refactor: model receives buffer & computes matches

This commit is contained in:
graelo 2021-03-22 08:48:51 +01:00
parent 893f80a1e2
commit 4ec240c2b1
4 changed files with 304 additions and 279 deletions

View file

@ -15,6 +15,7 @@ mod tests {
#[test]
fn match_reverse() {
let buffer = "lorem 127.0.0.1 lorem 255.255.255.255 lorem 127.0.0.1 lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -22,7 +23,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -30,7 +31,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.first().unwrap().hint, "a");
@ -40,6 +41,7 @@ mod tests {
#[test]
fn match_unique() {
let buffer = "lorem 127.0.0.1 lorem 255.255.255.255 lorem 127.0.0.1 lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -47,7 +49,7 @@ mod tests {
let reverse = false;
let unique_hint = true;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -55,7 +57,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.first().unwrap().hint, "a");
@ -65,6 +67,7 @@ mod tests {
#[test]
fn match_docker() {
let buffer = "latest sha256:30557a29d5abc51e5f1d5b472e79b7e296f595abcf19fe6b9199dbbc809c6ff4 20 hours ago";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -72,7 +75,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -80,7 +83,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
assert_eq!(
@ -93,6 +96,7 @@ mod tests {
fn match_ansi_colors() {
let buffer =
"path: /var/log/nginx.log\npath: test/log/nginx-2.log:32folder/.nginx@4df2.log";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -100,7 +104,7 @@ mod tests {
let reverse = true;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -108,7 +112,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.get(0).unwrap().text, "/var/log/nginx.log");
@ -120,6 +124,7 @@ mod tests {
fn match_paths() {
let buffer =
"Lorem /tmp/foo/bar_lol, lorem\n Lorem /var/log/boot-strap.log lorem ../log/kern.log lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -127,7 +132,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -135,7 +140,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.get(0).unwrap().text, "/tmp/foo/bar_lol");
@ -146,6 +151,7 @@ mod tests {
#[test]
fn match_home() {
let buffer = "Lorem ~/.gnu/.config.txt, lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -153,7 +159,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -161,7 +167,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
assert_eq!(results.get(0).unwrap().text, "~/.gnu/.config.txt");
@ -170,6 +176,7 @@ mod tests {
#[test]
fn match_uuids() {
let buffer = "Lorem ipsum 123e4567-e89b-12d3-a456-426655440000 lorem\n Lorem lorem lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -177,7 +184,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -185,7 +192,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
}
@ -193,6 +200,7 @@ mod tests {
#[test]
fn match_shas() {
let buffer = "Lorem fd70b5695 5246ddf f924213 lorem\n Lorem 973113963b491874ab2e372ee60d4b4cb75f717c lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -200,7 +208,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -208,7 +216,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 4);
assert_eq!(results.get(0).unwrap().text, "fd70b5695");
@ -223,6 +231,7 @@ mod tests {
#[test]
fn match_ipv4s() {
let buffer = "Lorem ipsum 127.0.0.1 lorem\n Lorem 255.255.10.255 lorem 127.0.0.1 lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -230,7 +239,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -238,7 +247,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.get(0).unwrap().pattern, "ipv4");
@ -252,6 +261,7 @@ mod tests {
#[test]
fn match_ipv6s() {
let buffer = "Lorem ipsum fe80::2:202:fe4 lorem\n Lorem 2001:67c:670:202:7ba8:5e41:1591:d723 lorem fe80::2:1 lorem ipsum fe80:22:312:fe::1%eth0";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -259,7 +269,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -267,7 +277,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 4);
assert_eq!(results.get(0).unwrap().text, "fe80::2:202:fe4");
@ -283,6 +293,7 @@ mod tests {
fn match_markdown_urls() {
let buffer =
"Lorem ipsum [link](https://github.io?foo=bar) ![](http://cdn.com/img.jpg) lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -290,7 +301,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -298,7 +309,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 2);
assert_eq!(results.get(0).unwrap().pattern, "markdown-url");
