update 125 and I don't think it could be done without adding significant complexity to the implementation and also affecting scanning performance for the vast majority of tagUpdate files that don't contain trailing blanks. This is because tagUpdate processing is done before leading and trailing blanks are stripped from tag values. For these reasons, I think it would be better to leave things as they are for now and require the tagUpdate value to match the tag value exactly, including any leading and trailing blanks if present.

Edit 2: On further investigation, the change is there in update 125 but it won't work as intended in some cases because of the order of processing (see Edit 1 above). For example, if the actual tag has a trailing blank and the corresponding tagUpdate line also has a trailing blank, MinimServer update 125 won't match the tagUpdate line with the actual tag. This is because at the time that the matching comparison is done, the trailing blank has been stripped from the tagUpdate line but has not yet been stripped from the actual tag. I will give this some further thought.