Pickaxe uses SQL statements combined with CSS selectors to pick out text from a web page. If you know SQL and a little about CSS selectors, this is the tool for you.
Pickaxe runs on linux, MacOS, and windows. Quickest way to get started is to run command below if docker is installed. To install locally, see installation instructions at bottom of page.
docker run -it bitsummation/pickaxe /bin/bash
Pickaxe is command line tool. Once installed or docker images is running you type pickaxe. If no arguments are given, it runs in interactive mode where code can be typed into the prompt and when a semicolon is typed the code is run. A file location to the source code can be passed or a url to source on the web. Run below to see if things are working.
pickaxe https://raw.githubusercontent.com/bitsummation/pickaxe/master/Examples/georgetown-airport.s
you should see a table like:
+---+-----+-------+------+-----------+----------+ | | day | time | wind | visibilty | weather | +---+-----+-------+------+-----------+----------+ | 1 | 21 | 11:56 | N 5 | 10.00 | Fog/Mist | | 2 | 21 | 10:56 | NE 3 | 10.00 | Fog/Mist | | 3 | 21 | 09:56 | Calm | 6.00 | Fog/Mist | | 4 | 21 | 08:56 | E 3 | 0.25 | Fog/Mist | | 5 | 21 | 07:56 | E 5 | 0.25 | Fog/Mist | | 6 | 21 | 06:53 | Calm | 0.75 | Fog/Mist | | 7 | 21 | 05:56 | Calm | 1.00 | Fog/Mist | +---+-----+-------+------+-----------+----------+
More examples are found here: https://github.com/bitsummation/pickaxe/tree/master/Examples
Pickaxe uses SQL like statements to select text from web pages. Instead of the SQL statements running against a database they run against live web pages.
Download page returns a table with columns url, nodes, date, size. The statement below downloads the entire web page but doesn't select any nodes in the where clause so not very useful.
select * from download page 'https://www.faa.gov/air_traffic/weather/asos/?state=TX'
Select the nodes we are interested in. To accomplish, set the nodes equal to a css expression. The css selector below gets all tr nodes that are under the table with class asos. In this case, multiple tr nodes are selected but nothing of interested is returned until we select nodes under each tr using the pick expression in the select statement.
select * from download page 'https://www.faa.gov/air_traffic/weather/asos/?state=TX' where nodes = 'table.asos tbody tr'
The pick expression picks out nodes under each node specified in the where clause. Pick takes a css selector. In this case, we are getting data in the td elements under each tr element. After the pick css selector, a part of the element can be specified.
select
pick 'td:nth-child(1) a' take attribute 'href' as details,
pick 'td:nth-child(1) a' as station,
pick 'td:nth-child(2)' as city,
pick 'td:nth-child(4)' as state
from download page 'https://www.faa.gov/air_traffic/weather/asos/?state=TX'
where nodes = 'table.asos tbody tr'
We create a memory table to store state strings then we insert states into it. The nested download select statement allows the download page statement to download multiple pages at once.
create buffer states(state string) insert into states select 'TX' insert into states select 'OR' select pick 'td:nth-child(1) a' take attribute 'href', --link to details pick 'td:nth-child(1) a', --station pick 'td:nth-child(2)', --city pick 'td:nth-child(4)' --state from download page (select 'https://www.faa.gov/air_traffic/weather/asos/?state=' + state from states) where nodes = 'table.asos tbody tr'Download Threads (make it faster)
A download page statement can use with (thread(2)) hint. The download page statement will then use the number of threads specified to download the pages resulting in much better performance.
create buffer states(state string) insert into states select 'TX' insert into states select 'OR' select pick 'td:nth-child(1) a' take attribute 'href', --link to details pick 'td:nth-child(1) a', --station pick 'td:nth-child(2)', --city pick 'td:nth-child(4)' --state from download page (select 'https://www.faa.gov/air_traffic/weather/asos/?state=' + state from states) with (thread(2)) where nodes = 'table.asos tbody tr'
Must be first statement in program. If the expression in the test block returns any rows, the proxy is considered good and all http requests will be routed through it. If more than one passes they are used in Round-robin fashion.
proxies ('104.156.252.188:8080', '75.64.204.199:8888', '107.191.49.249:8080')
with test {
select
pick '#tagline' take text
from download page 'http://vtss.brockreeve.com/'
}
Specify type as identity and it will auto increment.
create buffer temp(id identity, name string) insert into temp select 'test' insert into temp select 'test2' select * from temp
There are three different ways to store results. In memory, files, and sql databases. An example of each is listed below.
Store results in memory. The insert overwrite statement overwrites existing data in the buffer--if any--while insert into just appends to existing data.
create buffer results(type string, folder string, message string, changeDate string)
insert overwrite results
select
case pick '.icon .octicon-file-text'
when null then 'Folder'
else 'File'
end, --folder/file
pick '.content a', --name
pick '.message a', --comment
pick '.age span' --date
from download page 'https://github.com/bitsummation/pickaxe'
where nodes = 'table.files tr.js-navigation-item'
select *
from results
Store results into a file.
create file results(type string, folder string, message string, changeDate string)
with (
fieldterminator = '|',
rowterminator = '\r\n'
)
location 'C:\windows\temp\results.txt'
insert into results
select
case pick '.icon .octicon-file-text'
when null then 'Folder'
else 'File'
end, --folder/file
pick '.content a', --name
pick '.message a', --comment
pick '.age span' --date
from download page 'https://github.com/bitsummation/pickaxe'
where nodes = 'table.files tr.js-navigation-item'
Store results in Microsoft SQL Server. The mssql buffer definition must match the sql table structure.
