yequari
/
webweav.ing
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
webweav.ing/feed/feed_test.go

299 lines
12 KiB
Go
Raw Normal View History

restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
package feed_test
initial commit
2024-01-19 03:33:37 +00:00
rename module
2024-03-03 21:10:34 +00:00
import (
implement multiple article parsing
2024-03-15 23:52:03 +00:00
"strings"
"testing"
"time"
"git.32bit.cafe/yequari/webweav.ing/feed"
"golang.org/x/net/html"
rename module
2024-03-03 21:10:34 +00:00
)
initial commit
2024-01-19 03:33:37 +00:00
parse out titles and urls from heading tags
2024-04-11 03:49:54 +00:00
func TestTimeParsingFormat(t *testing.T) {
add more time formats
2024-03-06 20:27:37 +00:00
testDate, err := time.Parse("2006-Jan-02 15:04:05 -7", "2004-May-14 07:30:55 -7")
initial commit
2024-01-19 03:33:37 +00:00
if err != nil {
t.Errorf("creating test date failed: %s", err)
}
add more time formats
2024-03-06 20:27:37 +00:00
initial commit
2024-01-19 03:33:37 +00:00
var tests = []struct {
name string
input string
add more time formats
2024-03-06 20:27:37 +00:00
format string
initial commit
2024-01-19 03:33:37 +00:00
}{
{
add more time formats
2024-03-06 20:27:37 +00:00
"YYYY-MM-DD",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
`<article><time datetime="2004-05-14">May 14 2004</time>hello world</article>`,
add more time formats
2024-03-06 20:27:37 +00:00
time.DateOnly,
},
{
"YYYY-MM-DD HH:MM",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
`<article><time datetime="2004-05-14 07:30">May 14 2004</time>hello world</article>`,
add more time formats
2024-03-06 20:27:37 +00:00
"2006-01-02 15:04",
},
{
"YYYY-MM-DD HH:MM:SS",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
`<article><time datetime="2004-05-14 07:30:55">May 14 2004</time>hello world</article>`,
add more time formats
2024-03-06 20:27:37 +00:00
"2006-01-02 15:04:05",
initial commit
2024-01-19 03:33:37 +00:00
},
{
add more time formats
2024-03-06 20:27:37 +00:00
"YYYY-MM-DDTHH:MM:SS",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
`<article><time datetime="2004-05-14T07:30:55">May 14 2004</time>hello world</article>`,
add more time formats
2024-03-06 20:27:37 +00:00
"2006-01-02T15:04:05",
},
{
"YYYY-MM-DDTHH:MM",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
`<article><time datetime="2004-05-14T07:30">May 14 2004</time>hello world</article>`,
add more time formats
2024-03-06 20:27:37 +00:00
"2006-01-02T15:04",
initial commit
2024-01-19 03:33:37 +00:00
},
}
for _, tt := range tests {
t.Run(tt.name, func (t *testing.T) {
implement multiple article parsing
2024-03-15 23:52:03 +00:00
html, err := html.Parse(strings.NewReader(tt.input))
if err != nil {
t.Errorf("error: %s", err)
}
item, err := feed.NewFeedItem("", html)
initial commit
2024-01-19 03:33:37 +00:00
if err != nil {
t.Errorf("error: %s", err)
}
add more time formats
2024-03-06 20:27:37 +00:00
actualTime := item.PubTime.Format(tt.format)
expectedTime := testDate.Format(tt.format)
if actualTime != expectedTime {
t.Errorf("got %s, want %s", actualTime, expectedTime)
initial commit
2024-01-19 03:33:37 +00:00
}
})
}
}
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
parse out titles and urls from heading tags
2024-04-11 03:49:54 +00:00
func TestArticleTimeParsing(t *testing.T) {
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
testDate, err := time.Parse("2006-Jan-02", "2004-May-14")
if err != nil {
t.Errorf("creating test date failed: %s", err)
}
implement multiple article parsing
2024-03-15 23:52:03 +00:00
testDate2, err := time.Parse("2006-Jan-02", "2004-May-07")
if err != nil {
t.Errorf("creating test date failed: %s", err)
}
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
var tests = []struct {
name string
input string
implement multiple article parsing
2024-03-15 23:52:03 +00:00
want_time []*time.Time
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
}{
{
"article and time stripped out of basic HTML",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article></body></html>",
implement multiple article parsing
2024-03-15 23:52:03 +00:00
[]*time.Time{&testDate},
},
{
"multiple articles",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article><article><time datetime=\"2004-05-07\">May 7 2004</time>this is a second article</article></body></html>",
[]*time.Time{&testDate, &testDate2},
parse out titles and urls from heading tags
2024-04-11 03:49:54 +00:00
},
}
for _, tt := range tests {
t.Run(tt.name, func (t *testing.T) {
html, err := html.Parse(strings.NewReader(tt.input))
if err != nil {
t.Errorf("error: %s", err)
}
page := feed.SitePage{
Url: "",
Title: "",
Root: html,
Errors: make([]error, 0),
}
items, err := page.Parse()
if err != nil {
t.Errorf("error: %s", err)
}
for i, item := range items {
if tt.want_time[i] != nil && !item.PubTime.Equal(*tt.want_time[i]) {
t.Errorf("got %s, want %s", item.PubTime, *tt.want_time[i])
}
}
})
}
}
func TestArticleUrls(t *testing.T) {
var tests = []struct {
name string
input string
want_url []string
}{
{
"article without url",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article></body></html>",
[]string{""},
},
{
"one article with one url",
"<html><head></head><body><article><h1><a href=\"https://example.com\">Title</a></h1><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article></body></html>",
[]string{"https://example.com"},
},
{
"multiple articles",
