A command line helper for curl and web scraping.
The current version of curlyq is 0.0.16.
CurlyQ is a utility that provides a simple interface for curl, with additional features for things like extracting images and links, finding elements by CSS selector or XPath, getting detailed header info, and more. It’s designed to be part of a scripting pipeline, outputting everything as structured data (JSON or YAML). It also has rudimentary support for making calls to JSON endpoints easier, but it’s expected that you’ll use something like jq to parse the output.
Assuming you have Ruby and RubyGems installed, you can just run gem install curlyq. If you run into errors, try gem install --user-install curlyq, or use sudo gem install curlyq.
If you’re using Homebrew, you have the option to install via brew-gem:
brew install brew-gem
brew gem install curlyqIf you don’t have Ruby/RubyGems, you can install them pretty easily with Homebrew, rvm, or asdf.
Run curlyq help for a list of subcommands. Run curlyq help SUBCOMMAND for details on a particular subcommand and its options.
NAME
curlyq - A scriptable interface to curl
SYNOPSIS
curlyq [global options] command [command options] [arguments...]
VERSION
0.0.16
GLOBAL OPTIONS
--help - Show this message
--[no-]pretty - Output "pretty" JSON (default: enabled)
--version - Display the program version
-y, --[no-]yaml - Output YAML instead of json
COMMANDS
execute - Execute JavaScript on a URL
extract - Extract contents between two regular expressions
headlinks - Return all <head> links on URL's page
help - Shows a list of commands or help for one command
html, curl - Curl URL and output its elements, multiple URLs allowed
images - Extract all images from a URL
json - Get a JSON response from a URL, multiple URLs allowed
links - Return all links on a URL's page
scrape - Scrape a page using a web browser, for dynamic (JS) pages. Be sure to have the selected --browser installed.
screenshot - Save a screenshot of a URL
tags - Extract all instances of a tagYou can shape the results using --search (-s) and --query (-q) on some commands.
A search uses either CSS or XPath syntax to locate elements. For example, if you wanted to locate all of the <article> elements with a class of post inside of the div with an id of main, you would run --search '#main article.post'. Searches can target tags, ids, and classes, and can accept > to target direct descendents. You can also use XPaths, but I hate those so I’m not going to document them.
I’ve tried to make the query function useful, but if you want to do any kind of advanced shaping, you’re better off piping the JSON output to jq or yq.
Queries are specifically for shaping CurlyQ output. If you’re using the html command, it returns a key called images, so you can target just the images in the response with -q 'images'. The queries accept array syntax, so to get the first image, you would use -q 'images[0]'. Ranges are accepted as well, so -q 'images[1..4]' will return the 2nd through 5th images found on the page. You can also do comparisons, e.g. images[rel=me]' to target only images with a rel attribute of me.
The comparisons for the query flag are:
< less than> greater than<= less than or equal to>= greater than or equal to= or == is equal to*= contains text^= starts with text$= ends with textComparisons can be numeric or string comparisons. A numeric comparison like curlyq images -q '[width>500]' URL would return all of the images on the page with a width attribute greater than 500.
You can also use dot syntax inside of comparisons, e.g. [links.rel*=me] to target the links object (html command), and return only the links with a rel=me attribute. If the comparison is to an array object (like class or rel), it will match if any of the elements of the array match your comparison.
If you end the query with a specific key, only that key will be output. If there’s only one match, it will be output as a raw string. If there are multiple matches, output will be an array:
$ curlyq tags --search '#main .post h3' -q '[attrs.id*=what].source' 'https://brettterpstra.com/2024/01/10/introducing-curlyq-a-pipeline-oriented-curl-helper/'
<h3 id="whats-next">What’s Next</h3>curlyq makes use of subcommands, e.g. curlyq html [options] URL or curlyq extract [options] URL. Each subcommand takes its own options, but I’ve made an effort to standardize the choices between each command as much as possible.
Example:
$ curlyq extract -i -b 'Adding' -a 'accessing the source.' 'https://stackoverflow.com/questions/52428409/get-fully-rendered-html-using-selenium-webdriver-and-python'
[
"Adding <code>time.sleep(10)</code> in various places in case the page had not fully loaded when I was accessing the source."
]
This specifies a before and after string and includes them (-i) in the result.
NAME
extract - Extract contents between two regular expressions
SYNOPSIS
curlyq [global options] extract [command options] URL...
