Combine sentences in a .srt input subtitle. Lua script that works as a command line/terminal utility.

srt_to_sentences.lua

#!/usr/bin/env lua
-------------------------------------
-- This script takes an .srt file as an argument, and outputs an srt file with the sentences combined.
-- Output is sent to STDOUT.  So to convert a .srt file, you can do something like this
-- 
-- $ chmod +x ./srt_to_sentences.lua
-- $ ./srt_to_sentences.lua input.srt > input-sentences.srt
--
-- Optimized for my use case. Try tweaking the options below if you have issues.
-- note: the key property of this over other solutions I found is that the outputed srt timecodes are a strict
--       subset of the input timecodes.
--
-- based on this mpv script: https://github.com/kelciour/mpv-scripts/blob/master/sub-sentences.lua
--------------------------------------


local USE_CASING = true
local MAX_SENTENCE_TIME = 8
local COMBINE_NEWLINES = true


function round2(number, precision)
   local fmtStr = string.format('%%0.%sf',precision)
   number = string.format(fmtStr,number)
   return number
end

function srt_time_to_seconds(time)
    local major, minor = time:match("(%d%d:%d%d:%d%d),(%d%d%d)")
    local hours, mins, secs = major:match("(%d%d):(%d%d):(%d%d)")
    return hours * 3600 + mins * 60 + secs + minor / 1000
end

function seconds_to_srt_time(time)
    local hours = math.floor(time / 3600)
    local mins = math.floor(time / 60) % 60
    local secs = math.floor(time % 60)
    local milliseconds = (time * 1000) % 1000

    -- print("TIME=",hours, mins, secs, round2(milliseconds,0))
    return string.format("%02d:%02d:%02d,%03d", hours, mins, secs, round2(milliseconds,0))
end

function file_exists(file)
  local isExist = io.popen(
    '[[ -e "'.. tostring(file) ..'" ]] && { echo "true"; }')
  local isIt = isExist:read("*a")
  isExist:close()
  isIt = string.gsub(isIt, '^%s*(.-)%s*$', '%1')
  if isIt == "true" then
    return true
  end
end

function get_working_dir()
    -- local path = mp.get_property("path", "")
    local dir, filename = utils.split_path(path)
    return dir
end

function open_subtitles_file(srt_file_exts)
    --local srt_base_filename = mp.get_property("working-directory") .. "/" .. mp.get_property("filename/no-ext")
    local dir = get_working_dir()
    local srt_base_filename = utils.join_path(dir,mp.get_property("filename/no-ext"))
    
    for i, ext in ipairs(srt_file_exts) do
        local f, err = io.open(srt_base_filename .. ext, "r")
        --mp.msg.warn("Attempting to open file ",srt_base_filename .. ext)
        
        if f then
            --mp.msg.warn("Success! f=",f)
            return f
        else
            --mp.msg.warn("Failed to open...")
        end
    end

    return false
end

function read_subtitles(srt_path)
    -- local f = open_subtitles_file(srt_file_exts)
    local f = io.open(srt_path,"r")

    if not f then
        print("FAILED TO OPEN SRT FILE!", srt_path)
        return false
    end

    local data = f:read("*all")
    data = string.gsub(data, "\r\n", "\n")
    f:close()
    
    local subs = {}
    local subs_start = {}
    local subs_end = {}
    
    for start_time, end_time, text in string.gmatch(data, "(%d%d:%d%d:%d%d,%d%d%d) %-%-> (%d%d:%d%d:%d%d,%d%d%d)\n(.-)\n\n") do
      table.insert(subs, text)
      table.insert(subs_start, srt_time_to_seconds(start_time))
      table.insert(subs_end, srt_time_to_seconds(end_time))
    end

    return subs, subs_start, subs_end
end

function convert_into_sentences(subs, subs_start, subs_end)
    local sentences = {}
    local sentences_start = {}
    local sentences_end = {}

    for i, sub_text in ipairs(subs) do
        local sub_start = subs_start[i]
        local sub_end = subs_end[i]
        local sub_text = string.gsub(sub_text, "<[^>]+>", "")
        -- sub_text = string.gsub(sub_text,"%.%.%.","…")
        -- sub_text = string.gsub(sub_text,"%-%-","_____")

        if COMBINE_NEWLINES then
            sub_text = string.gsub(sub_text, "\n", "#")
            sub_text = string.gsub(sub_text, "#%-", "\n-")
            sub_text = string.gsub(sub_text, "#", " ")
        end

