如何解决wget镜像网站在Windows7的255字符路径限制问题？

Let's tackle this Windows 7 path length headache step by step— I've dealt with similar offline mirroring issues for legacy Windows systems, so here are actionable command-line solutions you can run directly on Debian:

First, let's refine your wget command to prevent overly deep directories in the first place. Your earlier -nH --cut-dirs=5 attempt failed because it flattened too much structure, causing index.html overlaps. Here's a balanced approach:

wget --mirror \
  --adjust-extension \
  --convert-links \
  --restrict-file-names=windows \
  --page-requisites \
  -e robots=off \
  --no-clobber \
  --no-parent \
  --base=./ \
  -nH \
  --cut-dirs=3 \  # Tweak this number based on the site's URL structure— start with 3 and test
  --directory-prefix=landesmuseum-mirror \  # Contains all files in a single root dir to avoid overwrites
  --no-if-modified-since \
  https://www.landesmuseum-mecklenburg.de/

• --cut-dirs=3: Removes the first 3 directory levels from the URL (e.g., www.landesmuseum-mecklenburg.de/de/exhibitions/long/path/ becomes exhibitions/long/path/). Adjust until your deepest paths stay under 200 characters (leaves buffer for Windows' 255 limit).
• For your extra resources in imagelist.txt, use this matching command to maintain the same structure:

wget -c -x -i imagelist.txt -P landesmuseum-mirror/ --restrict-file-names=windows

If you already have the mirrored files and don't want to re-download, use this two-step process to shorten paths with MD5 hashes and fix HTML links automatically:

Step 1: Rename long-path files to MD5 hashes #

Save this as rename-long-files.sh, make it executable, and run it to generate hashed filenames and log path changes:

#!/bin/bash
LOG_FILE="path-renames.log"
OUTPUT_DIR="landesmuseum-mirror"
MAX_PATH_LENGTH=200

# Find files with paths exceeding the limit, exclude directories
find "$OUTPUT_DIR" -type f -printf "%p\n" | while read -r FILE; do
  PATH_LENGTH=${#FILE}
  if [ "$PATH_LENGTH" -gt "$MAX_PATH_LENGTH" ]; then
    # Generate MD5 hash of the file's full path
    HASH=$(echo -n "$FILE" | md5sum | awk '{print $1}')
    # Keep the original file extension
    EXT="${FILE##*.}"
    NEW_FILE="$OUTPUT_DIR/$HASH.$EXT"
    # Rename and log the change
    mv "$FILE" "$NEW_FILE"
    echo "$FILE|$NEW_FILE" >> "$LOG_FILE"
    echo "Renamed: $FILE -> $NEW_FILE"
  fi
done

Execute with:

chmod +x rename-long-files.sh
./rename-long-files.sh

Step 2: Update HTML links to match new hashed paths #

Save this as update-links.sh, make it executable, and run to batch replace old paths in all HTML files:

#!/bin/bash
LOG_FILE="path-renames.log"
OUTPUT_DIR="landesmuseum-mirror"

# Read each log entry and replace old relative paths with new ones
while IFS="|" read -r OLD_PATH NEW_PATH; do
  OLD_REL=$(echo "$OLD_PATH" | sed "s|$OUTPUT_DIR/||")
  NEW_REL=$(echo "$NEW_PATH" | sed "s|$OUTPUT_DIR/||")
  # Replace in all HTML files
  find "$OUTPUT_DIR" -name "*.html" -exec sed -i "s|$OLD_REL|$NEW_REL|g" {} \;
done < "$LOG_FILE"

Execute with:

chmod +x update-links.sh
./update-links.sh

If you prefer DOS-style short names (e.g., longdirname → longdi~1), note this is less portable but works for Windows-only use:

Step 1: Generate DOS short names #

First install winepath with sudo apt install wine, then run:

find landesmuseum-mirror -type d -o -type f | while read -r PATH; do
  SHORT_PATH=$(winepath -w "$PATH" | sed 's|\\\\|/|g' | awk -F/ '{for(i=1;i<=NF;i++){if(length($i)>8){sub(/^(.{6}).*/,"\\1~1",$i)}};print "/" $0}')
  echo "$PATH|$SHORT_PATH" >> dos-short-names.log
done

Step 2: Update HTML links #

Use a similar sed script as Solution 2 to replace full paths with short paths in all HTML files.

After running any of these solutions, transfer the landesmuseum-mirror directory to your Windows 7 machine and test with MiniWebserver— the long path errors should be resolved.

内容的提问来源于stack exchange，提问作者Raphael