aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/3rdparty/libraries/MOBIClass
diff options
context:
space:
mode:
authorThomas Citharel <tcit@tcit.fr>2014-12-19 21:08:44 +0100
committerThomas Citharel <tcit@tcit.fr>2014-12-19 21:08:44 +0100
commit3dcd85c075998ecdf2d54c5661c42e24080067d5 (patch)
treeb939cdbf86c1be64ef64d8afb3efd0d55e8a945e /inc/3rdparty/libraries/MOBIClass
parent44b95cb81deae35f58e0058910afde2f2ffb9a60 (diff)
parente9a64ef8a9776becbe4c621ee4bd762f24b1bc3e (diff)
downloadwallabag-3dcd85c075998ecdf2d54c5661c42e24080067d5.tar.gz
wallabag-3dcd85c075998ecdf2d54c5661c42e24080067d5.tar.zst
wallabag-3dcd85c075998ecdf2d54c5661c42e24080067d5.zip
merge
Diffstat (limited to 'inc/3rdparty/libraries/MOBIClass')
-rw-r--r--inc/3rdparty/libraries/MOBIClass/CharacterEntities.php413
-rw-r--r--inc/3rdparty/libraries/MOBIClass/ContentProvider.php22
-rw-r--r--inc/3rdparty/libraries/MOBIClass/EXTHHelper.php132
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileByte.php41
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileDate.php40
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileElement.php89
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileInt.php40
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileObject.php168
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileRecord.php46
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileShort.php41
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileString.php83
-rw-r--r--inc/3rdparty/libraries/MOBIClass/FileTri.php41
-rw-r--r--inc/3rdparty/libraries/MOBIClass/Http.php171
-rw-r--r--inc/3rdparty/libraries/MOBIClass/ImageHandler.php40
-rw-r--r--inc/3rdparty/libraries/MOBIClass/MOBI.php192
-rw-r--r--inc/3rdparty/libraries/MOBIClass/MOBIFile.php157
-rw-r--r--inc/3rdparty/libraries/MOBIClass/MultipleFileHandler.php136
-rw-r--r--inc/3rdparty/libraries/MOBIClass/OnlineArticle.php116
-rw-r--r--inc/3rdparty/libraries/MOBIClass/PalmRecord.php136
-rw-r--r--inc/3rdparty/libraries/MOBIClass/Prc.php97
-rw-r--r--inc/3rdparty/libraries/MOBIClass/PreprocessedArticle.php89
-rw-r--r--inc/3rdparty/libraries/MOBIClass/RecognizeURL.php16
-rw-r--r--inc/3rdparty/libraries/MOBIClass/Record.php96
-rw-r--r--inc/3rdparty/libraries/MOBIClass/RecordFactory.php115
-rw-r--r--inc/3rdparty/libraries/MOBIClass/Settings.php97
-rw-r--r--inc/3rdparty/libraries/MOBIClass/constants.php11
-rw-r--r--inc/3rdparty/libraries/MOBIClass/downloaders/FanFictionNet.php125
-rw-r--r--inc/3rdparty/libraries/MOBIClass/http_build_url.php94
-rw-r--r--inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php110
-rw-r--r--inc/3rdparty/libraries/MOBIClass/readability/Readability.php1069
30 files changed, 4023 insertions, 0 deletions
diff --git a/inc/3rdparty/libraries/MOBIClass/CharacterEntities.php b/inc/3rdparty/libraries/MOBIClass/CharacterEntities.php
new file mode 100644
index 00000000..de896f87
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/CharacterEntities.php
@@ -0,0 +1,413 @@
1<?php
2
3/**
4 * Description of CharacterEntities
5 *
6 * @author Sander
7 */
8class CharacterEntities {
9 public static function convert($str){
10 //Assume the encoding is UTF-8 -> output is UTF-8
11 return $str;
12 //return utf8_encode($str);
13 //Convert to CP1252
14 list($from, $to) = CharacterEntities::generateTables();
15 return str_replace($from, $to, $str);
16 }
17
18 private static function generateTables(){
19 $from = array();
20 $to = array();
21
22 for($i = 0; $i < 256; $i++){
23 $from[$i] = $to[$i] = chr($i);
24 }
25
26 $from[0x80] = "€";
27 $from[0x82] = "‚";
28 $from[0x83] = "ƒ";
29 $from[0x84] = "„";
30 $from[0x85] = "…";
31 $from[0x86] = "†";
32 $from[0x87] = "‡";
33 $from[0x88] = "ˆ";
34 $from[0x89] = "‰";
35 $from[0x8A] = "Š";
36 $from[0x8B] = "‹";
37 $from[0x8C] = "Œ";
38 $from[0x8E] = "Ž";
39
40 $from[0x91] = "‘";
41 $from[0x92] = "’";
42 $from[0x93] = "“";
43 $from[0x94] = "”";
44 $from[0x95] = "•";
45 $from[0x96] = "–";
46 $from[0x97] = "—";
47 $from[0x98] = "˜";
48 $from[0x99] = "™";
49 $from[0x9A] = "š";
50 $from[0x9B] = "›";
51 $from[0x9C] = "œ";
52 $from[0x9E] = "ž";
53 $from[0x9F] = "Ÿ";
54
55 $from[0xA1] = "¡";
56 $from[0xA2] = "¢";
57 $from[0xA3] = "£";
58 $from[0xA4] = "¤";
59 $from[0xA5] = "¥";
60 $from[0xA6] = "¦";
61 $from[0xA7] = "§";
62 $from[0xA8] = "¨";
63 $from[0xA9] = "©";
64 $from[0xAA] = "ª";
65 $from[0xAB] = "«";
66 $from[0xAC] = "¬";
67 $from[0xAE] = "®";
68 $from[0xAF] = "¯";
69
70 $from[0xB0] = "°";
71 $from[0xB1] = "±";
72 $from[0xB2] = "²";
73 $from[0xB3] = "³";
74 $from[0xB4] = "´";
75 $from[0xB5] = "µ";
76 $from[0xB6] = "¶";
77 $from[0xB7] = "·";
78 $from[0xB8] = "¸";
79 $from[0xB9] = "¹";
80 $from[0xBA] = "º";
81 $from[0xBB] = "»";
82 $from[0xBC] = "¼";
83 $from[0xBD] = "½";
84 $from[0xBE] = "¾";
85 $from[0xBF] = "¿";
86
87 $from[0xC0] = "À";
88 $from[0xC1] = "Á";
89 $from[0xC2] = "Â";
90 $from[0xC3] = "Ã";
91 $from[0xC4] = "Ä";
92 $from[0xC5] = "Å";
93 $from[0xC6] = "Æ";
94 $from[0xC7] = "Ç";
95 $from[0xC8] = "È";
96 $from[0xC9] = "É";
97 $from[0xCA] = "Ê";
98 $from[0xCB] = "Ë";
99 $from[0xCC] = "Ì";
100 $from[0xCD] = "Í";
101 $from[0xCE] = "Î";
102 $from[0xCF] = "Ï";
103
104 $from[0xD0] = "Ð";
105 $from[0xD1] = "Ñ";
106 $from[0xD2] = "Ò";
107 $from[0xD3] = "Ó";
108 $from[0xD4] = "Ô";
109 $from[0xD5] = "Õ";
110 $from[0xD6] = "Ö";
111 $from[0xD7] = "×";
112 $from[0xD8] = "Ø";
113 $from[0xD9] = "Ù";
114 $from[0xDA] = "Ú";
115 $from[0xDB] = "Û";
116 $from[0xDC] = "Ü";
117 $from[0xDD] = "Ý";
118 $from[0xDE] = "Þ";
119 $from[0xDF] = "ß";
120
121 $from[0xE0] = "à";
122 $from[0xE1] = "á";
123 $from[0xE2] = "â";
124 $from[0xE3] = "ã";
125 $from[0xE4] = "ä";
126 $from[0xE5] = "å";
127 $from[0xE6] = "æ";
128 $from[0xE7] = "ç";
129 $from[0xE8] = "è";
130 $from[0xE9] = "é";
131 $from[0xEA] = "ê";
132 $from[0xEB] = "ë";
133 $from[0xEC] = "ì";
134 $from[0xED] = "í";
135 $from[0xEE] = "î";
136 $from[0xEF] = "ï";
137
138 $from[0xF0] = "ð";
139 $from[0xF1] = "ñ";
140 $from[0xF2] = "ò";
141 $from[0xF3] = "ó";
142 $from[0xF4] = "ô";
143 $from[0xF5] = "õ";
144 $from[0xF6] = "ö";
145 $from[0xF7] = "÷";
146 $from[0xF8] = "ø";
147 $from[0xF9] = "ù";
148 $from[0xFA] = "ú";
149 $from[0xFB] = "û";
150 $from[0xFC] = "ü";
151 $from[0xFD] = "ý";
152 $from[0xFE] = "þ";
153 $from[0xFF] = "ÿ";
154
155
156 return array($from, $to);
157 }
158 /*
159 00 = U+0000 : NULL
16001 = U+0001 : START OF HEADING
16102 = U+0002 : START OF TEXT
16203 = U+0003 : END OF TEXT
16304 = U+0004 : END OF TRANSMISSION
16405 = U+0005 : ENQUIRY
16506 = U+0006 : ACKNOWLEDGE
16607 = U+0007 : BELL
16708 = U+0008 : BACKSPACE
16809 = U+0009 : HORIZONTAL TABULATION
1690A = U+000A : LINE FEED
1700B = U+000B : VERTICAL TABULATION
1710C = U+000C : FORM FEED
1720D = U+000D : CARRIAGE RETURN
1730E = U+000E : SHIFT OUT
1740F = U+000F : SHIFT IN
17510 = U+0010 : DATA LINK ESCAPE
17611 = U+0011 : DEVICE CONTROL ONE
17712 = U+0012 : DEVICE CONTROL TWO
17813 = U+0013 : DEVICE CONTROL THREE
17914 = U+0014 : DEVICE CONTROL FOUR
18015 = U+0015 : NEGATIVE ACKNOWLEDGE
18116 = U+0016 : SYNCHRONOUS IDLE
18217 = U+0017 : END OF TRANSMISSION BLOCK
18318 = U+0018 : CANCEL
18419 = U+0019 : END OF MEDIUM
1851A = U+001A : SUBSTITUTE
1861B = U+001B : ESCAPE
1871C = U+001C : FILE SEPARATOR
1881D = U+001D : GROUP SEPARATOR
1891E = U+001E : RECORD SEPARATOR
1901F = U+001F : UNIT SEPARATOR
19120 = U+0020 : SPACE
19221 = U+0021 : EXCLAMATION MARK
19322 = U+0022 : QUOTATION MARK
19423 = U+0023 : NUMBER SIGN
19524 = U+0024 : DOLLAR SIGN
19625 = U+0025 : PERCENT SIGN
19726 = U+0026 : AMPERSAND
19827 = U+0027 : APOSTROPHE
19928 = U+0028 : LEFT PARENTHESIS
20029 = U+0029 : RIGHT PARENTHESIS
2012A = U+002A : ASTERISK
2022B = U+002B : PLUS SIGN
2032C = U+002C : COMMA
2042D = U+002D : HYPHEN-MINUS
2052E = U+002E : FULL STOP
2062F = U+002F : SOLIDUS
20730 = U+0030 : DIGIT ZERO
20831 = U+0031 : DIGIT ONE
20932 = U+0032 : DIGIT TWO
21033 = U+0033 : DIGIT THREE
21134 = U+0034 : DIGIT FOUR
21235 = U+0035 : DIGIT FIVE
21336 = U+0036 : DIGIT SIX
21437 = U+0037 : DIGIT SEVEN
21538 = U+0038 : DIGIT EIGHT
21639 = U+0039 : DIGIT NINE
2173A = U+003A : COLON
2183B = U+003B : SEMICOLON
2193C = U+003C : LESS-THAN SIGN
2203D = U+003D : EQUALS SIGN
2213E = U+003E : GREATER-THAN SIGN
2223F = U+003F : QUESTION MARK
22340 = U+0040 : COMMERCIAL AT
22441 = U+0041 : LATIN CAPITAL LETTER A
22542 = U+0042 : LATIN CAPITAL LETTER B
22643 = U+0043 : LATIN CAPITAL LETTER C
22744 = U+0044 : LATIN CAPITAL LETTER D
22845 = U+0045 : LATIN CAPITAL LETTER E
22946 = U+0046 : LATIN CAPITAL LETTER F
23047 = U+0047 : LATIN CAPITAL LETTER G
23148 = U+0048 : LATIN CAPITAL LETTER H
23249 = U+0049 : LATIN CAPITAL LETTER I
2334A = U+004A : LATIN CAPITAL LETTER J
2344B = U+004B : LATIN CAPITAL LETTER K
2354C = U+004C : LATIN CAPITAL LETTER L
2364D = U+004D : LATIN CAPITAL LETTER M
2374E = U+004E : LATIN CAPITAL LETTER N
2384F = U+004F : LATIN CAPITAL LETTER O
23950 = U+0050 : LATIN CAPITAL LETTER P
24051 = U+0051 : LATIN CAPITAL LETTER Q
24152 = U+0052 : LATIN CAPITAL LETTER R
24253 = U+0053 : LATIN CAPITAL LETTER S
24354 = U+0054 : LATIN CAPITAL LETTER T
24455 = U+0055 : LATIN CAPITAL LETTER U
24556 = U+0056 : LATIN CAPITAL LETTER V
24657 = U+0057 : LATIN CAPITAL LETTER W
24758 = U+0058 : LATIN CAPITAL LETTER X
24859 = U+0059 : LATIN CAPITAL LETTER Y
2495A = U+005A : LATIN CAPITAL LETTER Z
2505B = U+005B : LEFT SQUARE BRACKET
2515C = U+005C : REVERSE SOLIDUS
2525D = U+005D : RIGHT SQUARE BRACKET
2535E = U+005E : CIRCUMFLEX ACCENT
2545F = U+005F : LOW LINE
25560 = U+0060 : GRAVE ACCENT
25661 = U+0061 : LATIN SMALL LETTER A
25762 = U+0062 : LATIN SMALL LETTER B
25863 = U+0063 : LATIN SMALL LETTER C
25964 = U+0064 : LATIN SMALL LETTER D
26065 = U+0065 : LATIN SMALL LETTER E
26166 = U+0066 : LATIN SMALL LETTER F
26267 = U+0067 : LATIN SMALL LETTER G
26368 = U+0068 : LATIN SMALL LETTER H
26469 = U+0069 : LATIN SMALL LETTER I
2656A = U+006A : LATIN SMALL LETTER J
2666B = U+006B : LATIN SMALL LETTER K
2676C = U+006C : LATIN SMALL LETTER L
2686D = U+006D : LATIN SMALL LETTER M
2696E = U+006E : LATIN SMALL LETTER N
2706F = U+006F : LATIN SMALL LETTER O
27170 = U+0070 : LATIN SMALL LETTER P
27271 = U+0071 : LATIN SMALL LETTER Q
27372 = U+0072 : LATIN SMALL LETTER R
27473 = U+0073 : LATIN SMALL LETTER S
27574 = U+0074 : LATIN SMALL LETTER T
27675 = U+0075 : LATIN SMALL LETTER U
27776 = U+0076 : LATIN SMALL LETTER V
27877 = U+0077 : LATIN SMALL LETTER W
27978 = U+0078 : LATIN SMALL LETTER X
28079 = U+0079 : LATIN SMALL LETTER Y
2817A = U+007A : LATIN SMALL LETTER Z
2827B = U+007B : LEFT CURLY BRACKET
2837C = U+007C : VERTICAL LINE
2847D = U+007D : RIGHT CURLY BRACKET
2857E = U+007E : TILDE
2867F = U+007F : DELETE
28780 = U+20AC : EURO SIGN
28882 = U+201A : SINGLE LOW-9 QUOTATION MARK
28983 = U+0192 : LATIN SMALL LETTER F WITH HOOK
29084 = U+201E : DOUBLE LOW-9 QUOTATION MARK
29185 = U+2026 : HORIZONTAL ELLIPSIS
29286 = U+2020 : DAGGER
29387 = U+2021 : DOUBLE DAGGER
29488 = U+02C6 : MODIFIER LETTER CIRCUMFLEX ACCENT
29589 = U+2030 : PER MILLE SIGN
2968A = U+0160 : LATIN CAPITAL LETTER S WITH CARON
2978B = U+2039 : SINGLE LEFT-POINTING ANGLE QUOTATION MARK
2988C = U+0152 : LATIN CAPITAL LIGATURE OE
2998E = U+017D : LATIN CAPITAL LETTER Z WITH CARON
30091 = U+2018 : LEFT SINGLE QUOTATION MARK
30192 = U+2019 : RIGHT SINGLE QUOTATION MARK
30293 = U+201C : LEFT DOUBLE QUOTATION MARK
30394 = U+201D : RIGHT DOUBLE QUOTATION MARK
30495 = U+2022 : BULLET
30596 = U+2013 : EN DASH
30697 = U+2014 : EM DASH
30798 = U+02DC : SMALL TILDE
30899 = U+2122 : TRADE MARK SIGN
3099A = U+0161 : LATIN SMALL LETTER S WITH CARON
3109B = U+203A : SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
3119C = U+0153 : LATIN SMALL LIGATURE OE
3129E = U+017E : LATIN SMALL LETTER Z WITH CARON
3139F = U+0178 : LATIN CAPITAL LETTER Y WITH DIAERESIS
314A0 = U+00A0 : NO-BREAK SPACE
315A1 = U+00A1 : INVERTED EXCLAMATION MARK
316A2 = U+00A2 : CENT SIGN
317A3 = U+00A3 : POUND SIGN
318A4 = U+00A4 : CURRENCY SIGN
319A5 = U+00A5 : YEN SIGN
320A6 = U+00A6 : BROKEN BAR
321A7 = U+00A7 : SECTION SIGN
322A8 = U+00A8 : DIAERESIS
323A9 = U+00A9 : COPYRIGHT SIGN
324AA = U+00AA : FEMININE ORDINAL INDICATOR
325AB = U+00AB : LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
326AC = U+00AC : NOT SIGN
327AD = U+00AD : SOFT HYPHEN
328AE = U+00AE : REGISTERED SIGN
329AF = U+00AF : MACRON
330B0 = U+00B0 : DEGREE SIGN
331B1 = U+00B1 : PLUS-MINUS SIGN
332B2 = U+00B2 : SUPERSCRIPT TWO
333B3 = U+00B3 : SUPERSCRIPT THREE
334B4 = U+00B4 : ACUTE ACCENT
335B5 = U+00B5 : MICRO SIGN
336B6 = U+00B6 : PILCROW SIGN
337B7 = U+00B7 : MIDDLE DOT
338B8 = U+00B8 : CEDILLA
339B9 = U+00B9 : SUPERSCRIPT ONE
340BA = U+00BA : MASCULINE ORDINAL INDICATOR
341BB = U+00BB : RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
342BC = U+00BC : VULGAR FRACTION ONE QUARTER
343BD = U+00BD : VULGAR FRACTION ONE HALF
344BE = U+00BE : VULGAR FRACTION THREE QUARTERS
345BF = U+00BF : INVERTED QUESTION MARK
346C0 = U+00C0 : LATIN CAPITAL LETTER A WITH GRAVE
347C1 = U+00C1 : LATIN CAPITAL LETTER A WITH ACUTE
348C2 = U+00C2 : LATIN CAPITAL LETTER A WITH CIRCUMFLEX
349C3 = U+00C3 : LATIN CAPITAL LETTER A WITH TILDE
350C4 = U+00C4 : LATIN CAPITAL LETTER A WITH DIAERESIS
351C5 = U+00C5 : LATIN CAPITAL LETTER A WITH RING ABOVE
352C6 = U+00C6 : LATIN CAPITAL LETTER AE
353C7 = U+00C7 : LATIN CAPITAL LETTER C WITH CEDILLA
354C8 = U+00C8 : LATIN CAPITAL LETTER E WITH GRAVE
355C9 = U+00C9 : LATIN CAPITAL LETTER E WITH ACUTE
356CA = U+00CA : LATIN CAPITAL LETTER E WITH CIRCUMFLEX
357CB = U+00CB : LATIN CAPITAL LETTER E WITH DIAERESIS
358CC = U+00CC : LATIN CAPITAL LETTER I WITH GRAVE
359CD = U+00CD : LATIN CAPITAL LETTER I WITH ACUTE
360CE = U+00CE : LATIN CAPITAL LETTER I WITH CIRCUMFLEX
361CF = U+00CF : LATIN CAPITAL LETTER I WITH DIAERESIS
362D0 = U+00D0 : LATIN CAPITAL LETTER ETH
363D1 = U+00D1 : LATIN CAPITAL LETTER N WITH TILDE
364D2 = U+00D2 : LATIN CAPITAL LETTER O WITH GRAVE
365D3 = U+00D3 : LATIN CAPITAL LETTER O WITH ACUTE
366D4 = U+00D4 : LATIN CAPITAL LETTER O WITH CIRCUMFLEX
367D5 = U+00D5 : LATIN CAPITAL LETTER O WITH TILDE
368D6 = U+00D6 : LATIN CAPITAL LETTER O WITH DIAERESIS
369D7 = U+00D7 : MULTIPLICATION SIGN
370D8 = U+00D8 : LATIN CAPITAL LETTER O WITH STROKE
371D9 = U+00D9 : LATIN CAPITAL LETTER U WITH GRAVE
372DA = U+00DA : LATIN CAPITAL LETTER U WITH ACUTE
373DB = U+00DB : LATIN CAPITAL LETTER U WITH CIRCUMFLEX
374DC = U+00DC : LATIN CAPITAL LETTER U WITH DIAERESIS
375DD = U+00DD : LATIN CAPITAL LETTER Y WITH ACUTE
376DE = U+00DE : LATIN CAPITAL LETTER THORN
377DF = U+00DF : LATIN SMALL LETTER SHARP S
378E0 = U+00E0 : LATIN SMALL LETTER A WITH GRAVE
379E1 = U+00E1 : LATIN SMALL LETTER A WITH ACUTE
380E2 = U+00E2 : LATIN SMALL LETTER A WITH CIRCUMFLEX
381E3 = U+00E3 : LATIN SMALL LETTER A WITH TILDE
382E4 = U+00E4 : LATIN SMALL LETTER A WITH DIAERESIS
383E5 = U+00E5 : LATIN SMALL LETTER A WITH RING ABOVE
384E6 = U+00E6 : LATIN SMALL LETTER AE
385E7 = U+00E7 : LATIN SMALL LETTER C WITH CEDILLA
386E8 = U+00E8 : LATIN SMALL LETTER E WITH GRAVE
387E9 = U+00E9 : LATIN SMALL LETTER E WITH ACUTE
388EA = U+00EA : LATIN SMALL LETTER E WITH CIRCUMFLEX
389EB = U+00EB : LATIN SMALL LETTER E WITH DIAERESIS
390EC = U+00EC : LATIN SMALL LETTER I WITH GRAVE
391ED = U+00ED : LATIN SMALL LETTER I WITH ACUTE
392EE = U+00EE : LATIN SMALL LETTER I WITH CIRCUMFLEX
393EF = U+00EF : LATIN SMALL LETTER I WITH DIAERESIS
394F0 = U+00F0 : LATIN SMALL LETTER ETH
395F1 = U+00F1 : LATIN SMALL LETTER N WITH TILDE
396F2 = U+00F2 : LATIN SMALL LETTER O WITH GRAVE
397F3 = U+00F3 : LATIN SMALL LETTER O WITH ACUTE
398F4 = U+00F4 : LATIN SMALL LETTER O WITH CIRCUMFLEX
399F5 = U+00F5 : LATIN SMALL LETTER O WITH TILDE
400F6 = U+00F6 : LATIN SMALL LETTER O WITH DIAERESIS
401F7 = U+00F7 : DIVISION SIGN
402F8 = U+00F8 : LATIN SMALL LETTER O WITH STROKE
403F9 = U+00F9 : LATIN SMALL LETTER U WITH GRAVE
404FA = U+00FA : LATIN SMALL LETTER U WITH ACUTE
405FB = U+00FB : LATIN SMALL LETTER U WITH CIRCUMFLEX
406FC = U+00FC : LATIN SMALL LETTER U WITH DIAERESIS
407FD = U+00FD : LATIN SMALL LETTER Y WITH ACUTE
408FE = U+00FE : LATIN SMALL LETTER THORN
409FF = U+00FF : LATIN SMALL LETTER Y WITH DIAERESIS
410 *
411 */
412}
413?>
diff --git a/inc/3rdparty/libraries/MOBIClass/ContentProvider.php b/inc/3rdparty/libraries/MOBIClass/ContentProvider.php
new file mode 100644
index 00000000..dcf9c4de
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/ContentProvider.php
@@ -0,0 +1,22 @@
1<?php
2
3abstract class ContentProvider{
4 /**
5 * Get the text data to be integrated in the MOBI file
6 * @return string
7 */
8 public abstract function getTextData();
9 /**
10 * Get the images (an array containing the jpeg data). Array entry 0 will
11 * correspond to image record 0.
12 * @return array
13 */
14 public abstract function getImages();
15 /**
16 * Get the metadata in the form of a hashtable (for example, title or author).
17 * @return array
18 */
19 public abstract function getMetaData();
20}
21
22?>
diff --git a/inc/3rdparty/libraries/MOBIClass/EXTHHelper.php b/inc/3rdparty/libraries/MOBIClass/EXTHHelper.php
new file mode 100644
index 00000000..275142bf
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/EXTHHelper.php
@@ -0,0 +1,132 @@
1<?php
2//Reference: http://wiki.mobileread.com/wiki/MOBI
3
4class EXTHHelper{
5 static function typeToText($type){
6 $types = self::$types;
7 if(isset($types[$type])){
8 return $types[$type];
9 }
10 return $type;
11 }
12 static function textToType($text){
13 $text = strtolower($text);
14 if(isset(self::$flippedTypes[$text])){
15 return self::$flippedTypes[$text];
16 }
17 return false;
18 }
19
20 static function convert($n, $size){
21 $mask = 0xFF;
22 $out = "";
23 for($i = 0; $i < $size; $i++){
24 $out = chr(($n & $mask) >> (8*$i)).$out;
25 $mask = $mask << 8;
26 }
27 return $out;
28 }
29
30 static function getRightRepresentation($type, $value){
31 if($type >= 100 && $type < 200){
32 return $value;
33 }else{
34 return self::toHex($value);
35 }
36 }
37
38 static function toHex($value){
39 $out = "";
40 for($i = 0, $len = strlen($value); $i < $len; $i++){
41 if($i > 0) $out .= " ";
42 $hex = dechex(ord($value[$i]));
43 if(strlen($hex) < 2) $hex = "0".$hex;
44 $out .= $hex;
45 }
46 return $out;
47 }
48
49
50 static private $types = array(
51 1 => "drm server id",
52 2 => "drm commerce id",
53 3 => "drm ebookbase book id",
54 100 => "author",
55 101 => "publisher",
56 102 => "imprint",
57 103 => "description",
58 104 => "isbn",
59 105 => "subject",
60 106 => "publishingdate",
61 107 => "review",
62 108 => "contributor",
63 109 => "rights",
64 110 => "subjectcode",
65 111 => "type",
66 112 => "source",
67 113 => "asin",
68 114 => "versionnumber",
69 115 => "sample",
70 116 => "startreading",
71 118 => "retail price",
72 119 => "retail price currency",
73 201 => "coveroffset",
74 202 => "thumboffset",
75 203 => "hasfakecover",
76 204 => "Creator Software",
77 205 => "Creator Major Version",
78 206 => "Creator Minor Version",
79 207 => "Creator Build Number",
80 208 => "watermark",
81 209 => "tamper proof keys",
82 300 => "fontsignature",
83 401 => "clippinglimit",
84 402 => "publisherlimit",
85 403 => "403",
86 404 => "ttsflag",
87 501 => "cdetype",
88 502 => "lastupdatetime",
89 503 => "updatedtitle"
90 );
91 static private $flippedTypes = array(
92 "drm server id" => 1,
93 "drm commerce id" => 2,
94 "drm ebookbase book id" => 3,
95 "author" => 100,
96 "publisher" => 101,
97 "imprint" => 102,
98 "description" => 103,
99 "isbn" => 104,
100 "subject" => 105,
101 "publishingdate" => 106,
102 "review" => 107,
103 "contributor" => 108,
104 "rights" => 109,
105 "subjectcode" => 110,
106 "type" => 111,
107 "source" => 112,
108 "asin" => 113,
109 "versionnumber" => 114,
110 "sample" => 115,
111 "startreading" => 116,
112 "retail price" => 118,
113 "retail price currency" => 119,
114 "coveroffset" => 201,
115 "thumboffset" => 202,
116 "hasfakecover" => 203,
117 "Creator Software" => 204,
118 "Creator Major Version" => 205,
119 "Creator Minor Version" => 206,
120 "Creator Build Number" => 207,
121 "watermark" => 208,
122 "tamper proof keys" => 209,
123 "fontsignature" => 300,
124 "clippinglimit" => 401,
125 "publisherlimit" => 402,
126 "403" => 403,
127 "ttsflag" => 404,
128 "cdetype" => 501,
129 "lastupdatetime" => 502,
130 "updatedtitle" => 503
131 );
132} \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/FileByte.php b/inc/3rdparty/libraries/MOBIClass/FileByte.php
new file mode 100644
index 00000000..05fc7d04
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileByte.php
@@ -0,0 +1,41 @@
1<?php
2
3/**
4 * Description of FileByte
5 *
6 * @author Sander
7 */
8class FileByte extends FileObject {
9 private $data;
10
11 /**
12 * Make a short to be stored in a file
13 * @param short $n
14 */
15 public function __construct($n = 0){
16 parent::__construct(1);
17 $this->set($n);
18 }
19
20 public function get(){
21 return $this->data;
22 }
23
24 public function set($value){
25 $this->data = intval($value) & 0xFF;
26 }
27
28 public function serialize() {
29 return $this->byteToString($this->data);
30 }
31
32 public function unserialize($data) {
33 __construct($this->toInt($data));
34 }
35
36
37 public function __toString(){
38 return "FileByte: {".$this->byteAsString($this->data)."}";
39 }
40}
41?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileDate.php b/inc/3rdparty/libraries/MOBIClass/FileDate.php
new file mode 100644
index 00000000..2284eba2
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileDate.php
@@ -0,0 +1,40 @@
1<?php
2
3/**
4 * Description of FileDate
5 *
6 * @author Sander
7 */
8class FileDate extends FileObject {
9 private $data;
10
11 /**
12 * Make an integer to be stored in a file
13 * @param int $n
14 */
15 public function __construct($n = 0){
16 parent::__construct(4);
17 $this->set($n);
18 }
19
20 public function get(){
21 return $this->data;
22 }
23
24 public function set($value){
25 $this->data = intval($value);
26 }
27
28 public function serialize() {
29 return $this->intToString($this->data);
30 }
31
32 public function unserialize($data) {
33 __construct($this->toInt($data));
34 }
35
36 public function __toString(){
37 return "FileDate: {".(date("r", $this->data-94694400))."}";
38 }
39}
40?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileElement.php b/inc/3rdparty/libraries/MOBIClass/FileElement.php
new file mode 100644
index 00000000..552d04a8
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileElement.php
@@ -0,0 +1,89 @@
1<?php
2
3/**
4 * Description of FileElement
5 *
6 * @author Sander
7 */
8class FileElement {
9 /**
10 * @var FileObject
11 */
12 public $elements;
13
14 /**
15 * Make a record to be stored in a file
16 * @param Record $record
17 */
18 public function __construct($elements = array()){
19 $this->elements = $elements;
20 }
21
22 public function getByteLength(){
23 return $this->getLength();
24 }
25
26 public function getLength(){
27 $total = 0;
28 foreach($this->elements as $val){
29 $total += $val->getByteLength();
30 }
31 return $total;
32 }
33
34 public function offsetToEntry($name){
35 $pos = 0;
36 foreach($this->elements as $key=>$value){
37 if($name == $key){
38 break;
39 }
40 $pos += $value->getByteLength();
41 }
42 return $pos;
43 }
44
45 public function exists($key){
46 return isset($this->elements[$key]);
47 }
48 /**
49 * @param string $key
50 * @return FileObject
51 */
52 public function get($key){
53 return $this->elements[$key];
54 }
55
56 /**
57 * @param string $key
58 * @param FileObject $value
59 */
60 public function set($key, $value){
61 $this->elements[$key] = $value;
62 }
63
64 public function add($key, $value){
65 $this->elements[$key] = $value;
66 }
67
68 public function serialize() {
69 $result = "";
70 foreach($this->elements as $val){
71 $result .= $val->serialize();
72 }
73 return $result;
74 }
75
76 public function unserialize($data) {
77 //TODO: If reading is needed -> way more complex
78 }
79
80 public function __toString(){
81 $output = "FileElement (".$this->getByteLength()." bytes): {\n";
82 foreach($this->elements as $key=>$value){
83 $output .= "\t".$key.": ".$value."\n";
84 }
85 $output .= "}";
86 return $output;
87 }
88}
89?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileInt.php b/inc/3rdparty/libraries/MOBIClass/FileInt.php
new file mode 100644
index 00000000..ebe86f86
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileInt.php
@@ -0,0 +1,40 @@
1<?php
2
3/**
4 * Description of FileInt
5 *
6 * @author Sander
7 */
8class FileInt extends FileObject {
9 private $data;
10
11 /**
12 * Make an integer to be stored in a file
13 * @param int $n
14 */
15 public function __construct($n = 0){
16 parent::__construct(4);
17 $this->set($n);
18 }
19
20 public function get(){
21 return $this->data;
22 }
23
24 public function set($value){
25 $this->data = intval($value);
26 }
27
28 public function serialize() {
29 return $this->intToString($this->data);
30 }
31
32 public function unserialize($data) {
33 __construct($this->toInt($data));
34 }
35
36 public function __toString(){
37 return "FileInt: {".$this->intAsString($this->data)."}";
38 }
39}
40?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileObject.php b/inc/3rdparty/libraries/MOBIClass/FileObject.php
new file mode 100644
index 00000000..0df17df1
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileObject.php
@@ -0,0 +1,168 @@
1<?php
2
3/**
4 * Description of FileObject
5 *
6 * @author Sander
7 */
8abstract class FileObject {
9 private $byteLength = -1;
10
11 public function __construct($byteLength = -1){
12 $this->byteLength = $byteLength;
13 }
14
15 public function getByteLength(){
16 if($this->byteLength >= 0){
17 return $this->byteLength;
18 }
19 return $this->getLength();
20 }
21
22 public function getLength(){
23 throw new Exception("Sub-class needs to implement this if it doesn't have a fixed length");
24 }
25
26 /**
27 * Convert a string to byte format (maximum 4 bytes)
28 * @param string $string Input string
29 * @return int Output integer
30 */
31 public function toInt($string){
32 $out = 0;
33 for($i = 0, $len = min(4, strlen($string)); $i < $len; $i++){
34 $out = $out | (ord($string[$i]) << (($len-$i-1)*8));
35 }
36 return $out;
37 }
38
39 /**
40 * Convert a byte (stored in an integer) to a string
41 * @param byte $int
42 * @return string
43 */
44 public function byteToString($int){
45 return $this->toString($int, 1);
46 }
47
48 /**
49 * Convert a byte (stored in an integer) to a string
50 * @param byte $int
51 * @return string
52 */
53 public function byteAsString($int){
54 return $this->asString($int, 1);
55 }
56
57 /**
58 * Convert a short (stored in an integer) to a string
59 * @param short $int
60 * @return string
61 */
62 public function shortToString($int){
63 return $this->toString($int, 2);
64 }
65
66 /**
67 * Convert a short (stored in an integer) to a string
68 * @param short $int
69 * @return string
70 */
71 public function shortAsString($int){
72 return $this->asString($int, 2);
73 }
74
75 /**
76 * Convert a tri-byte (stored in an integer) to a string
77 * @param tri-byte $int
78 * @return string
79 */
80 public function triToString($int){
81 return $this->toString($int, 3);
82 }
83
84 /**
85 * Convert a tri-byte (stored in an integer) to a string
86 * @param tri-byte $int
87 * @return string
88 */
89 public function triAsString($int){
90 return $this->asString($int, 3);
91 }
92
93 /**
94 * Convert an integer to a string
95 * @param int $int
96 * @return string
97 */
98 public function intToString($int){
99 return $this->toString($int, 4);
100 }
101
102 /**
103 * Convert an integer to a string
104 * @param int $int
105 * @return string
106 */
107 public function intAsString($int){
108 return $this->asString($int, 4);
109 }
110
111 /**
112 * Convert a number of n bytes to a string
113 * @param int $int Number that should be converted
114 * @param int $size Number of bytes to convert
115 * @return string Output string
116 */
117 private function toString($int, $size){
118 $out = "";
119 for($i = 0; $i < $size; $i++){
120 $out = chr($int & 0xFF).$out;
121 $int = $int >> 8;
122 }
123 return $out;
124 }
125
126 /**
127 * Convert a number of n bytes to a string
128 * @param int $int Number that should be converted
129 * @param int $size Number of bytes to convert
130 * @return string Output string
131 */
132 private function asString($int, $size){
133 $out = "";
134 for($i = 0; $i < $size; $i++){
135 if($i > 0) $out = " ".$out;
136 $byte = dechex($int & 0xFF);
137 if(strlen($byte) == 1) $byte = "0".$byte;
138 $out = $byte.$out;
139 $int = $int >> 8;
140 }
141 return $out;
142 }
143
144 /**
145 * Get the value
146 * @return mixed Value to get
147 */
148 abstract public function get();
149
150 /**
151 * Set the value
152 * @return mixed Value to set
153 */
154 abstract public function set($value);
155
156 /**
157 * Serialize the object
158 * @return string String representation
159 */
160 abstract public function serialize();
161
162 /**
163 * Unserialize the object
164 * @param string $data String representation
165 */
166 abstract public function unserialize($data);
167}
168?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileRecord.php b/inc/3rdparty/libraries/MOBIClass/FileRecord.php
new file mode 100644
index 00000000..494a72e4
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileRecord.php
@@ -0,0 +1,46 @@
1<?php
2
3/**
4 * Description of FileRecord
5 *
6 * @author Sander
7 */
8class FileRecord extends FileObject {
9 /**
10 * @var Record
11 */
12 private $record;
13
14 /**
15 * Make a record to be stored in a file
16 * @param Record $record
17 */
18 public function __construct($record){
19 $this->record = $record;
20 }
21
22 public function getByteLength(){
23 return $this->getLength();
24 }
25
26 public function getLength(){
27 return $this->record->getLength();
28 }
29
30 public function get(){
31 return $this->record;
32 }
33
34 public function set($record){
35 $this->record = $record;
36 }
37
38 public function serialize() {
39 return $this->record->serialize();
40 }
41
42 public function unserialize($data) {
43 __construct($this->record->unserialize($data));
44 }
45}
46?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileShort.php b/inc/3rdparty/libraries/MOBIClass/FileShort.php
new file mode 100644
index 00000000..9921ea82
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileShort.php
@@ -0,0 +1,41 @@
1<?php
2
3/**
4 * Description of FileShort
5 *
6 * @author Sander
7 */
8class FileShort extends FileObject {
9 private $data;
10
11 /**
12 * Make a short to be stored in a file
13 * @param short $n
14 */
15 public function __construct($n = 0){
16 parent::__construct(2);
17 $this->set($n);
18 }
19
20 public function get(){
21 return $this->data;
22 }
23
24 public function set($value){
25 $this->data = intval($value) & 0xFFFF;
26 }
27
28 public function serialize() {
29 return $this->shortToString($this->data);
30 }
31
32 public function unserialize($data) {
33 __construct($this->toInt($data));
34 }
35
36
37 public function __toString(){
38 return "FileShort: {".$this->shortAsString($this->data)."}";
39 }
40}
41?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileString.php b/inc/3rdparty/libraries/MOBIClass/FileString.php
new file mode 100644
index 00000000..16e906a6
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileString.php
@@ -0,0 +1,83 @@
1<?php
2
3/**
4 * Description of FileString
5 *
6 * @author Sander
7 */
8class FileString extends FileObject {
9 private $forcedLength;
10 private $data;
11
12 /**
13 * Make a string to be stored in a file
14 * @param string|int $first Optional, if it is a string, it will be the contents,
15 * if it is a number, it will set the forced length.
