Bash: Split with 2 factors / variables - bash

Question

I want to split a file with the following algorithm.
This CSV has a 3600 lines previously ordered by Name alpabetically ( sort -k2 -n file.csv )
Currently I can run this command to split the file in equal number of lines:
split -l ${MAX_NUMBER_OF_LINES} filename.csv ${new_file_pattern}.
But the original requirement is:
Split into chunks of ${MAX_NUMBER_OF_LINES} UNLESS no more records with the first letter of the column 2 exists.
For example:
if I have ${MAX_NUMBER_OF_LINES} = 3, I can split the file in chunk of 300 lines if no more occurrencies of the last first letter of the column are found.
If the LINE 301 has a record with "Arboreal Peaches" the script has to add to the current chunk no matter the ${MAX_NUMBER_OF_LINE} was already reach.
Is sort of confusing explanation.. I hope any of you can help me (I already spent 2 days in this algorithm)
UPDATE
${MAX_NUMBER_OF_LINES} = 3
Example CSV (with fewer lines for exaple purpose).
Split command reaches ${MAX_NUMBER_OF_LINES}, but the line 4 already has a record with the letter A
'Aberdeen Research", 'Los Angeles', 'California'
'Aplueyo Labs", 'Los Angeles', 'US'
'Acar Media Group", 'Los Angeles', 'US'
'Aberdeen Research", 'San Jose', 'US'
'Beethoven Inc", 'San Jose', 'US'
EXPECTED RESULT
Splitted Files
1
'Aberdeen Research", 'Los Angeles', 'California'
'Aplueyo Labs", 'Los Angeles', 'US'
'Acar Media Group", 'Los Angeles', 'US'
'Aberdeen Research", 'San Jose', 'US'
2
'Beethoven Inc", 'San Jose', 'US'

Answer 1

Something like this? In awk:
$ cat split.awk
BEGIN {if(max=="")
print "Invalid numer of lines"; exit # exit if no max
}
(a=substr($0,2,1)) && ++c>=max && prev!=a { # first letter to a, if count >= max
c=0 # and first letter changes
fc++ # reset count and filemask counter
}
{
print $0 > (mask==""?"x":mask) (fc==""?0:fc) # write to file default mask x
prev=a # remember previous first letter
}
Run it:
$ awk -v max=3 -v mask="file" -f split.awk file.csv
$ cat file0
'Aberdeen Research", 'Los Angeles', 'California'
'Aplueyo Labs", 'Los Angeles', 'US'
'Acar Media Group", 'Los Angeles', 'US'
'Aberdeen Research", 'San Jose', 'US'
$ cat file1
'Beethoven Inc", 'San Jose', 'US'
mask is the filename prefix or $new_file_pattern and max is $MAX_NUMBER_OF_LINES, ie. in the command line set -v max=$MAX_NUMBER_OF_LINES -v mask=$new_file_pattern.