        if sub_text:find("^- ") ~= nil and sub_text:sub(3,3) ~= sub_text:sub(3,3):upper() then
           sub_text = string.gsub(sub_text, "^- ", "")
        end

        if #sentences > 0 then
            local prev_sub_start = sentences_start[#sentences]
            local prev_sub_end = sentences_end[#sentences]
            local prev_sub_text = sentences[#sentences]
            -- print("-->",sub_text,(sub_start - prev_sub_end) <= 1000 and sub_text:sub(1,1) ~= '-' and 
            --         sub_text:sub(1,1) ~= '"' and sub_text:sub(1,1) ~= "'" and sub_text:sub(1,1) ~= '(' and
            --         (prev_sub_text:sub(prev_sub_text:len()) ~= "." or prev_sub_text:sub(prev_sub_text:len()-2) == "...") and
            --         prev_sub_text:sub(prev_sub_text:len()) ~= "?" and prev_sub_text:sub(prev_sub_text:len()) ~= "!" and
            --         (sub_text:sub(1,1) == sub_text:sub(1,1):lower() or prev_sub_text:sub(prev_sub_text:len()) == ","))
            if USE_CASING then
                casing_result = sub_text:sub(1,1) == sub_text:sub(1,1):lower() or prev_sub_text:sub(prev_sub_text:len()) == "," or prev_sub_text:sub(prev_sub_text:len()-2) == "..."
            else
                casing_result = true
            end
            if (sub_start - prev_sub_end) <= MAX_SENTENCE_TIME and sub_text:sub(1,1) ~= '-' and 
                    sub_text:sub(1,1) ~= '"' and sub_text:sub(1,1) ~= "'" and sub_text:sub(1,1) ~= '(' and
                    (prev_sub_text:sub(prev_sub_text:len()) ~= "." or prev_sub_text:sub(prev_sub_text:len()-2) == "...") and
                    prev_sub_text:sub(prev_sub_text:len()) ~= "?" and prev_sub_text:sub(prev_sub_text:len()) ~= "!" and casing_result then
                
                local text = sentences[#sentences] .. " " .. sub_text
                text = string.gsub(text, "\n", "#")
                text = string.gsub(text, "%.%.%. %.%.%.", " ")
                text = string.gsub(text, "#%-", "\n-")
                text = string.gsub(text, "#", " ")
                if text:match("\n%-") ~= nil and text:match("^%-") == nil then
                    text = "- " .. text
                end
                
                sentences[#sentences] = text
                sentences_end[#sentences] = sub_end
            else
                table.insert(sentences, sub_text)
                table.insert(sentences_start, sub_start)
                table.insert(sentences_end, sub_end)    
            end
        else
            table.insert(sentences, sub_text)
            table.insert(sentences_start, sub_start)
            table.insert(sentences_end, sub_end)
        end
    end

    return sentences, sentences_start, sentences_end
end

function write_subtitles(subs, subs_start, subs_end)
	if subs ~= nil then

        for i, sub_text in ipairs(subs) do
            local sub_start = subs_start[i]
            local sub_end = subs_end[i]

			io.write(i, "\n")
            io.write(seconds_to_srt_time(sub_start) .. " --> " .. seconds_to_srt_time(sub_end), "\n")
            io.write(sub_text, "\n\n")
        end

        return true
    end
	return false
end

io.stderr:write("input_path=" .. arg[1] .. "\n")

local subs, subs_start, subs_end = read_subtitles(arg[1])
local sentences, sentences_start, sentences_end = convert_into_sentences(subs, subs_start, subs_end)

local ret = write_subtitles(sentences, sentences_start, sentences_end)

utf8 support please. renaming srt files to ansii pain

Error processing subtitle with Lua script: input_path=test\Böyle konusulur! Helal olsun [s4JHN9VPL8I].0.srt
lua: attempt to index a boolean value
stack traceback:
        [C]: in for iterator 'for iterator'
        C:\Users\Kaguya\scripts\srt_to_sentences.lua:112: in function 'convert_into_sentences'
        C:\Users\Kaguya\scripts\srt_to_sentences.lua:194: in main chunk
        [C]: in ?

Failed to process test\Böyle konuşulur! Helal olsun [s4JHN9VPL8I].0.srt with Lua script.
Processed Böyle konuşulur! Helal olsun [s4JHN9VPL8I].mkv