=================================================================== RCS file: /cvs/mandoc/Attic/makewhatis.c,v retrieving revision 1.6 retrieving revision 1.20 diff -u -p -r1.6 -r1.20 --- mandoc/Attic/makewhatis.c 2011/06/21 14:16:05 1.6 +++ mandoc/Attic/makewhatis.c 2011/07/12 10:03:02 1.20 @@ -1,4 +1,4 @@ -/* $Id: makewhatis.c,v 1.6 2011/06/21 14:16:05 kristaps Exp $ */ +/* $Id: makewhatis.c,v 1.20 2011/07/12 10:03:02 kristaps Exp $ */ /* * Copyright (c) 2011 Kristaps Dzonsons * @@ -21,11 +21,6 @@ #include #include -#ifdef __linux__ -# include -#else -# include -#endif #include #include #include @@ -33,6 +28,12 @@ #include #include +#ifdef __linux__ +# include +#else +# include +#endif + #include "man.h" #include "mdoc.h" #include "mandoc.h" @@ -44,15 +45,19 @@ /* Bit-fields. See makewhatis.1. */ -#define TYPE_NAME 0x01 -#define TYPE_FUNCTION 0x02 -#define TYPE_UTILITY 0x04 -#define TYPE_INCLUDES 0x08 -#define TYPE_VARIABLE 0x10 -#define TYPE_STANDARD 0x20 -#define TYPE_AUTHOR 0x40 -#define TYPE_CONFIG 0x80 -#define TYPE_DESC 0x100 +#define TYPE_NAME 0x01 +#define TYPE_FUNCTION 0x02 +#define TYPE_UTILITY 0x04 +#define TYPE_INCLUDES 0x08 +#define TYPE_VARIABLE 0x10 +#define TYPE_STANDARD 0x20 +#define TYPE_AUTHOR 0x40 +#define TYPE_CONFIG 0x80 +#define TYPE_DESC 0x100 +#define TYPE_XREF 0x200 +#define TYPE_PATH 0x400 +#define TYPE_ENV 0x800 +#define TYPE_ERR 0x1000 /* Buffer for storing growable data. */ @@ -62,6 +67,14 @@ struct buf { size_t size; }; +/* Operation we're going to perform. */ + +enum op { + OP_NEW = 0, /* new database */ + OP_UPDATE, /* update entries in existing database */ + OP_DELETE /* delete entries from existing database */ +}; + #define MAN_ARGS DB *hash, \ struct buf *buf, \ struct buf *dbuf, \ @@ -72,23 +85,32 @@ struct buf { const struct mdoc_node *n, \ const struct mdoc_meta *m +static void buf_appendmdoc(struct buf *, + const struct mdoc_node *, int); static void buf_append(struct buf *, const char *); static void buf_appendb(struct buf *, const void *, size_t); static void dbt_put(DB *, const char *, DBT *, DBT *); static void hash_put(DB *, const struct buf *, int); +static void hash_reset(DB **); +static void op_delete(const char *, int, DB *, + const char *, DB *, const char *); static int pman_node(MAN_ARGS); static void pmdoc_node(MDOC_ARGS); static void pmdoc_An(MDOC_ARGS); static void pmdoc_Cd(MDOC_ARGS); +static void pmdoc_Er(MDOC_ARGS); +static void pmdoc_Ev(MDOC_ARGS); static void pmdoc_Fd(MDOC_ARGS); static void pmdoc_In(MDOC_ARGS); static void pmdoc_Fn(MDOC_ARGS); static void pmdoc_Fo(MDOC_ARGS); static void pmdoc_Nd(MDOC_ARGS); static void pmdoc_Nm(MDOC_ARGS); +static void pmdoc_Pa(MDOC_ARGS); static void pmdoc_St(MDOC_ARGS); static void pmdoc_Vt(MDOC_ARGS); +static void pmdoc_Xr(MDOC_ARGS); static void usage(void); typedef void (*pmdoc_nf)(MDOC_ARGS); @@ -114,8 +136,8 @@ static const pmdoc_nf mdocs[MDOC_MAX] = { pmdoc_Cd, /* Cd */ NULL, /* Cm */ NULL, /* Dv */ - NULL, /* Er */ - NULL, /* Ev */ + pmdoc_Er, /* Er */ + pmdoc_Ev, /* Ev */ NULL, /* Ex */ NULL, /* Fa */ pmdoc_Fd, /* Fd */ @@ -129,12 +151,12 @@ static const pmdoc_nf mdocs[MDOC_MAX] = { pmdoc_Nm, /* Nm */ NULL, /* Op */ NULL, /* Ot */ - NULL, /* Pa */ + pmdoc_Pa, /* Pa */ NULL, /* Rv */ pmdoc_St, /* St */ pmdoc_Vt, /* Va */ pmdoc_Vt, /* Vt */ - NULL, /* Xr */ + pmdoc_Xr, /* Xr */ NULL, /* %A */ NULL, /* %B */ NULL, /* %D */ @@ -226,24 +248,27 @@ main(int argc, char *argv[]) struct mparse *mp; /* parse sequence */ struct mdoc *mdoc; /* resulting mdoc */ struct man *man; /* resulting man */ + enum op op; char *fn; /* current file being parsed */ const char *msec, /* manual section */ *mtitle, /* manual title */ *arch, /* manual architecture */ *dir; /* result dir (default: cwd) */ char ibuf[MAXPATHLEN], /* index fname */ - ibbuf[MAXPATHLEN], /* index backup fname */ fbuf[MAXPATHLEN], /* btree fname */ - fbbuf[MAXPATHLEN], /* btree backup fname */ vbuf[8]; /* stringified record number */ - int ch, seq; + int ch, seq, verb, i; DB *idx, /* index database */ *db, /* keyword database */ *hash; /* temporary keyword hashtable */ DBT key, val; - size_t sv, rsz; + enum mandoclevel ec; + size_t sv; BTREEINFO info; /* btree configuration */ - recno_t rec; /* current record number */ + recno_t rec, /* current record number */ + maxrec; + recno_t *recs; + size_t recsz; struct buf buf, /* keyword buffer */ dbuf; /* description buffer */ extern int optind; @@ -256,12 +281,33 @@ main(int argc, char *argv[]) ++progname; dir = ""; + verb = 0; + db = idx = NULL; + mp = NULL; + hash = NULL; + recs = NULL; + recsz = 0; + maxrec = 0; + op = OP_NEW; + ec = MANDOCLEVEL_SYSERR; - while (-1 != (ch = getopt(argc, argv, "d:"))) + memset(&buf, 0, sizeof(struct buf)); + memset(&dbuf, 0, sizeof(struct buf)); + + while (-1 != (ch = getopt(argc, argv, "d:ruv"))) switch (ch) { case ('d'): dir = optarg; break; + case ('r'): + op = OP_DELETE; + break; + case ('u'): + op = OP_UPDATE; + break; + case ('v'): + verb++; + break; default: usage(); return((int)MANDOCLEVEL_BADARG); @@ -270,116 +316,139 @@ main(int argc, char *argv[]) argc -= optind; argv += optind; - /* - * Set up temporary file-names into which we're going to write - * all of our data (both for the index and database). These - * will be securely renamed to the real file-names after we've - * written all of our data. - */ - ibuf[0] = ibuf[MAXPATHLEN - 2] = - ibbuf[0] = ibbuf[MAXPATHLEN - 2] = - fbuf[0] = fbuf[MAXPATHLEN - 2] = - fbbuf[0] = fbbuf[MAXPATHLEN - 2] = '\0'; + fbuf[0] = fbuf[MAXPATHLEN - 2] = '\0'; strlcat(fbuf, dir, MAXPATHLEN); strlcat(fbuf, MANDOC_DB, MAXPATHLEN); - strlcat(fbbuf, fbuf, MAXPATHLEN); - strlcat(fbbuf, "~", MAXPATHLEN); - strlcat(ibuf, dir, MAXPATHLEN); strlcat(ibuf, MANDOC_IDX, MAXPATHLEN); - strlcat(ibbuf, ibuf, MAXPATHLEN); - strlcat(ibbuf, "~", MAXPATHLEN); - if ('\0' != fbuf[MAXPATHLEN - 2] || - '\0' != fbbuf[MAXPATHLEN - 2] || - '\0' != ibuf[MAXPATHLEN - 2] || - '\0' != ibbuf[MAXPATHLEN - 2]) { - fprintf(stderr, "%s: Path too long\n", progname); - exit((int)MANDOCLEVEL_SYSERR); + '\0' != ibuf[MAXPATHLEN - 2]) { + fprintf(stderr, "%s: Path too long\n", dir); + goto out; } /* * For the keyword database, open a BTREE database that allows * duplicates. * For the index database, use a standard RECNO database type. - * For the temporary keyword hashtable, use the HASH database - * type. + * Truncate the database if we're creating a new one. */ - hash = dbopen(NULL, MANDOC_FLAGS, 0644, DB_HASH, NULL); - if (NULL == hash) { - perror("hash"); - exit((int)MANDOCLEVEL_SYSERR); - } - memset(&info, 0, sizeof(BTREEINFO)); info.flags = R_DUP; - db = dbopen(fbbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info); + if (OP_NEW == op) { + db = dbopen(fbuf, MANDOC_FLAGS, 0644, DB_BTREE, &info); + idx = dbopen(ibuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL); + } else { + db = dbopen(fbuf, O_CREAT|O_RDWR, 0644, DB_BTREE, &info); + idx = dbopen(ibuf, O_CREAT|O_RDWR, 0644, DB_RECNO, NULL); + } + if (NULL == db) { - perror(fbbuf); - (*hash->close)(hash); - exit((int)MANDOCLEVEL_SYSERR); + perror(fbuf); + goto out; + } else if (NULL == db) { + perror(ibuf); + goto out; } - idx = dbopen(ibbuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL); + /* + * If we're going to delete or update a database, remove the + * entries now. This doesn't actually remove them; it only sets + * their record value lengths to zero. + */ - if (NULL == db) { - perror(ibbuf); - (*db->close)(db); - (*hash->close)(hash); - exit((int)MANDOCLEVEL_SYSERR); + if (OP_DELETE == op || OP_UPDATE == op) + for (i = 0; i < argc; i++) + op_delete(argv[i], verb, idx, ibuf, db, fbuf); + + if (OP_DELETE == op) { + ec = MANDOCLEVEL_OK; + goto out; } /* - * Try parsing the manuals given on the command line. If we - * totally fail, then just keep on going. Take resulting trees - * and push them down into the database code. + * Compile a list of all available "empty" records to use. This + * keeps the size of the database small. + */ + + if (OP_UPDATE == op) { + i = 0; + seq = R_FIRST; + while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) { + seq = R_NEXT; + maxrec = *(recno_t *)key.data; + if (val.size > 0) + continue; + if ((size_t)i >= recsz) { + recsz += 1024; + recs = mandoc_realloc + (recs, recsz * sizeof(recno_t)); + } + recs[i++] = maxrec; + } + if (ch < 0) { + perror(ibuf); + exit((int)MANDOCLEVEL_SYSERR); + } + recsz = (size_t)i; + maxrec++; + assert(recsz < maxrec); + } + + /* + * Add records to the database. + * Try parsing each manual given on the command line. + * If we fail, then emit an error and keep on going. + * Take resulting trees and push them down into the database code. * Use the auto-parser and don't report any errors. */ mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL); - rec = 1; - rsz = 0; - - memset(&buf, 0, sizeof(struct buf)); - memset(&dbuf, 0, sizeof(struct buf)); - buf.size = dbuf.size = MANDOC_BUFSZ; - buf.cp = mandoc_malloc(buf.size); dbuf.cp = mandoc_malloc(dbuf.size); - while (NULL != (fn = *argv++)) { + for (rec = 0, i = 0; i < argc; i++) { + fn = argv[i]; + if (OP_UPDATE == op) { + if (recsz > 0) { + --recsz; + rec = recs[(int)recsz]; + } else if (maxrec > 0) { + rec = maxrec; + maxrec = 0; + } else + rec++; + } else + rec++; + mparse_reset(mp); + hash_reset(&hash); - /* Parse and get (non-empty) AST. */ - if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) { fprintf(stderr, "%s: Parse failure\n", fn); continue; } + mparse_result(mp, &mdoc, &man); if (NULL == mdoc && NULL == man) continue; - /* Manual section: can be empty string. */ - msec = NULL != mdoc ? - mdoc_meta(mdoc)->msec : - man_meta(man)->msec; + mdoc_meta(mdoc)->msec : man_meta(man)->msec; mtitle = NULL != mdoc ? - mdoc_meta(mdoc)->title : - man_meta(man)->title; + mdoc_meta(mdoc)->title : man_meta(man)->title; arch = NULL != mdoc ? mdoc_meta(mdoc)->arch : NULL; - assert(msec); - assert(mtitle); + if (NULL == arch) + arch = ""; /* * The index record value consists of a nil-terminated @@ -393,8 +462,7 @@ main(int argc, char *argv[]) buf_appendb(&dbuf, fn, strlen(fn) + 1); buf_appendb(&dbuf, msec, strlen(msec) + 1); buf_appendb(&dbuf, mtitle, strlen(mtitle) + 1); - buf_appendb(&dbuf, arch ? arch : "", - arch ? strlen(arch) + 1 : 1); + buf_appendb(&dbuf, arch, strlen(arch) + 1); sv = dbuf.len; @@ -416,18 +484,18 @@ main(int argc, char *argv[]) seq = R_FIRST; while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) { + seq = R_NEXT; + memcpy(vbuf, val.data, sizeof(uint32_t)); val.size = sizeof(vbuf); val.data = vbuf; - dbt_put(db, fbbuf, &key, &val); - if ((*hash->del)(hash, &key, 0) < 0) { - perror("hash"); - exit((int)MANDOCLEVEL_SYSERR); - } - seq = R_NEXT; + if (verb > 1) + printf("Indexed: %s, %s, 0x%x\n", + fn, (char *)key.data, + *(int *)val.data); + dbt_put(db, fbuf, &key, &val); } - if (ch < 0) { perror("hash"); exit((int)MANDOCLEVEL_SYSERR); @@ -447,27 +515,79 @@ main(int argc, char *argv[]) val.data = dbuf.cp; val.size = dbuf.len; - dbt_put(idx, ibbuf, &key, &val); - rec++; + if (verb > 0) + printf("Indexed: %s\n", fn); + + dbt_put(idx, ibuf, &key, &val); } - (*db->close)(db); - (*idx->close)(idx); - (*hash->close)(hash); + ec = MANDOCLEVEL_OK; +out: + if (db) + (*db->close)(db); + if (idx) + (*idx->close)(idx); + if (hash) + (*hash->close)(hash); + if (mp) + mparse_free(mp); - mparse_free(mp); - free(buf.cp); free(dbuf.cp); + free(recs); - /* Atomically replace the file with our temporary one. */ + return((int)ec); +} - if (-1 == rename(fbbuf, fbuf)) - perror(fbuf); - if (-1 == rename(ibbuf, ibuf)) - perror(fbuf); +static void +op_delete(const char *fn, int verb, DB *idx, + const char *ibuf, DB *db, const char *fbuf) +{ + int ch; + DBT