=================================================================== RCS file: /cvs/mandoc/Attic/mandoc-db.c,v retrieving revision 1.3 retrieving revision 1.13 diff -u -p -r1.3 -r1.13 --- mandoc/Attic/mandoc-db.c 2011/04/03 14:18:29 1.3 +++ mandoc/Attic/mandoc-db.c 2011/04/28 10:44:02 1.13 @@ -1,4 +1,4 @@ -/* $Id: mandoc-db.c,v 1.3 2011/04/03 14:18:29 kristaps Exp $ */ +/* $Id: mandoc-db.c,v 1.13 2011/04/28 10:44:02 kristaps Exp $ */ /* * Copyright (c) 2011 Kristaps Dzonsons * @@ -48,30 +48,42 @@ enum type { MANDOC_FUNCTION, MANDOC_UTILITY, MANDOC_INCLUDES, - MANDOC_VARIABLE + MANDOC_VARIABLE, + MANDOC_STANDARD }; +#define MAN_ARGS DB *db, \ + const char *dbn, \ + DBT *key, size_t *ksz, \ + DBT *val, \ + DBT *rval, size_t *rsz, \ + const struct man_node *n #define MDOC_ARGS DB *db, \ const char *dbn, \ DBT *key, size_t *ksz, \ DBT *val, \ + DBT *rval, size_t *rsz, \ const struct mdoc_node *n static void dbt_append(DBT *, size_t *, const char *); static void dbt_appendb(DBT *, size_t *, const void *, size_t); static void dbt_init(DBT *, size_t *); -static void version(void); +static void dbt_put(DB *, const char *, DBT *, DBT *); static void usage(void); -static void pmdoc(DB *, const char *, - DBT *, size_t *, DBT *, - const char *, struct mdoc *); +static void pman(DB *, const char *, DBT *, size_t *, + DBT *, DBT *, size_t *, struct man *); +static int pman_node(MAN_ARGS); +static void pmdoc(DB *, const char *, DBT *, size_t *, + DBT *, DBT *, size_t *, struct mdoc *); static void pmdoc_node(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_St(MDOC_ARGS); static void pmdoc_Vt(MDOC_ARGS); typedef void (*pmdoc_nf)(MDOC_ARGS); @@ -110,13 +122,13 @@ static const pmdoc_nf mdocs[MDOC_MAX] = { NULL, /* Ic */ pmdoc_In, /* In */ NULL, /* Li */ - NULL, /* Nd */ + pmdoc_Nd, /* Nd */ pmdoc_Nm, /* Nm */ NULL, /* Op */ NULL, /* Ot */ NULL, /* Pa */ NULL, /* Rv */ - NULL, /* St */ + pmdoc_St, /* St */ pmdoc_Vt, /* Va */ pmdoc_Vt, /* Vt */ NULL, /* Xr */ @@ -208,21 +220,24 @@ main(int argc, char *argv[]) { struct mparse *mp; /* parse sequence */ struct mdoc *mdoc; /* resulting mdoc */ - char *fn; - const char *dir; /* result dir (default: cwd) */ + struct man *man; /* resulting man */ + char *fn; /* current file being parsed */ + const char *msec, /* manual section */ + *dir; /* result dir (default: cwd) */ char ibuf[MAXPATHLEN], /* index fname */ ibbuf[MAXPATHLEN], /* index backup fname */ fbuf[MAXPATHLEN], /* btree fname */ fbbuf[MAXPATHLEN]; /* btree backup fname */ - int c; - DB *index, /* index database */ + int ch; + DB *idx, /* index database */ *db; /* keyword database */ DBT rkey, rval, /* recno entries */ key, val; /* persistent keyword entries */ - size_t ksz; /* entry buffer size */ - char vbuf[8]; + size_t sv, + ksz, rsz; /* entry buffer size */ + char vbuf[8]; /* stringified record number */ BTREEINFO info; /* btree configuration */ - recno_t rec; + recno_t rec; /* current record number */ extern int optind; extern char *optarg; @@ -234,14 +249,11 @@ main(int argc, char *argv[]) dir = ""; - while (-1 != (c = getopt(argc, argv, "d:V"))) - switch (c) { + while (-1 != (ch = getopt(argc, argv, "d:"))) + switch (ch) { case ('d'): dir = optarg; break; - case ('V'): - version(); - return((int)MANDOCLEVEL_OK); default: usage(); return((int)MANDOCLEVEL_BADARG); @@ -297,7 +309,7 @@ main(int argc, char *argv[]) exit((int)MANDOCLEVEL_SYSERR); } - index = dbopen(ibbuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL); + idx = dbopen(ibbuf, MANDOC_FLAGS, 0644, DB_RECNO, NULL); if (NULL == db) { perror(ibbuf); @@ -324,40 +336,76 @@ main(int argc, char *argv[]) rkey.size = sizeof(recno_t); rec = 1; - ksz = 0; + ksz = rsz = 0; while (NULL != (fn = *argv++)) { mparse_reset(mp); - if (mparse_readfd(mp, -1, fn) >= MANDOCLEVEL_FATAL) - continue; + /* Parse