COMMAND OPTIONS
-a, --after=arg - Text after extraction (default: none)
-b, --before=arg - Text before extraction (default: none)
-c, --[no-]compressed - Expect compressed results
--[no-]clean - Remove extra whitespace from results
-h, --header=arg - Define a header to send as key=value (may be used more than once, default: none)
-i, --[no-]include - Include the before/after matches in the result
-r, --[no-]regex - Process before/after strings as regular expressions
--[no-]strip - Strip HTML tags from resultsYou can execute JavaScript on a given web page using the execute subcommand.
Example:
curlyq execute -s "NiftyAPI.find('file/save').arrow().shoot('file-save')" file:///Users/ttscoff/Desktop/Code/niftymenu/dist/MultiMarkdown-Composer.htmlYou can specify an element id to wait for using --id, and define a pause to wait after executing a script with --wait (defaults to 2 seconds). Scripts can be read from the command line arguments with --script "SCRIPT", from STDIN with --script -, or from a file using --script PATH.
If you expect a return value, be sure to include a return statement in your executed script. Results will be output to STDOUT.
NAME
execute - Execute JavaScript on a URL
SYNOPSIS
curlyq [global options] execute [command options] URL...
COMMAND OPTIONS
-b, --browser=arg - Browser to use (firefox, chrome) (default: chrome)
-h, --header=arg - Define a header to send as key=value (may be used more than once, default: none)
-i, --id=arg - Element ID to wait for before executing (default: none)
-s, --script=arg - Script to execute, use - to read from STDIN (may be used more than once, default: none)
-w, --wait=arg - Seconds to wait after executing JS (default: 2)Example:
$ curlyq headlinks -q '[rel=stylesheet]' https://brettterpstra.com
{
"rel": "stylesheet",
"href": "https://cdn3.brettterpstra.com/stylesheets/screen.7261.css",
"type": "text/css",
"title": null
}
This pulls all <links> from the <head> of the page, and uses a query -q to only show links with rel="stylesheet".
NAME
headlinks - Return all <head> links on URL's page
SYNOPSIS
curlyq [global options] headlinks [command options] URL...
COMMAND OPTIONS
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)The html command (aliased as curl) gets the entire text of the web page and provides a JSON response with a breakdown of:
<head><body>You can add a query (-q) to only get the information needed, e.g. -q images[width>600].
Example:
$ curlyq html -s '#main article .aligncenter' -q 'images[1]' 'https://brettterpstra.com'
[
{
"class": "aligncenter",
"original": "https://cdn3.brettterpstra.com/uploads/2023/09/giveaway-keyboardmaestro2024-rb_tw.jpg",
"at2x": "https://cdn3.brettterpstra.com/uploads/2023/09/giveaway-keyboardmaestro2024-rb@2x.jpg",
"width": "800",
"height": "226",
"src": "https://cdn3.brettterpstra.com/uploads/2023/09/giveaway-keyboardmaestro2024-rb.jpg",
"alt": "Giveaway Robot with Keyboard Maestro icon",
"title": "Giveaway Robot with Keyboard Maestro icon"
}
]
The above example queries the full html of the page, but narrows the elements using --search and then takes the 2nd image from the results.
$ curlyq html -q 'meta.title' https://brettterpstra.com/2024/01/10/introducing-curlyq-a-pipeline-oriented-curl-helper/
Introducing CurlyQ, a pipeline-oriented curl helper - BrettTerpstra.comThe above example curls the page and returns the title attribute found in the meta (-q 'meta.title').
NAME
html - Curl URL and output its elements, multiple URLs allowed
SYNOPSIS
curlyq [global options] html [command options] URL...
COMMAND OPTIONS
-I, --info - Only retrieve headers/info
-b, --browser=arg - Use a browser to retrieve a dynamic web page (firefox, chrome) (default: none)
-c, --compressed - Expect compressed results
--[no-]clean - Remove extra whitespace from results
-f, --fallback=arg - If curl doesn't work, use a fallback browser (firefox, chrome) (default: none)
-h, --header=arg - Define a header to send as "key=value" (may be used more than once, default: none)
--[no-]ignore_fragments - Ignore fragment hrefs when gathering content links
--[no-]ignore_relative - Ignore relative hrefs when gathering content links
-l, --local_links_only - Only gather internal (same-site) links
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)
-r, --raw=arg - Output a raw value for a key (default: none)
-s, --search=arg - Regurn an array of matches to a CSS or XPath query (default: none)
-x, --external_links_only - Only gather external linksThe images command returns only the images on the page as an array of objects. It can be queried to match certain requirements (see Query and Search syntax above).
