12879605443

Committed 21 Jan 2025 03:31AM UTC coverage: 92.224% (+0.03%) from 92.197%

Build Type

push

github

Committed by

Commit Message

regexp: uses unmatched as null (BC break)

Run Details

14 of 14 new or added lines in 6 files covered. (100.0%)

101 existing lines in 14 files now uncovered.

2372 of 2572 relevant lines covered (92.22%)

0.92 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

95.8

/src/Texy/Modules/HtmlOutputModule.php

<?php

/**
 * This file is part of the Texy! (https://texy.info)
 * Copyright (c) 2004 David Grudl (https://davidgrudl.com)
 */

declare(strict_types=1);

namespace Texy\Modules;

use Texy;
use Texy\HtmlElement;
use Texy\Regexp;


/**
 * HTML output
 */
final class HtmlOutputModule extends Texy\Module
{
        /** indent HTML code? */
        public bool $indent = true;

        /** @var string[] */
        public array $preserveSpaces = ['textarea', 'pre', 'script', 'code', 'samp', 'kbd'];

        /** base indent level */
        public int $baseIndent = 0;

        /** wrap width, doesn't include indent space */
        public int $lineWrap = 80;

        /** indent space counter */
        private int $space = 0;

        /** @var array<string, int> */
        private array $tagUsed = [];

        /** @var array<int, array{tag: string, open: string, close: string, dtdContent: array<string, int>, indent: int}> */
        private array $tagStack = [];

        /** @var array<string, int>  content DTD used, when context is not defined */
        private array $baseDTD = [];


        public function __construct(Texy\Texy $texy)
        {
                $this->texy = $texy;
                $texy->addHandler('postProcess', $this->postProcess(...));
        }


        /**
         * Converts <strong><em> ... </strong> ... </em>.
         * into <strong><em> ... </em></strong><em> ... </em>
         */
        private function postProcess(Texy\Texy $texy, string &$s): void
        {
                $this->space = $this->baseIndent;
                $this->tagStack = [];
                $this->tagUsed = [];

                // special "base content"
                $dtd = $texy->getDTD();
                $this->baseDTD = $dtd['div'][1] + $dtd['html'][1] /*+ $dtd['head'][1]*/ + $dtd['body'][1] + ['html' => 1];

                // wellform and reformat
                $s = Regexp::replace(
                        $s . '</end/>',
                        '~
                                ( [^<]*+ )
                                < (?: (!--.*--) | (/?) ([a-z][a-z0-9._:-]*) (|[ \n].*) \s* (/?) ) >
                        ~Uis',
                        $this->cb(...),
                );

                // empty out stack
                foreach ($this->tagStack as $item) {
                        $s .= $item['close'];
                }

                // right trim
                $s = Regexp::replace($s, '~[\t ]+(\n|\r|$)~', '$1'); // right trim

                // join double \r to single \n
                $s = str_replace("\r\r", "\n", $s);
                $s = strtr($s, "\r", "\n");

                // greedy chars
                $s = Regexp::replace($s, '~\x07\ *~', '');
                // back-tabs
                $s = Regexp::replace($s, '~\t?\ *\x08~', '');

                // line wrap
                if ($this->lineWrap > 0) {
                        $s = Regexp::replace(
                                $s,
                                '~^(\t*)(.*)$~m',
                                $this->wrap(...),
                        );
                }
        }


        /**
         * Callback function: <tag> | </tag> | ....
         */
        private function cb(array $matches): string
        {
                // html tag
                [, $mText, $mComment, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;
                // [1] => text
                // [1] => !-- comment --
                // [2] => /
                // [3] => TAG
                // [4] => ... (attributes)
                // [5] => / (empty)

                $s = '';

                // phase #1 - stuff between tags
                if ($mText !== '') {
                        $item = reset($this->tagStack);
                        if ($item && !isset($item['dtdContent'][HtmlElement::InnerText])) {  // text not allowed?