create mssql results(type string, folder string, message string, changeDate string)
with (
connectionstring = 'Server=localhost;Database=scrape;Trusted_Connection=True;',
dbtable = 'Results'
)
insert into results
select
case pick '.icon .octicon-file-text'
when null then 'Folder'
else 'File'
end, --folder/file
pick '.content a', --name
pick '.message a', --comment
pick '.age span' --date
from download page 'https://github.com/bitsummation/pickaxe'
where nodes = 'table.files tr.js-navigation-item'
If a page renders the HTML client side with javascript a simple js hint is all that is needed. Only use if needed as performance is slower.
select pick '.main-link a' take attribute 'href', pick '.main-link a', pick '.posts span', --replies pick '.views span' --views from download page 'https://try.discourse.org/' with (js) where nodes = 'tr.topic-list-item'
For more control, specify the HTML element to wait for -- will wait until javscript renders the element -- and the time to wait before timing out (in seconds).
select
pick '.main-link a' take attribute 'href',
pick '.main-link a',
pick '.posts span', --replies
pick '.views span' --views
from download page 'https://try.discourse.org/' with (js('tr.topic-list-item', 5))
where nodes = 'tr.topic-list-item'
Run Javascript on the downloaded pages. Must return a javascript object or array of objects. The url variable is given by the framework and stores the url of the downloaded page.
select upc, u
from download page 'https://www.walmart.com/ip/Cheerios-Family-Size-Gluten-Free-Cereal-21-oz/33886599' with (js) => (
"
var primaryProductId = __WML_REDUX_INITIAL_STATE__.product.primaryProduct;
var primaryProduct = __WML_REDUX_INITIAL_STATE__.product.products[primaryProductId];
return { upc:primaryProduct.upc, u:url };
"
)
select p.title, u.upc
from (
select
pick '.prod-ProductTitle div' as title,
url
from download page 'https://www.walmart.com/ip/Cheerios-Family-Size-Gluten-Free-Cereal-21-oz/33886599'
) p
join (
select upc, url
from download page 'https://www.walmart.com/ip/Cheerios-Family-Size-Gluten-Free-Cereal-21-oz/33886599' with (js) =>
(
"
var primaryProductId = __WML_REDUX_INITIAL_STATE__.product.primaryProduct;
var primaryProduct = __WML_REDUX_INITIAL_STATE__.product.products[primaryProductId];
return { upc:primaryProduct.upc, url:url };
"
)
) u on u.url = p.url
Capture the commit information from this page.
select
case pick '.icon .octicon-file-text'
when null then 'Folder'
else 'File'
end as type, --folder/file
pick '.content a' as name, --name
pick '.message a' as comment, --comment
pick '.age span' as date --date
from download page 'https://github.com/bitsummation/pickaxe'
where nodes = 'table.files tr.js-navigation-item'
What's your WAN ip address?
select pick '#section_left div:nth-child(2) a' from download page 'http://whatismyipaddress.com/'
The match expression uses regular expressions to match text. In this case, it is just taking the numbers of the ip address.
select
pick '#section_left div:nth-child(2) a' match '\d'
from download page 'http://whatismyipaddress.com'
A match expression can be followed by a replace. In this case, we replace the dots with dashes.
select
pick '#section_left div:nth-child(2) a' match '\.' replace '---'
from download page 'http://whatismyipaddress.com'
Update table.
create buffer videos(video string, link string, title string, processed int) insert into videos select url, 'https://www.youtube.com' + pick '.content-wrapper a' take attribute 'href', pick '.content-wrapper a' take attribute 'title', 0 from download page 'https://www.youtube.com/watch?v=7NJqUN9TClM' where nodes = '#watch-related li.video-list-item' update videos set processed = 1 where link = 'https://www.youtube.com/watch?v=JYZMT8otKdI' select * from videos
insert file into 'C:\Windows\temp'
select
filename,
image
from download image 'http://brockreeve.com/image.axd?picture=2015%2f6%2fheadShot.jpg'
Use var to declare variable.
var startPage = 1 var endPage = 10
Used mostly to generate urls.
var startPage = 1
var endPage = 10
select
'http://example.com/p=' + value + '?t=All'
from expand (startPage to endPage){($*2) + 10}
Inner join tables.
create buffer a (id int, name string) create buffer b (id int, name string) insert into a select 1, 'first' insert into a select 2, 'first' insert into b select 1, 'second' select a.name, b.name from a join b on a.id = b.id
Case statement in select.
select
case when value < 20
then 'small' else 'large'
end
from expand (1 to 10){($*2) + 10}
Allows looping through tables.
create buffer temp(baseUrl string, startPage int, endPage int)
insert into temp
select 'http://test.com', 0, 10
insert into temp
select 'http://temp.com', 2, 10
create buffer urls(url string)
each(var t in temp){
insert into urls
select t.baseUrl + value + '?t=All'
from expand (t.startPage to t.endPage){($*2) + 10}
}
select *
from urls
The Pickaxe-Console.zip contains the command line version. Run pickaxe.exe without any arguments to run in interactive mode. Type in statements and run them by ending statement with a semicolon. Pass the file path to run the program. Command line arguments can be passed to the script.
pickaxe c:\test.s arg1 arg2
Get command line values in program. The ?? is used to assign a default value if args are not passed on command line.
@1 = @1 ?? 'first' @2 = @2 ?? 'second'Installation Instructions
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4