key, val; + recno_t rec; + unsigned int seq, sseq; - return((int)MANDOCLEVEL_OK); + seq = R_FIRST; + while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) { + seq = R_NEXT; + if (0 == val.size) + continue; + if (strcmp((char *)val.data, fn)) + continue; + + rec = *(recno_t *)key.data; + + sseq = R_FIRST; + while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) { + sseq = R_NEXT; + assert(8 == val.size); + if (rec != *(recno_t *)(val.data + 4)) + continue; + if (verb > 1) + printf("Deleted: %s, %s\n", + fn, (char *)key.data); + ch = (*db->del)(db, &key, R_CURSOR); + if (ch < 0) + break; + } + if (ch < 0) { + perror(fbuf); + exit((int)MANDOCLEVEL_SYSERR); + } + + val.size = 0; + if (verb) + printf("Deleted: %s\n", fn); + ch = (*idx->put) + (idx, &key, &val, R_CURSOR); + if (ch < 0) + break; + } + if (ch < 0) { + perror(ibuf); + exit((int)MANDOCLEVEL_SYSERR); + } } /* @@ -508,6 +628,33 @@ buf_append(struct buf *buf, const char *cp) buf_appendb(buf, cp, sz + 1); } +/* + * Recursively add all text from a given node. + * This is optimised for general mdoc nodes in this context, which do + * not consist of subexpressions and having a recursive call for n->next + * would be wasteful. + * The "f" variable should be 0 unless called from pmdoc_Nd for the + * description buffer, which does not start at the beginning of the + * buffer. + */ +static void +buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f) +{ + + for ( ; n; n = n->next) { + if (n->child) + buf_appendmdoc(buf, n->child, f); + + if (MDOC_TEXT == n->type && f) { + f = 0; + buf_appendb(buf, n->string, + strlen(n->string) + 1); + } else if (MDOC_TEXT == n->type) + buf_append(buf, n->string); + + } +} + /* ARGSUSED */ static void pmdoc_An(MDOC_ARGS) @@ -516,13 +663,25 @@ pmdoc_An(MDOC_ARGS) if (SEC_AUTHORS != n->sec) return; - for (n = n->child; n; n = n->next) - if (MDOC_TEXT == n->type) - buf_append(buf, n->string); - + buf_appendmdoc(buf, n->child, 0); hash_put(hash, buf, TYPE_AUTHOR); } +static void +hash_reset(DB **db) +{ + DB *hash; + + if (NULL != (hash = *db)) + (*hash->close)(hash); + + *db = dbopen(NULL, MANDOC_FLAGS, 0644, DB_HASH, NULL); + if (NULL == *db) { + perror("hash"); + exit((int)MANDOCLEVEL_SYSERR); + } +} + /* ARGSUSED */ static void pmdoc_Fd(MDOC_ARGS) @@ -577,10 +736,7 @@ pmdoc_Cd(MDOC_ARGS) if (SEC_SYNOPSIS != n->sec) return; - for (n = n->child; n; n = n->next) - if (MDOC_TEXT == n->type) - buf_append(buf, n->string); - + buf_appendmdoc(buf, n->child, 0); hash_put(hash, buf, TYPE_CONFIG); } @@ -640,6 +796,25 @@ pmdoc_St(MDOC_ARGS) /* ARGSUSED */ static void +pmdoc_Xr(MDOC_ARGS) +{ + + if (NULL == (n = n->child)) + return; + + buf_appendb(buf, n->string, strlen(n->string)); + + if (NULL != (n = n->next)) { + buf_appendb(buf, ".", 1); + buf_appendb(buf, n->string, strlen(n->string) + 1); + } else + buf_appendb(buf, ".", 2); + + hash_put(hash, buf, TYPE_XREF); +} + +/* ARGSUSED */ +static