                        } elseif (array_intersect(array_keys($this->tagUsed, true, false), $this->preserveSpaces)) { // inside pre & textarea preserve spaces
                                $s = Texy\Helpers::freezeSpaces($mText);

                        } else {
                                $s = Regexp::replace($mText, '~[ \n]+~', ' '); // otherwise shrink multiple spaces
                        }
                }

                // phase #2 - HTML comment
                if ($mComment) {
                        return $s . '<' . Texy\Helpers::freezeSpaces($mComment) . '>';
                }

                // phase #3 - HTML tag
                $mEmpty = $mEmpty || isset(HtmlElement::$emptyElements[$mTag]);
                if ($mEmpty && $mEnd) { // bad tag; /end/
                        return $s;
                } elseif ($mEnd) {
                        return $s . $this->processEndTag($mTag);
                } else {
                        return $this->processStartTag($mTag, $mEmpty, $mAttr, $s);
                }
        }


        private function processStartTag(string $tag, bool $empty, string $attr, string $s): string
        {
                $dtdContent = $this->baseDTD;
                $dtd = $this->texy->getDTD();

                if (!isset($dtd[$tag])) {
                        // unknown (non-html) tag
                        $allowed = true;
                        $item = reset($this->tagStack);
                        if ($item) {
                                $dtdContent = $item['dtdContent'];
                        }
                } else {
                        $s .= $this->closeOptionalTags($tag, $dtdContent);

                        // is tag allowed in this content?
                        $allowed = isset($dtdContent[$tag]);

                        // check deep element prohibitions
                        if ($allowed && isset(HtmlElement::$prohibits[$tag])) {
                                foreach (HtmlElement::$prohibits[$tag] as $pTag) {
                                        if (!empty($this->tagUsed[$pTag])) {
                                                $allowed = false;
                                                break;
                                        }
                                }
                        }
                }

                // empty elements se neukladaji do zasobniku
                if ($empty) {
                        if (!$allowed) {
                                return $s;
                        }

                        $indent = $this->indent && !array_intersect(array_keys($this->tagUsed, true, false), $this->preserveSpaces);

                        if ($indent && $tag === 'br') { // formatting exception
                                return rtrim($s) . '<' . $tag . $attr . ">\n" . str_repeat("\t", max(0, $this->space - 1)) . "\x07";

                        } elseif ($indent && !isset(HtmlElement::$inlineElements[$tag])) {
                                $space = "\r" . str_repeat("\t", $this->space);
                                return $s . $space . '<' . $tag . $attr . '>' . $space;

                        } else {
                                return $s . '<' . $tag . $attr . '>';
                        }
                }

                $open = null;
                $close = null;
                $indent = 0;

                if ($allowed) {
                        $open = '<' . $tag . $attr . '>';

                        // receive new content
                        if ($tagDTD = $dtd[$tag] ?? null) {
                                if (isset($tagDTD[1][HtmlElement::InnerTransparent])) {
                                        $dtdContent += $tagDTD[1];
                                        unset($dtdContent[HtmlElement::InnerTransparent]);
                                } else {
                                        $dtdContent = $tagDTD[1];
                                }
                        }

                        // format output
                        if ($this->indent && !isset(HtmlElement::$inlineElements[$tag])) {
                                $close = "\x08" . '</' . $tag . '>' . "\n" . str_repeat("\t", $this->space);
                                $s .= "\n" . str_repeat("\t", $this->space++) . $open . "\x07";
                                $indent = 1;
                        } else {
                                $close = '</' . $tag . '>';
                                $s .= $open;
                        }

                        // TODO: problematic formatting of select / options, object / params
                }

                // open tag, put to stack, increase counter
                $item = [
                        'tag' => $tag,
                        'open' => $open,
                        'close' => $close,
                        'dtdContent' => $dtdContent,
                        'indent' => $indent,
                ];
                array_unshift($this->tagStack, $item);
                $tmp = &$this->tagUsed[$tag];
                $tmp++;

                return $s;
        }


        private function processEndTag(string $tag): string
        {
                // has start tag?
                if (empty($this->tagUsed[$tag])) {
                        return '';
                }

                // autoclose tags
                $tmp = [];
                $back = true;
                $s = '';
                foreach ($this->tagStack as $i => $item) {
                        $itemTag = $item['tag'];
                        $s .= $item['close'];
                        $this->space -= $item['indent'];
                        $this->tagUsed[$itemTag]--;
                        $back = $back && isset(HtmlElement::$inlineElements[$itemTag]);
                        unset($this->tagStack[$i]);
                        if ($itemTag === $tag) {
                                break;
                        }

                        array_unshift($tmp, $item);
                }

                if (!$back || !$tmp) {
                        return $s;
                }

                // allowed-check (nejspis neni ani potreba)
                $item = reset($this->tagStack);
                $dtdContent = $item ? $item['dtdContent'] : $this->baseDTD;
                if (!isset($dtdContent[$tmp[0]['tag']])) {
                        return $s;
                }

