Skip to content
A webmining CLI tool & library for python.
Python Other
  1. Python 98.9%
  2. Other 1.1%
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs
ftest
hooks Reorganizing hooks Oct 18, 2019
img
minet Crawler State Oct 18, 2019
requirements
scripts
test
.gitignore
.travis.yml Windows... Oct 15, 2019
Makefile
README.md Adding insecure to multithreaded fetcher & resolver Oct 18, 2019
README.template.md Adding timeout kwargs to multithreaded fetchers Oct 18, 2019
setup.py

README.md

Build Status

Minet

minet is a webmining CLI tool & library for python. It adopts a lo-fi approach to various webmining problems by letting you perform a variety of actions from the comfort of your command line. No database needed: raw data files will get you going.

In addition, minet also exposes its high-level programmatic interface as a library so you can tweak its behavior at will.

Features

  • Multithreaded, memory-efficient fetching from the web.
  • Multiprocessed raw text content extraction from HTML pages.
  • Multiprocessed scraping from HTML pages using a comfy JSON DSL.
  • URL-related heuristics utilities such as normalization and matching.
  • Data collection from various APIs such as CrowdTangle.

Installation

minet can be installed using pip:

pip install minet

Usage

CLI

Global utilities

Basic commands

Platform-related commands

API


CLI

-h/--help

If you need help about a command, don't hesitate to use the -h/--help flag or the help command:

minet ct posts -h
# or:
minet ct posts --help
# or
minet help ct posts

To check the installed version of minet, you can use the --version flag:

minet --version
>>> minet x.x.x

fetch

usage: minet fetch [-h] [--compress] [--contents-in-report] [-d OUTPUT_DIR]
                   [-f FILENAME] [--filename-template FILENAME_TEMPLATE]
                   [-g {firefox,chrome}] [-H HEADERS] [--resume]
                   [--standardize-encoding] [-o OUTPUT] [-s SELECT] [-t THREADS]
                   [--throttle THROTTLE] [--total TOTAL]
                   [--url-template URL_TEMPLATE] [-X METHOD]
                   column [file]

Minet Fetch Command
===================

Use multiple threads to fetch batches of urls from a CSV file. The
command outputs a CSV report with additional metadata about the
HTTP calls and will generally write the retrieved files in a folder
given by the user.

positional arguments:
  column                                          Column of the CSV file containing urls to fetch.
  file                                            CSV file containing the urls to fetch.

optional arguments:
  -h, --help                                      show this help message and exit
  --compress                                      Whether to compress the contents.
  --contents-in-report, --no-contents-in-report   Whether to include retrieved contents, e.g. html, directly in the report
                                                  and avoid writing them in a separate folder. This requires to standardize
                                                  encoding and won't work on binary formats.
  -d OUTPUT_DIR, --output-dir OUTPUT_DIR          Directory where the fetched files will be written. Defaults to "content".
  -f FILENAME, --filename FILENAME                Name of the column used to build retrieved file names. Defaults to an uuid v4 with correct extension.
  --filename-template FILENAME_TEMPLATE           A template for the name of the fetched files.
  -g {firefox,chrome}, --grab-cookies {firefox,chrome}
                                                  Whether to attempt to grab cookies from your computer's browser.
  -H HEADERS, --header HEADERS                    Custom headers used with every requests.
  --resume                                        Whether to resume from an aborted report.
  --standardize-encoding                          Whether to systematically convert retrieved text to UTF-8.
  -o OUTPUT, --output OUTPUT                      Path to the output report file. By default, the report will be printed to stdout.
  -s SELECT, --select SELECT                      Columns to include in report (separated by `,`).
  -t THREADS, --threads THREADS                   Number of threads to use. Defaults to 25.
  --throttle THROTTLE                             Time to wait - in seconds - between 2 calls to the same domain. Defaults to 0.2.
  --total TOTAL                                   Total number of lines in CSV file. Necessary if you want to display a finite progress indicator.
  --url-template URL_TEMPLATE                     A template for the urls to fetch. Handy e.g. if you need to build urls from ids etc.
  -X METHOD, --request METHOD                     The http method to use. Will default to GET.

examples:

. Fetching a batch of url from existing CSV file:
    `minet fetch url_column file.csv > report.csv`

. CSV input from stdin:
    `xsv select url_column file.csv | minet fetch url_column > report.csv`

. Fetching a single url, useful to pipe into `minet scrape`:
    `minet fetch http://google.com | minet scrape ./scrape.json > scraped.csv`

extract

If you want to be able to use the extract command, you will need to install the dragnet library. Because it is a bit cumbersome to install, it's not included in minet's dependencies yet.

