webweav.ing/feed/feed_test.go

package feed_test

import (
	"strings"
	"testing"
	"time"

	"git.32bit.cafe/yequari/webweav.ing/feed"
	"golang.org/x/net/html"
)

func TestTimeParsing(t *testing.T) {
    testDate, err := time.Parse("2006-Jan-02 15:04:05 -7", "2004-May-14 07:30:55 -7")
    if err != nil {
        t.Errorf("creating test date failed: %s", err)
    }

    var tests = []struct {
        name string
        input string
        format string
    }{
            {
                "YYYY-MM-DD",
                `<article><time datetime="2004-05-14">May 14 2004</time>hello world</article>`,
                time.DateOnly,
            },
            {
                "YYYY-MM-DD HH:MM",
                `<article><time datetime="2004-05-14 07:30">May 14 2004</time>hello world</article>`,
                "2006-01-02 15:04",
            },
            {
                "YYYY-MM-DD HH:MM:SS",
                `<article><time datetime="2004-05-14 07:30:55">May 14 2004</time>hello world</article>`,
                "2006-01-02 15:04:05",
            },
            {
                "YYYY-MM-DDTHH:MM:SS",
                `<article><time datetime="2004-05-14T07:30:55">May 14 2004</time>hello world</article>`,
                "2006-01-02T15:04:05",
            },
            {
                "YYYY-MM-DDTHH:MM",
                `<article><time datetime="2004-05-14T07:30">May 14 2004</time>hello world</article>`,
                "2006-01-02T15:04",
            },
            {
                "YYYY-DD-MM HH:MM",
                `<article><time datetime="2004-14-05 07:30">May 14 2004</time>hello world</article>`,
                "2006-02-01 15:04",
            },
    }

    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)
            }
            item, err := feed.NewFeedItem("", html)
            if err != nil {
                t.Errorf("error: %s", err)
            }

            actualTime := item.PubTime.Format(tt.format)
            expectedTime := testDate.Format(tt.format)
            if actualTime != expectedTime {
                t.Errorf("got %s, want %s", actualTime, expectedTime)
            }
        })
    }
}

func TestArticleParsing(t *testing.T) {
    testDate, err := time.Parse("2006-Jan-02", "2004-May-14") 
    if err != nil {
        t.Errorf("creating test date failed: %s", err)
    }
    testDate2, err := time.Parse("2006-Jan-02", "2004-May-07") 
    if err != nil {
        t.Errorf("creating test date failed: %s", err)
    }
    var tests = []struct {
        name string
        input string
        want_time []*time.Time
        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>", 
                []*time.Time{&testDate}, 
                []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>", 
                []*time.Time{&testDate, &testDate2},
                []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])
                }
                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])
                }
            }
        })
    }
}
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
add more time formats 2024-03-06 20:27:37 +00:00			`func TestTimeParsing(t *testing.T) {`
			`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			`},`
implement basic multiple articles 2024-04-03 00:24:51 +00:00			`{`
			`"YYYY-DD-MM HH:MM",`
			`<article><time datetime="2004-14-05 07:30">May 14 2004</time>hello world</article>`,
			`"2006-02-01 15:04",`
			`},`
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
implement multiple article parsing 2024-03-15 23:52:03 +00:00			`func TestArticleParsing(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`
			`want_article []string`
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},`
			`[]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>",`
			`[]*time.Time{&testDate, &testDate2},`
			`[]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>"},`
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 {`
			`if item.RawText != tt.want_article[i] {`
			`t.Errorf("got %s, want %s", item.RawText, tt.want_article[i])`
			`}`
			`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])`
			`}`
restructure and implement feed.ParseContent 2024-02-29 05:41:36 +00:00			`}`
			`})`
			`}`
			`}`