                // autoopen tags
                foreach ($tmp as $item) {
                        $s .= $item['open'];
                        $this->space += $item['indent'];
                        $this->tagUsed[$item['tag']]++;
                        array_unshift($this->tagStack, $item);
                }

                return $s;
        }


        private function closeOptionalTags(string $tag, array &$dtdContent): string
        {
                $s = '';
                foreach ($this->tagStack as $i => $item) {
                        // is tag allowed here?
                        $dtdContent = $item['dtdContent'];
                        if (isset($dtdContent[$tag])) {
                                break;
                        }

                        $itemTag = $item['tag'];

                        // auto-close hidden, optional and inline tags
                        if (
                                $item['close']
                                && (
                                        !isset(HtmlElement::$optionalEnds[$itemTag])
                                        && !isset(HtmlElement::$inlineElements[$itemTag])
                                )
                        ) {
                                break;
                        }

                        // close it
                        $s .= $item['close'];
                        $this->space -= $item['indent'];
                        $this->tagUsed[$itemTag]--;
                        unset($this->tagStack[$i]);
                        $dtdContent = $this->baseDTD;
                }

                return $s;
        }


        /**
         * Callback function: wrap lines.
         */
        private function wrap(array $m): string
        {
                [, $space, $s] = $m;
                return $space . wordwrap($s, $this->lineWrap, "\n" . $space);
        }
}