Just run the following & in the same order (dragnet needs to have specific deps installed before it can be able to compile its native files):

pip install lxml numpy Cython
pip install dragnet
usage: minet extract [-h] [-e {dragnet,html2text}] [-i INPUT_DIRECTORY]
                     [-o OUTPUT] [-p PROCESSES] [-s SELECT] [--total TOTAL]
                     [report]

Minet Extract Command
=====================

Use multiple processes to extract raw text from a batch of HTML files.
This command can either work on a `minet fetch` report or on a bunch
of files. It will output an augmented report with the extracted text.

positional arguments:
  report                                          Input CSV fetch action report file.

optional arguments:
  -h, --help                                      show this help message and exit
  -e {dragnet,html2text}, --extractor {dragnet,html2text}
                                                  Extraction engine to use. Defaults to `dragnet`.
  -i INPUT_DIRECTORY, --input-directory INPUT_DIRECTORY
                                                  Directory where the HTML files are stored. Defaults to "content".
  -o OUTPUT, --output OUTPUT                      Path to the output report file. By default, the report will be printed to stdout.
  -p PROCESSES, --processes PROCESSES             Number of processes to use. Defaults to 4.
  -s SELECT, --select SELECT                      Columns to include in report (separated by `,`).
  --total TOTAL                                   Total number of HTML documents. Necessary if you want to display a finite progress indicator.

examples:

. Extracting raw text from a `minet fetch` report:
    `minet extract report.csv > extracted.csv`

. Working on a report from stdin:
    `minet fetch url_column file.csv | minet extract > extracted.csv`

. Extracting raw text from a bunch of files:
    `minet extract --glob "./content/*.html" > extracted.csv`

scrape

TODO: document the scraping DSL

usage: minet scrape [-h] [-f {csv,jsonl}] [-g GLOB] [-i INPUT_DIRECTORY]
                    [-o OUTPUT] [-p PROCESSES] [--total TOTAL]
                    scraper [report]

Minet Scrape Command
====================

Use multiple processes to scrape data from a batch of HTML files.
This command can either work on a `minet fetch` report or on a bunch
of files. It will output the scraped items.

positional arguments:
  scraper                                         Path to a scraper definition file.
  report                                          Input CSV fetch action report file.

optional arguments:
  -h, --help                                      show this help message and exit
  -f {csv,jsonl}, --format {csv,jsonl}            Output format.
  -g GLOB, --glob GLOB                            Whether to scrape a bunch of html files on disk matched by a glob pattern rather than sourcing them from a CSV report.
  -i INPUT_DIRECTORY, --input-directory INPUT_DIRECTORY
                                                  Directory where the HTML files are stored. Defaults to "content".
  -o OUTPUT, --output OUTPUT                      Path to the output report file. By default, the report will be printed to stdout.
  -p PROCESSES, --processes PROCESSES             Number of processes to use. Defaults to 4.
  --total TOTAL                                   Total number of HTML documents. Necessary if you want to display a finite progress indicator.

examples:

. Scraping item from a `minet fetch` report:
    `minet scrape scraper.json report.csv > scraped.csv`

. Working on a report from stdin:
    `minet fetch url_column file.csv | minet scrape scraper.json > scraped.csv`

. Scraping a single page from the web:
    `minet fetch https://news.ycombinator.com/ | minet scrape scraper.json > scraped.csv`

. Scraping items from a bunch of files:
    `minet scrape scraper.json --glob "./content/*.html" > scraped.csv`

url-join

usage: minet url-join [-h] [-o OUTPUT] [-s SELECT] column1 file1 column2 file2

Minet Url Join Command
======================

Join two CSV files by matching them on columns containing urls. In
fact, the command will index the first file's urls into a
hierchical trie before attempting to match the second file's ones.

positional arguments:
  column1                     Name of the url column in the first file.
  file1                       Path to the first file.
  column2                     Name of the url column in the second file.
  file2                       Path to the second file.

optional arguments:
  -h, --help                  show this help message and exit
  -o OUTPUT, --output OUTPUT  Path to the output joined file. By default, the join will be printed to stdout.
  -s SELECT, --select SELECT  Columns from the first file to keep, separated by comma.

examples:

. Joining two files:
    `minet url-join url webentities.csv post_url posts.csv > joined.csv`