void pmdoc_Vt(MDOC_ARGS) { const char *start; @@ -694,37 +869,59 @@ pmdoc_Fo(MDOC_ARGS) static void pmdoc_Nd(MDOC_ARGS) { - int first; - size_t sz; - - for (first = 1, n = n->child; n; n = n->next) { - if (MDOC_TEXT != n->type) - continue; - if (first) { - sz = strlen(n->string) + 1; - buf_appendb(dbuf, n->string, sz); - buf_appendb(buf, n->string, sz); - } else { - buf_append(dbuf, n->string); - buf_append(buf, n->string); - } + if (MDOC_BODY != n->type) + return; - first = 0; - } + buf_appendmdoc(dbuf, n->child, 1); + buf_appendmdoc(buf, n->child, 0); hash_put(hash, buf, TYPE_DESC); } /* ARGSUSED */ static void +pmdoc_Er(MDOC_ARGS) +{ + + if (SEC_ERRORS != n->sec) + return; + + buf_appendmdoc(buf, n->child, 0); + hash_put(hash, buf, TYPE_ERR); +} + +/* ARGSUSED */ +static void +pmdoc_Ev(MDOC_ARGS) +{ + + if (SEC_ENVIRONMENT != n->sec) + return; + + buf_appendmdoc(buf, n->child, 0); + hash_put(hash, buf, TYPE_ENV); +} + +/* ARGSUSED */ +static void +pmdoc_Pa(MDOC_ARGS) +{ + + if (SEC_FILES != n->sec) + return; + + buf_appendmdoc(buf, n->child, 0); + hash_put(hash, buf, TYPE_PATH); +} + +/* ARGSUSED */ +static void pmdoc_Nm(MDOC_ARGS) { if (SEC_NAME == n->sec) { - for (n = n->child; n; n = n->next) - if (MDOC_TEXT == n->type) - buf_append(buf, n->string); + buf_appendmdoc(buf, n->child, 0); hash_put(hash, buf, TYPE_NAME); return; } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) @@ -733,10 +930,7 @@ pmdoc_Nm(MDOC_ARGS) if (NULL == n->child) buf_append(buf, m->name); - for (n = n->child; n; n = n->next) - if (MDOC_TEXT == n->type) - buf_append(buf, n->string); - + buf_appendmdoc(buf, n->child, 0); hash_put(hash, buf, TYPE_UTILITY); } @@ -746,11 +940,12 @@ hash_put(DB *db, const struct buf *buf, int mask) DBT key, val; int rc; - key.data = buf->cp; - - if ((key.size = buf->len) < 2) + if (buf->len < 2) return; + key.data = buf->cp; + key.size = buf->len; + if ((rc = (*db->get)(db, &key, &val, 0)) < 0) { perror("hash"); exit((int)MANDOCLEVEL_SYSERR); @@ -760,9 +955,6 @@ hash_put(DB *db, const struct buf *buf, int mask) val.data = &mask; val.size = sizeof(int); - /*fprintf(stderr, "Hashing: [%s] (0x%x)\n", - (char *)key.data, mask);*/ - if ((rc = (*db->put)(db, &key, &val, 0)) < 0) { perror("hash"); exit((int)MANDOCLEVEL_SYSERR); @@ -773,9 +965,6 @@ static void dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) { - if (0 == key->size) - return; - assert(key->size); assert(val->size); @@ -885,8 +1074,9 @@ pman_node(MAN_ARGS) start++; } + buf->len = 0; + if (sv == start) { - buf->len = 0; buf_append(buf, start); return(1); } @@ -909,6 +1099,8 @@ pman_node(MAN_ARGS) sz = strlen(start) + 1; buf_appendb(dbuf, start, sz); buf_appendb(buf, start, sz); + + hash_put(hash, buf, TYPE_DESC); } } @@ -924,5 +1116,6 @@ static void usage(void) { - fprintf(stderr, "usage: %s [-d path] [file...]\n", progname); + fprintf(stderr, "usage: %s [-ruv] [-d path] [file...]\n", + progname); }