1 | /* ------------------------------------------------------------------- */
|
---|
2 | /* D2Mconv : Converts metadata present in a file to MARC */
|
---|
3 | /* Writes MARC record in "line format" */
|
---|
4 | /* */
|
---|
5 | /* Syntax : Arguments: */
|
---|
6 | /* */
|
---|
7 | /* char *dfile : input file */
|
---|
8 | /* DC metadata must be present in HTML 2.0 */
|
---|
9 | /* or HTML 4.0 syntax */
|
---|
10 | /* char *buffer: output record in line format */
|
---|
11 | /* Must be allocated in calling program! */
|
---|
12 | /* char *tfile : tracefile */
|
---|
13 | /* No trace if NULL */
|
---|
14 | /* char *url : URL for use in case no URL in DC.Identifier */
|
---|
15 | /* */
|
---|
16 | /* char *format: MARC format (see d2m.h) */
|
---|
17 | /* */
|
---|
18 | /* Author : Ole Husby, BIBSYS */
|
---|
19 | /* Updated : 1998-09-30 */
|
---|
20 | /* ------------------------------------------------------------------- */
|
---|
21 |
|
---|
22 | #include <stdlib.h>
|
---|
23 | #include <string.h>
|
---|
24 | #include <stdio.h>
|
---|
25 | #include <malloc.h>
|
---|
26 | #include <ctype.h>
|
---|
27 | #include <unistd.h>
|
---|
28 | #include <time.h>
|
---|
29 | #include "d2m.h"
|
---|
30 |
|
---|
31 |
|
---|
32 |
|
---|
33 |
|
---|
34 | /* ------------------------------------------------------------------- */
|
---|
35 | /* Checks a new line */
|
---|
36 | /* ------------------------------------------------------------------- */
|
---|
37 | /*
|
---|
38 | This is where starting and ending of a meta tag is parsed.
|
---|
39 | The whole metatag (resulting from one ore more lines) is put
|
---|
40 | into the buffer.
|
---|
41 |
|
---|
42 | Note: This function is called recursively!
|
---|
43 | */
|
---|
44 |
|
---|
45 | int mline(char *l, char *buf, int *meta,
|
---|
46 | struct marcrec *mrec, FILE *tf, int format)
|
---|
47 | {
|
---|
48 | char *p, *q, rc;
|
---|
49 | int diff;
|
---|
50 |
|
---|
51 | struct metatag *mtag;
|
---|
52 |
|
---|
53 | if (!l || !*l) return 1;
|
---|
54 |
|
---|
55 | q = l;
|
---|
56 |
|
---|
57 |
|
---|
58 | /* Looks for the end of an already started metatag buffer */
|
---|
59 | /* The end is normally '">' */
|
---|
60 |
|
---|
61 | if (*meta)
|
---|
62 | {
|
---|
63 | p = strstr(l, "\">");
|
---|
64 |
|
---|
65 | if (p)
|
---|
66 | {
|
---|
67 | if ( (char *) p != (char *) l)
|
---|
68 | {
|
---|
69 | diff = (int)(p - l);
|
---|
70 | if (*buf)
|
---|
71 | strcat(buf, " ");
|
---|
72 | strncat(buf, l, diff);
|
---|
73 | }
|
---|
74 |
|
---|
75 | /* Parsing the buffer for metadata */
|
---|
76 |
|
---|
77 | mtag = dc_parse(buf);
|
---|
78 |
|
---|
79 | if (mtag)
|
---|
80 | {
|
---|
81 | if (tf)
|
---|
82 | {
|
---|
83 | fprintf(tf, "Name : %s\n", mtag->name );
|
---|
84 | fprintf(tf, "Type : %s\n", mtag->type );
|
---|
85 | fprintf(tf, "Scheme: %s\n", mtag->scheme );
|
---|
86 | fprintf(tf, "Value : %s\n", mtag->value );
|
---|
87 | fprintf(tf, "\n");
|
---|
88 | }
|
---|
89 |
|
---|
90 | /* Convert to MARC */
|
---|
91 |
|
---|
92 | MARCmake(mtag, mrec, format);
|
---|
93 | }
|
---|
94 | free(mtag);
|
---|
95 |
|
---|
96 | *buf = 0;
|
---|
97 | *meta = FALSE;
|
---|
98 | q = p + 2;
|
---|
99 | }
|
---|
100 | else
|
---|
101 | {
|
---|
102 | strcat(buf, l);
|
---|
103 | q = NULL;
|
---|
104 | }
|
---|
105 | }
|
---|
106 |
|
---|
107 |
|
---|
108 | if (!q) return 1;
|
---|
109 |
|
---|
110 | p = (char *) cstr(q, "<meta ");
|
---|
111 |
|
---|
112 | if (p)
|
---|
113 | {
|
---|
114 | *meta = TRUE;
|
---|
115 | mline(p + 6, buf, meta, mrec, tf, format);
|
---|
116 | }
|
---|
117 |
|
---|
118 | return 1;
|
---|
119 | }
|
---|
120 |
|
---|
121 |
|
---|
122 |
|
---|
123 | int D2Mconv(char *dfile, char *buffer, char *tfile, char *url, int format)
|
---|
124 | {
|
---|
125 | struct marcrec *mrec;
|
---|
126 | struct metatag *mtag;
|
---|
127 | char line[50000], today[32];