16 * @param int $second Optional, will set the forced length. Can only be used when the
17 * first argument is contents.
18 */
19 public function __construct($first = null, $second = null){
20 $this->forcedLength = -1;
21 $this->data = "";
22
23 if($second != null){
24 $this->data = $first;
25 $this->forcedLength = $second;
26 }else if($first != null){
27 if(is_string($first)){
28 $this->data = $first;
29 }else{
30 $this->forcedLength = $first;
31 }
32 }
33 }
34
35 public function getByteLength(){
36 return $this->getLength();
37 }
38
39 public function getLength(){
40 if($this->forcedLength >= 0){
41 return $this->forcedLength;
42 }
43 return strlen($this->data);
44 }
45
46 public function get(){
47 return $this->data;
48 }
49
50 public function set($value){
51 $this->data = $value;
52 }
53
54 public function serialize() {
55 $output = $this->data;
56 $curLength = strlen($output);
57
58 if($this->forcedLength >= 0){
59 if($this->forcedLength > $curLength){
60 return str_pad($output, $this->forcedLength, "\0", STR_PAD_RIGHT);
61 }elseif($this->forcedLength == $curLength){
62 return $output;
63 }else{
64 return substr($output, 0, $this->forcedLength);
65 }
66 }
67 return $output;
68 }
69
70 public function unserialize($data) {
71 __construct($data);
72 }
73
74 public function __toString(){
75 $out = "FileString";
76 if($this->forcedLength >= 0){
77 $out .= " ".$this->forcedLength;
78 }
79 $out .= ": {\"".str_replace(array(" ", "\0"), "&nbsp;", $this->serialize())."\"}";
80 return $out;
81 }
82}
83?>
diff --git a/inc/3rdparty/libraries/MOBIClass/FileTri.php b/inc/3rdparty/libraries/MOBIClass/FileTri.php
new file mode 100644
index 00000000..6cacc0b0
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/FileTri.php
@@ -0,0 +1,41 @@
1<?php
2
3/**
4 * Description of FileTri
5 *
6 * @author Sander
7 */
8class FileTri extends FileObject {
9 private $data;
10
11 /**
12 * Make a tri-byte to be stored in a file
13 * @param tri-byte $n
14 */
15 public function __construct($n = 0){
16 parent::__construct(3);
17 $this->set($n);
18 }
19
20 public function get(){
21 return $this->data;
22 }
23
24 public function set($value){
25 $this->data = intval($value) & 0xFFFFFF;
26 }
27
28 public function serialize() {
29 return $this->triToString($this->data);
30 }
31
32 public function unserialize($data) {
33 __construct($this->toInt($data));
34 }
35
36
37 public function __toString(){
38 return "FileTri: {".$this->triAsString($this->data)."}";
39 }
40}
41?>
diff --git a/inc/3rdparty/libraries/MOBIClass/Http.php b/inc/3rdparty/libraries/MOBIClass/Http.php
new file mode 100644
index 00000000..9e5852e3
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/Http.php
@@ -0,0 +1,171 @@
1<?php
2class Http{
3 private static $cache = false;
4
5 public static function Request($url){
6 $url_parts = parse_url($url);
7 $url_parts["port"] = isset($url_parts["port"]) ? $url_parts["port"] : 80;
8 $url_parts["path"] = isset($url_parts["path"]) ? $url_parts["path"] : "/";
9
10 return self::FullRequest("GET", $url_parts["host"], $url_parts["port"], $url_parts["path"]);
11 }
12
13 public static function FullRequest(
14 $verb = 'GET', /* HTTP Request Method (GET and POST supported) */
15 $ip, /* Target IP/Hostname */
16 $port = 80, /* Target TCP port */
17 $uri = '/', /* Target URI */
18 $getdata = array(), /* HTTP GET Data ie. array('var1' => 'val1', 'var2' => 'val2') */
19 $postdata = array(), /* HTTP POST Data ie. array('var1' => 'val1', 'var2' => 'val2') */
20 $cookie = array(), /* HTTP Cookie Data ie. array('var1' => 'val1', 'var2' => 'val2') */
21 $custom_headers = array(), /* Custom HTTP headers ie. array('Referer: http://localhost/ */
22 $timeout = 1000, /* Socket timeout in milliseconds */
23 $req_hdr = false, /* Include HTTP request headers */
24 $res_hdr = false, /* Include HTTP response headers */
25 $depth = 4 /* Depth of the iteration left (to avoid redirection loops) */
26 )
27 {
28 if(self::$cache){
29 $cacheFile = "cache/".$ip."/".str_replace("/", "...", $uri);
30
31 if(is_file($cacheFile)){
32 $data = file_get_contents($cacheFile);
33
34 return self::resolveTruncated($data);
35 }
36 }
37 $ret = '';
38 $verb = strtoupper($verb);
39 $cookie_str = '';
40 $getdata_str = count($getdata) ? '?' : '';
41 $postdata_str = '';
42
43 foreach ($getdata as $k => $v)
44 $getdata_str .= urlencode($k) .'='. urlencode($v);
45
46 foreach ($postdata as $k => $v)
47 $postdata_str .= urlencode($k) .'='. urlencode($v) .'&';
48
49 foreach ($cookie as $k => $v)
50 $cookie_str .= urlencode($k) .'='. urlencode($v) .'; ';
51
52 $crlf = "\r\n";
53 $req = $verb .' '. $uri . $getdata_str .' HTTP/1.1' . $crlf;
54 $req .= 'Host: '. $ip . $crlf;
55 $req .= 'User-Agent: Mozilla/5.0 Firefox/3.6.12' . $crlf;
56 $req .= 'Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8' . $crlf;
57 $req .= 'Accept-Language: en-us,en;q=0.5' . $crlf;
58 $req .= 'Accept-Encoding: deflate' . $crlf;
59 $req .= 'Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7' . $crlf;
60
61
62 foreach ($custom_headers as $k => $v)
63 $req .= $k .': '. $v . $crlf;
64
65 if (!empty($cookie_str))
66 $req .= 'Cookie: '. substr($cookie_str, 0, -2) . $crlf;
67
68 if ($verb == 'POST' && !empty($postdata_str))
69 {
70 $postdata_str = substr($postdata_str, 0, -1);
71 $req .= 'Content-Type: application/x-www-form-urlencoded' . $crlf;
72 $req .= 'Content-Length: '. strlen($postdata_str) . $crlf . $crlf;
73 $req .= $postdata_str;
74 }
75 else $req .= $crlf;
76
77 if ($req_hdr)
78 $ret .= $req;
79
80 if (($fp = @fsockopen($ip, $port, $errno, $errstr)) == false)
81 return "Error $errno: $errstr\n";
82
83 stream_set_timeout($fp, 0, $timeout * 1000);
84
85 fputs($fp, $req);
86 $ret .= stream_get_contents($fp);
87 fclose($fp);
88
89 $headerSplit = strpos($ret, "\r\n\r\n");
90 $header = substr($ret, 0, $headerSplit);
91
92 $redirectURL = self::CheckForRedirect($header);
93
94 if($redirectURL !== false){
95 if($depth > 0){
96 $url_parts = parse_url($redirectURL);
97 $url_parts["port"] = isset($url_parts["port"]) ? $url_parts["port"] : 80;
98 $url_parts["path"] = isset($url_parts["path"]) ? $url_parts["path"] : "/";
99
100 return self::FullRequest($verb, $url_parts["host"], $url_parts["port"], $url_parts["path"], $getdata, $postdata, $cookie, $custom_headers, $timeout, $req_hdr, $res_hdr, $depth-1);
101 }else{
102 return "Redirect loop, stopping...";
103 }
104 }
105
106 $truncated = false;
107 $headerLines = explode("\r\n", $header);
108 foreach($headerLines as $line){
109 list($name, $value) = explode(":", $line);
110 $name = trim($name);
111 $value = trim($value);
112
113 if(strtolower($name) == "transfer-encoding" && strtolower($value) == "chunked"){ //TODO: Put right values!
114 $truncated = true;
115 }
116 }
117
118 if (!$res_hdr)
119 $ret = substr($ret, $headerSplit + 4);
120
121 if($truncated){
122 $ret = self::resolveTruncated($ret);
123 }
124 if(self::$cache){
125 if(!is_dir("cache")){
126 mkdir("cache");
127 }
128 if(!is_dir("cache/".$ip)){
129 mkdir("cache/".$ip);
130 }
131 if(!is_file("cache/".$ip."/".str_replace("/", "...", $uri))){
132 $h = fopen("cache/".$ip."/".str_replace("/", "...", $uri), "w");
133 fwrite($h, $ret);
134 fclose($h);
135 }
136 }
137
138 return $ret;
139 }
140
141 private static function resolveTruncated($data){
142 $pos = 0;
143 $end = strlen($data);
144 $out = "";
145
146 while($pos < $end){
147 $endVal = strpos($data, "\r\n", $pos);
148 $value = hexdec(substr($data, $pos, $endVal-$pos));
149 $out .= substr($data, $endVal+2, $value);
150 $pos = $endVal+2+$value;
151 }
152
153 return $out;
154 }
155
156 private static function CheckForRedirect($header){
157 $firstLine = substr($header, 0, strpos($header, "\r\n"));
158 list($httpVersion, $statusCode, $message) = explode(" ", $firstLine);
159
160 if(substr($statusCode, 0, 1) == "3"){
161 $part = substr($header, strpos(strtolower($header), "location: ")+strlen("location: "));
162 $location = trim(substr($part, 0, strpos($part, "\r\n")));
163
164 if(strlen($location) > 0){
165 return $location;
166 }
167 }
168 return false;
169 }
170}
171?> \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/ImageHandler.php b/inc/3rdparty/libraries/MOBIClass/ImageHandler.php
new file mode 100644
index 00000000..bcb48e9f
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/ImageHandler.php
@@ -0,0 +1,40 @@
1<?php
2
3class ImageHandler {
4 /**
5 * Download an image
6 * @param string $url Url to the image
7 * @return false|string False if failed, else the data of the image (converted to grayscale jpeg)
8 */
9 public static function DownloadImage($url){
10 $data = Http::Request($url);
11 $imgFile = @imagecreatefromstring($data);
12
13 if($imgFile !== false){
14 $result = self::CreateImage($imgFile);
15 imagedestroy($imgFile);
16 return $result;
17 }
18 return false;
19 }
20 /**
21 * Create an image
22 * @param resource $img Create an image created with createimagetruecolor
23 * @return false|string False if failed, else the data of the image (converted to grayscale jpeg)
24 */
25 public static function CreateImage($img){
26 try{
27 imagefilter($img, IMG_FILTER_GRAYSCALE);
28
29 ob_start();
30 imagejpeg($img);
31 $image = ob_get_contents();
32 ob_end_clean();
33
34 return $image;
35 }catch(Exception $e){
36 return false;
37 }
38 }
39}
40?>
diff --git a/inc/3rdparty/libraries/MOBIClass/MOBI.php b/inc/3rdparty/libraries/MOBIClass/MOBI.php
new file mode 100644
index 00000000..17e718c1
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/MOBI.php
@@ -0,0 +1,192 @@
1<?php
2require_once(dirname(__FILE__)."/readability/Readability.php");
3require_once(dirname(__FILE__).'/CharacterEntities.php');
4require_once(dirname(__FILE__).'/constants.php');
5require_once(dirname(__FILE__).'/ContentProvider.php');
6require_once(dirname(__FILE__).'/MultipleFileHandler.php');
7require_once(dirname(__FILE__)."/downloaders/FanFictionNet.php");
8require_once(dirname(__FILE__).'/EXTHHelper.php');
9require_once(dirname(__FILE__).'/FileObject.php');
10require_once(dirname(__FILE__).'/FileByte.php');
11require_once(dirname(__FILE__).'/FileDate.php');
12require_once(dirname(__FILE__).'/FileElement.php');
13require_once(dirname(__FILE__).'/FileInt.php');
14require_once(dirname(__FILE__).'/FileRecord.php');
15require_once(dirname(__FILE__).'/FileShort.php');
16require_once(dirname(__FILE__).'/FileString.php');
17require_once(dirname(__FILE__).'/FileTri.php');
18require_once(dirname(__FILE__).'/Http.php');
19require_once(dirname(__FILE__).'/http_build_url.php');
20require_once(dirname(__FILE__).'/ImageHandler.php');
21require_once(dirname(__FILE__).'/MOBIFile.php');
22require_once(dirname(__FILE__).'/OnlineArticle.php');
23require_once(dirname(__FILE__).'/PalmRecord.php');
24require_once(dirname(__FILE__).'/Prc.php');
25require_once(dirname(__FILE__).'/PreprocessedArticle.php');
26require_once(dirname(__FILE__).'/RecognizeURL.php');
27require_once(dirname(__FILE__).'/Record.php');
28require_once(dirname(__FILE__).'/RecordFactory.php');
29require_once(dirname(__FILE__).'/Settings.php');
30
31/**
32 * Description of MOBI.
33 *
34 * Usage:
35 * include("MOBIClass/MOBI.php");
36 *
37 * $mobi = new MOBI();
38 *
39 * //Then use one of the following ways to prepare information (it should be in the form of valid html)
40 * $mobi->setInternetSource($url); //Load URL, the result will be cleaned using a Readability port
41 * $mobi->setFileSource($file); //Load a local file without any extra changes
42 * $mobi->setData($data); //Load data
43 *
44 * //If you want, you can set some optional settings (see Settings.php for all recognized settings)
45 * $options = array(
46 * "title"=>"Insert title here",
47 * "author"=>"Author"
48 * );
49 * $mobi->setOptions($options);
50 *
51 * //Then there are two ways to output it:
52 * $mobi->save($file); //Save the file locally
53 * $mobi->download($name); //Let the client download the file, make sure the page
54 * //that calls it doesn't output anything, otherwise it might
55 * //conflict with the download. $name contains the file name,
56 * //usually something like "title.mobi" (where the title should
57 * //be cleaned so as not to contain illegal characters).