1	<?php
2
3	/**
4	* This file is part of the Texy! (https://texy.info)
5	* Copyright (c) 2004 David Grudl (https://davidgrudl.com)
6	*/
7
8	declare(strict_types=1);
9
10	namespace Texy\Modules;
11
12	use Texy;
13	use Texy\HtmlElement;
14	use Texy\Regexp;
15
16
17	/**
18	* HTML output
19	*/
20	final class HtmlOutputModule extends Texy\Module
21	{
22	/** indent HTML code? */
23	public bool $indent = true;
24
25	/** @var string[] */
26	public array $preserveSpaces = ['textarea', 'pre', 'script', 'code', 'samp', 'kbd'];
27
28	/** base indent level */
29	public int $baseIndent = 0;
30
31	/** wrap width, doesn't include indent space */
32	public int $lineWrap = 80;
33
34	/** indent space counter */
35	private int $space = 0;
36
37	/** @var array<string, int> */
38	private array $tagUsed = [];
39
40	/** @var array<int, array{tag: string, open: string, close: string, dtdContent: array<string, int>, indent: int}> */
41	private array $tagStack = [];
42
43	/** @var array<string, int> content DTD used, when context is not defined */
44	private array $baseDTD = [];
45
46
47	public function __construct(Texy\Texy $texy)	1✔
48	{
49	$this->texy = $texy;	1✔
50	$texy->addHandler('postProcess', $this->postProcess(...));	1✔
51	}	1✔
52
53
54	/**
55	* Converts <strong><em> ... </strong> ... </em>.
56	* into <strong><em> ... </em></strong><em> ... </em>
57	*/
58	private function postProcess(Texy\Texy $texy, string &$s): void	1✔
59	{
60	$this->space = $this->baseIndent;	1✔
61	$this->tagStack = [];	1✔
62	$this->tagUsed = [];	1✔
63
64	// special "base content"
65	$dtd = $texy->getDTD();	1✔
66	$this->baseDTD = $dtd['div'][1] + $dtd['html'][1] /+ $dtd['head'][1]/ + $dtd['body'][1] + ['html' => 1];	1✔
67
68	// wellform and reformat
69	$s = Regexp::replace(	1✔
70	$s . '</end/>',	1✔
71	'~	1✔
72	( [^<]*+ )
73	< (?: (!--.--) \| (/?) ([a-z][a-z0-9._:-]) (\|[ \n].) \s (/?) ) >
74	~Uis',
75	$this->cb(...),	1✔
76	);
77
78	// empty out stack
79	foreach ($this->tagStack as $item) {	1✔
80	$s .= $item['close'];	1✔
81	}
82
83	// right trim
84	$s = Regexp::replace($s, '~[\t ]+(\n\|\r\|$)~', '$1'); // right trim	1✔
85
86	// join double \r to single \n
87	$s = str_replace("\r\r", "\n", $s);	1✔
88	$s = strtr($s, "\r", "\n");	1✔
89
90	// greedy chars
91	$s = Regexp::replace($s, '~\x07\ *~', '');	1✔
92	// back-tabs
93	$s = Regexp::replace($s, '~\t?\ *\x08~', '');	1✔
94
95	// line wrap
96	if ($this->lineWrap > 0) {	1✔
97	$s = Regexp::replace(	1✔
98	$s,	1✔
99	'~^(\t)(.)$~m',	1✔
100	$this->wrap(...),	1✔
101	);
102	}
103	}	1✔
104
105
106	/**
107	* Callback function: <tag> \| </tag> \| ....
108	*/
109	private function cb(array $matches): string	1✔
110	{
111	// html tag
112	[, $mText, $mComment, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
113	// [1] => text
114	// [1] => !-- comment --
115	// [2] => /
116	// [3] => TAG
117	// [4] => ... (attributes)
118	// [5] => / (empty)
119
120	$s = '';	1✔
121
122	// phase #1 - stuff between tags
123	if ($mText !== '') {	1✔
124	$item = reset($this->tagStack);	1✔
125	if ($item && !isset($item['dtdContent'][HtmlElement::InnerText])) { // text not allowed?	1✔
126
127	} elseif (array_intersect(array_keys($this->tagUsed, true, false), $this->preserveSpaces)) { // inside pre & textarea preserve spaces	1✔
128	$s = Texy\Helpers::freezeSpaces($mText);	1✔
129
130	} else {
131	$s = Regexp::replace($mText, '~[ \n]+~', ' '); // otherwise shrink multiple spaces	1✔
132	}
133	}
134
135	// phase #2 - HTML comment
136	if ($mComment) {	1✔
137	return $s . '<' . Texy\Helpers::freezeSpaces($mComment) . '>';	1✔
138	}
139
140	// phase #3 - HTML tag
141	$mEmpty = $mEmpty \|\| isset(HtmlElement::$emptyElements[$mTag]);	1✔
142	if ($mEmpty && $mEnd) { // bad tag; /end/	1✔
143	return $s;	1✔
144	} elseif ($mEnd) {	1✔
145	return $s . $this->processEndTag($mTag);	1✔
146	} else {
147	return $this->processStartTag($mTag, $mEmpty, $mAttr, $s);	1✔
148	}
149	}
150
151
152	private function processStartTag(string $tag, bool $empty, string $attr, string $s): string	1✔
153	{
154	$dtdContent = $this->baseDTD;	1✔
155	$dtd = $this->texy->getDTD();	1✔
156
157	if (!isset($dtd[$tag])) {	1✔
158	// unknown (non-html) tag
159	$allowed = true;	×
UNCOV 160	$item = reset($this->tagStack);	×
UNCOV 161	if ($item) {	×
UNCOV 162	$dtdContent = $item['dtdContent'];	×
163	}
164	} else {
165	$s .= $this->closeOptionalTags($tag, $dtdContent);	1✔
166
167	// is tag allowed in this content?
168	$allowed = isset($dtdContent[$tag]);	1✔
