arno
/
pup
mirror of https://github.com/ericchiang/pup
1
0
Fork 0
Code Issues Releases Wiki Activity

Merge pull request #7 from EricChiang/dev

Browse Source 
Dev
pull/9/head
Eric Chiang 10 years ago
parent abd4c986bf 8f8d21ab92
commit af61952d1c
4 changed files with 93 additions and 9 deletions
Show Stats Download Patch File Download Diff File

4
main.go
Unescape View File

@ -11,7 +11,7 @@ import (
"strings" "strings"
) )
const VERSION string = "0.1.1" const VERSION string = "0.1.2a"
var ( var (
// Flags // Flags
@ -125,7 +125,7 @@ func main() {
PrintNode(root, 0) PrintNode(root, 0)
os.Exit(0) os.Exit(0)
} }
selectors := make([]*selector.Selector, len(cmds)) selectors := make([]selector.Selector, len(cmds))
for i, cmd := range cmds { for i, cmd := range cmds {
if i+1 == len(cmds) { if i+1 == len(cmds) {
d, err := funcs.NewDisplayFunc(cmd) d, err := funcs.NewDisplayFunc(cmd)

23
selector/selector.go
Unescape View File

@ -8,11 +8,20 @@ import (
) )
// A CSS Selector // A CSS Selector
type Selector struct { type BasicSelector struct {
Name *regexp.Regexp Name *regexp.Regexp
Attrs map[string]*regexp.Regexp Attrs map[string]*regexp.Regexp
} }
type Selector interface {
// Does this selector match a given node?
Match(node *html.Node) bool
// Find all nodes which match a selector. May return itself.
FindAll(node *html.Node) []*html.Node
// Find all child nodes which match a selector.
FindAllChildren(node *html.Node) []*html.Node
}
type selectorField int type selectorField int
const ( const (
@ -63,7 +72,7 @@ func parseAttrField(command string) (attrKey string, matcher *regexp.Regexp,
} }
// Set a field of this selector. // Set a field of this selector.
func (s *Selector) setFieldValue(f selectorField, v string) error { func (s *BasicSelector) setFieldValue(f selectorField, v string) error {
if v == "" { if v == "" {
return nil return nil
} }
@ -98,9 +107,9 @@ func (s *Selector) setFieldValue(f selectorField, v string) error {
} }
// Convert a string to a selector. // Convert a string to a selector.
func NewSelector(s string) (*Selector, error) { func NewSelector(s string) (Selector, error) {
attrs := map[string]*regexp.Regexp{} attrs := map[string]*regexp.Regexp{}
selector := &Selector{nil, attrs} selector := BasicSelector{nil, attrs}
nextField := NameField nextField := NameField
start := 0 start := 0
// Parse the selector character by character // Parse the selector character by character
@ -153,7 +162,7 @@ func NewSelector(s string) (*Selector, error) {
} }
// Find all nodes which match a selector. // Find all nodes which match a selector.
func (sel *Selector) FindAllChildren(node *html.Node) []*html.Node { func (sel BasicSelector) FindAllChildren(node *html.Node) []*html.Node {
selected := []*html.Node{} selected := []*html.Node{}
child := node.FirstChild child := node.FirstChild
for child != nil { for child != nil {
@ -165,7 +174,7 @@ func (sel *Selector) FindAllChildren(node *html.Node) []*html.Node {
} }
// Find all nodes which match a selector. May return itself. // Find all nodes which match a selector. May return itself.
func (sel *Selector) FindAll(node *html.Node) []*html.Node { func (sel BasicSelector) FindAll(node *html.Node) []*html.Node {
selected := []*html.Node{} selected := []*html.Node{}
if sel.Match(node) { if sel.Match(node) {
return []*html.Node{node} return []*html.Node{node}
@ -180,7 +189,7 @@ func (sel *Selector) FindAll(node *html.Node) []*html.Node {
} }
// Does this selector match a given node? // Does this selector match a given node?
func (sel *Selector) Match(node *html.Node) bool { func (sel BasicSelector) Match(node *html.Node) bool {
if node.Type != html.ElementNode { if node.Type != html.ElementNode {
return false return false
} }

19
tests/README.md
Unescape View File

@ -0,0 +1,19 @@
# Tests
A simple set of tests to help maintain sanity.
The tests themselves are written in Python and can be run using the [nose](
https://nose.readthedocs.org/en/latest/) tool.
Install with:
```bash
$ pip install nose
```
Run the following command from either the base directory or this one to perform
the tests:
```bash
$ nosetests
```

56
tests/tests.py
Unescape View File

@ -0,0 +1,56 @@
from subprocess import Popen, PIPE, STDOUT
example_data = """
<html>
<head>
</head>
<body>
<div>
<div class="nav clearfix">
My data
</div>
<p>Some other data</p>
</div>
</body>
</html>
"""
# run a pup command as a subprocess
def run_pup(args, input_html):
cmd = ["pup"]
cmd.extend(args)
p = Popen(cmd, stdout=PIPE, stdin=PIPE, stderr=PIPE)
stdout_data = p.communicate(input=input_html)[0]
p.wait()
return stdout_data
# simply count the number of lines returned by this pup command
def run_pup_count(args, input_html):
pup_output = run_pup(args, input_html)
lines = [l for l in pup_output.split("\n") if l]
return len(lines)
def test_class_selector():
assert run_pup_count([".nav"], example_data) == 3
def test_attr_eq():
assert run_pup_count(["[class=nav]"], example_data) == 0
def test_attr_pre():
assert run_pup_count(["[class^=nav]"], example_data) == 3
assert run_pup_count(["[class^=clearfix]"], example_data) == 0
def test_attr_post():
assert run_pup_count(["[class$=nav]"], example_data) == 0
assert run_pup_count(["[class$=clearfix]"], example_data) == 3
def test_attr_func():
result = run_pup(["div", "attr{class}"], example_data).strip()
assert result == ""
result = run_pup(["div", "div", "attr{class}"], example_data).strip()
assert result == "nav clearfix"
def test_text_func():
result = run_pup(["p", "text{}"], example_data).strip()
assert result == "Some other data"
Write Preview
Loading…
Cancel
Save