changes - hope they work
This commit is contained in:
@ -2,13 +2,13 @@
|
||||
set -euo pipefail
|
||||
|
||||
CSV_DEFAULT="wunf_scenes.csv"
|
||||
DRY_RUN=0
|
||||
|
||||
usage() {
|
||||
echo "Användning:"
|
||||
echo " $0 <filnamn> [csv-fil]"
|
||||
echo " $0 --all <katalog> [csv-fil]"
|
||||
echo " $0 --list <textfil> [csv-fil] # en väg per rad"
|
||||
echo " ls -1 --quoting-style=literal \[WUNF\]_*.mp4 > filenames.txt && ./script/wunf_upd_prefix.sh --list filenames.txt wunf_scenes.csv && rm filenames.txt"
|
||||
echo " $0 [--dry-run] <filnamn> [csv-fil]"
|
||||
echo " $0 [--dry-run] --all <katalog> [csv-fil]"
|
||||
echo " $0 [--dry-run] --list <textfil> [csv-fil] # en väg per rad"
|
||||
exit 1
|
||||
}
|
||||
|
||||
@ -16,77 +16,167 @@ rename_file() {
|
||||
local FILE="$1"
|
||||
local CSV="$2"
|
||||
|
||||
local base dir rest first second remainder name code code_us newbase newpath
|
||||
|
||||
if [[ ! -f "$FILE" ]]; then
|
||||
echo "Fel: Filen finns inte: $FILE" >&2
|
||||
return 1
|
||||
fi
|
||||
|
||||
local base dir rest name_segment code newbase newpath
|
||||
|
||||
base="$(basename -- "$FILE")"
|
||||
dir="$(dirname -- "$FILE")"
|
||||
|
||||
# Kontrollera prefix
|
||||
if [[ "$base" != "[WUNF]"* ]]; then
|
||||
# Hoppa över om den redan har index, t.ex. [WUNF_427]...
|
||||
if [[ "${base:0:7}" == "[WUNF_" ]]; then
|
||||
echo "Hoppar över (har redan index): $base" >&2
|
||||
return 0
|
||||
fi
|
||||
|
||||
# Kräver exakt prefixet "[WUNF]"
|
||||
if [[ "${base:0:6}" != "[WUNF]" ]]; then
|
||||
echo "Hoppar över (ej [WUNF]-prefix): $base" >&2
|
||||
return 0
|
||||
fi
|
||||
|
||||
# Plocka ut allt efter prefixet [WUNF]_
|
||||
rest="${base#"[WUNF]"}"
|
||||
rest="${rest#_}"
|
||||
# Ta bort prefixet "[WUNF]"
|
||||
rest="${base:6}"
|
||||
|
||||
IFS='_' read -r first second remainder <<< "$rest"
|
||||
# Ta bort ett eller flera ledande underscores efter prefixet
|
||||
while [[ "${rest:0:1}" == "_" ]]; do
|
||||
rest="${rest:1}"
|
||||
done
|
||||
|
||||
if [[ -z "${first:-}" || -z "${second:-}" ]]; then
|
||||
echo "Fel: Kunde inte extrahera två namn-delar i: $base" >&2
|
||||
return 1
|
||||
# Ta bort extension (.mp4, .mkv osv)
|
||||
name_segment="${rest%.*}"
|
||||
|
||||
if [[ -z "$name_segment" ]]; then
|
||||
echo "Ingen namn-del efter prefix i: $base" >&2
|
||||
return 0
|
||||
fi
|
||||
|
||||
# Namn = två första token, med '_' -> ' '
|
||||
name="${first//_/ } ${second//_/ }"
|
||||
name="$(printf '%s' "$name" | sed -E 's/^[[:space:]]+|[[:space:]]+$//g')"
|
||||
|
||||
# slå upp i CSV (format Namn;Kod;Tid)
|
||||
# Slå upp kod i CSV via tokeniserad matchning
|
||||
code="$(
|
||||
awk -F';' -v target="$name" '
|
||||
BEGIN{IGNORECASE=1}
|
||||
awk -F';' -v fstring="$name_segment" '
|
||||
BEGIN {
|
||||
# 1) Tokenisera filnamnets namn-del
|
||||
s = fstring
|
||||
gsub(/_+/, " ", s) # "_" -> mellanslag
|
||||
gsub(/[[:space:]]+/, " ", s) # normalisera whitespace
|
||||
sub(/^[[:space:]]+/, "", s)
|
||||
sub(/[[:space:]]+$/, "", s)
|
||||
s = tolower(s)
|
||||
|
||||
if (length(s) == 0) { exit 0 }
|
||||
|
||||
n_ft = split(s, ft, " ") # ft[] = filtokens
|
||||
|
||||
bestScore = 0
|
||||
bestCode = ""
|
||||
ambiguous = 0
|
||||
}
|
||||
{
|
||||
n=$1; gsub(/^[ \t]+|[ \t]+$/, "", n);
|
||||
if (n==target) {
|
||||
c=$2; gsub(/^[ \t]+|[ \t]+$/, "", c);
|
||||
print c; exit
|
||||
# 2) Tokenisera namn-kolumnen i CSV
|
||||
name = $1
|
||||
gsub(/,/, " ", name) # kommatecken ignoreras
|
||||
gsub(/[[:space:]]+/, " ", name)
|
||||
sub(/^[[:space:]]+/, "", name)
|
||||
sub(/[[:space:]]+$/, "", name)
|
||||
lname = tolower(name)
|
||||
if (length(lname) == 0) next