. Keeping only some columns from first file:
    `minet url-join url webentities.csv post_url posts.csv -s url,id > joined.csv`

url-parse

usage: minet url-parse [-h] [-o OUTPUT] [-s SELECT] [--separator SEPARATOR]
                       [--total TOTAL]
                       column [file]

Minet Url Parse Command
=======================

Overload a CSV file containing urls with a selection of additional
metadata such as their normalized version, domain name etc.

positional arguments:
  column                      Name of the column containing urls.
  file                        Target CSV file.

optional arguments:
  -h, --help                  show this help message and exit
  -o OUTPUT, --output OUTPUT  Path to the output file. By default, the result will be printed to stdout.
  -s SELECT, --select SELECT  Columns to keep in output, separated by comma.
  --separator SEPARATOR       Split url column by a separator?
  --total TOTAL               Total number of lines in CSV file. Necessary if you want to display a finite progress indicator.

examples:

. Creating a report about a file's urls:
    `minet url-report url posts.csv > report.csv`

. Keeping only selected columns from the input file:
    `minet url-report url posts.csv -s id,url,title > report.csv`

. Multiple urls joined by separator:
    `minet url-report urls posts.csv --separator "|" > report.csv`

CrowdTangle

usage: minet crowdtangle [-h] [--rate-limit RATE_LIMIT] [-o OUTPUT] [-t TOKEN]
                         {leaderboard,lists,posts,search} ...

Minet Crowdtangle Command
=========================

Gather data from the CrowdTangle APIs easily and efficiently.

optional arguments:
  -h, --help                        show this help message and exit
  --rate-limit RATE_LIMIT           Authorized number of hits by minutes. Defaults to 6.
  -o OUTPUT, --output OUTPUT        Path to the output file. By default, everything will be printed to stdout.
  -t TOKEN, --token TOKEN           CrowdTangle dashboard API token.

actions:
  {leaderboard,lists,posts,search}  Action to perform using the CrowdTangle API.

leaderboard

usage: minet crowdtangle leaderboard [-h] [--rate-limit RATE_LIMIT] [-o OUTPUT]
                                     [-t TOKEN] [--no-breakdown]
                                     [-f {csv,jsonl}] [-l LIMIT]
                                     [--list-id LIST_ID]

Minet CrowdTangle Leaderboard Command
=====================================

Gather information and aggregated stats about pages and groups of
the designated dashboard (indicated by a given token).

optional arguments:
  -h, --help                            show this help message and exit
  --rate-limit RATE_LIMIT               Authorized number of hits by minutes. Defaults to 6.
  -o OUTPUT, --output OUTPUT            Path to the output file. By default, everything will be printed to stdout.
  -t TOKEN, --token TOKEN               CrowdTangle dashboard API token.
  --no-breakdown                        Whether to skip statistics breakdown by post type in the CSV output.
  -f {csv,jsonl}, --format {csv,jsonl}  Output format. Defaults to `csv`.
  -l LIMIT, --limit LIMIT               Maximum number of posts to retrieve. Will fetch every post by default.
  --list-id LIST_ID                     Optional list id from which to retrieve accounts.

examples:

. Fetching accounts statistics for every account in your dashboard:
    `minet ct leaderboard --token YOUR_TOKEN > accounts-stats.csv`

lists

usage: minet crowdtangle lists [-h] [--rate-limit RATE_LIMIT] [-o OUTPUT]
                               [-t TOKEN]

Minet CrowdTangle Lists Command
===============================

Retrieve the lists from a CrowdTangle dashboard (indicated by a
given token).

optional arguments:
  -h, --help                  show this help message and exit
  --rate-limit RATE_LIMIT     Authorized number of hits by minutes. Defaults to 6.
  -o OUTPUT, --output OUTPUT  Path to the output file. By default, everything will be printed to stdout.
  -t TOKEN, --token TOKEN     CrowdTangle dashboard API token.

examples:

. Fetching a dashboard's lists:
    `minet ct lists --token YOUR_TOKEN > lists.csv`

posts

usage: minet crowdtangle posts [-h] [--rate-limit RATE_LIMIT] [-o OUTPUT]
                               [-t TOKEN] [--end-date END_DATE] [-f {csv,jsonl}]
                               [--language LANGUAGE] [-l LIMIT]
                               [--list-ids LIST_IDS]
                               [--partition-strategy PARTITION_STRATEGY]
                               [--resume]
                               [--sort-by {date,interaction_rate,overperforming,total_interactions,underperforming}]
                               [--start-date START_DATE]
                               [--url-report URL_REPORT]

Minet CrowdTangle Posts Command
===============================

Gather post data from the designated dashboard (indicated by
a given token).

optional arguments:
  -h, --help                                      show this help message and exit
  --rate-limit RATE_LIMIT                         Authorized number of hits by minutes. Defaults to 6.
  -o OUTPUT, --output OUTPUT                      Path to the output file. By default, everything will be printed to stdout.
  -t TOKEN, --token TOKEN                         CrowdTangle dashboard API token.
  --end-date END_DATE                             The latest date at which a post could be posted (UTC!).
  -f {csv,jsonl}, --format {csv,jsonl}            Output format. Defaults to `csv`.
  --language LANGUAGE                             Language of posts to retrieve.
  -l LIMIT, --limit LIMIT                         Maximum number of posts to retrieve. Will fetch every post by default.
  --list-ids LIST_IDS                             Ids of the lists from which to retrieve posts, separated by commas.
  --partition-strategy PARTITION_STRATEGY         Query partition strategy to use to overcome the API search result limits. Should either be `day` or a number of posts.
  --resume                                        Whether to resume an interrupted collection. Requires -o/--output & --sort-by date
  --sort-by {date,interaction_rate,overperforming,total_interactions,underperforming}
                                                  The order in which to retrieve posts. Defaults to `date`.
  --start-date START_DATE                         The earliest date at which a post could be posted (UTC!).
  --url-report URL_REPORT                         Path to an optional report file to write about urls found in posts.

examples:

. Fetching the 500 most latest posts from a dashboard:
    `minet ct posts --token YOUR_TOKEN --limit 500 > latest-posts.csv`

search

usage: minet crowdtangle posts [-h] [--rate-limit RATE_LIMIT] [-o OUTPUT]
                               [-t TOKEN] [--end-date END_DATE] [-f {csv,jsonl}]
                               [--language LANGUAGE] [-l LIMIT]
                               [--list-ids LIST_IDS]
                               [--partition-strategy PARTITION_STRATEGY]
                               [--resume]
                               [--sort-by {date,interaction_rate,overperforming,total_interactions,underperforming}]
                               [--start-date START_DATE]
                               [--url-report URL_REPORT]

Minet CrowdTangle Posts Command
===============================

Gather post data from the designated dashboard (indicated by
a given token).

optional arguments:
  -h, --help                                      show this help message and exit
  --rate-limit RATE_LIMIT                         Authorized number of hits by minutes. Defaults to 6.
  -o OUTPUT, --output OUTPUT                      Path to the output file. By default, everything will be printed to stdout.
  -t TOKEN, --token TOKEN                         CrowdTangle dashboard API token.
  --end-date END_DATE                             The latest date at which a post could be posted (UTC!).
  -f {csv,jsonl}, --format {csv,jsonl}            Output format. Defaults to `csv`.
  --language LANGUAGE                             Language of posts to retrieve.
  -l LIMIT, --limit LIMIT                         Maximum number of posts to retrieve. Will fetch every post by default.
  --list-ids LIST_IDS                             Ids of the lists from which to retrieve posts, separated by commas.
  --partition-strategy PARTITION_STRATEGY         Query partition strategy to use to overcome the API search result limits. Should either be `day` or a number of posts.
  --resume                                        Whether to resume an interrupted collection. Requires -o/--output & --sort-by date
  --sort-by {date,interaction_rate,overperforming,total_interactions,underperforming}
                                                  The order in which to retrieve posts. Defaults to `date`.
  --start-date START_DATE                         The earliest date at which a post could be posted (UTC!).
  --url-report URL_REPORT                         Path to an optional report file to write about urls found in posts.

examples:

. Fetching the 500 most latest posts from a dashboard:
    `minet ct posts --token YOUR_TOKEN --limit 500 > latest-posts.csv`

Facebook

usage: minet facebook [-h] {comments} ...

Minet Facebook Command
======================

Collects data from Facebook.

optional arguments:
  -h, --help  show this help message and exit

actions:
  {comments}  Action to perform to collect data on Facebook

comments

usage: minet facebook comments [-h] [-c COOKIE] [-o OUTPUT] url

Minet Facebook Comments Command
===============================

Scrape series of comments on Facebook.

positional arguments:
  url                         Url of the post from which to scrape comments.

optional arguments:
  -h, --help                  show this help message and exit
  -c COOKIE, --cookie COOKIE  Authenticated cookie to use or browser from which to extract it (support "firefox" and "chrome").
  -o OUTPUT, --output OUTPUT  Path to the output report file. By default, the report will be printed to stdout.

examples:

. Fetching a dashboard's lists:
    `minet fb comments`

Mediacloud

topic

stories

usage: minet mediacloud topic stories [-h] [-t TOKEN] [-o OUTPUT] topic_id

Minet Mediacloud Topic Stories Command
======================================

Retrieves the list of stories from a mediacloud topic.

positional arguments:
  topic_id                    Id of the topic

optional arguments:
  -h, --help                  show this help message and exit
  -t TOKEN, --token TOKEN     Mediacloud API token (also called key).
  -o OUTPUT, --output OUTPUT  Path to the output report file. By default, the report will be printed to stdout.


API

multithreaded_fetch

Function fetching urls in a multithreaded fashion.

from minet import multithreaded_fetch

# Most basic usage
urls = ['https://google.com', 'https://twitter.com']

for result in multithreaded_fetch(urls):
  print(result.url, result.response.status)

# Using a list of dicts

urls = [
  {
    'url': 'https://google.com',
    'label': 'Google'
  },
  {
    'url': 'https://twitter.com',
    'label': 'Twitter'
  }
]

for result in multithreaded_fetch(urls, key=lambda x: x['url']):
  print(result.item['label'], result.response.status)

Arguments:

  • iterator iterable: An iterator over urls or arbitrary items, if you provide a key argument along with it.
  • key ?callable: A function extracting the url to fetch from the items yielded by the provided iterator.
  • request_args ?callable: A function returning arguments to pass to the internal request helper for a call.
  • threads ?int [25]: Number of threads to use.
  • throttle ?float|callable [0.2]: Per-domain throttle in seconds. Or a function taking the domain and current item and returning the throttle to apply.
  • guess_extension ?bool [True]: Whether to attempt to guess the resource's extension.
  • guess_encoding ?bool [True]: Whether to attempt to guess the resource's encoding.
  • buffer_size ?int [25]: Max number of items per domain to enqueue into memory in hope of finding a new domain that can be processed immediately.
  • insecure ?bool [False]: Whether to ignore SSL certification errors when performing requests.

Yields:

A FetchWorkerResult having the following attributes:

  • url ?string: the fetched url.
  • item any: original item from the iterator.
  • error ?Exception: an error.
  • response ?urllib3.HTTPResponse: the http response.
  • meta ?dict: additional metadata:
    • mime ?string: resource's mimetype.
    • ext ?string: resource's extension.
    • encoding ?string: resource's encoding.

multithreaded_resolve

Function resolving url redirections in a multithreaded fashion.

from minet import multithreaded_resolve

# Most basic usage
urls = ['https://bit.ly/whatever', 'https://t.co/whatever']

for result in multithreaded_resolve(urls):
  print(result.stack)

# Using a list of dicts

urls = [
  {
    'url': 'https://bit.ly/whatever',
    'label': 'Bit.ly'
  },
  {
    'url': 'https://t.co/whatever',
    'label': 'Twitter'
  }
]

for result in multithreaded_resolve(urls, key=lambda x: x['url']):
  print(result.stack)

Arguments:

  • iterator iterable: An iterator over urls or arbitrary items, if you provide a key argument along with it.
  • key ?callable: A function extracting the url to fetch from the items yielded by the provided iterator.
  • resolve_args ?callable: A function returning arguments to pass to the internal resolve helper for a call.
  • threads ?int [25]: Number of threads to use.
  • throttle ?float|callable [0.2]: Per-domain throttle in seconds. Or a function taking the domain and current item and returning the throttle to apply.
  • max_redirects ?int [5]: Max number of redirections to follow.
  • follow_refresh_header ?bool [False]: Whether to follow Refresh headers or not.
  • follow_meta_refresh ?bool [False]: Whether to follow meta refresh tags. It's more costly because we need to stream the start of the response's body and cannot rely on headers alone.
  • buffer_size ?int [25]: Max number of items per domain to enqueue into memory in hope of finding a new domain that can be processed immediately.
  • insecure ?bool [False]: Whether to ignore SSL certification errors when performing requests.

Yields:

A ResolveWorkerResult having the following attributes:

  • url ?string: the fetched url.
  • item any: original item from the iterator.
  • error ?Exception: an error.
  • stack ?list: the redirection stack.
You can’t perform that action at this time.