|
1 | 1 | # Introduction to Here Documents |
2 | 2 |
|
3 | | -TODO |
| 3 | +In Bash scripting, a "here document" (or "heredoc") redirects multiple lines of input to a command or program, as if you were typing them directly into the terminal. |
| 4 | +It's a powerful tool for embedding multi-line text within your scripts without needing external files or complex string manipulation. |
| 5 | + |
| 6 | +## Key Features and Syntax |
| 7 | + |
| 8 | +1. Delimiter: a heredoc starts with the `<<` operator followed by a delimiter word (often called the "marker" or "terminator"). |
| 9 | + This delimiter can be any word you choose, but it's common to use something like `EOF`, `END`, or `TEXT` for clarity. |
| 10 | + For more readable code, you can use something descriptive as the delimiter, for example `END_INSTALLATION_INSTRUCTIONS`. |
| 11 | + |
| 12 | +1. Content: after the initial `<< DELIMITER`, you write the content you want to redirect. |
| 13 | + This can be multiple lines of text, code, or anything else. |
| 14 | + |
| 15 | +1. Termination: the heredoc ends when the delimiter word appears again on a line by itself, with no leading or trailing whitespace. |
| 16 | + |
| 17 | +## Basic Syntax |
| 18 | + |
| 19 | +```bash |
| 20 | +command << DELIMITER |
| 21 | + Content line 1 |
| 22 | + Content line 2 |
| 23 | + ... |
| 24 | + Content line N |
| 25 | +DELIMITER |
| 26 | +``` |
| 27 | + |
| 28 | +## How it Works |
| 29 | + |
| 30 | +* Bash reads all the lines between the starting `<< DELIMITER` and the ending `DELIMITER`. |
| 31 | +* Bash connects this content to the command's standard input. |
| 32 | +* The command processes this input as if it were coming from the keyboard. |
| 33 | + |
| 34 | +### Example 1: Simple Text Output |
| 35 | + |
| 36 | +```bash |
| 37 | +cat << EOF |
| 38 | +This is the first line. |
| 39 | +This is the second line. |
| 40 | +This is the third line. |
| 41 | +EOF |
| 42 | +``` |
| 43 | + |
| 44 | +Output: |
| 45 | + |
| 46 | +```plaintext |
| 47 | +This is the first line. |
| 48 | +This is the second line. |
| 49 | +This is the third line. |
| 50 | +``` |
| 51 | + |
| 52 | +In this example: |
| 53 | + |
| 54 | +* `cat` is the command. |
| 55 | +* `<< EOF` starts the heredoc with `EOF` as the delimiter. |
| 56 | +* The three lines of text are the content. |
| 57 | +* `EOF` on its own line ends the heredoc. |
| 58 | +* `cat` then outputs the content it received. |
| 59 | + |
| 60 | +### Example 2: Using with wc (Word Count) |
| 61 | + |
| 62 | +```bash |
| 63 | +wc -l << END |
| 64 | +Line 1 |
| 65 | +Line 2 |
| 66 | +Line 3 |
| 67 | +END |
| 68 | +``` |
| 69 | + |
| 70 | +Output: |
| 71 | + |
| 72 | +```plaintext |
| 73 | +3 |
| 74 | +``` |
| 75 | + |
| 76 | +Here, `wc -l` counts the number of lines. |
| 77 | +The heredoc provides the three lines as input. |
| 78 | + |
| 79 | +### Example 3: Passing data to a script |
| 80 | + |
| 81 | +The script: |
| 82 | + |
| 83 | +```bash |
| 84 | +#!/usr/bin/env bash |
| 85 | + |
| 86 | +# Script to process input |
| 87 | +while IFS= read -r line; do |
| 88 | + echo "Processing: $line" |
| 89 | +done |
| 90 | +``` |
| 91 | + |
| 92 | +Call the script from an interactive bash prompt with a heredoc: |
| 93 | + |
| 94 | +```bash |
| 95 | +./your_script << MY_DATA |
| 96 | +Item 1 |
| 97 | +Item 2 |
| 98 | +Item 3 |
| 99 | +MY_DATA |
| 100 | +``` |
| 101 | + |
| 102 | +Output: |
| 103 | + |
| 104 | +```plaintext |
| 105 | +Processing: Item 1 |
| 106 | +Processing: Item 2 |
| 107 | +Processing: Item 3 |
| 108 | +``` |
| 109 | + |
| 110 | +## Variations and Advanced Features |
| 111 | + |
| 112 | +### Literal Content |
| 113 | + |
| 114 | +Bash performs variable expansion, command substitution, and arithmetic expansion within a heredoc. |
| 115 | +In this sense, heredocs act like double quoted strings. |