169
170	// check deep element prohibitions
171	if ($allowed && isset(HtmlElement::$prohibits[$tag])) {	1✔
172	foreach (HtmlElement::$prohibits[$tag] as $pTag) {	1✔
173	if (!empty($this->tagUsed[$pTag])) {	1✔
174	$allowed = false;	1✔
175	break;	1✔
176	}
177	}
178	}
179	}
180
181	// empty elements se neukladaji do zasobniku
182	if ($empty) {	1✔
183	if (!$allowed) {	1✔
UNCOV 184	return $s;	×
185	}
186
187	$indent = $this->indent && !array_intersect(array_keys($this->tagUsed, true, false), $this->preserveSpaces);	1✔
188
189	if ($indent && $tag === 'br') { // formatting exception	1✔
190	return rtrim($s) . '<' . $tag . $attr . ">\n" . str_repeat("\t", max(0, $this->space - 1)) . "\x07";	1✔
191
192	} elseif ($indent && !isset(HtmlElement::$inlineElements[$tag])) {	1✔
193	$space = "\r" . str_repeat("\t", $this->space);	1✔
194	return $s . $space . '<' . $tag . $attr . '>' . $space;	1✔
195
196	} else {
197	return $s . '<' . $tag . $attr . '>';	1✔
198	}
199	}
200
201	$open = null;	1✔
202	$close = null;	1✔
203	$indent = 0;	1✔
204
205	if ($allowed) {	1✔
206	$open = '<' . $tag . $attr . '>';	1✔
207
208	// receive new content
209	if ($tagDTD = $dtd[$tag] ?? null) {	1✔
210	if (isset($tagDTD[1][HtmlElement::InnerTransparent])) {	1✔
211	$dtdContent += $tagDTD[1];	1✔
212	unset($dtdContent[HtmlElement::InnerTransparent]);	1✔
213	} else {
214	$dtdContent = $tagDTD[1];	1✔
215	}
216	}
217
218	// format output
219	if ($this->indent && !isset(HtmlElement::$inlineElements[$tag])) {	1✔
220	$close = "\x08" . '</' . $tag . '>' . "\n" . str_repeat("\t", $this->space);	1✔
221	$s .= "\n" . str_repeat("\t", $this->space++) . $open . "\x07";	1✔
222	$indent = 1;	1✔
223	} else {
224	$close = '</' . $tag . '>';	1✔
225	$s .= $open;	1✔
226	}
227
228	// TODO: problematic formatting of select / options, object / params
229	}
230
231	// open tag, put to stack, increase counter
232	$item = [	1✔
233	'tag' => $tag,	1✔
234	'open' => $open,	1✔
235	'close' => $close,	1✔
236	'dtdContent' => $dtdContent,	1✔
237	'indent' => $indent,	1✔
238	];
239	array_unshift($this->tagStack, $item);	1✔
240	$tmp = &$this->tagUsed[$tag];	1✔
241	$tmp++;	1✔
242
243	return $s;	1✔
244	}
245
246
247	private function processEndTag(string $tag): string	1✔
248	{
249	// has start tag?
250	if (empty($this->tagUsed[$tag])) {	1✔
251	return '';	1✔
252	}
253
254	// autoclose tags
255	$tmp = [];	1✔
256	$back = true;	1✔
257	$s = '';	1✔
258	foreach ($this->tagStack as $i => $item) {	1✔
259	$itemTag = $item['tag'];	1✔
260	$s .= $item['close'];	1✔
261	$this->space -= $item['indent'];	1✔
262	$this->tagUsed[$itemTag]--;	1✔
263	$back = $back && isset(HtmlElement::$inlineElements[$itemTag]);	1✔
264	unset($this->tagStack[$i]);	1✔
265	if ($itemTag === $tag) {	1✔
266	break;	1✔
267	}
268
269	array_unshift($tmp, $item);	1✔
270	}
271
272	if (!$back \|\| !$tmp) {	1✔
273	return $s;	1✔
274	}
275
276	// allowed-check (nejspis neni ani potreba)
277	$item = reset($this->tagStack);	1✔
278	$dtdContent = $item ? $item['dtdContent'] : $this->baseDTD;	1✔
279	if (!isset($dtdContent[$tmp[0]['tag']])) {	1✔
UNCOV 280	return $s;	×
281	}
282
283	// autoopen tags
284	foreach ($tmp as $item) {	1✔
285	$s .= $item['open'];	1✔
286	$this->space += $item['indent'];	1✔
287	$this->tagUsed[$item['tag']]++;	1✔
288	array_unshift($this->tagStack, $item);	1✔
289	}
290
291	return $s;	1✔
292	}
293
294
295	private function closeOptionalTags(string $tag, array &$dtdContent): string	1✔
296	{
297	$s = '';	1✔
298	foreach ($this->tagStack as $i => $item) {	1✔
299	// is tag allowed here?
300	$dtdContent = $item['dtdContent'];	1✔
301	if (isset($dtdContent[$tag])) {	1✔
302	break;	1✔
303	}
304
305	$itemTag = $item['tag'];	1✔
306
307	// auto-close hidden, optional and inline tags
308	if (
309	$item['close']	1✔
310	&& (
311	!isset(HtmlElement::$optionalEnds[$itemTag])	1✔
312	&& !isset(HtmlElement::$inlineElements[$itemTag])	1✔
313	)
314	) {
315	break;	1✔
316	}
317
318	// close it
319	$s .= $item['close'];	1✔
320	$this->space -= $item['indent'];	1✔
321	$this->tagUsed[$itemTag]--;	1✔
322	unset($this->tagStack[$i]);	1✔
323	$dtdContent = $this->baseDTD;	1✔
324	}
325
326	return $s;	1✔
327	}
328
329
330	/**
331	* Callback function: wrap lines.
332	*/
333	private function wrap(array $m): string	1✔
334	{
335	[, $space, $s] = $m;	1✔
336	return $space . wordwrap($s, $this->lineWrap, "\n" . $space);	1✔
337	}
338	}

dg / texy / 12879605443

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous