Annotation of docbook2mdoc/statistics.c, Revision 1.4
1.4 ! schwarze 1: /* $Id: statistics.c,v 1.3 2019/04/03 08:39:53 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2019 Ingo Schwarze <schwarze@openbsd.org>
4: *
5: * Permission to use, copy, modify, and distribute this software for any
6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
8: *
9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16: */
17: #include <assert.h>
18: #include <ctype.h>
19: #include <err.h>
20: #include <fcntl.h>
1.2 schwarze 21: #include <getopt.h>
1.1 schwarze 22: #include <stdio.h>
23: #include <stdlib.h>
24: #include <string.h>
25: #include <unistd.h>
26:
27: /*
28: * Count parent-child element relations in a corpus of DocBook documents.
29: *
30: * Read absolute or relative input file names from standard input,
31: * one per line.
32: * For each parent-child relation, print the total number of occurrences,
33: * the parent name, and the child name, separated by tab characters
34: * and followed by a newline character.
35: *
36: * Typical usage:
37: * statistics < filenames.txt | sort -n
38: * statistics < filenames.txt | grep '\<listitem\>' | sort -n
1.4 ! schwarze 39: *
! 40: * Relations already fully implemented are excluded by default.
! 41: * The option -a shows all relations.
! 42: *
! 43: * If two arguments (parent and child) are given, a histogram
! 44: * of the number of children of the kind in each parent is given
! 45: * in addition to the normal output.
! 46: *
! 47: * Example usage:
! 48: * statistics tgroup colspec < filenames.txt | grep colspec
1.1 schwarze 49: */
50:
51: struct entry {
52: char *parent;
53: char *child;
54: int count;
55: };
56:
57: static struct entry *table;
58: static size_t tablesz;
59: static size_t tablei;
60:
61: static char **stack;
62: static size_t stacksz;
63: static size_t stacki;
64:
1.4 ! schwarze 65: static const int nchildsz = 8;
! 66: struct nchild {
! 67: char *parent;
! 68: char *child;
! 69: int freq[nchildsz];
! 70: int count;
! 71: };
! 72:
! 73: static struct nchild nchild;
! 74: static char *fname;
! 75:
1.1 schwarze 76:
77: /*
78: * Count one instance of a parent-child relation.
1.2 schwarze 79: * Before the special call table_add(NULL, NULL),
80: * mark relations to not be counted;
81: * in that phase, child can be NULL as a wildcard.
1.1 schwarze 82: */
83: static void
84: table_add(const char *parent, const char *child)
85: {
1.2 schwarze 86: static int init_done;
87: size_t i;
88:
89: if (parent == NULL && child == NULL) {
90: init_done = 1;
91: return;
92: }
1.1 schwarze 93:
1.4 ! schwarze 94: /* Optional parent-child histogram. */
! 95:
! 96: if (init_done && parent != NULL && child != NULL &&
! 97: nchild.parent != NULL && nchild.child != NULL &&
! 98: strcmp(parent, nchild.parent) == 0 &&
! 99: strcmp(child, nchild.child) == 0) {
! 100: if (nchild.count < nchildsz) {
! 101: nchild.freq[nchild.count]++;
! 102: if (nchild.count > 0)
! 103: nchild.freq[nchild.count - 1]--;
! 104: } else if (nchild.count == nchildsz)
! 105: puts(fname);
! 106: nchild.count++;
! 107: }
! 108:
1.1 schwarze 109: /* If the table entry already exists, increment its count. */
110:
111: for (i = 0; i < tablei; i++) {
112: if (strcmp(parent, table[i].parent) == 0 &&
1.2 schwarze 113: (child == NULL || table[i].child == NULL ||
114: strcmp(child, table[i].child) == 0)) {
115: assert(init_done);
116: if (table[i].count != -1)
117: table[i].count++;
1.1 schwarze 118: return;
119: }
120: }
121:
122: /* If the table is full, make room. */
123:
124: if (tablei == tablesz) {
125: tablesz += 64;
126: table = reallocarray(table, tablesz, sizeof(*table));
127: if (table == NULL)
128: err(1, NULL);
129: }
130:
131: /* Add a new entry to the table. */
132:
133: if ((table[tablei].parent = strdup(parent)) == NULL)
134: err(1, NULL);
1.2 schwarze 135: if (child == NULL)
136: table[tablei].child = NULL;
137: else if ((table[tablei].child = strdup(child)) == NULL)
1.1 schwarze 138: err(1, NULL);
1.2 schwarze 139: table[tablei++].count = init_done ? 1 : -1;
1.1 schwarze 140: }
141:
142: /*
143: * Enter an element.