| 116 | + |
| 117 | +```bash |
| 118 | +cat << EOF |
| 119 | +The value of HOME is $HOME |
| 120 | +The current date is $(date) |
| 121 | +Two plus two is $((2 + 2)) |
| 122 | +EOF |
| 123 | +``` |
| 124 | + |
| 125 | +Output: |
| 126 | + |
| 127 | +```plaintext |
| 128 | +The value of HOME is /home/glennj |
| 129 | +The current date is Thu Apr 24 13:47:32 EDT 2025 |
| 130 | +Two plus two is 4 |
| 131 | +``` |
| 132 | + |
| 133 | +When the delimiter is quoting (using single or double quotes), these expansions are prevented. |
| 134 | +The content is taken literally. |
| 135 | +This is like single quoted strings. |
| 136 | + |
| 137 | +```bash |
| 138 | +cat << 'EOF' |
| 139 | +The value of $HOME is not expanded here. |
| 140 | +The result of $(date) is not executed. |
| 141 | +Two plus two is calculated by $((2 + 2)) |
| 142 | +EOF |
| 143 | +``` |
| 144 | + |
| 145 | +Output: |
| 146 | + |
| 147 | +```plaintext |
| 148 | +The value of $HOME is not expanded here. |
| 149 | +The result of $(date) is not executed. |
| 150 | +Two plus two is calculated by $((2 + 2)) |
| 151 | +``` |
| 152 | + |
| 153 | +### Stripping Leading Tabs |
| 154 | + |
| 155 | +If you use `<<-` (with a trailing hyphen) instead of `<<`, Bash will strip any leading _tab characters_ from each line of the heredoc. |
| 156 | +This is useful for indenting the heredoc content within your script without affecting the output. |
| 157 | + |
| 158 | +```bash |
| 159 | +# Note, the leading whitespace is tab characters only, not spaces! |
| 160 | +# The ending delimiter can have leading tabs as well. |
| 161 | +cat <<- END |
| 162 | + This line has 1 leading tab. |
| 163 | + This line has a leading tab and some spaces. |
| 164 | + This line 2 leading tabs. |
| 165 | + END |
| 166 | +``` |
| 167 | + |
| 168 | +The output is printed with all the leading tabs removed: |
| 169 | + |
| 170 | +```plaintext |
| 171 | +This line has 1 leading tab. |
| 172 | + This line has a leading tab and some spaces. |
| 173 | +This line 2 leading tabs. |
| 174 | +``` |
| 175 | + |
| 176 | +~~~~exercism/caution |
| 177 | +The author doesn't recommend this usage. |
| 178 | +While it can improve the readability of the script, |
| 179 | +
|
| 180 | +1. it's easy to accidentally replace the tab characters with spaces (your editor may do this automatically), and |
| 181 | +1. it's hard to spot the difference between spaces and tabs. |
| 182 | +~~~~ |
| 183 | + |
| 184 | +## When to Use Here Documents |
| 185 | + |
| 186 | +* Multi-line input: when you need to provide multiple lines of text to a command. |
| 187 | +* Configuration files: embedding small configuration snippets within a script. |
| 188 | +* Generating code: creating code on the fly within a script. |
| 189 | +* Scripting interactions: simulating user input for interactive programs. |
| 190 | +* Avoiding external files: when you want to avoid creating temporary files. |
| 191 | + |
| 192 | +A typical usage might be to provide some help text: |
| 193 | + |
| 194 | +```bash |
| 195 | +#!/usr/bin/env bash |
| 196 | + |
| 197 | +usage() { |
| 198 | + cat << END_USAGE |
| 199 | +Refresh database tables. |
| 200 | +
|
| 201 | +usage: ${0##*/} [-h|--help] [-A|--no-archive] |
| 202 | +
|
| 203 | +where: --no-archive flag will _skip_ the archive jobs |
| 204 | +END_USAGE |
| 205 | +} |
| 206 | + |
| 207 | +# ... parsing command line options here ... |
| 208 | + |
| 209 | +if [[ $flag_help == "true" ]]; then |
| 210 | + usage |
| 211 | + exit 0 |
| 212 | +fi |
| 213 | +``` |
| 214 | + |
| 215 | +## Possible Drawbacks |
| 216 | + |
| 217 | +* Large embedded documents can make your code harder to read. |
| 218 | + It can be better to deploy your script with documentation in separate files. |
| 219 | +* Here documents can break the flow of the code. |
