Extract a single file from a zip file only knowing the extension

Question

My first part of the requirement:

I would like to extract a single file from ex1234.zip. The structure and contents of ex1234.zip:

ex1234 (directory)
    directory1
    ex1234 (directory)
    directory2
    ex1234.csv

I want to be able to extract only ex1234.csv file but will not know the name.

Second part is to able to do this for all exXXXX.zip that sit in the same directory.

ex1234.zip
ex3245.zip
ex8829.zip
exXXXX.zip…

Output will be:

ex1234.csv
ex3245.csv
ex8829.csv
exXXXX.csv

Real sample:

$ less CW2178470.zip
Archive:  CW2178470.zip
Zip file size: 26108 bytes, number of entries: 26
-rw----     2.0 fat      108 bl defN 15-Aug-04 09:37 CW2178470/CW2178470.csv
-rw----     2.0 fat     1363 bl defN 15-Aug-04 09:37 CW2178470/config/BusinessContactApprovers.csv
-rw----     2.0 fat      158 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/announcements.xml
-rw----     2.0 fat     1037 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/Plan/plan.xml
-rw----     2.0 fat      141 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/Plan/tasks.xml
-rw----     2.0 fat     2408 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/FI_Doc208411460_doc.xml
-rw----     2.0 fat      215 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/MessageBoard/nb_27482kst.26ihyzj_.htm
-rw----     2.0 fat     2364 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/MessageBoard/messageboard.xml
-rw----     2.0 fat     1250 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/team.xml
-rw----     2.0 fat    22016 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/Doc208411460.doc
-rw----     2.0 fat     9973 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/audithistory.xml
-rw----     2.0 fat     6731 bl defN 15-Aug-04 09:37 CW2178470/CW2178470/ws.xml
-rw----     2.0 fat      308 bl defN 15-Aug-04 09:37 CW2178470/xsd/WSFolder.xsd
-rw----     2.0 fat     4897 bl defN 15-Aug-04 09:37 CW2178470/xsd/Task.xsd
-rw----     2.0 fat      770 bl defN 15-Aug-04 09:37 CW2178470/xsd/ContractWorkspace.xsd
-rw----     2.0 fat     4754 bl defN 15-Aug-04 09:37 CW2178470/xsd/AuditHistory.xsd
-rw----     2.0 fat    25564 bl defN 15-Aug-04 09:37 CW2178470/xsd/CommonTypes.xsd
-rw----     2.0 fat     5657 bl defN 15-Aug-04 09:37 CW2178470/xsd/MessageBoard.xsd
-rw----     2.0 fat     2471 bl defN 15-Aug-04 09:37 CW2178470/xsd/Plan.xsd
-rw----     2.0 fat      337 bl defN 15-Aug-04 09:37 CW2178470/xsd/InternalContractWorkspace.xsd
-rw----     2.0 fat     1045 bl defN 15-Aug-04 09:37 CW2178470/xsd/SalesContractRequest.xsd
-rw----     2.0 fat     3133 bl defN 15-Aug-04 09:37 CW2178470/xsd/FolderItem.xsd
-rw----     2.0 fat      906 bl defN 15-Aug-04 09:37 CW2178470/xsd/ContractRequest.xsd
-rw----     2.0 fat     8973 bl defN 15-Aug-04 09:37 CW2178470/xsd/WorkspaceTypes.xsd
-rw----     2.0 fat     4645 bl defN 15-Aug-04 09:37 CW2178470/xsd/Team.xsd
-rw----     2.0 fat      781 bl defN 15-Aug-04 09:37 CW2178470/xsd/SalesContractWorkspace.xsd
26 files, 112005 bytes uncompressed, 21940 bytes compressed:  80.4%
(END)

Answer 1

You could use unzip like this:

unzip -j file[.zip] [file] [-x xfile]

where -j means junk paths, file[.zip] is your archive name, [file] is the archive member to be processed and [-x xfile] is the list of archive members to be excluded from processing. All these options are described in detail in the man page.
So in your case, running for example:

unzip -j ex1234.zip '*/*.csv' -x '*/*/*'

will extract in the current directory all files matching *.csv from depth level 2 in the ex1234.zip archive (excluding archive members from depth level 3 and below as '*/*/*' means paths that match at least two /).