|
---|
128 | char *l, *p;
|
---|
129 | time_t d;
|
---|
130 | struct tm *time_struct;
|
---|
131 |
|
---|
132 | FILE *df, *tf;
|
---|
133 | int *meta;
|
---|
134 |
|
---|
135 | putenv("TZ=NFT-1DFT");
|
---|
136 | d = time(NULL);
|
---|
137 | time_struct = localtime(&d);
|
---|
138 | strftime(today, 16, "%y%m%d", time_struct);
|
---|
139 |
|
---|
140 | mrec = malloc(sizeof(*mrec));
|
---|
141 |
|
---|
142 | mrec->marcline = malloc(100000);
|
---|
143 | mrec->partitle = malloc(10000);
|
---|
144 | mrec->subtitle = malloc(10000);
|
---|
145 | mrec->year = malloc(1000);
|
---|
146 | mrec->url = malloc(1000);
|
---|
147 | mrec->fmat = malloc(1000);
|
---|
148 | mrec->s008 = malloc(41);
|
---|
149 |
|
---|
150 | mrec->ncreators = 0;
|
---|
151 | mrec->ntitles = 0;
|
---|
152 | *mrec->marcline = 0;
|
---|
153 | *mrec->subtitle = 0;
|
---|
154 | *mrec->partitle = 0;
|
---|
155 | *mrec->year = 0;
|
---|
156 | *mrec->url = 0;
|
---|
157 | *mrec->fmat = 0;
|
---|
158 | strcpy(mrec->s008, " ");
|
---|
159 |
|
---|
160 | meta = malloc(sizeof(int));
|
---|
161 |
|
---|
162 | tf = (FILE *) NULL;
|
---|
163 |
|
---|
164 |
|
---|
165 | /* Open inputfile */
|
---|
166 |
|
---|
167 | df = fopen(dfile, "r");
|
---|
168 | if (!df)
|
---|
169 | return 0; /* Error: Unable to read data */
|
---|
170 |
|
---|
171 |
|
---|
172 |
|
---|
173 | /* Open tracefile */
|
---|
174 |
|
---|
175 | if (*tfile)
|
---|
176 | {
|
---|
177 | tf = fopen(tfile, "w");
|
---|
178 | if (!tf)
|
---|
179 | {
|
---|
180 | fclose(df);
|
---|
181 | return 0; /* Error: Unable to create trace */
|
---|
182 | }
|
---|
183 | }
|
---|
184 |
|
---|
185 | *meta = FALSE;
|
---|
186 |
|
---|
187 |
|
---|
188 | /* Reads the file */
|
---|
189 |
|
---|
190 |
|
---|
191 | while (l = fgets(line, 9999, df))
|
---|
192 | {
|
---|
193 |
|
---|
194 | /* Remove trailing LF and / or CR */
|
---|
195 |
|
---|
196 | while ( (l[strlen(l) - 1] == '\n') || (l[strlen(l) - 1] == '\r') )
|
---|
197 | l[strlen(l) -1] = '\0';
|
---|
198 |
|
---|
199 |
|
---|
200 | /* printf("... main %s\n",l); */
|
---|
201 |
|
---|
202 | /* Stops reading file if "</head" or "<body" is encountered */
|
---|
203 |
|
---|
204 | p = cstr(l, (char *) "</head");
|
---|
205 | if (p) break;
|
---|
206 |
|
---|
207 | p = cstr(l, (char *) "<body");
|
---|
208 | if (p) break;
|
---|
209 |
|
---|
210 | mline(l, (char *) buffer, (int *) meta, mrec, tf, format);
|
---|
211 | }
|
---|
212 |
|
---|
213 |
|
---|
214 |
|
---|
215 | /* Check if remaining data in buffer */
|
---|
216 |
|
---|
217 | mtag = dc_parse((char *) buffer);
|
---|
218 |
|
---|
219 | if (mtag)
|
---|
220 | {
|
---|
221 | if (tf)
|
---|
222 | {
|
---|
223 | fprintf(tf, "Name : %s\n", mtag->name );
|
---|
224 | fprintf(tf, "Type : %s\n", mtag->type );
|
---|
225 | fprintf(tf, "Scheme: %s\n", mtag->scheme );
|
---|
226 | fprintf(tf, "Value : %s\n", mtag->value );
|
---|
227 | fprintf(tf, "\n");
|
---|
228 | }
|
---|
229 | MARCmake(mtag, mrec, format);
|
---|
230 | }
|
---|
231 | free(mtag);
|
---|
232 |
|
---|
233 |
|
---|
234 | /* Close files */
|
---|
235 |
|
---|
236 | fclose(df);
|
---|
237 |
|
---|
238 | if (tf)
|
---|
239 | fclose(tf);
|
---|
240 |
|
---|
241 | /* Put URL into mrec (if not present) */
|
---|
242 |
|
---|
243 | if (!*mrec->url)
|
---|
244 | strcpy(mrec->url, url);
|
---|
245 |
|
---|
246 | /* Put date into 008 */
|
---|
247 |
|
---|
248 | put008(mrec->s008, today, F008_DATE_ENTERED);
|
---|
249 | put008(mrec->s008, "s", F008_TYPE_OF_DATE);
|
---|
250 |
|
---|
251 | switch (format)
|
---|
252 | {
|
---|
253 | case ISMARC :
|
---|
254 | {
|
---|
255 | put008(mrec->s008, "k", F008_FORM_OF_PUBLICATION);
|
---|
256 | break;
|
---|
257 | }
|
---|
258 | }
|
---|
259 |
|
---|
260 | /* Tidy up the MARC buffer */
|
---|
261 |
|
---|
262 | MARCtidy(mrec, buffer, format);
|
---|
263 |
|
---|
264 | return 1;
|
---|
265 | }
|
---|