source: gsdl/trunk/packages/d2m/dc_parse.c@ 15554

Last change on this file since 15554 was 10365, checked in by kjdon, 19 years ago

changed my mind, now adding these all individually instead of in a tar file
  • Property svn:keywords set to Author Date Id Revision
File size: 6.2 KB
Line 
1/* ------------------------------------------------------------------- */
2/* dc_parse : parse the meta tag, build a metatag struct */
3/* Author : Ole Husby */
4/* Updated : 1998-09-30 */
5/* */
6/* Syntax : #include "d2m.h" */
7/* dc_parse(char * string) */
8/* */
9/* Returns pointer to struct metatag */
10/* */
11/* Memory for the struct is allocated in the program! */
12/* */
13/* String includes a meta tag, after the initial */
14/* "<meta " and the ending ">" are stripped away. */
15/* */
16/* The program accepts the tag if name starts with "dc." */
17/* (case insensitive), else NULL is returned. */
18/* */
19/* The parsing looks for the HTML attributes "name", */
20/* "scheme", and "content" only. Other attributes are */
21/* ignored. */
22/* */
23/* The DC "type" will be recognized if suffixed */
24/* "name" by "dot notation", or contained within */
25/* parentheses within "content" (old notation). */
26/* (type = subelement) */
27/* */
28/* The DC "scheme" will be recognized if */
29/* present in "scheme" attribute (HTML 4.0), or if */
30/* contained within aprentheses in "content" (HTML 2.0) */
31/* */
32/* Examples: */
33/* */
34/* meta name="DC.Creator.Personalname" */
35/* content="Smith, John" */
36/* */
37/* meta name="dc.Creator" */
38/* content="(type=Personalname) Smith, John"> */
39/* */
40/* meta name="DC.Subject" scheme="LCSH" */
41/* content="Cataloguing" */
42/* */
43/* meta name="DC.Subject" */
44/* content="(scheme=LCSH) Cataloguing" */
45/* */
46/* The values of the DC attributes are NOT validated! */
47/* */
48/* ------------------------------------------------------------------- */
49
50#include <stdlib.h>
51#include <stdio.h>
52#include <string.h>
53#include <malloc.h>
54#include "d2m.h"
55
56
57
58
59struct metatag *dc_parse(char *buf)
60{
61 int more;
62 char *p, *q;
63 struct metatag *mtag;
64 int strict = TRUE; /* Indicates that DC. prefix is mandatory */
65 char attrname[16];
66
67/* Allocates and initializes metatag struct */
68
69 mtag = malloc(sizeof(*mtag));
70
71 mtag->name = malloc(256);
72 mtag->type = malloc(256);
73 mtag->scheme = malloc(256);
74 mtag->value = malloc(10000);
75
76 *mtag->name = 0;
77 *mtag->type = 0;
78 *mtag->scheme = 0;
79 *mtag->value = 0;
80
81
82
83
84/* Parse attribute name/value pairs: */
85
86 if (!buf || !*buf) return NULL;
87
88 p = strtok(buf, "\"");
89
90
91
92 while (p)
93 {
94/* Look for attribute name */
95
96 *attrname = 0;
97
98 while (*p && ( p[strlen(p) - 1] == ' ' || p[strlen(p) - 1] == '='))
99 p[strlen(p) - 1] = '\0';
100
101 if (*p && strlen(p) < 10)
102 strcpy(attrname, p);
103
104/* Look for attribute value */
105
106 p = strtok(NULL, "\"");
107
108 while(*p && p[0] == ' ')
109 p++;
110
111 if (!p || !*p)
112 break;
113
114 if (strncasecmp(attrname, "name", 4) == 0)
115 strcpy(mtag->name, p);
116 else if (strncasecmp(attrname, "scheme", 6) == 0)
117 strcpy(mtag->scheme, p);
118 else if (strncasecmp(attrname, "content", 7) == 0)
119 strcpy(mtag->value, p);
120 p = strtok(NULL, "\"");
121
122 if (p)
123 {
124 while (*p && p[0] == ' ')
125 p++;
126 if (!*p) p = NULL;
127 }
128 }
129
130
131
132/* Check the name attribute value */
133
134 p = mtag->name;
135
136 if (!p || !*p)
137 return NULL;
138
139/* Check for leading "dc." and suffixed type */
140
141 if (strncasecmp(p, "dc.", 3) == 0)
142 {
143 p += 3;
144 q = strstr(p, ".");
145 if (q)
146 {
147 strcpy(mtag->type, q + 1);
148 *q = 0;
149 }
150 mtag->name = p;
151 }
152 else
153 {
154 if (strict)
155 return NULL;
156 }
157
158
159/* Check the content attribute value */
160/* Extracting type and scheme */
161
162
163 p = mtag->value;
164
165 more = 1;
166
167 while (more)
168 {
169 if (strncasecmp(p, "(type=", 6) == 0)
170 {
171 p += 6;
172 q = strtok(p, ")");
173 if (!q)
174 return NULL;
175 else
176 {
177 if (!*mtag->type)
178 strcpy(mtag->type, q);
179 q = strtok(NULL, "\0");
180 p = q;
181 while (p && *p && p[0] == ' ') p++;
182
183 }
184 }
185
186 else if (strncasecmp(p, "(scheme=", 8) == 0)
187 {
188 p += 8;
189 q = strtok(p, ")");
190 if (!q)
191 return NULL;
192 else
193 {
194 if (!*mtag->scheme)
195 strcpy(mtag->scheme, q);
196 q = strtok(NULL, "\0");
197 p = q;
198 while (p && *p && p[0] == ' ') p++;
199 }
200 }
201
202 else
203 {
204 mtag->value = p;
205 more = 0;
206 }
207
208 }
209
210/* Finished */
211
212 if (!*mtag->value)
213 mtag = (struct metatag *) NULL;
214
215 return mtag;
216}
Note: See TracBrowser for help on using the repository browser.