Search
PublicWWW searches the source code of web pages, not only the text visitors see. A query is the fragment of markup, script, stylesheet or text you expect to find - a tracker id, a library filename, a meta tag, an affiliate code or a phrase from the page.
What a search goes through
By default a search covers the home pages of websites; depth: reaches
their indexed internal pages - see Operators.
Response headers
Not only the page source is searched, but also the HTTP headers the server
sent with the page - Server, X-Powered-By and the
like, written as name: value.
In what order
Results come in order of popularity rank, the most popular websites first.
like: puts them in order of similarity instead.
Websites by topic
To find websites by what they are about rather than by their code, search with
like:"words", in any language, or with like:example.com for
websites like a given one - the most similar first. See
Websites by topic.
Which characters are searched
Only the Latin letters a-z, the digits 0-9 and the
underscore _ are searched. Every other character works like a
space - punctuation, accented letters, and the letters of other alphabets:
Cyrillic, Greek, Arabic, Chinese and the rest. Words written in other alphabets
are therefore moved into like:"..." automatically: similarity search takes words
in any language and alphabet and finds websites about that topic - see
Websites by topic. To match the code itself, look for the
Latin traces a page carries - a transliterated word, the language in the
<html> tag, the country domain. More in Languages.
The count above the results
It counts websites: one row is one website - its home page, or with
depth:N its page N. With depth:all it counts pages, and a
website can give several. A plain number is exact. 1,000,000+ means the
search stopped counting at a million - there are more. of about N
means the search ran out of time before it went through the whole index, as a
guest's search can, and N is an estimate of the total.
The count is what the index holds, not what your plan shows: rows hidden on your
plan - masked names with Sign up to view or Upgrade to view - are
counted too, and so are the rows past those a search may return. A cluster made
from depth:all results keeps each domain once.
Pages of results
A page shows 20 results. Without signing in, only the first page opens: the numbers after it are inactive, and the note next to them links to a free sign-up. A free account opens the first 50 pages (1,000 results); the pages after them open on the paid plans. A download takes as many rows as your plan opens - see Downloading results.
Rows also open only down to a place in the popularity ranking: the top 1,000,000 websites without signing in, the top 3,000,000 on a free account and the whole index on the paid plans - a paid account that has spent the day's searches sees the free depth until the reset. A row past that place shows Sign up to view or Upgrade to view instead of the website, and the count includes it.
In this section
- Signatures - the code a product leaves on its customers' websites, and how to turn it into a query.
- Product traces - the kinds of code to look for, from the most reliable to the least.
- Checking a signature - what a query finds, what does not belong, and what its number counts.
- Build signatures - give a list of websites, get code that finds more websites like them.
- Downloading results - the URLs, CSV and CSV+snippets files: format, separators, how many rows.
Where to go next
- Query Syntax - every operator and modifier, one line each.
- Technologies - ready-made queries by topic.
- Clusters - keep and compare sets of results, and read fields off every site in them.
- API - the same searches from your own code.