144: */
145: static void
146: stack_push(const char *name)
147: {
1.4 ! schwarze 148: if (nchild.parent != NULL && strcmp(name, nchild.parent) == 0)
! 149: nchild.count = 0;
! 150:
1.1 schwarze 151: if (stacki == stacksz) {
152: stacksz += 8;
153: stack = reallocarray(stack, stacksz, sizeof(*stack));
154: if (stack == NULL)
155: err(1, NULL);
156: }
157: if ((stack[stacki++] = strdup(name)) == NULL)
158: err(1, NULL);
159: }
160:
161: /*
162: * Exit an element.
163: */
164: static void
165: stack_pop(const char *name)
166: {
167: if (stacki > 0 && (name == NULL ||
168: strcmp(name, stack[stacki - 1]) == 0))
169: free(stack[--stacki]);
170: }
171:
172: /*
173: * Simplified version from parse.c.
174: */
175: static int
176: advance(char *b, size_t rlen, size_t *pend, const char *charset)
177: {
178: int space;
179:
180: if (*charset == ' ') {
181: space = 1;
182: charset++;
183: } else
184: space = 0;
185:
186: while (*pend < rlen) {
187: if (space && isspace((unsigned char)b[*pend]))
188: break;
189: if (strchr(charset, b[*pend]) != NULL)
190: break;
191: ++*pend;
192: }
193: if (*pend == rlen) {
194: b[rlen] = '\0';
195: return 1;
196: } else
197: return 0;
198: }
199:
200: /*
201: * Simplified version from parse.c.
202: */
203: static void
204: parse_file(int fd, char *fname)
205: {
206: char b[4096];
1.3 schwarze 207: char *cp;
1.1 schwarze 208: ssize_t rsz; /* Return value from read(2). */
209: size_t rlen; /* Number of bytes in b[]. */
210: size_t poff; /* Parse offset in b[]. */
211: size_t pend; /* Offset of the end of the current word. */
212: int in_tag, in_arg, in_quotes, elem_end;
213:
214: rlen = 0;
215: in_tag = in_arg = in_quotes = 0;
216: while ((rsz = read(fd, b + rlen, sizeof(b) - rlen - 1)) >= 0) {
217: if ((rlen += rsz) == 0)
218: break;
219: pend = 0;
220: for (;;) {
221: if ((poff = pend) == rlen)
222: break;
223: if (isspace((unsigned char)b[pend])) {
224: pend++;
225: continue;
226: }
227: if (in_arg) {
228: if (in_quotes == 0 && b[pend] == '"') {
229: in_quotes = 1;
230: pend++;
231: continue;
232: }
233: if (advance(b, rlen, &pend,
234: in_quotes ? "\"" : " >") && rsz > 0)
235: break;
236: in_arg = in_quotes = elem_end = 0;
237: if (b[pend] == '>') {
238: in_tag = 0;
239: if (pend > 0 && b[pend - 1] == '/') {
240: b[pend - 1] = '\0';
241: elem_end = 1;
242: }
243: }
244: b[pend] = '\0';
245: if (pend < rlen)
246: pend++;
247: if (elem_end)
248: stack_pop(NULL);
249: } else if (in_tag) {
250: if (advance(b, rlen, &pend, " =>") && rsz > 0)
251: break;
252: elem_end = 0;
253: switch (b[pend]) {
254: case '>':
255: in_tag = 0;
256: if (pend > 0 && b[pend - 1] == '/') {
257: b[pend - 1] = '\0';
258: elem_end = 1;
259: }
260: break;
261: case '=':
262: in_arg = 1;
263: break;
264: default:
265: break;
266: }
267: b[pend] = '\0';
268: if (pend < rlen)
269: pend++;
270: if (elem_end)