@ -310,6 +321,7 @@ mod tests {
#[test]
fn match_urls() {
let buffer = "Lorem ipsum https://www.rust-lang.org/tools lorem\n Lorem ipsumhttps://crates.io lorem https://github.io?foo=bar lorem ssh://github.io";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -317,7 +329,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -325,7 +337,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 4);
assert_eq!(
@ -345,6 +357,7 @@ mod tests {
fn match_emails() {
let buffer =
"Lorem ipsum <first.last+social@example.com> john@server.department.company.com lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -352,7 +365,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -360,7 +373,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 2);
assert_eq!(results.get(0).unwrap().pattern, "email");
@ -378,6 +391,7 @@ mod tests {
#[test]
fn match_addresses() {
let buffer = "Lorem 0xfd70b5695 0x5246ddf lorem\n Lorem 0x973113tlorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -385,7 +399,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -393,7 +407,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 3);
assert_eq!(results.get(0).unwrap().pattern, "mem-address");
@ -407,6 +421,7 @@ mod tests {
#[test]
fn match_hex_colors() {
let buffer = "Lorem #fd7b56 lorem #FF00FF\n Lorem #00fF05 lorem #abcd00 lorem #afRR00";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -414,7 +429,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -422,7 +437,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 4);
assert_eq!(results.get(0).unwrap().text, "#fd7b56");
@ -434,6 +449,7 @@ mod tests {
#[test]
fn match_ipfs() {
let buffer = "Lorem QmRdbNSxDJBXmssAc9fvTtux4duptMvfSGiGuq6yHAQVKQ lorem Qmfoobar";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -441,7 +457,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -449,7 +465,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
assert_eq!(
@ -461,6 +477,7 @@ mod tests {
#[test]
fn match_process_port() {
let buffer = "Lorem 5695 52463 lorem\n Lorem 973113 lorem 99999 lorem 8888 lorem\n 23456 lorem 5432 lorem 23444";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -468,7 +485,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -476,7 +493,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 8);
}
@ -484,6 +501,7 @@ mod tests {
#[test]
fn match_diff_a() {
let buffer = "Lorem lorem\n--- a/src/main.rs";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -491,7 +509,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -499,7 +517,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
assert_eq!(results.get(0).unwrap().pattern, "diff-a");
@ -509,6 +527,7 @@ mod tests {
#[test]
fn match_diff_b() {
let buffer = "Lorem lorem\n+++ b/src/main.rs";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom = vec![];
@ -516,7 +535,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -524,7 +543,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 1);
assert_eq!(results.get(0).unwrap().pattern, "diff-b");
@ -534,6 +553,7 @@ mod tests {
#[test]
fn priority_between_regexes() {
let buffer = "Lorem [link](http://foo.bar) ipsum CUSTOM-52463 lorem ISSUE-123 lorem\nLorem /var/fd70b569/9999.log 52463 lorem\n Lorem 973113 lorem 123e4567-e89b-12d3-a456-426655440000 lorem 8888 lorem\n https://crates.io/23456/fd70b569 lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = true;
let named_pat = vec![];
let custom: Vec<String> = ["CUSTOM-[0-9]{4,}", "ISSUE-[0-9]{3}"]
@ -544,7 +564,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -552,7 +572,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 9);
assert_eq!(results.get(0).unwrap().text, "http://foo.bar");
@ -575,6 +595,7 @@ mod tests {
#[test]
fn named_patterns() {
let buffer = "Lorem [link](http://foo.bar) ipsum CUSTOM-52463 lorem ISSUE-123 lorem\nLorem /var/fd70b569/9999.log 52463 lorem\n Lorem 973113 lorem 123e4567-e89b-12d3-a456-426655440000 lorem 8888 lorem\n https://crates.io/23456/fd70b569 lorem";
let lines = buffer.split('\n').collect::<Vec<_>>();
let use_all_patterns = false;
use crate::textbuf::regexes::parse_pattern_name;
@ -585,7 +606,7 @@ mod tests {
let reverse = false;
let unique_hint = false;
let results = Model::new(
buffer,
&lines,
&alphabet,
use_all_patterns,
&named_pat,
@ -593,7 +614,7 @@ mod tests {
reverse,
unique_hint,
)
.matches();
.matches;
assert_eq!(results.len(), 2);
assert_eq!(results.get(0).unwrap().text, "http://foo.bar");