58 *
59 *
60 * @author Sander Kromwijk
61 */
62class MOBI {
63 private $source = false;
64 private $images = array();
65 private $optional = array();
66 private $imgCounter = 0;
67 private $debug = false;
68 private $prc = false;
69
70 public function __construct(){
71
72 }
73
74 public function getTitle(){
75 if(isset($this->optional["title"])){
76 return $this->optional["title"];
77 }
78 return false;
79 }
80
81 /**
82 * Set a content provider as source
83 * @param ContentProvider $content Content Provider to use
84 */
85 public function setContentProvider($content){
86 $this->setOptions($content->getMetaData());
87 $this->setImages($content->getImages());
88 $this->setData($content->getTextData());
89 }
90
91 /**
92 * Set a local file as source
93 * @param string $file Path to the file
94 */
95 public function setFileSource($file){
96 $this->setData(file_get_contents($file));
97 }
98
99 /**
100 * Set the data to use
101 * @param string $data Data to put in the file
102 */
103 public function setData($data){
104 //$data = utf8_encode($data);
105 $data = CharacterEntities::convert($data);
106 //$data = utf8_decode($data);
107 //$this->source = iconv('UTF-8', 'ISO-8859-1//TRANSLIT', $data);
108 $this->source = $data;
109 $this->prc = false;
110 }
111
112 /**
113 * Set the images to use
114 * @param array $data Data to put in the file
115 */
116 public function setImages($data){
117 $this->images = $data;
118 $this->prc = false;
119 }
120
121 /**
122 * Set options, usually for things like titles, authors, etc...
123 * @param array $options Options to set
124 */
125 public function setOptions($options){
126 $this->optional = $options;
127 $this->prc = false;
128 }
129
130 /**
131 * Prepare the prc file
132 * @return Prc The file that can be used to be saved/downloaded
133 */
134 private function preparePRC(){
135 if($this->source === false){
136 throw new Exception("No data set");
137 }
138 if($this->prc !== false) return $this->prc;
139
140 $data = $this->source;
141 $len = strlen($data);
142
143 $settings = new Settings($this->optional);
144 $rec = new RecordFactory($settings);
145 $dataRecords = $rec->createRecords($data);
146 $nRecords = sizeof($dataRecords);
147 $mobiHeader = new PalmRecord($settings, $dataRecords, $nRecords, $len, sizeof($this->images));
148 array_unshift($dataRecords, $mobiHeader);
149 $dataRecords = array_merge($dataRecords, $this->images);
150 $dataRecords[] = $rec->createFLISRecord();
151 $dataRecords[] = $rec->createFCISRecord($len);
152 $dataRecords[] = $rec->createEOFRecord();
153 $this->prc = new Prc($settings, $dataRecords);
154 return $this->prc;
155 }
156
157 /**
158 * Save the file locally
159 * @param string $filename Path to save the file
160 */
161 public function save($filename){
162 $prc = $this->preparePRC();
163 $prc->save($filename);
164 }
165
166 /**
167 * Let the client download the file. Warning! No data should be
168 * outputted before or after.
169 * @param string $name Name used for download, usually "title.mobi"
170 */
171 public function download($name){
172 $prc = $this->preparePRC();
173 $data = $prc->serialize();
174 $length = strlen($data);
175
176 if($this->debug) return; //In debug mode, don't start the download
177
178 header("Content-Type: application/x-mobipocket-ebook");
179 header("Content-Disposition: attachment; filename=\"".$name."\"");
180 header("Content-Transfer-Encoding: binary");
181 header("Accept-Ranges: bytes");
182 header("Cache-control: private");
183 header('Pragma: private');
184 header("Expires: Mon, 26 Jul 1997 05:00:00 GMT");
185 header("Content-Length: ".$length);
186
187 echo $data;
188 //Finished!
189 }
190
191}
192?> \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/MOBIFile.php b/inc/3rdparty/libraries/MOBIClass/MOBIFile.php
new file mode 100644
index 00000000..349227ae
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/MOBIFile.php
@@ -0,0 +1,157 @@
1<?php
2/**
3 * This is the way MOBI files should be created if you want all features (TOC, images).
4 *
5 * File modified by Dawson for use in eBook Creator
6 * Added pagebreaks and a setting to remove table of contents.
7 */
8
9class MOBIFile extends ContentProvider {
10 const PARAGRAPH = 0;
11 const H2 = 1;
12 const H3 = 2;
13 const IMAGE = 3;
14 const PAGEBREAK = 4;
15
16 private $settings = array("title" => "Unknown Title", "toc" => true);
17 private $parts = array();
18 private $images = array();
19
20 /**
21 * Get the text data (the "html" code)
22 */
23 public function getTextData(){
24 $prefix = "<html><head><guide><reference title='CONTENT' type='toc' filepos=0000000000 /></guide></head><body>";
25
26 $title = "<h1>".$this->settings["title"]."</h1>";
27
28 list($text, $entries) = $this->generateText();
29
30 if($this->settings["toc"]) {
31 $toc = $this->generateTOC($entries); //Generate TOC to get the right length
32 $toc = $this->generateTOC($entries, strlen($prefix)+strlen($toc)+strlen($title)); //Generate the real TOC
33 }
34
35 $suffix = "</body></html>";
36
37 return $prefix.$toc.$title.$text.$suffix;
38 }
39
40 /**
41 * Generate the body's text and the chapter entries
42 * @return array($string, $entries) $string is the html data, $entries
43 * contains the level, the title and the position of the titles.
44 */
45 public function generateText(){
46 $str = "";
47 $entries = array();
48
49 for($i = 0; $i < sizeof($this->parts); $i++){
50 list($type, $data) = $this->parts[$i];
51 $id = "title_".$i;
52 switch($type){
53 case self::PARAGRAPH:
54 $str .= "<p>".$data."</p>";
55 break;
56 case self::PAGEBREAK:
57 $str .= '<mbp:pagebreak/>';
58 break;
59 case self::H2:
60 $entries[] = array("level" => 2, "position" => strlen($str), "title" => $data, "id" => $id);
61 $str .= "<h2 id='" . $id . "'>".$data."</h2>";
62 break;
63 case self::H3:
64 $entries[] = array("level" => 3, "position" => strlen($str), "title" => $data, "id" => $id);
65 $str .= "<h3 id='" . $id . "'>".$data."</h3>";
66 break;
67 case self::IMAGE:
68 $str .= "<img recindex=".str_pad($data+1, 10, "0", STR_PAD_LEFT)." />";
69 break;
70 }
71 }
72 return array($str, $entries);
73 }
74
75 /**
76 * Generate a TOC
77 * @param $entries The entries array generated by generateText
78 * @param $base The zero position
79 */
80 public function generateTOC($entries, $base = 0){
81 $toc = "<h2>Contents</h2>";
82 $toc .= "<blockquote><table summary='Table of Contents'><col/><tbody>";
83 for($i = 0, $len = sizeof($entries); $i < $len; $i++){
84 $entry = $entries[$i];
85 $pos = str_pad($entry["position"]+$base, 10, "0", STR_PAD_LEFT);
86 $toc .= "<tr><td><a href='#".$entry["id"]."' filepos='".$pos."'>".$entry["title"]."</a></td></tr>";
87 }
88 return $toc."</tbody></b></table></blockquote><mbp:pagebreak/>";
89 }
90
91 /**
92 * Get the file records of the images
93 */
94 public function getImages(){
95 return $this->images;
96 }
97
98 /**
99 * Get the metadata
100 */
101 public function getMetaData(){
102 return $this->settings;
103 }
104
105 /**
106 * Change the file's settings. For example set("author", "John Doe") or set("title", "The adventures of John Doe").
107 * @param $key Key of the setting to insert.
108 */
109 public function set($key, $value){
110 $this->settings[$key] = $value;
111 }
112
113 /**
114 * Get the file's settings.
115 */
116 public function get($key){
117 return $this->settings[$key];
118 }
119
120 /**
121 * Append a paragraph of text to the file.
122 * @param string $text The text to insert.
123 */
124 public function appendParagraph($text){
125 $this->parts[] = array(self::PARAGRAPH, $text);
126 }
127
128 /**
129 * Append a chapter title (H2)
130 * @param string $title The title to insert.
131 */
132 public function appendChapterTitle($title){
133 $this->parts[] = array(self::H2, $title);
134 }
135
136 /**
137 * Append a section title (H3)
138 * @param string $title The title to insert.
139 */
140 public function appendSectionTitle($title){
141 $this->parts[] = array(self::H3, $title);
142 }
143
144 public function appendPageBreak() {
145 $this->parts[] = array(self::PAGEBREAK, null);
146 }
147
148 /**
149 * Append an image.
150 * @param resource $img An image file (for example, created by `imagecreate`)
151 */
152 public function appendImage($img){
153 $imgIndex = sizeof($this->images);
154 $this->images[] = new FileRecord(new Record(ImageHandler::CreateImage($img)));
155 $this->parts[] = array(self::IMAGE, $imgIndex);
156 }
157} \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/MultipleFileHandler.php b/inc/3rdparty/libraries/MOBIClass/MultipleFileHandler.php
new file mode 100644
index 00000000..e9792816
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/MultipleFileHandler.php
@@ -0,0 +1,136 @@
1<?php
2
3/**
4 * Description of MultipleFileHandler
5 *
6 * @author Sander
7 */
8abstract class MultipleFileHandler extends ContentProvider {
9 /**
10 * @var array
11 */
12 private $files = array();
13 /**
14 * @var array
15 */
16 private $images = array();
17 /**
18 * @var array
19 */
20 private $metadata = array();
21
22 private $toc = array();
23
24 /**
25 * Add a page to the file
26 * @param string $contents Contents of the chapter/page
27 * @param string $title Optional, title of the chapter/page. Will automatically add a h2
28 * before the contents
29 */
30 public function addPage($contents, $title = ""){
31 if($title != ""){
32 //TODO: Add in TOC (and add a way of generating it
33 $contents = "<h2>".$title."</h2>".$contents."<mbp:pagebreak>";
34 }
35 $pos = 0;
36
37 if(sizeof($this->toc) > 0){
38 $lastToc = $this->toc[sizeof($this->toc)-1];
39 $lastFile = $this->files[sizeof($this->files)-1];
40 $pos = $lastToc["pos"] + strlen($lastFile) + 1;
41 }
42
43 $this->files[] = $contents;
44 $this->toc[] = array("title"=>$title, "pos"=>$pos);
45 }
46
47 /**
48 * Add an image to the file
49 * @param string $imageContents Data string containing the binary data of the image
50 * @return int The reference of the image
51 */
52 public function addImage($imageContents){
53 $this->images[] = $imageContents;
54 return sizeof($this->images)-1;
55 }
56
57 /**
58 * Add an image to the file
59 * @param string $url Url to the image
60 * @return int The reference of the image, false if the image couldn't be downloaded
61 */
62 public function addImageFromUrl($url){
63 $image = ImageHandler::DownloadImage($url);
64
65 if($image === false) return false;
66 return $this->addImage($image);
67 }
68
69 /**
70 * Set the metadata
71 * @param string $key Key
72 * @param string $value Value
73 */
74 public function setMetadata($key, $value){
75 $this->metadata[$key] = $value;
76 }
77
78 /**
79 * Get the text data to be integrated in the MOBI file
80 * @return string
81 */
82 public function getTextData(){
83 $data = implode("\n", $this->files);
84 $begin = "<html><head><guide><reference title='CONTENT' type='toc' filepos=0000000000 /></guide></head><body>";
85 $beforeTOC = $begin.$data;
86
87 $tocPos = strlen($beforeTOC);
88
89 $toc = $this->generateTOC(strlen($begin));
90
91 $customBegin = "<html><head><guide><reference title='CONTENT' type='toc' filepos=".$this->forceLength($tocPos, 10)." /></guide></head><body>";
92 $data = $customBegin.$data.$toc."</body></html>";
93 return $data;
94 }
95
96 public function forceLength($n, $l){
97 $str = $n."";
98 $cur = strlen($str);
99 while($cur < $l){
100 $str = "0".$str;
101 $cur++;
102 }
103 return $str;
104 }
105
106 public function generateTOC($base = 0){
107 $toc = "<h2>Contents</h2>";
108 $toc .= "<blockquote><table summary='Table of Contents'><b><col/><col/><tbody>";
109 for($i = 0, $len = sizeof($this->toc); $i < $len; $i++){
110 $entry = $this->toc[$i];
111 $position = $entry["pos"]+$base;
112 $toc .= "<tr><td>".($i+1).".</td><td><a filepos=".$position.">".$entry["title"]."</a></td></tr>";
113 }
114 $toc .= "</tbody></b></table></blockquote>";
115
116 return $toc;
117 }
118 /**
119 * Get the images (an array containing the jpeg data). Array entry 0 will
120 * correspond to image record 0.
121 * @return array
122 */
123 public function getImages(){
124 return $this->images;
125 }
126
127 /**
128 * Get the metadata in the form of a hashtable (for example, title or author).
129 * @return array
130 */
131 public function getMetaData(){
132 return $this->metadata;
133 }
134
135}
136?>
diff --git a/inc/3rdparty/libraries/MOBIClass/OnlineArticle.php b/inc/3rdparty/libraries/MOBIClass/OnlineArticle.php
new file mode 100644
index 00000000..ec3182fe
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/OnlineArticle.php
@@ -0,0 +1,116 @@
1<?php
2
3/**
4 * Description of OnlineArticle
5 *
6 * @author Sander
7 */
8class OnlineArticle extends ContentProvider {
9 private $text;
10 private $images;
11 private $metadata = array();
12 private $imgCounter = 0;
13
14 public function __construct($url) {
15 if (!preg_match('!^https?://!i', $url)) $url = 'http://'.$url;
16
17 $data = Http::Request($url);
18 //$enc = mb_detect_encoding($str, "UTF-8,ISO-8859-1,ASCII");
19 $html = mb_convert_encoding($data, "UTF-8", "UTF-8,ISO-8859-1,ASCII");
20 //$html = utf8_encode($html);
21 $r = new Readability($html, $url);
22 $r->init();
23 if(!isset($this->metadata["title"])){
24 $this->metadata["title"] = CharacterEntities::convert(strip_tags($r->getTitle()->innerHTML));
25 }
26 if(!isset($this->metadata["author"])){
27 $parts = parse_url($url);
28 $this->metadata["author"] = $parts["host"];
29 }
30
31 $article = $r->getContent()->innerHTML;
32 if(substr($article, 0, 5) == "<body"){
33 $article = "<html><head><meta http-equiv='Content-Type' content='text/html; charset=UTF-8'/></head>".$article."</html>";
34 }else{
35 $article = "<html><head><meta http-equiv='Content-Type' content='text/html; charset=UTF-8'/></head><body>".$article."</body></html>";
36 }
37 $doc = new DOMDocument();
38 @$doc->loadHTML($article) or die($article);
39 $doc->normalizeDocument();
40
41 $this->images = $this->handleImages($doc, $url);
42 $this->text = $doc->saveHTML();
43 }
44
45 /**
46 * Get the text data to be integrated in the MOBI file
47 * @return string
48 */
49 public function getTextData(){
50 return $this->text;
51 }
52 /**
53 * Get the images (an array containing the jpeg data). Array entry 0 will
54 * correspond to image record 0.
55 * @return array
56 */
57 public function getImages(){
58 return $this->images;
59 }
60 /**
61 * Get the metadata in the form of a hashtable (for example, title or author).