| 220 | + You might be in a deeply nested section of code when you want to pass some text to a program. |
| 221 | + The heredoc's indentation can look jarring compared to the surrounding code. |
| 222 | + |
| 223 | +## Here Strings |
| 224 | + |
| 225 | +Like here documents, _here strings_ (or "herestrings") provide input to a command. |
| 226 | +However, while heredocs are given as a block of text, herestrings are given as a single string of text. |
| 227 | +Here strings use the `<<< "text"` syntax. |
| 228 | + |
| 229 | +```bash |
| 230 | +tr 'a-z' 'A-Z' <<< "upper case this string" |
| 231 | +``` |
| 232 | + |
| 233 | +Output: |
| 234 | + |
| 235 | +```plaintext |
| 236 | +UPPER CASE THIS STRING |
| 237 | +``` |
| 238 | + |
| 239 | +Unlike heredocs, no ending delimiter is required. |
| 240 | + |
| 241 | +### Why Use Here Strings? |
| 242 | + |
| 243 | +A pipeline can be used instead of a here string: |
| 244 | + |
| 245 | +```bash |
| 246 | +echo "upper case this string" | tr 'a-z' 'A-Z' |
| 247 | +``` |
| 248 | + |
| 249 | +So why use a here string? |
| 250 | + |
| 251 | +Consider the case where you get the string as output from a long-running computation, and you want to feed the result to two separate commands. |
| 252 | +Using pipelines, you have to execute the computation twice: |
| 253 | + |
| 254 | +```bash |
| 255 | +some_long_running_calculation | first_command |
| 256 | +some_long_running_calculation | second_command |
| 257 | +``` |
| 258 | + |
| 259 | +A more efficient approach is to capture the output of the computation (using command substutition), and use here strings to provide input to the two subsequent commands: |
| 260 | + |
| 261 | +```bash |
| 262 | +result=$( some_long_running_calculation ) |
| 263 | +first_command <<< "$result" |
| 264 | +second_command <<< "$result" |
| 265 | +``` |
| 266 | + |
| 267 | +Here's a real-world application of that example: |
| 268 | + |
| 269 | +* capture the JSON response to a REST API query (that is paginated), |
| 270 | +* provide the JSON data to a jq program to parse the results and output that to a file, and then |
| 271 | +* provide the JSON data to another jq program to determine the URL of the next query. |
| 272 | + |
| 273 | +```bash |
| 274 | +# initialize the output CSV file |
| 275 | +echo "ID,VALUE" > data.csv |
| 276 | + |
| 277 | +url='https//example.com/api/query?page=1' |
| 278 | + |
| 279 | +while true; do |
| 280 | + json=$( curl "$url" ) |
| 281 | + |
| 282 | + # convert the results part of the response into CSV |
| 283 | + jq -r '.results[] | [.id, .value] | @csv' <<< "$json" |
| 284 | + |
| 285 | + # get the URL for the next page |
| 286 | + url=$( jq -r '.next_url // ""' <<< "$json" ) |
| 287 | + if [[ "$url" == "" ]]; then |
| 288 | + break |
| 289 | + fi |
| 290 | +done >> data.csv |
| 291 | +``` |
| 292 | + |
| 293 | +Note the position of the output redirection. |
| 294 | +All output from the while loop will be appended to the file `data.csv`. |
| 295 | + |
| 296 | +## Heredocs and Herestrings as Redirection |
| 297 | + |
| 298 | +Because these are just forms of redirection, they can be combined with other redirection operations: |
| 299 | + |
| 300 | +```bash |
| 301 | +cat <<< END_OF_TEXT > output.txt |
| 302 | +This is my important text. |
| 303 | +END_OF_TEXT |
| 304 | + |
| 305 | +awk '...' <<< "$my_var" >> result.csv |
| 306 | +``` |
| 307 | + |
| 308 | +## In Summary |
| 309 | + |
| 310 | +Here documents (or "heredocs") are a flexible and convenient way to manage multi-line input in Bash scripts. |
| 311 | +They simplify the process of embedding text and data directly within your scripts, making them more self-contained and easier to read. |
| 312 | + |
| 313 | +Here strings (or "herestrings") are like here documents, but offer a simpler, more dynamic syntax. |
0 commit comments