271: stack_pop(NULL);
272: } else if (b[poff] == '<') {
273: if (advance(b, rlen, &pend, " >") && rsz > 0)
274: break;
1.3 schwarze 275: if (pend > poff + 3 &&
276: strncmp(b + poff, "<!--", 4) == 0) {
277: /* Skip a comment. */
278: cp = strstr(b + pend - 2, "-->");
279: if (cp == NULL) {
280: pend = rlen;
281: if (rsz > 0)
282: break;
283: } else
284: pend = cp + 3 - b;
285: continue;
286: }
1.1 schwarze 287: elem_end = 0;
288: if (b[pend] != '>')
289: in_tag = 1;
290: else if (pend > 0 && b[pend - 1] == '/') {
291: b[pend - 1] = '\0';
292: elem_end = 1;
293: }
294: b[pend] = '\0';
295: if (pend < rlen)
296: pend++;
297: if (b[++poff] == '/') {
298: elem_end = 1;
299: poff++;
300: } else if (b[poff] != '!' && b[poff] != '?') {
301: table_add(stacki > 0 ?
1.2 schwarze 302: stack[stacki - 1] : "ROOT",
1.1 schwarze 303: b + poff);
304: stack_push(b + poff);
305: }
306: if (elem_end)
307: stack_pop(b + poff);
308: } else {
309: advance(b, rlen, &pend, "<");
310: if (stacki > 0)
311: table_add(stack[stacki - 1], "TEXT");
312: }
313: }
314: assert(poff > 0);
315: memmove(b, b + poff, rlen - poff);
316: rlen -= poff;
317: }
318: if (rsz < 0)
319: perror(fname);
320: }
321:
322: int
323: main(int argc, char *argv[])
324: {
325: size_t fsz, i;
326: ssize_t rsz;
1.2 schwarze 327: int ch, fd, show_all;
328:
329: show_all = 0;
330: while ((ch = getopt(argc, argv, "a")) != -1) {
331: switch (ch) {
332: case 'a':
333: show_all = 1;
334: break;
335: default:
336: return 1;
337: }
338: }
1.4 ! schwarze 339: argc -= optind;
! 340: argv += optind;
! 341:
! 342: if (argc > 1) {
! 343: nchild.parent = argv[0];
! 344: nchild.child = argv[1];
! 345: }
1.1 schwarze 346:
1.2 schwarze 347: /* Exclude relations that are already fully implemented. */
348: if (show_all == 0) {
349: table_add("para", NULL);
350: }
351: table_add(NULL, NULL);
352:
353: /* Loop over input files. */
1.1 schwarze 354: fd = -1;
355: fname = NULL;
356: while ((rsz = getline(&fname, &fsz, stdin)) != -1) {
357: if (fname[rsz - 1] == '\n')
358: fname[--rsz] = '\0';
359: if ((fd = open(fname, O_RDONLY, 0)) == -1)
360: err(1, "%s", fname);
361: parse_file(fd, fname);
362: close(fd);
363: }
364:
365: /* Cleanup and error handling. */
366: free(fname);
367: if (ferror(stdin))
368: err(1, "standard input");
369: if (fd == -1)
370: errx(1, "No input file names found on standard input");
371:
372: /* Dump results. */
373: for (i = 0; i < tablei; i++)
1.2 schwarze 374: if (table[i].count != -1)
375: printf("%d\t%s\t%s\n", table[i].count,
376: table[i].parent, table[i].child);
1.4 ! schwarze 377:
! 378: /* Optional parent-child histogram. */
! 379: if (nchild.parent != NULL) {
! 380: printf("%s %s", nchild.parent, nchild.child);
! 381: for (i = 0; i < nchildsz; i++)
! 382: printf(" %d", nchild.freq[i]);
! 383: putchar('\n');
! 384: }
1.1 schwarze 385: return 0;
386: }
CVSweb