62 * @return array
63 */
64 public function getMetaData(){
65 return $this->metadata;
66 }
67 /**
68 *
69 * @param DOMElement $dom
70 * @return array
71 */
72 private function handleImages($dom, $url){
73 $images = array();
74
75 $parts = parse_url($url);
76
77 $savedImages = array();
78
79 $imgElements = $dom->getElementsByTagName('img');
80 foreach($imgElements as $img) {
81 $src = $img->getAttribute("src");
82
83 $is_root = false;
84 if(substr($src, 0, 1) == "/"){
85 $is_root = true;
86 }
87
88 $parsed = parse_url($src);
89
90 if(!isset($parsed["host"])){
91 if($is_root){
92 $src = http_build_url($url, $parsed, HTTP_URL_REPLACE);
93 }else{
94 $src = http_build_url($url, $parsed, HTTP_URL_JOIN_PATH);
95 }
96 }
97 $img->setAttribute("src", "");
98 if(isset($savedImages[$src])){
99 $img->setAttribute("recindex", $savedImages[$src]);
100 }else{
101 $image = ImageHandler::DownloadImage($src);
102
103 if($image !== false){
104 $images[$this->imgCounter] = new FileRecord(new Record($image));
105
106 $img->setAttribute("recindex", $this->imgCounter);
107 $savedImages[$src] = $this->imgCounter;
108 $this->imgCounter++;
109 }
110 }
111 }
112
113 return $images;
114 }
115}
116?>
diff --git a/inc/3rdparty/libraries/MOBIClass/PalmRecord.php b/inc/3rdparty/libraries/MOBIClass/PalmRecord.php
new file mode 100644
index 00000000..d0de8dfe
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/PalmRecord.php
@@ -0,0 +1,136 @@
1<?php
2/**
3 * A Record of a PDB file
4 *
5 * @author Sander
6 */
7class PalmRecord extends FileObject {
8 /**
9 * @var FileElement
10 */
11 private $elements;
12
13 public function __construct($settings, $records, $textRecords, $textLength, $images){
14 $this->elements = new FileElement(array(
15 "compression"=>new FileShort(),
16 "unused"=>new FileShort(),
17 "textLength"=>new FileInt(),
18 "recordCount"=>new FileShort(),
19 "recordSize"=>new FileShort(),
20 "encryptionType"=>new FileShort(),
21 "unused2"=>new FileShort(),
22 //MOBI Header
23 "mobiIdentifier"=>new FileString("MOBI", 4),
24 "mobiHeaderLength"=>new FileInt(),
25 "mobiType"=>new FileInt(),
26 "textEncoding"=>new FileInt(),
27 "uniqueID"=>new FileInt(),
28 "fileVersion"=>new FileInt(),
29 "reserved"=>new FileString(40),
30 "firstNonBookIndex"=>new FileInt(),
31 "fullNameOffset"=>new FileInt(),
32 "fullNameLength"=>new FileInt(),
33 "locale"=>new FileInt(),
34 "inputLanguage"=>new FileInt(),
35 "outputLanguage"=>new FileInt(),
36 "minimumVersion"=>new FileInt(),
37 "firstImageIndex"=>new FileInt(),
38 "huffmanRecordOffset"=>new FileInt(),
39 "huffmanRecordCount"=>new FileInt(),
40 "unused3"=>new FileString(8),
41 "exthFlags"=>new FileInt(0x40),
42 "unknown"=>new FileString(32),
43 "drmOffset"=>new FileInt(0xFFFFFFFF),
44 "drmCount"=>new FileShort(0xFFFFFFFF),
45 "drmSize"=>new FileShort(),
46 "drmFlags"=>new FileInt(),
47 "mobiFiller"=>new FileString(72),
48 //EXTH Header
49 "exthIdentifier"=>new FileString("EXTH", 4),
50 "exthHeaderLength"=>new FileInt(),
51 "exthRecordCount"=>new FileInt(),
52 "exthRecords"=>new FileElement(),
53 "exthPadding"=>new FileString(),
54 //"fullNamePadding"=>new FileString(100),
55 "fullName"=>new FileString()
56 ));
57
58 //Set values from the info block
59 foreach($settings->values as $name => $val){
60 //echo $name.", ";
61 if($this->elements->exists($name)){
62 $this->elements->get($name)->set($settings->get($name));
63 }
64 }
65
66 $els = $settings->values;
67
68 $exthElems = new FileElement();
69 $i = 0;
70 $l = 0;
71 foreach($els as $name=>$val){
72 $type = EXTHHelper::textToType($name);
73 if($type !== false){
74 $type = new FileInt($type);
75 $length = new FileInt(8+strlen($val));
76 $data = new FileString($val);
77 $l += 8+strlen($val);
78 $exthElems->add("type".$i, $type);
79 $exthElems->add("length".$i, $length);
80 $exthElems->add("data".$i, $data);
81 $i++;
82 }
83 }
84
85 if($images > 0){
86 $this->elements->get("firstImageIndex")->set($textRecords+1);
87 }
88 $this->elements->get("firstNonBookIndex")->set($textRecords+2+$images);
89 $this->elements->get("reserved")->set(str_pad("", 40, chr(255), STR_PAD_RIGHT));
90 $this->elements->get("exthRecordCount")->set($i);
91 $this->elements->set("exthRecords", $exthElems);
92 $pad = $l%4;
93 $pad = (4-$pad)%4;
94 $this->elements->get("exthPadding")->set(str_pad("", $pad, "\0", STR_PAD_RIGHT));
95 $this->elements->get("exthHeaderLength")->set(12+$l+$pad);
96
97
98 $this->elements->get("recordCount")->set($textRecords);
99
100 $this->elements->get("fullNameOffset")->set($this->elements->offsetToEntry("fullName"));
101 $this->elements->get("fullNameLength")->set(strlen($settings->get("title")));
102 $this->elements->get("fullName")->set($settings->get("title"));
103 $this->elements->get("textLength")->set($textLength);
104 }
105
106 public function getByteLength(){
107 return $this->getLength();
108 }
109
110 public function getLength(){
111 return $this->elements->getByteLength();
112 }
113
114 public function get(){
115 return $this;
116 }
117
118 public function set($elements){
119 throw new Exception("Unallowed set");
120 }
121
122 public function serialize() {
123 return $this->elements->serialize();
124 }
125
126 public function unserialize($data) {
127 $this->elements->unserialize($data);
128 }
129
130 public function __toString(){
131 $output = "PalmDoc Record (".$this->getByteLength()." bytes):\n";
132 $output .= $this->elements;
133 return $output;
134 }
135}
136?>
diff --git a/inc/3rdparty/libraries/MOBIClass/Prc.php b/inc/3rdparty/libraries/MOBIClass/Prc.php
new file mode 100644
index 00000000..c40b5805
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/Prc.php
@@ -0,0 +1,97 @@
1<?php
2
3/**
4 * Description of Prc
5 *
6 * @author Sander
7 */
8class Prc extends FileElement {
9 public function __construct($settings, $records){
10 parent::__construct(array(
11 "title"=>new FileString(32),
12 "attributes"=>new FileShort(),
13 "version"=>new FileShort(),
14 "creationTime"=>new FileDate(),
15 "modificationTime"=>new FileDate(),
16 "backupTime"=>new FileDate(),
17 "modificationNumber"=>new FileInt(),
18 "appInfoID"=>new FileInt(),
19 "sortInfoID"=>new FileInt(),
20 "prcType"=>new FileString(4),
21 "creator"=>new FileString(4),
22 "uniqueIDSeed"=>new FileInt(),
23 "nextRecordListID"=>new FileInt(),
24 "numberRecords"=>new FileShort(),
25 "recordList"=>new FileElement(),
26 "filler"=>new FileShort(),
27 "records"=>new FileElement()
28 ));
29
30 //Set values from the info block
31 foreach($this->elements as $name => $val){
32 if($settings->exists($name)){
33 $this->get($name)->set($settings->get($name));
34 }
35 }
36
37 $this->get("numberRecords")->set(sizeof($records));
38
39 $i = 0;
40 foreach($records as $record){
41 $offset = new FileInt();
42 $attr = new FileByte();
43 $uniqueID = new FileTri($i);
44
45 $this->elements["recordList"]->add("Rec".$i, new FileElement(array(
46 "offset"=>$offset,
47 "attribute"=>$attr,
48 "uniqueID"=>$uniqueID
49 )));
50
51 $this->elements["records"]->add("Rec".$i, $record);
52 $i++;
53 }
54
55 $this->updateOffsets($records);
56 }
57
58 public function getByteLength(){
59 throw new Exception("Test");
60 }
61
62 public function updateOffsets($records){
63 $base = $this->offsetToEntry("records");
64
65 $i = 0;
66
67 foreach($records as $record){
68 $el = $this->elements["recordList"]->get("Rec".$i);
69
70 $local = $this->elements["records"]->offsetToEntry("Rec".$i);
71
72 $el->get("offset")->set($base+$local);
73
74 $i++;
75 }
76 }
77
78 public function save($file){
79 $handle = fopen($file, "w");
80 fwrite($handle, $this->serialize());
81 fclose($handle);
82 }
83
84 public function output(){
85 echo $this->serialize();
86 }
87
88 public function __toString(){
89 $output = "Prc (".$this->getByteLength()." bytes): {\n";
90 foreach($this->elements as $key=>$value){
91 $output .= "\t".$key.": ".$value."\n";
92 }
93 $output .= "}";
94 return $output;
95 }
96}
97?>
diff --git a/inc/3rdparty/libraries/MOBIClass/PreprocessedArticle.php b/inc/3rdparty/libraries/MOBIClass/PreprocessedArticle.php
new file mode 100644
index 00000000..2e992404
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/PreprocessedArticle.php
@@ -0,0 +1,89 @@
1<?php
2
3/**
4 * Description of OnlineArticle
5 *
6 * @author Sander
7 */
8class PreprocessedArticle extends ContentProvider {
9 private $text;
10 private $images;
11 private $metadata = array();
12 private $imgCounter = 0;
13
14 public function __construct($textData, $imageLinks, $metadata) {
15 $this->text = $textData;
16 $this->metadata = $metadata;
17
18 $this->images = $this->downloadImages($imageLinks);
19 }
20
21 /**
22 * Create a Preprocessed article from a json string
23 * @param string $json JSON data. Should be of the following format:
24 * {"text": "TEXT", "images: ["imageURL1", "imageURL2"], "metadata": {"key": "value"}}
25 *
26 * Note: Any image tags should have the recindex attribute set to the appropriate index (the
27 * same index as the image in the array)
28 * @return PreprocessedArticle The generated preprocessed array
29 */
30 static public function CreateFromJson($json){
31 $data = json_decode($json);
32 return new PreprocessedArticle($data["text"], $data["images"], $data["metadata"]);
33 }
34
35 /**
36 * Get the text data to be integrated in the MOBI file
37 * @return string
38 */
39 public function getTextData(){
40 return $this->text;
41 }
42 /**
43 * Get the images (an array containing the jpeg data). Array entry 0 will
44 * correspond to image record 0.
45 * @return array
46 */
47 public function getImages(){
48 return $this->images;
49 }
50 /**
51 * Get the metadata in the form of a hashtable (for example, title or author).
52 * @return array
53 */
54 public function getMetaData(){
55 return $this->metadata;
56 }
57 /**
58 *
59 * @param DOMElement $dom
60 * @return array
61 */
62 private function downloadImages($links){
63 $images = array();
64 foreach($links as $link) {
65 $imgFile = @imagecreatefromstring(Http::Request($link));
66
67 if($imgFile === false){
68 $imgFile = @imagecreate(1, 1);
69 $black = @imagecolorallocate($imgFile, 255, 255, 255);
70 }
71 if($imgFile !== false){
72 @imagefilter($imgFile, IMG_FILTER_GRAYSCALE);
73
74 ob_start();
75 @imagejpeg($imgFile);
76 $image = ob_get_contents();
77 ob_end_clean();
78
79 $images[$this->imgCounter] = new FileRecord(new Record($image));
80 imagedestroy($imgFile);
81
82 $this->imgCounter++;
83 }
84 }
85
86 return $images;
87 }
88}
89?>
diff --git a/inc/3rdparty/libraries/MOBIClass/RecognizeURL.php b/inc/3rdparty/libraries/MOBIClass/RecognizeURL.php
new file mode 100644
index 00000000..6319ed57
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/RecognizeURL.php
@@ -0,0 +1,16 @@
1<?php
2
3/**
4 * Description of RecognizeURL
5 *
6 * @author Sander
7 */
8class RecognizeURL {
9 public static function GetContentHandler($url){
10 if(FanFictionNet::Matches($url)){
11 return new FanFictionNet($url);
12 }
13 return null;
14 }
15}
16?>
diff --git a/inc/3rdparty/libraries/MOBIClass/Record.php b/inc/3rdparty/libraries/MOBIClass/Record.php
new file mode 100644
index 00000000..3cb39582
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/Record.php
@@ -0,0 +1,96 @@
1<?php
2/**
3 * A Record of a PDB file
4 *
5 * @author Sander
6 */
7class Record extends FileObject {
8 /**
9 * Data in the record
10 * @var string
11 */
12 private $data;
13 /**
14 * Length of the record
15 * @var int
16 */
17 private $length;
18
19 /**
20 * Create a record
21 * @param string $data Data contained in the record
22 * @param int $length Length of the record (if set to -1,
23 * the length of $data will be taken)
24 */
25 public function __construct($data = "", $length = -1){
26 $this->data = $data;
27 if($length >= 0){
28 $this->length = $length;
29 }else{
30 $this->length = strlen($data);
31 }
32 }
33
34 public function compress($compression_method){
35 switch($compression_method){
36 case NO_COMPRESSION:
37 //Finished!
38 break;
39 case PALMDOC_COMPRESSION:
40 throw new Exception("Not implemented yet");
41 break;
42 case HUFF:
43 throw new Exception("Not implemented yet");
44 break;
45 default:
46 throw new Exception("Invalid argument");
47 }
48 }
49
50 public function getByteLength(){
51 return $this->getLength();
52 }
53
54 /**
55 * Get the length of the record
56 * @return int Length of the data
57 */
58 public function getLength(){
59 return $this->length;
60 }
61
62 /**
63 * Get the data contained in the record
64 * @return string Data contained in the record
65 */
66 public function get(){
67 return $this->data;
68 }
69
70 /**
71 * Set the data contained in the record
72 * @param string $value Data contained in the record
73 */
74 public function set($value){
75 $this->data = $value;
76 }
77
78 public function serialize(){
79 return $this->data;
80 }
81 public function unserialize($data){
82 __construct($data);
83 }
84
85 public function __toString() {
86 $toShow = $this->data;
87 if(strlen($this->data) > 103){
88 $toShow = substr($this->data, 0, 100)."...";
89 }
90 $out = "Record: {\n";
91 $out .= "\t".htmlspecialchars($toShow)."\n";
92 $out .= "}";
93 return $out;
94 }
95}
96?>
diff --git a/inc/3rdparty/libraries/MOBIClass/RecordFactory.php b/inc/3rdparty/libraries/MOBIClass/RecordFactory.php
new file mode 100644
index 00000000..12806fe3
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/RecordFactory.php
@@ -0,0 +1,115 @@
1<?php
2
3/**
4 * Helper class to help with creating records from a
5 * long data stream
6 *
7 * @author Sander
8 */
9class RecordFactory {
10 /**
11 * Settings for the record factory
12 * @var Settings
13 */
14 private $settings;
15
16 /**
17 * Create the helper class
18 * @param Settings $settings The Settings to be used for the records
19 */
20 public function __construct($settings){
21 $this->settings = $settings;
22 }
23
24 /**
25 * Create records from a data string
26 * @param string $data
27 * @return array(Record)
28 */
29 public function createRecords($data){
30 $records = array();
31 $size = $this->settings->get("recordSize");
32 $compression = $this->settings->get("compression");
33
34 $dataEntries = mb_str_split($data, $size);
35
36 for($i = 0, $len = sizeof($dataEntries); $i < $len; $i++){
37 $records[$i] = new Record($dataEntries[$i]);
38 $records[$i]->compress($compression);
39 }
40
41 return $records;
42 }
43
44 public function createEOFRecord(){
45 return new Record(0xe98e0d0a);
46 }
47
48 public function createFCISRecord($textLength){
49 $r = "FCIS";
50 $r .= $this->asString(20, 4);
51 $r .= $this->asString(16, 4);
52 $r .= $this->asString(1, 4);
53 $r .= $this->asString(0, 4);
54 $r .= $this->asString($textLength, 4);
55 $r .= $this->asString(0, 4);
56 $r .= $this->asString(32, 4);
57 $r .= $this->asString(8, 4);
58 $r .= $this->asString(1, 2);
59 $r .= $this->asString(1, 2);
60 $r .= $this->asString(0, 4);
61 return new Record($r);
62 }
63
64 public function createFLISRecord(){
65 $r = "FLIS";
66 $r .= $this->asString(8, 4);
67 $r .= $this->asString(65, 2);
68 $r .= $this->asString(0, 2);
69 $r .= $this->asString(0, 4);
70 $r .= $this->asString(-1, 4);
71 $r .= $this->asString(1, 2);
72 $r .= $this->asString(3, 2);
73 $r .= $this->asString(3, 4);
74 $r .= $this->asString(1, 4);
75 $r .= $this->asString(-1, 4);
76 return new Record($r);
77 }
78
79 private function asString($int, $size){
80 $out = "";
81 for($i = 0; $i < $size; $i++){
82 if($i > 0) $out = " ".$out;
83 $byte = dechex($int & 0xFF);
84 if(strlen($byte) == 1) $byte = "0".$byte;
85 $out = $byte.$out;
86 $int = $int >> 8;
87 }
88 return $out;
89 }
90
91 public function __toString() {
92 $out = "Record Factory: {\n";
93 $out .= "\tRecord Size: ".$this->settings->get("recordSize")."\n";
94 $out .= "\tCompression: ".$this->settings->get("compression")."\n";
95 $out .= "}";
96 return $out;
97 }
98}
99function mb_str_split($string, $split_length = 1){
100 mb_internal_encoding('UTF-8');
101 mb_regex_encoding('UTF-8');
102
103 $split_length = ($split_length <= 0) ? 1 : $split_length;
104
105 $mb_strlen = mb_strlen($string, 'utf-8');
106
107 $array = array();
108
109 for($i = 0; $i < $mb_strlen; $i += $split_length){
110 $array[] = mb_substr($string, $i, $split_length);
111 }
112
113 return $array;
114}
115?>
diff --git a/inc/3rdparty/libraries/MOBIClass/Settings.php b/inc/3rdparty/libraries/MOBIClass/Settings.php
new file mode 100644
index 00000000..ddcf2054
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/Settings.php
@@ -0,0 +1,97 @@
1<?php
2
3/**
4 * Description of Settings
5 *
6 * @author Sander
7 */
8class Settings {
9 /**
10 * Values of the settings
11 * @var array
12 */
13 public $values;
14
15 /**
16 * Construct a Settings object with the default settings. If necessary,
17 * those settings can be extended with additional settings
18 * @param array $additionalSettings Additional settings to add (should
19 * be added with a key/value pair format.
20 */
21 public function __construct($additionalSettings = array()) {
22 // Most values shouldn't be changed (the result will be an invalid file)
23 $this->values = array(
24 "attributes"=>0,
25 "version"=>0,
26 "creationTime"=>time()+94694400,
27 "modificationTime"=>time()+94694400,
28 "backupTime"=>0,
29 "modificationNumber"=>0,
30 "appInfoID"=>0,
31 "sortInfoID"=>0,
32 "prcType"=>"BOOK",
33 "creator"=>"MOBI",
34 "uniqueIDSeed"=>rand(),
35 "nextRecordListID"=>0,
36 "recordAttributes"=>0,
37 "compression"=>NO_COMPRESSION,
38 "recordSize"=>RECORD_SIZE,
39 "encryptionType"=>NO_ENCRYPTION,
40 "mobiIdentifier"=>"MOBI",
41 "mobiHeaderLength"=>0xe8,
42 "mobiType"=>MOBIPOCKET_BOOK,
43 "textEncoding"=>UTF8,
44 "uniqueID"=>rand(),
45 "fileVersion"=>6,
46 "locale"=>0x09,
47 "inputLanguage"=>0,
48 "outputLanguage"=>0,
49 "minimumVersion"=>6,
50 "huffmanRecordOffset"=>0,
51 "huffmanRecordCount"=>0,
52 "exthFlags"=>0x40,
53 "drmOffset"=>0xFFFFFFFF,
54 "drmCount"=>0,
55 "drmSize"=>0,
56 "drmFlags"=>0,
57 "extraDataFlags"=>0,
58 "exthIdentifier"=>"EXTH",
59 // These can be changed without any risk
60 "title"=>"Unknown title",
61 "author"=>"Unknown author",
62 "subject"=>"Unknown subject"
63 );
64
65 foreach($additionalSettings as $key=>$value){
66 $this->values[$key] = $value;
67 }
68 }
69
70 /**
71 * Get a value from the settings
72 * @param string $key Key of the setting
73 * @return mixed The value of the setting
74 */
75 public function get($key){
76 return $this->values[$key];
77 }
78
79 /**
80 * Checks if a value is set
81 * @param string $key Key of the setting
82 * @return bool True if the value exists
83 */
84 public function exists($key){
85 return isset($this->values[$key]);
86 }
87
88 public function __toString() {
89 $out = "Settings: {\n";
90 foreach($this->values as $key=>$value){
91 $out .= "\t".$key.": ".$value."\n";
92 }
93 $out .= "}";
94 return $out;
95 }
96}
97?>
diff --git a/inc/3rdparty/libraries/MOBIClass/constants.php b/inc/3rdparty/libraries/MOBIClass/constants.php
new file mode 100644
index 00000000..bd363118
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/constants.php
@@ -0,0 +1,11 @@
1<?php
2define("NO_COMPRESSION", 1);
3define("PALMDOC_COMPRESSION", 2);
4define("HUFF", 17480);
5define("RECORD_SIZE", 4096);
6
7define("NO_ENCRYPTION", 0);
8
9define("MOBIPOCKET_BOOK", 2);
10define("CP1252", 1252);
11define("UTF8", 65001); \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/downloaders/FanFictionNet.php b/inc/3rdparty/libraries/MOBIClass/downloaders/FanFictionNet.php
new file mode 100644
index 00000000..65d5a466
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/downloaders/FanFictionNet.php
@@ -0,0 +1,125 @@
1<?php
2
3/**
4 * Description of FanFictionNet
5 *
6 * @author Sander
7 */
8class FanFictionNet extends MultipleFileHandler {
9 private static $prefix = "http://www.fanfiction.net/s/";
10 private $downloadedMetadata = false;
11 private $id = 0;
12 private $chapterCount = -1;
13
14 public function __construct($url) {
15 $ending = substr($url, strlen(self::$prefix));
16 $this->id = intval(substr($ending, 0, strpos($ending, "/")));
17
18 for($i = 1; $i <= max(1, $this->chapterCount); $i++){
19 $this->addChapter($i);
20 }
21 }
22
23 private function addChapter($n){
24 $doc = new DOMDocument();
25 $file = Http::Request(self::$prefix.$this->id."/".$n."/");
26 @$doc->loadHTML($file) or die($file);
27
28 if(!$this->downloadedMetadata){
29 $this->loadMetadata($doc);
30 $this->downloadedMetadata = true;
31 }
32 if($this->chapterCount < 0){
33 $this->chapterCount = $this->getNumberChapters($doc);
34
35 if($this->chapterCount > 4){
36 die("Too many files to download, don't use php for this!");
37 }
38 }
39
40 $textEl = $doc->getElementById("storytext");
41 if($textEl == null) die("Error: ".$doc->saveHTML());
42 $horizontalRulebars = $doc->getElementsByTagName('hr');
43 /**
44 * @var DOMNode
45 */
46 $hr;
47 foreach($horizontalRulebars as $hr) {
48 $hr->setAttribute("size", null);
49 $hr->setAttribute("noshade", null);
50 }
51 $text = $this->innerHtml($textEl);
52
53 $title = "";
54 $selects = $doc->getElementsByTagName('select');
55 foreach($selects as $select) {
56 if($select->hasAttribute("name") && $select->getAttribute("name") == "chapter"){
57 $options = $select->getElementsByTagName("option");
58
59 $test = $n.". ";
60 foreach($options as $option){
61 $val = $option->nodeValue;
62 if(substr($val, 0, strlen($test)) == $test){
63 $title = substr($val, strlen($test));
64 break;
65 }
66 }
67 break;
68 }
69 }
70 $this->addPage($text, $title);
71 }
72
73 private function getNumberChapters($doc){
74 $selects = $doc->getElementsByTagName('select');
75 foreach($selects as $select) {
76 if($select->hasAttribute("name") && $select->getAttribute("name") == "chapter"){
77 $options = $select->getElementsByTagName("option");
78
79 $count = $options->length;
80 return $count;
81 }
82 }
83 }
84
85 private function loadMetadata($doc){
86 //Author
87 $links = $doc->getElementsByTagName('a');
88 foreach($links as $link) {
89 if($link == null){
90 var_dump($link);
91 }
92 if($link->hasAttribute("href") && substr($link->getAttribute("href"), 0, 3) == "/u/"){
93 $this->setMetadata("author", $link->nodeValue);
94 }
95 }
96 //Title
97 /*
98 $links = $doc->getElementsByTagName('link');
99 foreach($links as $link) {
100 if($link->hasAttribute("rel") && $link->getAttribute("rel") == "canonical"){
101 $url = $link->getAttribute("href");
102 $title = str_replace("_", " ", substr($url, strrpos($url, "/")+1));
103 $this->setMetadata("title", $title);
104 }
105 }*/
106
107 //TODO: Find a more reliable way to extract the title
108 $title = $doc->getElementsByTagName("b")->item(0)->nodeValue;
109 $this->setMetadata("title", $title);
110 }
111
112 private function innerHtml($node){
113 $doc = new DOMDocument();
114 foreach ($node->childNodes as $child)
115 $doc->appendChild($doc->importNode($child, true));
116
117 return $doc->saveHTML();
118 }
119
120 public static function Matches($url){
121 //TODO: Implement with regex
122 return strpos($url, self::$prefix) !== false;
123 }
124}
125?>
diff --git a/inc/3rdparty/libraries/MOBIClass/http_build_url.php b/inc/3rdparty/libraries/MOBIClass/http_build_url.php
new file mode 100644
index 00000000..b475edb0
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/http_build_url.php
@@ -0,0 +1,94 @@
1<?php
2if(!is_callable("http_build_url")){
3 define('HTTP_URL_REPLACE', 1); // Replace every part of the first URL when there's one of the second URL
4 define('HTTP_URL_JOIN_PATH', 2); // Join relative paths
5 define('HTTP_URL_JOIN_QUERY', 4); // Join query strings
6 define('HTTP_URL_STRIP_USER', 8); // Strip any user authentication information
7 define('HTTP_URL_STRIP_PASS', 16); // Strip any password authentication information
8 define('HTTP_URL_STRIP_AUTH', 32); // Strip any authentication information
9 define('HTTP_URL_STRIP_PORT', 64); // Strip explicit port numbers
10 define('HTTP_URL_STRIP_PATH', 128); // Strip complete path
11 define('HTTP_URL_STRIP_QUERY', 256); // Strip query string
12 define('HTTP_URL_STRIP_FRAGMENT', 512); // Strip any fragments (#identifier)
13 define('HTTP_URL_STRIP_ALL', 1024); // Strip anything but scheme and host
14
15 // Build an URL
16 // The parts of the second URL will be merged into the first according to the flags argument.
17 //
18 // @param mixed (Part(s) of) an URL in form of a string or associative array like parse_url() returns
19 // @param mixed Same as the first argument
20 // @param int A bitmask of binary or'ed HTTP_URL constants (Optional)HTTP_URL_REPLACE is the default
21 // @param array If set, it will be filled with the parts of the composed url like parse_url() would return
22 function http_build_url($url, $parts = array (), $flags = HTTP_URL_REPLACE, &$new_url = false) {
23 $keys = array (
24 'user',
25 'pass',
26 'port',
27 'path',
28 'query',
29 'fragment'
30 );
31
32 // HTTP_URL_STRIP_ALL becomes all the HTTP_URL_STRIP_Xs
33 if ($flags & HTTP_URL_STRIP_ALL) {
34 $flags |= HTTP_URL_STRIP_USER;
35 $flags |= HTTP_URL_STRIP_PASS;
36 $flags |= HTTP_URL_STRIP_PORT;
37 $flags |= HTTP_URL_STRIP_PATH;
38 $flags |= HTTP_URL_STRIP_QUERY;
39 $flags |= HTTP_URL_STRIP_FRAGMENT;
40 }
41 // HTTP_URL_STRIP_AUTH becomes HTTP_URL_STRIP_USER and HTTP_URL_STRIP_PASS
42 else if ($flags & HTTP_URL_STRIP_AUTH) {
43 $flags |= HTTP_URL_STRIP_USER;
44 $flags |= HTTP_URL_STRIP_PASS;
45 }
46
47 // Parse the original URL
48 $parse_url = parse_url($url);
49
50 // Scheme and Host are always replaced
51 if (isset($parts['scheme']))
52 $parse_url['scheme'] = $parts['scheme'];
53
54 if (isset($parts['host']))
55 $parse_url['host'] = $parts['host'];
56
57 // (If applicable) Replace the original URL with it's new parts
58 if ($flags & HTTP_URL_REPLACE) {
59 foreach ($keys as $key) {
60 if (isset($parts[$key]))
61 $parse_url[$key] = $parts[$key];
62 }
63 } else {
64 // Join the original URL path with the new path
65 if (isset($parts['path']) && ($flags & HTTP_URL_JOIN_PATH)) {
66 if (isset($parse_url['path']))
67 $parse_url['path'] = rtrim(str_replace(basename($parse_url['path']), '', $parse_url['path']), '/') . '/' . ltrim($parts['path'], '/');
68 else
69 $parse_url['path'] = $parts['path'];
70 }
71
72 // Join the original query string with the new query string
73 if (isset($parts['query']) && ($flags & HTTP_URL_JOIN_QUERY)) {
74 if (isset($parse_url['query']))
75 $parse_url['query'] .= '&' . $parts['query'];
76 else
77 $parse_url['query'] = $parts['query'];
78 }
79 }
80
81 // Strips all the applicable sections of the URL
82 // Note: Scheme and Host are never stripped
83 foreach ($keys as $key) {
84 if ($flags & (int)constant('HTTP_URL_STRIP_' . strtoupper($key)))
85 unset($parse_url[$key]);
86 }
87
88 $new_url = $parse_url;
89
90 return ((isset($parse_url['scheme'])) ? $parse_url['scheme'] . '://' : '') . ((isset($parse_url['user'])) ? $parse_url['user'] . ((isset($parse_url['pass'])) ? ':' . $parse_url['pass'] : '') . '@' : '')
91 . ((isset($parse_url['host'])) ? $parse_url['host'] : '') . ((isset($parse_url['port'])) ? ':' . $parse_url['port'] : '') . ((isset($parse_url['path'])) ? $parse_url['path'] : '')
92 . ((isset($parse_url['query'])) ? '?' . $parse_url['query'] : '') . ((isset($parse_url['fragment'])) ? '#' . $parse_url['fragment'] : '');
93 }
94}
diff --git a/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php b/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php
new file mode 100644
index 00000000..1a8ec88c
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/readability/JSLikeHTMLElement.php
@@ -0,0 +1,110 @@
1<?php
2/**
3* JavaScript-like HTML DOM Element
4*
5* This class extends PHP's DOMElement to allow
6* users to get and set the innerHTML property of
7* HTML elements in the same way it's done in
8* JavaScript.
9*
10* Example usage:
11* @code
12* require_once 'JSLikeHTMLElement.php';
13* header('Content-Type: text/plain');
14* $doc = new DOMDocument();
15* $doc->registerNodeClass('DOMElement', 'JSLikeHTMLElement');
16* $doc->loadHTML('<div><p>Para 1</p><p>Para 2</p></div>');
17* $elem = $doc->getElementsByTagName('div')->item(0);
18*
19* // print innerHTML
20* echo $elem->innerHTML; // prints '<p>Para 1</p><p>Para 2</p>'
21* echo "\n\n";
22*
23* // set innerHTML
24* $elem->innerHTML = '<a href="http://fivefilters.org">FiveFilters.org</a>';
25* echo $elem->innerHTML; // prints '<a href="http://fivefilters.org">FiveFilters.org</a>'
26* echo "\n\n";
27*
28* // print document (with our changes)
29* echo $doc->saveXML();
30* @endcode
31*
32* @author Keyvan Minoukadeh - http://www.keyvan.net - keyvan@keyvan.net
33* @see http://fivefilters.org (the project this was written for)
34*/
35class JSLikeHTMLElement extends DOMElement
36{
37 /**
38 * Used for setting innerHTML like it's done in JavaScript:
39 * @code
40 * $div->innerHTML = '<h2>Chapter 2</h2><p>The story begins...</p>';
41 * @endcode
42 */
43 public function __set($name, $value) {
44 if ($name == 'innerHTML') {
45 // first, empty the element
46 for ($x=$this->childNodes->length-1; $x>=0; $x--) {
47 $this->removeChild($this->childNodes->item($x));
48 }
49 // $value holds our new inner HTML
50 if ($value != '') {
51 $f = $this->ownerDocument->createDocumentFragment();
52 // appendXML() expects well-formed markup (XHTML)
53 $result = @$f->appendXML($value); // @ to suppress PHP warnings
54 if ($result) {
55 if ($f->hasChildNodes()) $this->appendChild($f);
56 } else {
57 // $value is probably ill-formed
58 $f = new DOMDocument();
59 $value = mb_convert_encoding($value, 'HTML-ENTITIES', 'UTF-8');
60 // Using <htmlfragment> will generate a warning, but so will bad HTML
61 // (and by this point, bad HTML is what we've got).
62 // We use it (and suppress the warning) because an HTML fragment will
63 // be wrapped around <html><body> tags which we don't really want to keep.
64 // Note: despite the warning, if loadHTML succeeds it will return true.
65 $result = @$f->loadHTML('<htmlfragment>'.$value.'</htmlfragment>');
66 if ($result) {
67 $import = $f->getElementsByTagName('htmlfragment')->item(0);
68 foreach ($import->childNodes as $child) {
69 $importedNode = $this->ownerDocument->importNode($child, true);
70 $this->appendChild($importedNode);
71 }
72 } else {
73 // oh well, we tried, we really did. :(
74 // this element is now empty
75 }
76 }
77 }
78 } else {
79 $trace = debug_backtrace();
80 trigger_error('Undefined property via __set(): '.$name.' in '.$trace[0]['file'].' on line '.$trace[0]['line'], E_USER_NOTICE);
81 }
82 }
83
84 /**
85 * Used for getting innerHTML like it's done in JavaScript:
86 * @code
87 * $string = $div->innerHTML;
88 * @endcode
89 */
90 public function __get($name)
91 {
92 if ($name == 'innerHTML') {
93 $inner = '';
94 foreach ($this->childNodes as $child) {
95 $inner .= $this->ownerDocument->saveXML($child);
96 }
97 return $inner;
98 }
99
100 $trace = debug_backtrace();
101 trigger_error('Undefined property via __get(): '.$name.' in '.$trace[0]['file'].' on line '.$trace[0]['line'], E_USER_NOTICE);
102 return null;
103 }
104
105 public function __toString()
106 {
107 return '['.$this->tagName.']';
108 }
109}
110?> \ No newline at end of file
diff --git a/inc/3rdparty/libraries/MOBIClass/readability/Readability.php b/inc/3rdparty/libraries/MOBIClass/readability/Readability.php
new file mode 100644
index 00000000..91554243
--- /dev/null
+++ b/inc/3rdparty/libraries/MOBIClass/readability/Readability.php
@@ -0,0 +1,1069 @@
1<?php
2/**
3* Arc90's Readability ported to PHP for FiveFilters.org
4* Based on readability.js version 1.7.1 (without multi-page support)
5* ------------------------------------------------------
6* Original URL: http://lab.arc90.com/experiments/readability/js/readability.js
7* Arc90's project URL: http://lab.arc90.com/experiments/readability/
8* JS Source: http://code.google.com/p/arc90labs-readability
9* Ported by: Keyvan Minoukadeh, http://www.keyvan.net
10* More information: http://fivefilters.org/content-only/
11* License: Apache License, Version 2.0
12* Requires: PHP5
13* Date: 2010-10-29
14*
15* Differences between the PHP port and the original
16* ------------------------------------------------------
17* Arc90's Readability is designed to run in the browser. It works on the DOM
18* tree (the parsed HTML) after the page's CSS styles have been applied and
19* Javascript code executed. This PHP port does not run inside a browser.
20* We use PHP's ability to parse HTML to build our DOM tree, but we cannot
21* rely on CSS or Javascript support. As such, the results will not always
22* match Arc90's Readability. (For example, if a web page contains CSS style
23* rules or Javascript code which hide certain HTML elements from display,
24* Arc90's Readability will dismiss those from consideration but our PHP port,
25* unable to understand CSS or Javascript, will not know any better.)
26*
27* Another significant difference is that the aim of Arc90's Readability is
28* to re-present the main content block of a given web page so users can
29* read it more easily in their browsers. Correct identification, clean up,
30* and separation of the content block is only a part of this process.
31* This PHP port is only concerned with this part, it does not include code
32* that relates to presentation in the browser - Arc90 already do
33* that extremely well, and for PDF output there's FiveFilters.org's
34* PDF Newspaper: http://fivefilters.org/pdf-newspaper/.
35*
36* Finally, this class contains methods that might be useful for developers
37* working on HTML document fragments. So without deviating too much from
38* the original code (which I don't want to do because it makes debugging
39* and updating more difficult), I've tried to make it a little more
40* developer friendly. You should be able to use the methods here on
41* existing DOMElement objects without passing an entire HTML document to
42* be parsed.
43*/
44
45// This class allows us to do JavaScript like assignements to innerHTML
46require_once(dirname(__FILE__).'/JSLikeHTMLElement.php');
47
48// Alternative usage (for testing only!)
49// uncomment the lins below and call Readability.php in your browser
50// passing it the URL of the page you'd like content from, e.g.:
51// Readability.php?url=http://medialens.org/alerts/09/090615_the_guardian_climate.php
52
53/*
54if (!isset($_GET['url']) || $_GET['url'] == '') {
55 die('Please pass a URL to the script. E.g. Readability.php?url=bla.com/story.html');
56}
57$url = $_GET['url'];
58if (!preg_match('!^https?://!i', $url)) $url = 'http://'.$url;
59$html = file_get_contents($url);
60$r = new Readability($html, $url);
61$r->init();
62echo $r->articleContent->innerHTML;
63*/
64
65class Readability
66{
67 public $version = '1.7.1-without-multi-page';
68 public $convertLinksToFootnotes = false;
69 public $revertForcedParagraphElements = true;
70 public $articleTitle;
71 public $articleContent;
72 public $dom;
73 public $url = null; // optional - URL where HTML was retrieved
74 public $debug = false;
75 protected $body = null; //
76 protected $bodyCache = null; // Cache the body HTML in case we need to re-use it later
77 protected $flags = 7; // 1 | 2 | 4; // Start with all flags set.
78 protected $success = false; // indicates whether we were able to extract or not
79
80 /**
81 * All of the regular expressions in use within readability.
82 * Defined up here so we don't instantiate them repeatedly in loops.
