Commit 990da46c4f4d2b08eccf96f7c6ce27e72d0d8881 - golang-github-jaytaylor-html2text

Import upstream version 0.0~git20190408.0.01ec452, md5 eaf728835f9afeb014fa23f4db014449 Debian Janitor 4 years ago

4 changed file(s) with 172 addition(s) and 16 deletion(s). Raw diff Collapse all Expand all

+24

-0

.gitignore less more

	0	# Compiled Object files, Static and Dynamic libs (Shared Objects)
	1	*.o
	2	*.a
	3	*.so
	4
	5	# Folders
	6	_obj
	7	_test
	8
	9	# Architecture specific extensions/prefixes
	10	*.[568vq]
	11	[568vq].out
	12
	13	*.cgo1.go
	14	*.cgo2.c
	15	_cgo_defun.c
	16	_cgo_gotypes.go
	17	_cgo_export.*
	18
	19	_testmain.go
	20
	21	*.exe
	22	*.test
	23	*.prof

-5

README.md less more

3	3	[![Build Status](https://travis-ci.org/jaytaylor/html2text.svg?branch=master)](https://travis-ci.org/jaytaylor/html2text)
4	4	[![Report Card](https://goreportcard.com/badge/github.com/jaytaylor/html2text)](https://goreportcard.com/report/github.com/jaytaylor/html2text)
5	5
6		### Converts HTML into text
	6	### Converts HTML into text of the markdown-flavored variety
7	7
8	8
9	9	## Introduction
10	10
11	11	Ensure your emails are readable by all!
12	12
13		Turns HTML into raw text, useful for sending fancy HTML emails with a equivalently nicely formatted TXT document as a fallback (e.g. for people who don't allow HTML emails or have other display issues).
	13	Turns HTML into raw text, useful for sending fancy HTML emails with an equivalently nicely formatted TXT document as a fallback (e.g. for people who don't allow HTML emails or have other display issues).
14	14
15	15	html2text is a simple golang package for rendering HTML into plaintext.
16	16

22	22	## Download the package
23	23
24	24	```bash
25		go get github.com/jaytaylor/html2text
	25	go get jaytaylor.com/html2text
26	26	```
27	27
28	28	## Example usage

33	33	import (
34	34	"fmt"
35	35
36		"github.com/jaytaylor/html2text"
	36	"jaytaylor.com/html2text"
37	37	)
38	38
39	39	func main() {

77	77	</body>
78	78	</html>`
79	79
80		text, err := FromString(inputHTML, Options{PrettyTables: true})
	80	text, err := html2text.FromString(inputHTML, html2text.Options{PrettyTables: true})
81	81	if err != nil {
82	82	panic(err)
83	83	}

+78

-3

html2text.go less more

14	14
15	15	// Options provide toggles and overrides to control specific rendering behaviors.
16	16	type Options struct {
17		PrettyTables bool // Turns on pretty ASCII rendering for table elements.
	17	PrettyTables bool // Turns on pretty ASCII rendering for table elements.
	18	PrettyTablesOptions *PrettyTablesOptions // Configures pretty ASCII rendering for table elements.
	19	OmitLinks bool // Turns on omitting links
	20	}
	21
	22	// PrettyTablesOptions overrides tablewriter behaviors
	23	type PrettyTablesOptions struct {
	24	AutoFormatHeader bool
	25	AutoWrapText bool
	26	ReflowDuringAutoWrap bool
	27	ColWidth int
	28	ColumnSeparator string
	29	RowSeparator string
	30	CenterSeparator string
	31	HeaderAlignment int
	32	FooterAlignment int
	33	Alignment int
	34	ColumnAlignment []int
	35	NewLine string
	36	HeaderLine bool
	37	RowLine bool
	38	AutoMergeCells bool
	39	Borders tablewriter.Border
	40	}
	41
	42	// NewPrettyTablesOptions creates PrettyTablesOptions with default settings
	43	func NewPrettyTablesOptions() *PrettyTablesOptions {
	44	return &PrettyTablesOptions{
	45	AutoFormatHeader: true,
	46	AutoWrapText: true,
	47	ReflowDuringAutoWrap: true,
	48	ColWidth: tablewriter.MAX_ROW_WIDTH,
	49	ColumnSeparator: tablewriter.COLUMN,
	50	RowSeparator: tablewriter.ROW,
	51	CenterSeparator: tablewriter.CENTER,
	52	HeaderAlignment: tablewriter.ALIGN_DEFAULT,
	53	FooterAlignment: tablewriter.ALIGN_DEFAULT,
	54	Alignment: tablewriter.ALIGN_DEFAULT,
	55	ColumnAlignment: []int{},
	56	NewLine: tablewriter.NEWLINE,
	57	HeaderLine: true,
	58	RowLine: false,
	59	AutoMergeCells: false,
	60	Borders: tablewriter.Border{Left: true, Right: true, Bottom: true, Top: true},
	61	}
18	62	}
19	63
20	64	// FromHTMLNode renders text output from a pre-parsed HTML document.

78	122	justClosedDiv bool
79	123	blockquoteLevel int
80	124	lineLength int
	125	isPre bool
81	126	}
82	127
83	128	// tableTraverseContext holds table ASCII-form related context.

208	253	if attrVal := getAttrVal(node, "href"); attrVal != "" {
209	254	attrVal = ctx.normalizeHrefLink(attrVal)
210	255	// Don't print link href if it matches link element content or if the link is empty.
211		if attrVal != "" && linkText != attrVal {
	256	if !ctx.options.OmitLinks && attrVal != "" && linkText != attrVal {
212	257	hrefLink = "( " + attrVal + " )"
213	258	}
214	259	}

226	271	}
227	272	return ctx.traverseChildren(node)
228	273
	274	case atom.Pre:
	275	ctx.isPre = true
	276	err := ctx.traverseChildren(node)
	277	ctx.isPre = false
	278	return err
	279
229	280	case atom.Style, atom.Script, atom.Head:
230	281	// Ignore the subtree.
231	282	return nil

268	319
269	320	buf := &bytes.Buffer{}
270	321	table := tablewriter.NewWriter(buf)
	322	if ctx.options.PrettyTablesOptions != nil {
	323	options := ctx.options.PrettyTablesOptions
	324	table.SetAutoFormatHeaders(options.AutoFormatHeader)
	325	table.SetAutoWrapText(options.AutoWrapText)
	326	table.SetReflowDuringAutoWrap(options.ReflowDuringAutoWrap)
	327	table.SetColWidth(options.ColWidth)
	328	table.SetColumnSeparator(options.ColumnSeparator)
	329	table.SetRowSeparator(options.RowSeparator)
	330	table.SetCenterSeparator(options.CenterSeparator)
	331	table.SetHeaderAlignment(options.HeaderAlignment)
	332	table.SetFooterAlignment(options.FooterAlignment)
	333	table.SetAlignment(options.Alignment)
	334	table.SetColumnAlignment(options.ColumnAlignment)
	335	table.SetNewLine(options.NewLine)
	336	table.SetHeaderLine(options.HeaderLine)
	337	table.SetRowLine(options.RowLine)
	338	table.SetAutoMergeCells(options.AutoMergeCells)
	339	table.SetBorders(options.Borders)
	340	}
271	341	table.SetHeader(ctx.tableCtx.header)
272	342	table.SetFooter(ctx.tableCtx.footer)
273	343	table.AppendBulk(ctx.tableCtx.body)

324	394	return ctx.traverseChildren(node)
325	395
326	396	case html.TextNode:
327		data := strings.Trim(spacingRe.ReplaceAllString(node.Data, " "), " ")
	397	var data string
	398	if ctx.isPre {
	399	data = node.Data
	400	} else {
	401	data = strings.TrimSpace(spacingRe.ReplaceAllString(node.Data, " "))
	402	}
328	403	return ctx.emit(data)
329	404
330	405	case html.ElementNode:

+65

-8

html2text_test.go less more

138	138	"Test text<br><BR />Test text",
139	139	"Test text\n\nTest text",
140	140	},
	141	{
	142	"<pre>test1\ntest 2\n\ntest 3</pre>",
	143	"test1\ntest 2\n\ntest 3",
	144	},
141	145	}
142	146
143	147	for _, testCase := range testCases {

184	188	{
185	189	`<table>
186	190	<tbody>
187		<tr><td><p>Row-1-Col-1-Msg1</p><p>Row-1-Col-1-Msg2</p></td><td>Row-1-Col-2</td></tr>
	191	<tr><td><p>Row-1-Col-1-Msg123456789012345</p><p>Row-1-Col-1-Msg2</p></td><td>Row-1-Col-2</td></tr>
188	192	<tr><td>Row-2-Col-1</td><td>Row-2-Col-2</td></tr>
189	193	</tbody>
190	194	</table>`,
191	195	// +--------------------------------+-------------+
192		// \| Row-1-Col-1-Msg1 \| Row-1-Col-2 \|
	196	// \| Row-1-Col-1-Msg123456789012345 \| Row-1-Col-2 \|
193	197	// \| Row-1-Col-1-Msg2 \| \|
194	198	// \| Row-2-Col-1 \| Row-2-Col-2 \|
195	199	// +--------------------------------+-------------+
196	200	`+--------------------------------+-------------+
197		\| Row-1-Col-1-Msg1 \| Row-1-Col-2 \|
	201	\| Row-1-Col-1-Msg123456789012345 \| Row-1-Col-2 \|
198	202	\| Row-1-Col-1-Msg2 \| \|
199	203	\| Row-2-Col-1 \| Row-2-Col-2 \|
200	204	+--------------------------------+-------------+`,
201		`Row-1-Col-1-Msg1
	205	`Row-1-Col-1-Msg123456789012345
202	206
203	207	Row-1-Col-1-Msg2
204	208

240	244	"Header 1 Header 2 Footer 1 Footer 2 Row 1 Col 1 Row 1 Col 2 Row 2 Col 1 Row 2 Col 2",
241	245	},
242	246	// Two tables in same HTML (goal is to test that context is
243		// reinitalized correctly).
	247	// reinitialized correctly).
244	248	{
245	249	`<p>
246	250	<table>

329	333
330	334	for _, testCase := range testCases {
331	335	options := Options{
332		PrettyTables: true,
	336	PrettyTables: true,
	337	PrettyTablesOptions: NewPrettyTablesOptions(),
333	338	}
334	339	// Check pretty tabular ASCII version.
335	340	if msg, err := wantString(testCase.input, testCase.tabularOutput, options); err != nil {

467	472	}
468	473	}
469	474
470		func TestImageAltTags(t *testing.T) {
	475	func TestOmitLinks(t *testing.T) {
471	476	testCases := []struct {
472	477	input string
473	478	output string
474	479	}{
475	480	{
	481	`<a></a>`,
	482	``,
	483	},
	484	{
	485	`<a href=""></a>`,
	486	``,
	487	},
	488	{
	489	`<a href="http://example.com/"></a>`,
	490	``,
	491	},
	492	{
	493	`<a href="">Link</a>`,
	494	`Link`,
	495	},
	496	{
	497	`<a href="http://example.com/">Link</a>`,
	498	`Link`,
	499	},
	500	{
	501	`<a href="http://example.com/"><span class="a">Link</span></a>`,
	502	`Link`,
	503	},
	504	{
	505	"<a href='http://example.com/'>\n\t<span class='a'>Link</span>\n\t</a>",
	506	`Link`,
	507	},
	508	{
	509	`<a href="http://example.com/"><img src="http://example.ru/hello.jpg" alt="Example"></a>`,
	510	`Example`,
	511	},
	512	}
	513
	514	for _, testCase := range testCases {
	515	if msg, err := wantString(testCase.input, testCase.output, Options{OmitLinks: true}); err != nil {
	516	t.Error(err)
	517	} else if len(msg) > 0 {
	518	t.Log(msg)
	519	}
	520	}
	521	}
	522
	523	func TestImageAltTags(t *testing.T) {
	524	testCases := []struct {
	525	input string
	526	output string
	527	}{
	528	{
476	529	`<img />`,
477	530	``,
478	531	},

614	667	{
615	668	"Test 1<div>Test 2</div> <div>Test 3</div>Test 4",
616	669	"Test 1\nTest 2\nTest 3\nTest 4",
	670	},
	671	{
	672	"Test 1<div> Test 2 </div>",
	673	"Test 1\nTest 2",
617	674	},
618	675	}
619	676

743	800	`hi
744	801
745	802	<br>
746
	803
747	804	hello <a href="https://google.com">google</a>
748	805	<br><br>
749	806	test<p>List:</p>