The base command will return all images on the page, including OpenGraph images from the head, <img> tags from the body, and <srcset> tags along with their child images.
OpenGraph images will be returned with the structure:
{
"type": "opengraph",
"attrs": null,
"src": "https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb_tw.jpg"
}
img tags will be returned with the structure:
{
"type": "img",
"src": "https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb.jpg",
"width": "800",
"height": "226",
"alt": "Banner image for CurlyQ",
"title": "CurlyQ, curl better",
"attrs": [
{
"class": [
"aligncenter"
], // all attributes included
}
]
}
srcset images will be returned with the structure:
{
"type": "srcset",
"attrs": [
{
"key": "srcset",
"value": "https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb_tw.jpg 1x, https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb@2x.jpg 2x"
}
],
"images": [
{
"src": "https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb_tw.jpg",
"media": "1x"
},
{
"src": "https://cdn3.brettterpstra.com/uploads/2024/01/curlyq_header-rb@2x.jpg",
"media": "2x"
}
]
}
Example:
$ curlyq images -t img -q '[alt$=screenshot]' https://brettterpstra.comThis will return an array of images that are <img> tags, and only show the ones that have an alt attribute that ends with screenshot.
$ curlyq images -q '[width>750]' https://brettterpstra.comThis example will only return images that have a width greater than 750 pixels. This query depends on the images having proper width attributes set on them in the source.
NAME
images - Extract all images from a URL
SYNOPSIS
curlyq [global options] images [command options] URL...
COMMAND OPTIONS
-c, --[no-]compressed - Expect compressed results
--[no-]clean - Remove extra whitespace from results
-h, --header=arg - Define a header to send as key=value (may be used more than once, default: none)
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)
-t, --type=arg - Type of images to return (img, srcset, opengraph, all) (may be used more than once, default: ["all"])The json command just returns an object with header/response info, and the contents of the JSON response after it’s been read by the Ruby JSON library and output. If there are fetching or parsing errors it will fail gracefully with an error code.
NAME
json - Get a JSON response from a URL, multiple URLs allowed
SYNOPSIS
curlyq [global options] json [command options] URL...
COMMAND OPTIONS
-c, --[no-]compressed - Expect compressed results
-h, --header=arg - Define a header to send as key=value (may be used more than once, default: none)
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)Returns all the links on the page, which can be queried on any attribute.
Example:
$ curlyq links -q '[content*=twitter]' 'https://stackoverflow.com/questions/52428409/get-fully-rendered-html-using-selenium-webdriver-and-python'
[
{
"href": "https://twitter.com/stackoverflow",
"title": null,
"rel": null,
"content": "Twitter",
"class": [
"-link",
"js-gps-track"
]
}
]
This example gets all links from the page but only returns ones with link content containing ‘twitter’ (-q '[content*=twitter]').
NAME
links - Return all links on a URL's page
SYNOPSIS
curlyq [global options] links [command options] URL...
COMMAND OPTIONS
-d, --[no-]dedup - Filter out duplicate links, preserving only first one
--[no-]ignore_fragments - Ignore fragment hrefs when gathering content links
--[no-]ignore_relative - Ignore relative hrefs when gathering content links
-l, --local_links_only - Only gather internal (same-site) links
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)
-x, --external_links_only - Only gather external linksLoads the page in a web browser, allowing scraping of dynamically loaded pages that return nothing but scripts when curled. The -b (--browser) option is required and should be ‘chrome’ or ‘firefox’ (or just ‘c’ or ‘f’). The selected browser must be installed on your system.
Example:
$ curlyq scrape -b firefox -q 'links[rel=me&content*=mastodon][0]' https://brettterpstra.com/2024/01/10/introducing-curlyq-a-pipeline-oriented-curl-helper/
{
"href": "https://nojack.easydns.ca/@ttscoff",
"title": null,
"rel": [
"me"
],
"content": "Mastodon",
"class": [
"u-url"
]
}
This example scrapes the page using firefox and finds the first link with a rel of ‘me’ and text containing ‘mastodon’.
NAME
scrape - Scrape a page using a web browser, for dynamic (JS) pages. Be sure to have the selected --browser installed.
SYNOPSIS
curlyq [global options] scrape [command options] URL...
COMMAND OPTIONS
-b, --browser=arg - Browser to use (firefox, chrome) (required, default: none)
--[no-]clean - Remove extra whitespace from results
-h, --header=arg - Define a header to send as "key=value" (may be used more than once, default: none)
-q, --query, --filter=arg - Filter output using dot-syntax path (default: none)
-r, --raw=arg - Output a raw value for a key (default: none)
--search=arg - Regurn an array of matches to a CSS or XPath query (default: none)Full-page screenshots require Firefox, installed and specified with --browser firefox.
Type defaults to full, but will only work if -b is Firefox. If you want to use Chrome, you must specify a --type as ‘visible’ or ‘print’.
The -o (--output) flag is required. It should be a path to a target PNG file (or PDF for -t print output). Extension will be modified automatically, all you need is the base name.
Example:
$ curlyq screenshot -b f -o ~/Desktop/test https://brettterpstra.com/2024/01/10/introducing-curlyq-a-pipeline-oriented-curl-helper/
Screenshot saved to /Users/ttscoff/Desktop/test.pngYou can wait for an element ID to be visible using --id. This can be any #ID on the page. If the ID doesn’t exist on the page, though, the screenshot will hang for a timeout of 10 seconds.
You can execute a script before taking the screenshot with the --script flag. If this is set to -, it will read the script from STDIN. If it’s set to an existing file path, that file will be read for script input. Specify an interval (in seconds) to wait after executing the script with --wait.
NAME
screenshot - Save a screenshot of a URL
SYNOPSIS
curlyq [global options] screenshot [command options] URL...
COMMAND OPTIONS
-b, --browser=arg - Browser to use (firefox, chrome) (default: chrome)
-h, --header=arg - Define a header to send as key=value (may be used more than once, default: none)
-i, --id=arg - Element ID to wait for before taking screenshot (default: none)
-o, --out, --file=arg - File destination (required, default: none)
-s, --script=arg - Script to execute before taking screenshot (may be used more than once, default: none)
-t, --type=arg - Type of screenshot to save (full (requires firefox), print, visible) (default: visible)
-w, --wait=arg - Time to wait before taking screenshot (default: 0)Return a hierarchy of all tags in a page. Use -t to limit to a specific tag.
$ curlyq tags –search ‘#main .post h3’ -q ‘[attrs.id*=what]’ https://brettterpstra.com/2024/01/10/introducing-curlyq-a-pipeline-oriented-curl-helper/
[
{
"tag": "h3",
"source": "<h3 id=\"whats-next\">What’s Next</h3>",
"attrs": [
{
"id": "whats-next"
}
],
"content": "What’s Next",
"tags": []
}
]
The above command filters the tags based on a CSS query, then further filters them to just tags with an id containing ‘what’.
NAME
tags - Extract all instances of a tag
SYNOPSIS
curlyq [global options] tags [command options] URL...
COMMAND OPTIONS
-c, --[no-]compressed - Expect compressed results
--[no-]clean - Remove extra whitespace from results
-h, --header=KEY=VAL - Define a header to send as key=value (may be used more than once, default: none)
-q, --query, --filter=DOT_SYNTAX - Dot syntax query to filter results (default: none)
--search=CSS/XPATH - Regurn an array of matches to a CSS or XPath query (default: none)
--[no-]source, --[no-]html - Output the HTML source of the results
-t, --tag=TAG - Specify a tag to collect (may be used more than once, default: none)2024-11-07 06:45
2024-10-25 10:31
2024-10-25 10:26
2024-10-25 10:23
2024-04-04 13:06
execute command for executing JavaScript on a page2024-01-21 15:29
2024-01-17 13:50
2024-01-16 12:38
[attrs.id*=what])2024-01-15 16:45
2024-01-12 17:03
2024-01-12 14:44
2024-01-11 18:06
2024-01-10 13:54
2024-01-10 13:38
2024-01-10 09:18
2024-01-10 08:20
Speaking of CurlyQ…
Brett is a writer and developer living in Minnesota, USA. You can follow him as ttscoff on Twitter, GitHub, Mastodon, and elsewhere.
Sign up for the email newsletter, and keep up with this blog by adding it to your favorite news reader, or following it on Bluesky or Mastodon at @brettterpstra.com@web.brid.gy.