83 **/
84 public $regexps = array(
85 'unlikelyCandidates' => '/combx|comment|community|disqus|extra|foot|header|menu|remark|rss|shoutbox|sidebar|sponsor|ad-break|agegate|pagination|pager|popup|tweet|twitter/i',
86 'okMaybeItsACandidate' => '/and|article|body|column|main|shadow/i',
87 'positive' => '/article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i',
88 'negative' => '/combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget/i',
89 'divToPElements' => '/<(a|blockquote|dl|div|img|ol|p|pre|table|ul)/i',
90 'replaceBrs' => '/(<br[^>]*>[ \n\r\t]*){2,}/i',
91 'replaceFonts' => '/<(\/?)font[^>]*>/i',
92 // 'trimRe' => '/^\s+|\s+$/g', // PHP has trim()
93 'normalize' => '/\s{2,}/',
94 'killBreaks' => '/(<br\s*\/?>(\s|&nbsp;?)*){1,}/',
95 'video' => '/http:\/\/(www\.)?(youtube|vimeo)\.com/i',
96 'skipFootnoteLink' => '/^\s*(\[?[a-z0-9]{1,2}\]?|^|edit|citation needed)\s*$/i'
97 );
98
99 /* constants */
100 const FLAG_STRIP_UNLIKELYS = 1;
101 const FLAG_WEIGHT_CLASSES = 2;
102 const FLAG_CLEAN_CONDITIONALLY = 4;
103
104 /**
105 * Create instance of Readability
106 * @param string UTF-8 encoded string
107 * @param string (optional) URL associated with HTML (used for footnotes)
108 */
109 function __construct($html, $url=null)
110 {
111 /* Turn all double br's into p's */
112 /* Note, this is pretty costly as far as processing goes. Maybe optimize later. */
113 $html = preg_replace($this->regexps['replaceBrs'], '</p><p>', $html);
114 $html = preg_replace($this->regexps['replaceFonts'], '<$1span>', $html);
115 $html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
116 $this->dom = new DOMDocument();
117 $this->dom->preserveWhiteSpace = false;
118 $this->dom->registerNodeClass('DOMElement', 'JSLikeHTMLElement');
119 @$this->dom->loadHTML($html);
120 $this->url = $url;
121 }
122
123 /**
124 * Get article title element
125 * @return DOMElement
126 */
127 public function getTitle() {
128 return $this->articleTitle;
129 }
130
131 /**
132 * Get article content element
133 * @return DOMElement
134 */
135 public function getContent() {
136 return $this->articleContent;
137 }
138
139 /**
140 * Runs readability.
141 *
142 * Workflow:
143 * 1. Prep the document by removing script tags, css, etc.
144 * 2. Build readability's DOM tree.
145 * 3. Grab the article content from the current dom tree.
146 * 4. Replace the current DOM tree with the new one.
147 * 5. Read peacefully.
148 *
149 * @return boolean true if we found content, false otherwise
150 **/
151 public function init()
152 {
153 $this->removeScripts($this->dom);
154
155 // Assume successful outcome
156 $this->success = true;
157
158 $bodyElems = $this->dom->getElementsByTagName('body');
159 if ($bodyElems->length > 0) {
160 if ($this->bodyCache == null) {
161 $this->bodyCache = $bodyElems->item(0)->innerHTML;
162 }
163 if ($this->body == null) {
164 $this->body = $bodyElems->item(0);
165 }
166 }
167
168 $this->prepDocument();
169
170 //die($this->dom->documentElement->parentNode->nodeType);
171 //$this->setInnerHTML($this->dom->documentElement, $this->getInnerHTML($this->dom->documentElement));
172 //die($this->getInnerHTML($this->dom->documentElement));
173
174 /* Build readability's DOM tree */
175 $overlay = $this->dom->createElement('div');
176 $innerDiv = $this->dom->createElement('div');
177 $articleTitle = $this->getArticleTitle();
178 $articleContent = $this->grabArticle();
179
180 if (!$articleContent) {
181 $this->success = false;
182 $articleContent = $this->dom->createElement('div');
183 $articleContent->setAttribute('id', 'readability-content');
184 $articleContent->innerHTML = '<p>Sorry, Readability was unable to parse this page for content.</p>';
185 }
186
187 $overlay->setAttribute('id', 'readOverlay');
188 $innerDiv->setAttribute('id', 'readInner');
189
190 /* Glue the structure of our document together. */
191 $innerDiv->appendChild($articleTitle);
192 $innerDiv->appendChild($articleContent);
193 $overlay->appendChild($innerDiv);
194
195 /* Clear the old HTML, insert the new content. */
196 $this->body->innerHTML = '';
197 $this->body->appendChild($overlay);
198 //document.body.insertBefore(overlay, document.body.firstChild);
199 $this->body->removeAttribute('style');
200
201 $this->postProcessContent($articleContent);
202
203 // Set title and content instance variables
204 $this->articleTitle = $articleTitle;
205 $this->articleContent = $articleContent;
206
207 return $this->success;
208 }
209
210 /**
211 * Debug
212 */
213 protected function dbg($msg) {
214 if ($this->debug) echo '* ',$msg, '<br />', "\n";
215 }
216
217 /**
218 * Run any post-process modifications to article content as necessary.
219 *
220 * @param DOMElement
221 * @return void
222 */
223 public function postProcessContent($articleContent) {
224 if ($this->convertLinksToFootnotes && !preg_match('/wikipedia\.org/', @$this->url)) {
225 $this->addFootnotes($articleContent);
226 }
227 }
228
229 /**
230 * Get the article title as an H1.
231 *
232 * @return DOMElement
233 */
234 protected function getArticleTitle() {
235 $curTitle = '';
236 $origTitle = '';
237
238 try {
239 $curTitle = $origTitle = $this->getInnerText($this->dom->getElementsByTagName('title')->item(0));
240 } catch(Exception $e) {}
241
242 if (preg_match('/ [\|\-] /', $curTitle))
243 {
244 $curTitle = preg_replace('/(.*)[\|\-] .*/i', '$1', $origTitle);
245
246 if (count(explode(' ', $curTitle)) < 3) {
247 $curTitle = preg_replace('/[^\|\-]*[\|\-](.*)/i', '$1', $origTitle);
248 }
249 }
250 else if (strpos($curTitle, ': ') !== false)
251 {
252 $curTitle = preg_replace('/.*:(.*)/i', '$1', $origTitle);
253
254 if (count(explode(' ', $curTitle)) < 3) {
255 $curTitle = preg_replace('/[^:]*[:](.*)/i','$1', $origTitle);
256 }
257 }
258 else if(strlen($curTitle) > 150 || strlen($curTitle) < 15)
259 {
260 $hOnes = $this->dom->getElementsByTagName('h1');
261 if($hOnes->length == 1)
262 {
263 $curTitle = $this->getInnerText($hOnes->item(0));
264 }
265 }
266
267 $curTitle = trim($curTitle);
268
269 if (count(explode(' ', $curTitle)) <= 4) {
270 $curTitle = $origTitle;
271 }
272
273 $articleTitle = $this->dom->createElement('h1');
274 $articleTitle->innerHTML = $curTitle;
275
276 return $articleTitle;
277 }
278
279 /**
280 * Prepare the HTML document for readability to scrape it.
281 * This includes things like stripping javascript, CSS, and handling terrible markup.
282 *
283 * @return void
284 **/
285 protected function prepDocument() {
286 /**
287 * In some cases a body element can't be found (if the HTML is totally hosed for example)
288 * so we create a new body node and append it to the document.
289 */
290 if($this->dom->documentElement == null){
291 throw new Exception("No document element");
292 }
293 if ($this->body == null)
294 {
295 $this->body = $this->dom->createElement('body');
296 $this->dom->documentElement->appendChild($this->body);
297 }
298
299 $this->body->setAttribute('id', 'readabilityBody');
300
301 /* Remove all style tags in head */
302 $styleTags = $this->dom->getElementsByTagName('style');
303 for ($i = $styleTags->length-1; $i >= 0; $i--)
304 {
305 $styleTags->item($i)->parentNode->removeChild($styleTags->item($i));
306 }
307
308 /* Turn all double br's into p's */
309 /* Note, this is pretty costly as far as processing goes. Maybe optimize later. */
310 //document.body.innerHTML = document.body.innerHTML.replace(readability.regexps.replaceBrs, '</p><p>').replace(readability.regexps.replaceFonts, '<$1span>');
311 // We do this in the constructor for PHP as that's when we have raw HTML - before parsing it into a DOM tree.
312 // Manipulating innerHTML as it's done in JS is not possible in PHP.
313 }
314
315 /**
316 * For easier reading, convert this document to have footnotes at the bottom rather than inline links.
317 * @see http://www.roughtype.com/archives/2010/05/experiments_in.php
318 *
319 * @return void
320 **/
321 public function addFootnotes($articleContent) {
322 $footnotesWrapper = $this->dom->createElement('div');
323 $footnotesWrapper->setAttribute('id', 'readability-footnotes');
324 $footnotesWrapper->innerHTML = '<h3>References</h3>';
325
326 $articleFootnotes = $this->dom->createElement('ol');
327 $articleFootnotes->setAttribute('id', 'readability-footnotes-list');
328 $footnotesWrapper->appendChild($articleFootnotes);
329
330 $articleLinks = $articleContent->getElementsByTagName('a');
331
332 $linkCount = 0;
333 for ($i = 0; $i < $articleLinks->length; $i++)
334 {
335 $articleLink = $articleLinks->item($i);
336 $footnoteLink = $articleLink->cloneNode(true);
337 $refLink = $this->dom->createElement('a');
338 $footnote = $this->dom->createElement('li');
339 $linkDomain = @parse_url($footnoteLink->getAttribute('href'), PHP_URL_HOST);
340 if (!$linkDomain && isset($this->url)) $linkDomain = @parse_url($this->url, PHP_URL_HOST);
341 //linkDomain = footnoteLink.host ? footnoteLink.host : document.location.host,
342 $linkText = $this->getInnerText($articleLink);
343
344 if ((strpos($articleLink->getAttribute('class'), 'readability-DoNotFootnote') !== false) || preg_match($this->regexps['skipFootnoteLink'], $linkText)) {
345 continue;
346 }
347
348 $linkCount++;
349
350 /** Add a superscript reference after the article link */
351 $refLink->setAttribute('href', '#readabilityFootnoteLink-' . $linkCount);
352 $refLink->innerHTML = '<small><sup>[' . $linkCount . ']</sup></small>';
353 $refLink->setAttribute('class', 'readability-DoNotFootnote');
354 $refLink->setAttribute('style', 'color: inherit;');
355
356 //TODO: does this work or should we use DOMNode.isSameNode()?
357 if ($articleLink->parentNode->lastChild == $articleLink) {
358 $articleLink->parentNode->appendChild($refLink);
359 } else {
360 $articleLink->parentNode->insertBefore($refLink, $articleLink->nextSibling);
361 }
362
363 $articleLink->setAttribute('style', 'color: inherit; text-decoration: none;');
364 $articleLink->setAttribute('name', 'readabilityLink-' . $linkCount);
365
366 $footnote->innerHTML = '<small><sup><a href="#readabilityLink-' . $linkCount . '" title="Jump to Link in Article">^</a></sup></small> ';
367
368 $footnoteLink->innerHTML = ($footnoteLink->getAttribute('title') != '' ? $footnoteLink->getAttribute('title') : $linkText);
369 $footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount);
370
371 $footnote->appendChild($footnoteLink);
372 if ($linkDomain) $footnote->innerHTML = $footnote->innerHTML . '<small> (' . $linkDomain . ')</small>';
373
374 $articleFootnotes->appendChild($footnote);
375 }
376
377 if ($linkCount > 0) {
378 $articleContent->appendChild($footnotesWrapper);
379 }
380 }
381
382 /**
383 * Reverts P elements with class 'readability-styled'
384 * to text nodes - which is what they were before.
385 *
386 * @param DOMElement
387 * @return void
388 */
389 function revertReadabilityStyledElements($articleContent) {
390 $xpath = new DOMXPath($articleContent->ownerDocument);
391 $elems = $xpath->query('.//p[@class="readability-styled"]', $articleContent);
392 //$elems = $articleContent->getElementsByTagName('p');
393 for ($i = $elems->length-1; $i >= 0; $i--) {
394 $e = $elems->item($i);
395 $e->parentNode->replaceChild($articleContent->ownerDocument->createTextNode($e->textContent), $e);
396 //if ($e->hasAttribute('class') && $e->getAttribute('class') == 'readability-styled') {
397 // $e->parentNode->replaceChild($this->dom->createTextNode($e->textContent), $e);
398 //}
399 }
400 }
401
402 /**
403 * Prepare the article node for display. Clean out any inline styles,
404 * iframes, forms, strip extraneous <p> tags, etc.
405 *
406 * @param DOMElement
407 * @return void
408 */
409 function prepArticle($articleContent) {
410 $this->cleanStyles($articleContent);
411 $this->killBreaks($articleContent);
412 if ($this->revertForcedParagraphElements) {
413 $this->revertReadabilityStyledElements($articleContent);
414 }
415
416 /* Clean out junk from the article content */
417 $this->cleanConditionally($articleContent, 'form');
418 $this->clean($articleContent, 'object');
419 $this->clean($articleContent, 'h1');
420
421 /**
422 * If there is only one h2, they are probably using it
423 * as a header and not a subheader, so remove it since we already have a header.
424 ***/
425 if ($articleContent->getElementsByTagName('h2')->length == 1) {
426 $this->clean($articleContent, 'h2');
427 }
428 $this->clean($articleContent, 'iframe');
429
430 $this->cleanHeaders($articleContent);
431
432 /* Do these last as the previous stuff may have removed junk that will affect these */
433 $this->cleanConditionally($articleContent, 'table');
434 $this->cleanConditionally($articleContent, 'ul');
435 $this->cleanConditionally($articleContent, 'div');
436
437 /* Remove extra paragraphs */
438 $articleParagraphs = $articleContent->getElementsByTagName('p');
439 for ($i = $articleParagraphs->length-1; $i >= 0; $i--)
440 {
441 $imgCount = $articleParagraphs->item($i)->getElementsByTagName('img')->length;
442 $embedCount = $articleParagraphs->item($i)->getElementsByTagName('embed')->length;
443 $objectCount = $articleParagraphs->item($i)->getElementsByTagName('object')->length;
444
445 if ($imgCount === 0 && $embedCount === 0 && $objectCount === 0 && $this->getInnerText($articleParagraphs->item($i), false) == '')
446 {
447 $articleParagraphs->item($i)->parentNode->removeChild($articleParagraphs->item($i));
448 }
449 }
450
451 try {
452 $articleContent->innerHTML = preg_replace('/<br[^>]*>\s*<p/i', '<p', $articleContent->innerHTML);
453 //articleContent.innerHTML = articleContent.innerHTML.replace(/<br[^>]*>\s*<p/gi, '<p');
454 }
455 catch (Exception $e) {
456 $this->dbg("Cleaning innerHTML of breaks failed. This is an IE strict-block-elements bug. Ignoring.: " . $e);
457 }
458 }
459
460 /**
461 * Initialize a node with the readability object. Also checks the
462 * className/id for special names to add to its score.
463 *
464 * @param Element
465 * @return void
466 **/
467 protected function initializeNode($node) {
468 $readability = $this->dom->createAttribute('readability');
469 $readability->value = 0; // this is our contentScore
470 $node->setAttributeNode($readability);
471
472 switch (strtoupper($node->tagName)) { // unsure if strtoupper is needed, but using it just in case
473 case 'DIV':
474 $readability->value += 5;
475 break;
476
477 case 'PRE':
478 case 'TD':
479 case 'BLOCKQUOTE':
480 $readability->value += 3;
481 break;
482
483 case 'ADDRESS':
484 case 'OL':
485 case 'UL':
486 case 'DL':
487 case 'DD':
488 case 'DT':
489 case 'LI':
490 case 'FORM':
491 $readability->value -= 3;
492 break;
493
494 case 'H1':
495 case 'H2':
496 case 'H3':
497 case 'H4':
498 case 'H5':
499 case 'H6':
500 case 'TH':
501 $readability->value -= 5;
502 break;
503 }
504 $readability->value += $this->getClassWeight($node);
505 }
506
507 /***
508 * grabArticle - Using a variety of metrics (content score, classname, element types), find the content that is
509 * most likely to be the stuff a user wants to read. Then return it wrapped up in a div.
510 *
511 * @return DOMElement
512 **/
513 protected function grabArticle($page=null) {
514 $stripUnlikelyCandidates = $this->flagIsActive(self::FLAG_STRIP_UNLIKELYS);
515 if (!$page) $page = $this->dom;
516 $allElements = $page->getElementsByTagName('*');
517 /**
518 * First, node prepping. Trash nodes that look cruddy (like ones with the class name "comment", etc), and turn divs
519 * into P tags where they have been used inappropriately (as in, where they contain no other block level elements.)
520 *
521 * Note: Assignment from index for performance. See http://www.peachpit.com/articles/article.aspx?p=31567&seqNum=5
522 * TODO: Shouldn't this be a reverse traversal?
523 **/
524 $node = null;
525 $nodesToScore = array();
526 for ($nodeIndex = 0; ($node = $allElements->item($nodeIndex)); $nodeIndex++) {
527 //for ($nodeIndex=$targetList->length-1; $nodeIndex >= 0; $nodeIndex--) {
528 //$node = $targetList->item($nodeIndex);
529 $tagName = strtoupper($node->tagName);
530 /* Remove unlikely candidates */
531 if ($stripUnlikelyCandidates) {
532 $unlikelyMatchString = $node->getAttribute('class') . $node->getAttribute('id');
533 if (
534 preg_match($this->regexps['unlikelyCandidates'], $unlikelyMatchString) &&
535 !preg_match($this->regexps['okMaybeItsACandidate'], $unlikelyMatchString) &&
536 $tagName != 'BODY'
537 )
538 {
539 $this->dbg('Removing unlikely candidate - ' . $unlikelyMatchString);
540 //$nodesToRemove[] = $node;
541 $node->parentNode->removeChild($node);
542 $nodeIndex--;
543 continue;
544 }
545 }
546
547 if ($tagName == 'P' || $tagName == 'TD' || $tagName == 'PRE') {
548 $nodesToScore[] = $node;
549 }
550
551 /* Turn all divs that don't have children block level elements into p's */
552 if ($tagName == 'DIV') {
553 if (!preg_match($this->regexps['divToPElements'], $node->innerHTML)) {
554 //$this->dbg('Altering div to p');
555 $newNode = $this->dom->createElement('p');
556 try {
557 $newNode->innerHTML = $node->innerHTML;
558 //$nodesToReplace[] = array('new'=>$newNode, 'old'=>$node);
559 $node->parentNode->replaceChild($newNode, $node);
560 $nodeIndex--;
561 $nodesToScore[] = $node; // or $newNode?
