Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
att
GitHub Repository: att/ast
 Path: blob/master/src/cmd/html/html2db.c
1808 views
1
/***********************************************************************

2
*                                                                      *

3
*               This software is part of the ast package               *

4
*          Copyright (c) 1996-2011 AT&T Intellectual Property          *

5
*                      and is licensed under the                       *

6
*                 Eclipse Public License, Version 1.0                  *

7
*                    by AT&T Intellectual Property                     *

8
*                                                                      *

9
*                A copy of the License is available at                 *

10
*          http://www.eclipse.org/org/documents/epl-v10.html           *

11
*         (with md5 checksum b35adb5213ca9657e911e9befb180842)         *

12
*                                                                      *

13
*              Information and Software Systems Research               *

14
*                            AT&T Research                             *

15
*                           Florham Park NJ                            *

16
*                                                                      *

17
*                 Glenn Fowler <[email protected]>                  *

18
*                                                                      *

19
***********************************************************************/

20
#pragma prototyped

21
/*

22
 * Glenn Fowler

23
 * AT&T Research

24
 *

25
 * html2db - extract flat file database from html tables

26
 */

27


28
static const char usage[] =

29
"[-?\n@(#)$Id: html2db (AT&T Research) 1998-11-10 $\n]"

30
USAGE_LICENSE

31
"[+NAME?html2db - extract flat file database from html tables]"

32
"[+DESCRIPTION?\bhtml2db\b extracts a flat file database from tables in the"

33
"	input \bhtml\b \afile\as. If \afile\a is not specified then the"

34
"	standard input is read. The \bhtml\b parse is rudimentary; don't use"

35
"	\bhtml2db\b to detect valid \bhtml\b files.]"

36


37
"\n"

38
"\n[ file ... ]\n"

39
"\n"

40


41
"[+SEE ALSO?\bhtml2rtf\b(1)]"

42
;

43
#include <ast.h>

44
#include <ctype.h>

45
#include <error.h>

46


47
static void

48
flatten(const char* path, Sfio_t* ip, Sfio_t* op)

49
{

50
	register int	c;

51
	register int	q;

52
	register int	p;

53
	register int	b;

54
	register char*	s;

55


56
	char		tag[256];

57


58
	b = p = 0;

59
	for (;;)

60
	{

61
		switch (c = sfgetc(ip))

62
		{

63
		case EOF:

64
			break;

65
		case '<':

66
			q = 0;

67
			s = tag;

68
			for (;;)

69
			{

70
				switch (c = sfgetc(ip))

71
				{

72
				case EOF:

73
					return;

74
				case '>':

75
					sfungetc(ip, c);

76
					break;

77
				default:

78
					if (isspace(c))

79
						break;

80
					if (s < &tag[sizeof(tag)-1])

81
						*s++ = islower(c) ? toupper(c) : c;

82
					continue;

83
				}

84
				break;

85
			}

86
			*s = 0;

87
			q = 0;

88
			for (;;)

89
			{

90
				switch (c = sfgetc(ip))

91
				{

92
				case EOF:

93
					return;

94
				case '\'':

95
				case '"':

96
					if (q == c)

97
						q = 0;

98
					else if (q == 0)

99
						q = c;

100
					continue;

101
				case '>':

102
					if (q == 0)

103
						break;

104
					continue;

105
				default:

106
					continue;

107
				}

108
				break;

109
			}

110
			s = tag;

111
			if (s[0] == 'T' && s[1] == 'D' && s[2] == 0)

112
				p = 1;

113
			else if (s[0] == '/' && s[1] == 'T')

114
			{

115
				if (s[2] == 'D' && s[3] == 0)

116
				{

117
					b = p = 0;

118
					sfputc(op, ';');

119
				}

120
				else if (s[2] == 'R' && s[3] == 0)

121
					sfputc(op, '\n');

122
			}

123
			continue;

124
		default:

125
			if (p)

126
			{

127
				if (isspace(c))

128
				{

129
					if (b)

130
						continue;

131
					b = 1;

132
					c = ' ';

133
				}

134
				else

135
					b = 0;

136
				sfputc(op, c);

137
			}

138
			continue;

139
		}

140
		break;

141
	}

142
}

143


144
int

145
main(int argc, char** argv)

146
{

147
	register char*		s;

148
	register Sfio_t*	ip;

149


150
	NoP(argc);

151
	error_info.id = "html2db";

152
	for (;;)

153
	{

154
		switch (optget(argv, usage))

155
		{

156
		case '?':

157
			error(ERROR_USAGE|4, "%s", opt_info.arg);

158
			continue;

159
		case ':':

160
			error(2, "%s", opt_info.arg);

161
			continue;

162
		}

163
		break;

164
	}

165
	argv += opt_info.index;

166
	if (error_info.errors)

167
		error(ERROR_USAGE|4, "%s", optusage(NiL));

168
	do

169
	{

170
		if (!(s = *argv) || streq(s, "-") || streq(s, "/dev/stdin") || streq(s, "/dev/fd/0"))

171
		{

172
			s = "/dev/stdin";

173
			ip = sfstdin;

174
		}

175
		else if (!(ip = sfopen(NiL, s, "r")))

176
		{

177
			error(ERROR_SYSTEM|2, "%s: cannot read", s);

178
			continue;

179
		}

180
		flatten(s, ip, sfstdout);

181
		if (ip != sfstdin)

182
			sfclose(ip);

183
	} while (*argv++);

184
	return error_info.errors != 0;

185
}

186


187