Wecr - simple web crawler

Overview

Just a simple HTML web spider with no dependencies. It is possible to search for pages with a text on them or for the text itself, extract images, video, audio and save pages that satisfy the criteria along the way.

Configuration

The flow of work fully depends on the configuration file. By default conf.json is used as a configuration file, but the name can be changed via -conf flag. The default configuration is embedded in the program so on the first launch or by simply deleting the file, a new conf.json will be created in the same directory as the executable itself unless the -wDir (working directory) flag is set to some other value. To see al available flags run wecr -h.

The configuration is split into different branches like requests (how requests are made, ie: request timeout, wait time, user agent), logging (use logs, output to a file), save (output file|directory, save pages or not) or search (use regexp, query string) each of which contain tweakable parameters. There are global ones as well such as workers (working threads that make requests in parallel) and depth (literally, how deep the recursive search should go). The names are simple and self-explanatory so no attribute-by-attribute explanation needed for most of them.

The parsing starts from initial_pages and goes deeper while ignoring the pages on domains that are in blacklisted_domains or are NOT in allowed_domains. If all initial pages are happen to be on blacklisted domains or are not in the allowed list - the program will get stuck. It is important to note that *_domains should be specified with an existing scheme (ie: https://en.wikipedia.org). Subdomains and ports matter: https://unbewohnte.su:3000/ and https://unbewohnte.su/ are different.

Previous versions stored the entire visit queue in memory, resulting in gigabytes of memory usage but as of v0.2.4 it is possible to offload the queue to the persistent storage via in_memory_visit_queue option (false by default).

Search query

There are some special query values:

• email - tells wecr to scrape email addresses and output to output_file
• images - find all images on pages and output to the corresponding directory in output_dir (IMPORTANT: set content_fetch_timeout_ms to 0 so the images (and other content below) load fully)
• videos - find and fetch files that look like videos
• audio - find and fetch files that look like audio
• documents - find and fetch files that look like a document
• everything - find and fetch images, audio, video, documents and email addresses

When is_regexp is enabled, the query is treated as a regexp string and pages will be scanned for matches that satisfy it.

Output

By default, if the query is not something of special values all the matches and other data will be outputted to output.json file as separate continuous JSON objects, but if save_pages is set to true and|or query is set to images, videos, audio, etc. - the additional contents will be put in the corresponding directories inside output_dir, which is neatly created by the executable's side.

The output almost certainly contains some duplicates and is not easy to work with programmatically, so you can use -extractData with the output JSON file argument (like output.json, which is the default output file name) to extract the actual data, filter out the duplicates and put each entry on its new line in a new text file.

Build

If you're on *nix - it's as easy as make.

Otherwise - go build in the src directory to build wecr.

Examples

See page on my website for some basic examples.

License

AGPLv3