562 }
563 catch(Exception $e) {
564 $this->dbg('Could not alter div to p, reverting back to div.: ' . $e);
565 }
566 }
567 else
568 {
569 /* EXPERIMENTAL */
570 // TODO: change these p elements back to text nodes after processing
571 for ($i = 0, $il = $node->childNodes->length; $i < $il; $i++) {
572 $childNode = $node->childNodes->item($i);
573 if ($childNode->nodeType == 3) { // XML_TEXT_NODE
574 //$this->dbg('replacing text node with a p tag with the same content.');
575 $p = $this->dom->createElement('p');
576 $p->innerHTML = $childNode->nodeValue;
577 $p->setAttribute('style', 'display: inline;');
578 $p->setAttribute('class', 'readability-styled');
579 $childNode->parentNode->replaceChild($p, $childNode);
580 }
581 }
582 }
583 }
584 }
585
586 /**
587 * Loop through all paragraphs, and assign a score to them based on how content-y they look.
588 * Then add their score to their parent node.
589 *
590 * A score is determined by things like number of commas, class names, etc. Maybe eventually link density.
591 **/
592 $candidates = array();
593 for ($pt=0; $pt < count($nodesToScore); $pt++) {
594 $parentNode = $nodesToScore[$pt]->parentNode;
595 // $grandParentNode = $parentNode ? $parentNode->parentNode : null;
596 $grandParentNode = !$parentNode ? null : (($parentNode->parentNode instanceof DOMElement) ? $parentNode->parentNode : null);
597 $innerText = $this->getInnerText($nodesToScore[$pt]);
598
599 if (!$parentNode || !isset($parentNode->tagName)) {
600 continue;
601 }
602
603 /* If this paragraph is less than 25 characters, don't even count it. */
604 if(strlen($innerText) < 25) {
605 continue;
606 }
607
608 /* Initialize readability data for the parent. */
609 if (!$parentNode->hasAttribute('readability'))
610 {
611 $this->initializeNode($parentNode);
612 $candidates[] = $parentNode;
613 }
614
615 /* Initialize readability data for the grandparent. */
616 if ($grandParentNode && !$grandParentNode->hasAttribute('readability') && isset($grandParentNode->tagName))
617 {
618 $this->initializeNode($grandParentNode);
619 $candidates[] = $grandParentNode;
620 }
621
622 $contentScore = 0;
623
624 /* Add a point for the paragraph itself as a base. */
625 $contentScore++;
626
627 /* Add points for any commas within this paragraph */
628 $contentScore += count(explode(',', $innerText));
629
630 /* For every 100 characters in this paragraph, add another point. Up to 3 points. */
631 $contentScore += min(floor(strlen($innerText) / 100), 3);
632
633 /* Add the score to the parent. The grandparent gets half. */
634 $parentNode->getAttributeNode('readability')->value += $contentScore;
635
636 if ($grandParentNode) {
637 $grandParentNode->getAttributeNode('readability')->value += $contentScore/2;
638 }
639 }
640
641 /**
642 * After we've calculated scores, loop through all of the possible candidate nodes we found
643 * and find the one with the highest score.
644 **/
645 $topCandidate = null;
646 for ($c=0, $cl=count($candidates); $c < $cl; $c++)
647 {
648 /**
649 * Scale the final candidates score based on link density. Good content should have a
650 * relatively small link density (5% or less) and be mostly unaffected by this operation.
651 **/
652 $readability = $candidates[$c]->getAttributeNode('readability');
653 $readability->value = $readability->value * (1-$this->getLinkDensity($candidates[$c]));
654
655 $this->dbg('Candidate: ' . $candidates[$c]->tagName . ' (' . $candidates[$c]->getAttribute('class') . ':' . $candidates[$c]->getAttribute('id') . ') with score ' . $readability->value);
656
657 if (!$topCandidate || $readability->value > (int)$topCandidate->getAttribute('readability')) {
658 $topCandidate = $candidates[$c];
659 }
660 }
661
662 /**
663 * If we still have no top candidate, just use the body as a last resort.
664 * We also have to copy the body node so it is something we can modify.
665 **/
666 if ($topCandidate === null || strtoupper($topCandidate->tagName) == 'BODY')
667 {
668 $topCandidate = $this->dom->createElement('div');
669 $topCandidate->innerHTML = ($page instanceof DOMDocument) ? $page->saveXML($page->documentElement) : $page->innerHTML;
670 $page->innerHTML = '';
671 $page->appendChild($topCandidate);
672 $this->initializeNode($topCandidate);
673 }
674
675 /**
676 * Now that we have the top candidate, look through its siblings for content that might also be related.
677 * Things like preambles, content split by ads that we removed, etc.
678 **/
679 $articleContent = $this->dom->createElement('div');
680 $articleContent->setAttribute('id', 'readability-content');
681 $siblingScoreThreshold = max(10, ((int)$topCandidate->getAttribute('readability')) * 0.2);
682 $siblingNodes = $topCandidate->parentNode->childNodes;
683
684 for ($s=0, $sl=$siblingNodes->length; $s < $sl; $s++)
685 {
686 $siblingNode = $siblingNodes->item($s);
687 $append = false;
688
689 $this->dbg('Looking at sibling node: ' . $siblingNode->nodeName . (($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability')) ? (' with score ' . $siblingNode->getAttribute('readability')) : ''));
690
691 //dbg('Sibling has score ' . ($siblingNode->readability ? siblingNode.readability.contentScore : 'Unknown'));
692
693 if ($siblingNode === $topCandidate)
694 // or if ($siblingNode->isSameNode($topCandidate))
695 {
696 $append = true;
697 }
698
699 $contentBonus = 0;
700 /* Give a bonus if sibling nodes and top candidates have the example same classname */
701 if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->getAttribute('class') == $topCandidate->getAttribute('class') && $topCandidate->getAttribute('class') != '') {
702 $contentBonus += ((int)$topCandidate->getAttribute('readability')) * 0.2;
703 }
704
705 if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability') && (((int)$siblingNode->getAttribute('readability')) + $contentBonus) >= $siblingScoreThreshold)
706 {
707 $append = true;
708 }
709
710 if (strtoupper($siblingNode->nodeName) == 'P') {
711 $linkDensity = $this->getLinkDensity($siblingNode);
712 $nodeContent = $this->getInnerText($siblingNode);
713 $nodeLength = strlen($nodeContent);
714
715 if ($nodeLength > 80 && $linkDensity < 0.25)
716 {
717 $append = true;
718 }
719 else if ($nodeLength < 80 && $linkDensity === 0 && preg_match('/\.( |$)/', $nodeContent))
720 {
721 $append = true;
722 }
723 }
724
725 if ($append)
726 {
727 $this->dbg('Appending node: ' . $siblingNode->nodeName);
728
729 $nodeToAppend = null;
730 $sibNodeName = strtoupper($siblingNode->nodeName);
731 if ($sibNodeName != 'DIV' && $sibNodeName != 'P') {
732 /* We have a node that isn't a common block level element, like a form or td tag. Turn it into a div so it doesn't get filtered out later by accident. */
733
734 $this->dbg('Altering siblingNode of ' . $sibNodeName . ' to div.');
735 $nodeToAppend = $this->dom->createElement('div');
736 try {
737 $nodeToAppend->setAttribute('id', $siblingNode->getAttribute('id'));
738 $nodeToAppend->innerHTML = $siblingNode->innerHTML;
739 }
740 catch(Exception $e)
741 {
742 $this->dbg('Could not alter siblingNode to div, reverting back to original.');
743 $nodeToAppend = $siblingNode;
744 $s--;
745 $sl--;
746 }
747 } else {
748 $nodeToAppend = $siblingNode;
749 $s--;
750 $sl--;
751 }
752
753 /* To ensure a node does not interfere with readability styles, remove its classnames */
754 $nodeToAppend->removeAttribute('class');
755
756 /* Append sibling and subtract from our list because it removes the node when you append to another node */
757 $articleContent->appendChild($nodeToAppend);
758 }
759 }
760
761 /**
762 * So we have all of the content that we need. Now we clean it up for presentation.
763 **/
764 $this->prepArticle($articleContent);
765
766 /**
767 * Now that we've gone through the full algorithm, check to see if we got any meaningful content.
768 * If we didn't, we may need to re-run grabArticle with different flags set. This gives us a higher
769 * likelihood of finding the content, and the sieve approach gives us a higher likelihood of
770 * finding the -right- content.
771 **/
772 if (strlen($this->getInnerText($articleContent, false)) < 250)
773 {
774 $this->body->innerHTML = $this->bodyCache;
775
776 if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS)) {
777 $this->removeFlag(self::FLAG_STRIP_UNLIKELYS);
778 return $this->grabArticle($this->body);
779 }
780 else if ($this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
781 $this->removeFlag(self::FLAG_WEIGHT_CLASSES);
782 return $this->grabArticle($this->body);
783 }
784 else if ($this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
785 $this->removeFlag(self::FLAG_CLEAN_CONDITIONALLY);
786 return $this->grabArticle($this->body);
787 }
788 else {
789 return false;
790 }
791 }
792 return $articleContent;
793 }
794
795 /**
796 * Remove script tags from document
797 *
798 * @param DOMElement
799 * @return void
800 */
801 public function removeScripts($doc) {
802 $scripts = $doc->getElementsByTagName('script');
803 for($i = $scripts->length-1; $i >= 0; $i--)
804 {
805 $scripts->item($i)->parentNode->removeChild($scripts->item($i));
806 }
807 }
808
809 /**
810 * Get the inner text of a node.
811 * This also strips out any excess whitespace to be found.
812 *
813 * @param DOMElement $
814 * @param boolean $normalizeSpaces (default: true)
815 * @return string
816 **/
817 public function getInnerText($e, $normalizeSpaces=true) {
818 $textContent = '';
819
820 if (!isset($e->textContent) || $e->textContent == '') {
821 return '';
822 }
823
824 $textContent = trim($e->textContent);
825
826 if ($normalizeSpaces) {
827 return preg_replace($this->regexps['normalize'], ' ', $textContent);
828 } else {
829 return $textContent;
830 }
831 }
832
833 /**
834 * Get the number of times a string $s appears in the node $e.
835 *
836 * @param DOMElement $e
837 * @param string - what to count. Default is ","
838 * @return number (integer)
839 **/
840 public function getCharCount($e, $s=',') {
841 return substr_count($this->getInnerText($e), $s);
842 }
843
844 /**
845 * Remove the style attribute on every $e and under.
846 *
847 * @param DOMElement $e
848 * @return void
849 */
850 public function cleanStyles($e) {
851 $elems = $e->getElementsByTagName('*');
852 foreach ($elems as $elem) {
853 $elem->removeAttribute('style');
854 }
855 }
856
857 /**
858 * Get the density of links as a percentage of the content
859 * This is the amount of text that is inside a link divided by the total text in the node.
860 *
861 * @param DOMElement $e
862 * @return number (float)
863 */
864 public function getLinkDensity($e) {
865 $links = $e->getElementsByTagName('a');
866 $textLength = strlen($this->getInnerText($e));
867 $linkLength = 0;
868 for ($i=0, $il=$links->length; $i < $il; $i++)
869 {
870 $linkLength += strlen($this->getInnerText($links->item($i)));
871 }
872 if ($textLength > 0) {
873 return $linkLength / $textLength;
874 } else {
875 return 0;
876 }
877 }
878
879 /**
880 * Get an elements class/id weight. Uses regular expressions to tell if this
881 * element looks good or bad.
882 *
883 * @param DOMElement $e
884 * @return number (Integer)
885 */
886 public function getClassWeight($e) {
887 if(!$this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
888 return 0;
889 }
890
891 $weight = 0;
892
893 /* Look for a special classname */
894 if ($e->hasAttribute('class') && $e->getAttribute('class') != '')
895 {
896 if (preg_match($this->regexps['negative'], $e->getAttribute('class'))) {
897 $weight -= 25;
898 }
899 if (preg_match($this->regexps['positive'], $e->getAttribute('class'))) {
900 $weight += 25;
901 }
902 }
903
904 /* Look for a special ID */
905 if ($e->hasAttribute('id') && $e->getAttribute('id') != '')
906 {
907 if (preg_match($this->regexps['negative'], $e->getAttribute('id'))) {
908 $weight -= 25;
909 }
910 if (preg_match($this->regexps['positive'], $e->getAttribute('id'))) {
911 $weight += 25;
912 }
913 }
914 return $weight;
915 }
916
917 /**
918 * Remove extraneous break tags from a node.
919 *
920 * @param DOMElement $node
921 * @return void
922 */
923 public function killBreaks($node) {
924 $html = $node->innerHTML;
925 $html = preg_replace($this->regexps['killBreaks'], '<br />', $html);
926 $node->innerHTML = $html;
927 }
928
929 /**
930 * Clean a node of all elements of type "tag".
931 * (Unless it's a youtube/vimeo video. People love movies.)
932 *
933 * @param DOMElement $e
934 * @param string $tag
935 * @return void
936 */
937 public function clean($e, $tag) {
938 $targetList = $e->getElementsByTagName($tag);
939 $isEmbed = ($tag == 'object' || $tag == 'embed');
940
941 for ($y=$targetList->length-1; $y >= 0; $y--) {
942 /* Allow youtube and vimeo videos through as people usually want to see those. */
943 if ($isEmbed) {
944 $attributeValues = '';
945 for ($i=0, $il=$targetList->item($y)->attributes->length; $i < $il; $i++) {
946 $attributeValues .= $targetList->item($y)->attributes->item($i)->value . '|'; // DOMAttr? (TODO: test)
947 }
948
949 /* First, check the elements attributes to see if any of them contain youtube or vimeo */
950 if (preg_match($this->regexps['video'], $attributeValues)) {
951 continue;
952 }
953
954 /* Then check the elements inside this element for the same. */
955 if (preg_match($this->regexps['video'], $targetList->item($y)->innerHTML)) {
956 continue;
957 }
958 }
959 $targetList->item($y)->parentNode->removeChild($targetList->item($y));
960 }
961 }
962
963 /**
964 * Clean an element of all tags of type "tag" if they look fishy.
965 * "Fishy" is an algorithm based on content length, classnames,
966 * link density, number of images & embeds, etc.
967 *
968 * @param DOMElement $e
969 * @param string $tag
970 * @return void
971 */
972 public function cleanConditionally($e, $tag) {
973 if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
974 return;
975 }
976
977 $tagsList = $e->getElementsByTagName($tag);
978 $curTagsLength = $tagsList->length;
979
980 /**
981 * Gather counts for other typical elements embedded within.
982 * Traverse backwards so we can remove nodes at the same time without effecting the traversal.
983 *
984 * TODO: Consider taking into account original contentScore here.
985 */
986 for ($i=$curTagsLength-1; $i >= 0; $i--) {
987 $weight = $this->getClassWeight($tagsList->item($i));
988 $contentScore = ($tagsList->item($i)->hasAttribute('readability')) ? (int)$tagsList->item($i)->getAttribute('readability') : 0;
989
990 $this->dbg('Cleaning Conditionally ' . $tagsList->item($i)->tagName . ' (' . $tagsList->item($i)->getAttribute('class') . ':' . $tagsList->item($i)->getAttribute('id') . ')' . (($tagsList->item($i)->hasAttribute('readability')) ? (' with score ' . $tagsList->item($i)->getAttribute('readability')) : ''));
991
992 if ($weight + $contentScore < 0) {
993 $tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
994 }
995 else if ( $this->getCharCount($tagsList->item($i), ',') < 10) {
996 /**
997 * If there are not very many commas, and the number of
998 * non-paragraph elements is more than paragraphs or other ominous signs, remove the element.
999 **/
1000 $p = $tagsList->item($i)->getElementsByTagName('p')->length;
1001 $img = $tagsList->item($i)->getElementsByTagName('img')->length;
1002 $li = $tagsList->item($i)->getElementsByTagName('li')->length-100;
1003 $input = $tagsList->item($i)->getElementsByTagName('input')->length;
1004
1005 $embedCount = 0;
1006 $embeds = $tagsList->item($i)->getElementsByTagName('embed');
1007 for ($ei=0, $il=$embeds->length; $ei < $il; $ei++) {
1008 if (preg_match($this->regexps['video'], $embeds->item($ei)->getAttribute('src'))) {
1009 $embedCount++;
1010 }
1011 }
1012
1013 $linkDensity = $this->getLinkDensity($tagsList->item($i));
1014 $contentLength = strlen($this->getInnerText($tagsList->item($i)));
1015 $toRemove = false;
1016
1017 if ( $img > $p ) {
1018 $toRemove = true;
1019 } else if ($li > $p && $tag != 'ul' && $tag != 'ol') {
1020 $toRemove = true;
1021 } else if ( $input > floor($p/3) ) {
1022 $toRemove = true;
1023 } else if ($contentLength < 25 && ($img === 0 || $img > 2) ) {
1024 $toRemove = true;
1025 } else if($weight < 25 && $linkDensity > 0.2) {
1026 $toRemove = true;
1027 } else if($weight >= 25 && $linkDensity > 0.5) {
1028 $toRemove = true;
1029 } else if(($embedCount == 1 && $contentLength < 75) || $embedCount > 1) {
1030 $toRemove = true;
1031 }
1032
1033 if ($toRemove) {
1034 $tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
1035 }
1036 }
1037 }
1038 }
1039
1040 /**
1041 * Clean out spurious headers from an Element. Checks things like classnames and link density.
1042 *
1043 * @param DOMElement $e
1044 * @return void
1045 */
1046 public function cleanHeaders($e) {
1047 for ($headerIndex = 1; $headerIndex < 3; $headerIndex++) {
1048 $headers = $e->getElementsByTagName('h' . $headerIndex);
1049 for ($i=$headers->length-1; $i >=0; $i--) {
1050 if ($this->getClassWeight($headers->item($i)) < 0 || $this->getLinkDensity($headers->item($i)) > 0.33) {
1051 $headers->item($i)->parentNode->removeChild($headers->item($i));
1052 }
1053 }
1054 }
1055 }
1056
1057 public function flagIsActive($flag) {
1058 return ($this->flags & $flag) > 0;
1059 }
1060
1061 public function addFlag($flag) {
1062 $this->flags = $this->flags | $flag;
1063 }
1064
1065 public function removeFlag($flag) {
1066 $this->flags = $this->flags & ~$flag;
1067 }
1068}
1069?> \ No newline at end of file
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-03 07:25:10 +0000