|
||||
|
||||
n_rt = split(lname, rt, " ") # rt[] = CSV-tokens
|
||||
|
||||
# kan inte vara prefix om den är längre än filens tokenlista
|
||||
if (n_rt > n_ft) next
|
||||
|
||||
# 3) Kolla om CSV-tokenlistan är prefix till filnamnets tokens
|
||||
matchOK = 1
|
||||
for (i = 1; i <= n_rt; i++) {
|
||||
if (rt[i] != ft[i]) { matchOK = 0; break }
|
||||
}
|
||||
}' "$CSV"
|
||||
if (!matchOK) next
|
||||
|
||||
matchLen = n_rt
|
||||
|
||||
if (matchLen > 0) {
|
||||
if (matchLen > bestScore) {
|
||||
bestScore = matchLen
|
||||
bestCode = $2
|
||||
ambiguous = 0
|
||||
} else if (matchLen == bestScore) {
|
||||
# Fler rader med samma bästa längd => tvetydigt
|
||||
ambiguous = 1
|
||||
}
|
||||
}
|
||||
}
|
||||
END {
|
||||
if (bestScore > 0 && !ambiguous && length(bestCode) > 0) {
|
||||
c = bestCode
|
||||
gsub(/^[ \t]+|[ \t]+$/, "", c)
|
||||
print c
|
||||
}
|
||||
}
|
||||
' "$CSV"
|
||||
)"
|
||||
|
||||
if [[ -z "$code" ]]; then
|
||||
echo "Ingen kod för \"$name\" hittades i $CSV (fil: $base)" >&2
|
||||
return 1
|
||||
echo "Ingen unik match för filen: $base" >&2
|
||||
return 0
|
||||
fi
|
||||
|
||||
code_us="${code// /_}"
|
||||
# Bygg prefix med paddad index:
|
||||
# förväntat format: t.ex. "WUNF 1", "WUNF 88", "WUNF 427"
|
||||
local trimmed_code prefix number number_padded code_us
|
||||
trimmed_code="$(printf '%s' "$code" | sed -E 's/^[[:space:]]+|[[:space:]]+$//g')"
|
||||
|
||||
if [[ "$trimmed_code" =~ ^([A-Za-z]+)[[:space:]]*([0-9]+)$ ]]; then
|
||||
prefix="${BASH_REMATCH[1]}"
|
||||
number="${BASH_REMATCH[2]}"
|
||||
printf -v number_padded "%03d" "$number"
|
||||
code_us="${prefix}_${number_padded}"
|
||||
else
|
||||
# Fallback om formatet skulle avvika – använd gammal logik
|
||||
code_us="${trimmed_code// /_}"
|
||||
fi
|
||||
|
||||
# Ersätt bara själva prefixet [WUNF] i början
|
||||
newbase="$(printf '%s' "$base" | sed -E "s/^\[WUNF\]/[$code_us]/")"
|
||||
newpath="$dir/$newbase"
|
||||
|
||||
if [[ "$base" == "$newbase" ]]; then
|
||||
echo "Hoppar över (redan rätt namn): $base" >&2
|
||||
echo "Hoppar över (redan rätt prefix): $base" >&2
|
||||
return 0
|
||||
fi
|
||||
|
||||
if [[ -e "$newpath" ]]; then
|
||||
if [[ -e "$newpath" && $DRY_RUN -eq 0 ]]; then
|
||||
echo "Fel: Målfilen finns redan: $newpath" >&2
|
||||
return 1
|
||||
fi
|
||||
|
||||
mv -- "$dir/$base" "$newpath"
|
||||
echo "Bytt namn:"
|
||||
echo " Från: $base"
|
||||
echo " Till: $newbase"
|
||||
if [[ $DRY_RUN -eq 1 ]]; then
|
||||
echo "DRY-RUN: skulle byta namn:"
|
||||
echo " Från: $base"
|
||||
echo " Till: $newbase"
|
||||
else
|
||||
mv -- "$dir/$base" "$newpath"
|
||||
echo "Bytt namn:"
|
||||
echo " Från: $base"
|
||||
echo " Till: $newbase"
|
||||
fi
|
||||
}
|
||||
|
||||
### Huvudprogram
|
||||
|
||||
if [[ $# -lt 1 ]]; then
|
||||
usage
|
||||
fi
|
||||
|
||||
# Global flagga --dry-run
|
||||
if [[ "${1:-}" == "--dry-run" ]]; then
|
||||
DRY_RUN=1
|
||||
shift
|
||||
fi
|
||||
|
||||
if [[ $# -lt 1 ]]; then
|
||||
usage
|
||||
fi
|
||||
@ -128,14 +218,9 @@ case "$mode" in
|
||||
exit 1
|
||||
fi
|
||||
|
||||
# Läs en rad i taget; ignorera tomma rader och kommentarer
|
||||
# Hanterar vägar med mellanslag korrekt
|
||||
while IFS= read -r line || [[ -n "$line" ]]; do
|
||||
# Trim CR (om filen är CRLF)
|
||||
line="${line%$'\r'}"
|
||||
# Trim whitespace runt
|
||||
trimmed="$(printf '%s' "$line" | sed -E 's/^[[:space:]]+|[[:space:]]+$//g')"
|
||||
# Hoppa över tomma rader och kommentarer
|
||||
[[ -z "$trimmed" || "$trimmed" =~ ^# ]] && continue
|
||||
rename_file "$trimmed" "$CSV" || true
|
||||
done < "$LIST"
|
||||
|
||||
Reference in New Issue
Block a user