Now, to process all the archives in the current directory you could run:

for zipfile in *.zip; do unzip -j "$zipfile" '*/*.csv' -x '*/*/*'; done

which extracts the .csv file from each archive in the current directory (that's why -j is needed).
In your particular case, there's no .csv on level 1 depth so you could also run:

for zipfile in *.zip; do unzip -j "$zipfile" '*.csv' -x '*/*/*'; done

which should yield the same result.
To dry-run and see which files will be extracted (their archive paths) without actually extracting them, replace -j with -qql:

for zipfile in *.zip; do unzip -qql "$zipfile" '*/*.csv' -x '*/*/*'; done

---

As a side note, the -j option could be omitted iff the .csv files to be extracted were on depth level 1 (i.e. no parent dir); in that case you could simply run:

for zipfile in *.zip; do unzip "$zipfile" '*.csv' -x '*/*'; done

Answer 2

Tried with unzip provided by Debian:

UnZip 6.00 of 20 April 2009, by Debian. Original by Info-ZIP.

for file in ex*.zip
do
  unzip -j $file '*.csv'
done

Answer 3

Use a FUSE-based filesystem to access the zip file as a directory tree. Mount each zip file, then access it with normal method (shell wildcards, cp command, etc.).

With fuse-zip:

mkdir mnt
for z in *.zip; do
  fuse-zip -- "$z" mnt
  set mnt/*.csv
  if [ $# -gt 1 ]; then
    echo "Skipping $z because it contains multiple .csv files"
  elif ! [ -e "$1" ]; then
    echo "Skipping $z because it does not contain a .csv file"
  else
    cp -- "$1" "${z%.zip}.csv"
  fi
  fusermount -u mnt
done

You can do the same with archivemount instead of fuse-zip.

There's also AVFS which works differently: it creates a view of the whole filesystem under ~/.avfs; in this view, if you have an archive file /path/to/foo.zip, you can access it as a directory under the name ~/.avfs/path/to/foo.zip#.

mountavfs
cd "$HOME/.avfs$PWD"
for z in *.zip; do
  set -- "$z#/"*.csv
  if [ $# -gt 1 ]; then
    echo "Skipping $z because it contains multiple .csv files"
  elif ! [ -e "$1" ]; then
    echo "Skipping $z because it does not contain a .csv file"
  else
    cp "$1" "${z%.zip}.csv"
  fi
done

If you use a shell with arrays and with a way to get an empty list when a wildcard doesn't match, you can get a slightly more readable script. For example, in ksh93, with fuse-zip:

#!/bin/ksh
mkdir mnt
for z in *.zip; do
  fuse-zip -- "$z" mnt
  csv=(~(N)"$z/"*.csv)
  if ((${#csv[@]} > 1)); then
    echo "Skipping $z because it contains multiple .csv files"
  elif ((${#csv[@]} == 0)); then
    echo "Skipping $z because it does not contain a .csv file"
  else
    cp -- "$1" "${z%.zip}.csv"
  fi
  fusermount -u mnt
done

In zsh, use csv=($z/*.csv(N)). In bash, use csv=($z/*.csv) but first run shopt -s nullglob.

Answer 4

You suggest that all the names have to match:

• The ZIP filename
• The directory name inside the ZIP file
• The CSV file inside the directory

In that case, try:

for zipfile in ./*.zip; do
    base="$(basename "$zipfile" .zip)"
    unzip "$zipfile" "$base/$base.csv"
done

Answer 5

On the assumption that all the files match this pattern - from CW2178470.zip you need to always extract CW2178470/CW2178470.csv

This is relatively easy:

for i in ./*.zip
do
   SERIAL=$(echo "$i" | sed -e 's,^.*/,,' -e 's,.zip$,,' )
   unzip "$i" "${SERIAL}/${SERIAL}.csv"
done

If you need smarter logic than that, I'd probably start looking at perl and Archive::Zip to extract.