"<html><head></head><body><article><h1><a href=\"https://example.com/example1\">Title</a></h1><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article><article><h1><a href=\"https://example.com/example2\">Title</a></h1><time datetime=\"2004-05-07\">May 7 2004</time>this is a second article</article></body></html>",
[]string{"https://example.com/example1", "https://example.com/example2"},
},
{
"one article with multiple urls",
"<html><head></head><body><article><h1><a href=\"https://example.com\">Title</a></h1><time datetime=\"2004-05-14\">May 14 2004</time><a href=\"https://google.com\">click me</a><p>hello world</p></article></body></html>",
[]string{"https://example.com"},
},
{
"one article with multiple h1 urls",
"<html><head></head><body><article><h1><a href=\"https://example.com\">Title</a></h1><time datetime=\"2004-05-14\">May 14 2004</time><h1><a href=\"https://google.com\">click me</a></h1><p>hello world</p></article></body></html>",
[]string{"https://example.com"},
},
}
for _, tt := range tests {
t.Run(tt.name, func (t *testing.T) {
html, err := html.Parse(strings.NewReader(tt.input))
if err != nil {
t.Errorf("error: %s", err)
}
page := feed.SitePage{
Url: "",
Title: "",
Root: html,
Errors: make([]error, 0),
}
items, err := page.Parse()
if err != nil {
t.Errorf("error: %s", err)
}
for i, item := range items {
if item.Url != tt.want_url[i] {
t.Errorf("got %s, want %s", item.Url, tt.want_url[i])
}
}
})
}
}
func TestArticleTitles(t *testing.T) {
var tests = []struct {
name string
input string
want_title []string
}{
{
"article and time stripped out of basic HTML",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article></body></html>",
[]string{"May 14 2004"},
},
{
"multiple articles",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article><article><time datetime=\"2004-05-07\">May 7 2004</time>this is a second article</article></body></html>",
parse headers
2024-04-10 01:43:27 +00:00
[]string{"May 14 2004", "May 07 2004"},
},
parse out titles and urls from heading tags
2024-04-11 03:49:54 +00:00
{
"multiple articles with h1",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h1>Hello</h1>hello world</article><article><time datetime=\"2004-05-07\"><h1>World</h1>May 7 2004</time>this is a second article</article></body></html>",
[]string{"Hello", "World"},
},
parse headers
2024-04-10 01:43:27 +00:00
{
"article with h1",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h1>Hello</h1>hello world</article></body></html>",
[]string{"Hello"},
},
{
"article with h2",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h2>Hello</h2>hello world</article></body></html>",
[]string{"Hello"},
},
{
"article with h3",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h3>Hello</h3>hello world</article></body></html>",
[]string{"Hello"},
},
{
"article with h1 and h2",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h1>Hello</h1><h2>World</h2>hello world</article></body></html>",
[]string{"Hello"},
},
{
"article with h2 and h3",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h3>Hello</h3><h2>World</h2>hello world</article></body></html>",
[]string{"World"},
},
{
"article with multiple h1",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time><h1>Hello</h1><h1>World</h1>hello world</article></body></html>",
[]string{"Hello"},
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
},
}
for _, tt := range tests {
t.Run(tt.name, func (t *testing.T) {
implement multiple article parsing
2024-03-15 23:52:03 +00:00
html, err := html.Parse(strings.NewReader(tt.input))
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
if err != nil {
t.Errorf("error: %s", err)
}
implement multiple article parsing
2024-03-15 23:52:03 +00:00
page := feed.SitePage{
Url: "",
Title: "",
Root: html,
Errors: make([]error, 0),
}
items, err := page.Parse()
if err != nil {
t.Errorf("error: %s", err)
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
}
implement multiple article parsing
2024-03-15 23:52:03 +00:00
for i, item := range items {
parse headers
2024-04-10 01:43:27 +00:00
if item.Title != tt.want_title[i] {
t.Errorf("got %s, want %s", item.Title, tt.want_title[i])
}
restructure and implement feed.ParseContent
2024-02-29 05:41:36 +00:00
}
})
}
}
parse out titles and urls from heading tags
2024-04-11 03:49:54 +00:00
func TestArticleParsing(t *testing.T) {
var tests = []struct {
name string
input string
want_article []string
}{
{
"article and time stripped out of basic HTML",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article></body></html>",
[]string{"<article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article>"},
},
{
"multiple articles",
"<html><head></head><body><article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article><article><time datetime=\"2004-05-07\">May 7 2004</time>this is a second article</article></body></html>",
[]string{"<article><time datetime=\"2004-05-14\">May 14 2004</time>hello world</article>", "<article><time datetime=\"2004-05-07\">May 7 2004</time>this is a second article</article>"},
},
}
for _, tt := range tests {
t.Run(tt.name, func (t *testing.T) {
html, err := html.Parse(strings.NewReader(tt.input))
if err != nil {
t.Errorf("error: %s", err)
}
page := feed.SitePage{
Url: "",
Title: "",
Root: html,
Errors: make([]error, 0),
}
items, err := page.Parse()
if err != nil {
t.Errorf("error: %s", err)
}
for i, item := range items {
if item.RawText != tt.want_article[i] {
t.Errorf("got %s, want %s", item.RawText, tt.want_article[i])
}
}
})
}
}