FFOPTS and default behavior doesn't re-encode.

README.md

@@ -32,22 +32,58 @@ The *time-file* must have this format:
 
 # Environment variables
 
+The behavior of the script can be changed by several environment variables.
+
 * **SIMULATION** [empty or not]\
     do not invoke ffmpeg
-* **NONUMBER** [empty or 1]\
+
-    do not write song numbers
 * **FORMAT** [mp3,ogg,opus,…]\
     see the ffmpeg documentation for the output formats available
-* **SEPARATOR** [separator] (default: ' - ')\
+* **FFOPTS** [any ffmpeg options] *(default: '-c:a copy')*\
+    ffmpeg options, can be used to change audio quality\
+    can be **required** to change in case input and output file formats differ\
+    see the ffmpeg documentation for available parameters
+
+* **NONUMBER** [empty or 1]\
+    do not write song numbers
+* **SEPARATOR** [separator] *(default: ' - ')*\
     separator between number and name\
     example with SEPARATOR='_': 01_intro.opus 02_blah.opus…
+
 * **HEADERS** [empty or 1]\
     print environment parameters (verbosity, simulation, etc.)
-* **VERBOSITY** [0-3] (default: 1)\
+* **VERBOSITY** [0-3] *(default: 1)*\
     0: no output except errors from ffmpeg\
     1: simple indications on the current track being extracted\
     2: print actual ffmpeg commands the script currently runs
 
+# Different input and output file formats
+
+In case you want to change the file format, let's say from `flac` to `opus`, you need to override the default ffmpeg options provided by `get-tracks.sh`.
+This is done through the `FFOPTS` environment variable, **which needs to NOT be empty** in order to replace the default `get-tracks.sh` behavior (which is `-c:a copy`).
+By default, `ffmpeg` performs re-encoding by itself.
+
+```Bash
+FORMAT=opus FFOPTS=" " get-tracks.sh cd.flac cd.txt
+```
+
+### Warning: sometimes you don't even need to
+
+You may encounter files in some format like `webm` and you want to convert the output files in `opus`.
+But, inside the `webm` format, you **may** have `opus`-encoded audio.
+In these cases, no re-encoding is necessary, and you can do something like:
+
+```Bash
+FORMAT=opus get-tracks.sh cd.webm cd.txt
+```
+
+You'll have a warning mentionning FFOPTS (based on different formats).
+But the generated audio files won't have any quality loss.
+This happens sometimes with the `youtube-dl` utility.
+
+In case there are actual `ffmpeg` errors, and you don't have output audio files, then the contained audio hadn't the right format.
+You'll have to re-encode.
+
 # More
 
-Run `get-track.sh` without arguments.
+You can get some help by running `get-track.sh` without arguments.

get-tracks.sh

@@ -146,7 +146,7 @@ run_ffmpeg(){
 			echo "extracting '$final_title'"
 			;;
 		v2)
-			echo "ffmpeg $LOG_LEVEL $FROM $TO -i $INPUT_FILE '$OUTPUT_FILE'"
+			echo "ffmpeg $LOG_LEVEL $FROM $TO -i $INPUT_FILE $FFOPTS '$OUTPUT_FILE'"
 			;;
 		*)
 			echo "verbosity is not set properly" >&2
@@ -155,7 +155,7 @@ run_ffmpeg(){
 	esac
 
 	if [ "$SIMULATION" = "" ]; then
-		ffmpeg $LOG_LEVEL $FROM $TO -i "$INPUT_FILE" "$OUTPUT_FILE"
+		ffmpeg $LOG_LEVEL $FROM $TO -i "$INPUT_FILE" $FFOPTS "$OUTPUT_FILE"
 	fi
 }
 
@@ -195,17 +195,17 @@ Format for <song-list>:
   1:30 Second track
 
 Environment variables:
-- SIMULATION [empty or not]
+- SIMULATION [empty or not]         do not invoke ffmpeg
-    do not invoke ffmpeg
+
-- NONUMBER [empty or 1]
+- FORMAT [mp3,ogg,opus,…]           see ffmpeg documentation
-    do not write song numbers
+- FFOPTS (default: '-c:a copy')     see ffmpeg documentation
-- FORMAT [mp3,ogg,opus,…]
+
-    see the ffmpeg documentation
+- NONUMBER [empty or 1]             do not write song numbers
 - SEPARATOR [separator] (default: ' - ')
     separator between number and name
     example with SEPARATOR='_': 01_intro.opus 02_blah.opus…
-- HEADERS [empty or 1]
+
-    print environment parameters (verbosity, simulation, etc.)
+- HEADERS [empty or 1]             print env params (verbosity, quality, etc.)
 - VERBOSITY [0-3] (default: 1)
     0: no output except errors from ffmpeg
     1: simple indications on the current track being extracted
@@ -219,13 +219,36 @@ header(){
 	fi
 }
 
+warning(){
+	echo "WARNING: $*"
+}
+
+# Default output format is based on the extension of the input audio file.
+if [ $# -eq 2 ]; then
+	DEFAULT_FORMAT="$(echo $1 | awk -F . '{print $NF}')"
+else
+	header "no default FORMAT selected"
+fi
+
 if [ "$FORMAT" = "" ]; then
-	header "default FORMAT: opus"
+	FORMAT="$DEFAULT_FORMAT"
-	FORMAT="opus"
+	header "default FORMAT: ${FORMAT}"
 else
 	header "FORMAT: $FORMAT"
 fi
 
+# For unexperienced users, print a warning when input and output formats differ.
+# In case FFOPTS is set, encoding is expected to be handled, drop the warning.
+# Example (remove the get-tracks.sh default behavior, perform re-encoding):
+#   FFOPTS=" "
+if [ "$FFOPTS" = "" ] && [ "$FORMAT" != "$DEFAULT_FORMAT" ]; then
+	warning "input and output formats seem to differ"
+	warning "1. re-encoding may be required (through the FFOPTS envvar)"
+	warning "2. FFOPTS represents ffmpeg options, directly given to ffmpeg"
+	warning '   (default: "-c:a copy" = copy without re-encoding)'
+	warning '   You can put FFOPTS=" " if you want to perform re-encoding.'
+fi
+
 if [ "$VERBOSITY" = "" ]; then
 	header "default VERBOSITY: 1"
 	VERBOSITY=1
@@ -249,6 +272,13 @@ else
 	SEPARATOR=""
 fi
 
+if [ "$FFOPTS" != "" ]; then
+	header "FFOPTS envvar is set: ${FFOPTS}."
+else
+	FFOPTS="-c:a copy"
+	header "default FFOPTS: ${FFOPTS}"
+fi
+
 if [ "$SIMULATION" != "" ]; then
 	header "SIMULATION envvar is set: this is a simulation."
 fi