Some Bash Tricks

To open this blog, a few bash tricks used to manage it, that could be useful by occasion on many other cases. This blog is runned with jekyll, and I created some scripts to create posts, and create/remove categories (tags) associated to posts.

Iterate through properties in files

Each category is associated with a file in the _category folder. In each of those files, there is a line starting with category: followed by the category (name). We want to retrieve the file name and the category name it contains:

1
2
3
4
5
6
7
8
9
10

categories=()
category_files=()
for filename in ../_category/*; do
  if [ -f "${filename}" ]; then
    category_name=$( grep -m 1  '^category: ' ${filename} \
        | sed "s/^category: //g" | sed "s/ /-/g" )
    categories+=( ${category_name} )
    category_files+=( ${filename} )
  fi
done

On l.1-2 we initialize the array that will contain the category names and the files they are in. On l.3 we iterate through the files in the _category folder (and check for each one if it’s actually a file in the following line). The real trick is on l.5. There are 3 commands piped with another inside $( ):

  1. grep -m 1 '^category: ' ${filename}: Take the first line in the file starting with category
  2. sed "s/^category: //g": Remove category: at the beginning of the line, resp. replace it with empty string.
  3. sed "s/ /-/g: Replace spaces by - (otherwise it confuses the array) The following lines just append the extracted data to the lists and close the loop.

Check a user entry

When a user wants to remove a category, the categories are listed with an associated number:

1
2
3

-first category [0]
-category 2 [1]
- etc.

There are 3 constraints for the user entry:

  1. it must be a number (an integer)
  2. it must be larger or equal 0
  3. it must be less or equal the number of categories

This is handled in l.4 of the following snippet:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

article_categories=()
while true; do
  read entry
  if [[ "$entry" =~ ^[0-9]+$ ]] \
        && [[ "$(( entry ))" -ge 0 ]] \
        && [[ "$(( entry ))" -lt ${#categories[@]} ]]; then
    category_to_remove=${categories[$(( entry ))]}
    file_to_remove=${category_files[$(( entry ))]}
    break;
  else
    echo "invalid entry: must be an integer \
            between 0 and $(( ${#categories[@]} - 1 ))"
  fi
done

The first condition in if checks by regular expression that $entry has the form of an integer (must be done, since also letters can be interpreted as numbers). The second checks the second contraint, same for the third.

Checking and transforming strings

When a user enters a name for a category, this name is processed to generate category_name, title_name and file_name

1
2
3
4
5
6
7
8
9
10
11
12
13
14

read name
category_name=$( echo ${name} \
    | sed 's/[^a-zA-Z0-9 ]\+/-/g' \
    | sed 's/[ ]\+/ /g' \
    | sed 's/[ ]\+-//g' )
title_name=$( echo ${category_name} \
    | sed 's/[ ]\+/-/g' \
    | tr [:upper:] [:lower:]  )
file_name=$( echo ${title_name} \
    | iconv -f utf8 -t ascii//TRANSLIT//IGNORE )
if [[ -z $( echo ${title_name} | tr -d '-' ) ]]; then
  echo "category name must contain at least \
        one alphanumerical character, \
        please enter another name:"

On l.2 we have sed 's/[^a-zA-Z0-9 ]\+/-/g' replacing all non-alphanumerical characters by -, followed by sed 's/[ ]\+/ /g' replacing all multiple spaces by single spaces, itself followed by sed 's/[ ]\+-//g' removing - (and the spaces before) as it can occur at the end of the string. For example, Ca&tegory name would be transformed in Category name.

On l.3 the spaces are replaced by - and the letter in upper case by lower cases. Category name gets category-name. And on l.4, the result of the previous transformation is formatted to be compatible with a file name (format conversion).

If the entered category name was empty or contained just non-alphanumerical characters, then $title_name is empty or contains just - characters, and the user has to enter another name.

Check if a date is in the valid format

When a user creates an article, the default date is today. However s/he can enter another date if wished, but this date has to be in the format ‘YYYY-MM-DD’. The trick here is to try to transform the entry to a date in this format, and if it fails, it means the entry was not in the right format:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

default_date=$(date +"%Y-%m-%d")
article_date=$default_date
while true; do
  echo Enter date \(default: ${default_date}\):
  read line
  if [ ! -z "$line" ]; then
    if date -d "$line" +"%Y-%m-%d" > /dev/null 2>&1; then
      article_date=$( date -d "$line" +"%Y-%m-%d" )
      break
    else
      echo invalid date format
    fi
  else
    break
  fi
done

On l.1-2, the default date (today) is created, and assigned per default to the article date. If the user entered an non-empty string (l.6), then we try to transform it to a date of the expected format (l.7). If this successes, the entry (formatted) is assigned to the article date.

The complete scripts are available in a Github repo folder. Hope it has been useful for you, and if you have any question, don’t hesitate to ask.

Share this: