When you need to convert an html file to a pdf, often the default pdf printer of your OS is enough. For those times when you need a much better rendering, you need a better tool.

I had to convert my html/css resume to a pdf file today. The default pdf printer on Ubuntu was discarding the background color and adding useless margins.

I finally resorted to using wkhtmltopdf. It does the job, but needs a few tweaks. First, you have to add -T 0 -R 0 -B 0 -L 0 to remove the margins. Then, you have to use the file:// notation to target a local file. Also, it does not understand the css properties page-break-after nor page- break-before. I had to manually add padding in my elements to stop them from being cut in between two pages. But worst of all, it chokes on UTF-8 characters in filenames.

In the end, I wrote a small ruby wrapper around wkhtmltopdf to work around all those issues for me and called it [html2pdf](https://github.com/pixelastic/oro shi/blob/master/scripts/bin/html2pdf).

Everytime I install Ubuntu on a fresh computer and try to define custom keybindings that use the meta (ie. Windows) key, I run into issues.

The Ubuntu UI uses the meta key to open what they call the "Dash". I never use it and want my meta key back to define my own keybindings.

There is no way to disable it from the default Ubuntu options. You have to install compizconfig-settings-manager and launch it using ccsm. There, go to the Ubuntu Unity Plugin and disable the "Key to show the launcher" binding.

You will now be free to use the meta key in your keybindings, again.

Ever deleted a file several commits away and then suddenly realizing that you actually need it? Well, here's an easy way to get your file back.

First of all, we'll get the commit in which the file has been deleted with: git log --diff-filter=D --format=format:%H -- {your_filename}

--diff-filter=D will only keep commits where files gets deleted, --format=format:%H will only display the commit hash and finally the -- is a separator between your options and the argument. Argument should be your filename, and it even accepts * as a wildcard.

Once you got the commit hash, it's just a matter of checking the file out using git checkout {hash}~1 {your_filename}. The ~1 targets the previous commit, right before the file gets deleted.

All wrapped up in a nice shell script, here is what it looks like:

#!/usr/bin/env bash
filename=$1
# We first find the commit where the file was deleted
hash=$(git log --diff-filter=D --format=format:%H -- $filename)
# We then bring it back from the dead
git checkout $hash~1 $filename

I've aliased mine to git resurrect.

When moving files from my hard-drive (formatted in ext3) to my mp3 music player or to any drive formatted in FAT32, some files never get to the destination due to illegal characters in their filename.

Those includes characters like question marks, colons or quotes which can be quite common in media filenames.

So I wrote myself a [tiny command-line script](https://github.com/pixelastic/oroshi/blob/master/scripts/bin/rename- fat32) to make the filenames FTA32 compliant. Note that this is a destructive operation as it will simply remove the illegal characters.

Converting to and from charset encoding is boring. Life would be much easier if everything was saved in proper UTF-8.

I just coded a couple of scripts that will help be read file encoding from files and convert them to UTF-8.

First, not all encoding can be easily read. Some uses markers that are easy to check, other requires heuristic and guessing. I won't deal with the second part. Most of the files I'm currently handling are either Latin1 (ISO-8859-1) or Latin2 (ISO-8859-2), so I'll limit the scope of my scripts to those two sets.

Fortunatly, the default file command can do that.

$ file -bi file.txt
text/plain; charset=us-ascii
$ file -bi bad.html
text/html; charset=iso-8859-1
$ file -bi good.html
text/html; charset=utf-8

Now that we have a way to know the input encoding, we can use recode to convert to UTF-8. You can install recode easily with sudo apt-get install recode.

Once done, it is just a matter of convert latin1..utf8 bad.html

I wrapped those methods in two scripts: [encoding](https://github.com/pixelas tic/oroshi/blob/master/scripts/bin/encoding) and utf8 that respectively output the file encoding and convert the file to utf8.