and get (non-empty) AST. */ - mparse_result(mp, &mdoc, NULL); - if (NULL == mdoc) + 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; - rkey.data = &rec; - rval.data = fn; - rval.size = strlen(fn) + 1; + /* Manual section: can be empty string. */ - if (-1 == (*index->put)(index, &rkey, &rval, 0)) { - perror(ibbuf); - break; - } + msec = NULL != mdoc ? + mdoc_meta(mdoc)->msec : + man_meta(man)->msec; + assert(msec); + + /* + * The index record value consists of a nil-terminated + * filename, a nil-terminated manual section, and a + * nil-terminated description. Since the description + * may not be set, we set a sentinel to see if we're + * going to write a nil byte in its place. + */ + + dbt_init(&rval, &rsz); + dbt_appendb(&rval, &rsz, fn, strlen(fn) + 1); + dbt_appendb(&rval, &rsz, msec, strlen(msec) + 1); + sv = rval.size; + + /* Fix the record number in the btree value. */ + memset(val.data, 0, sizeof(uint32_t)); memcpy(val.data + 4, &rec, sizeof(uint32_t)); - pmdoc(db, fbbuf, &key, &ksz, &val, fn, mdoc); + if (mdoc) + pmdoc(db, fbbuf, &key, &ksz, + &val, &rval, &rsz, mdoc); + else + pman(db, fbbuf, &key, &ksz, + &val, &rval, &rsz, man); + + /* + * Apply this to the index. If we haven't had a + * description set, put an empty one in now. + */ + + if (rval.size == sv) + dbt_appendb(&rval, &rsz, "", 1); + + rkey.data = &rec; + dbt_put(idx, ibbuf, &rkey, &rval); + + printf("Indexed: %s\n", fn); rec++; } (*db->close)(db); - (*index->close)(index); + (*idx->close)(idx); mparse_free(mp); free(key.data); + free(rval.data); /* Atomically replace the file with our temporary one. */ @@ -437,29 +485,39 @@ pmdoc_Fd(MDOC_ARGS) uint32_t fl; const char *start, *end; size_t sz; - char nil; + const char nil = '\0'; if (SEC_SYNOPSIS != n->sec) return; if (NULL == (n = n->child) || MDOC_TEXT != n->type) return; + + /* + * Only consider those `Fd' macro fields that begin with an + * "inclusion" token (versus, e.g., #define). + */ if (strcmp("#include", n->string)) return; + if (NULL == (n = n->next) || MDOC_TEXT != n->type) return; + /* + * Strip away the enclosing angle brackets and make sure we're + * not zero-length. + */ + start = n->string; - if ('<' == *start) + if ('<' == *start || '"' == *start) start++; if (0 == (sz = strlen(start))) return; end = &start[(int)sz - 1]; - if ('>' == *end) + if ('>' == *end || '"' == *end) end--; - nil = '\0'; dbt_appendb(key, ksz, start, end - start + 1); dbt_appendb(key, ksz, &nil, 1); @@ -501,7 +559,7 @@ pmdoc_Fn(MDOC_ARGS) if (NULL == cp) cp = n->child->string; - /* Ignore pointers. */ + /* Strip away pointer symbol. */ while ('*' == *cp) cp++; @@ -513,22 +571,38 @@ pmdoc_Fn(MDOC_ARGS) /* ARGSUSED */ static void +pmdoc_St(MDOC_ARGS) +{ + uint32_t fl; + + if (SEC_STANDARDS != n->sec) + return; + if (NULL == n->child || MDOC_TEXT != n->child->type) + return; + + dbt_append(key, ksz, n->child->string); + fl = MANDOC_STANDARD; + memcpy(val->data, &fl, 4); +} + +/* ARGSUSED */ +static void pmdoc_Vt(MDOC_ARGS) { uint32_t fl; const char *start, *end; size_t sz; - char nil; + const char nil = '\0'; if (SEC_SYNOPSIS != n->sec) return; if (MDOC_Vt == n->tok && MDOC_BODY != n->type) return; - if (NULL == n->child || MDOC_TEXT != n->child->type) + if (NULL == n->last || MDOC_TEXT != n->last->type) return; /* - * Strip away leading '*' and trailing ';'. + * Strip away leading pointer symbol '*' and trailing ';'. */ start = n->last->string; @@ -546,7 +620,6 @@ pmdoc_Vt(MDOC_ARGS) if (end == start) return; - nil = '\0'; dbt_appendb(key, ksz, start, end - start + 1); dbt_appendb(key, ksz, &nil, 1); fl = MANDOC_VARIABLE; @@ -569,8 +642,27 @@ pmdoc_Fo(MDOC_ARGS) memcpy(val->data, &fl, 4); } + /* ARGSUSED */ static void +pmdoc_Nd(MDOC_ARGS) +{ + int first; + + for (first = 1, n = n->child; n; n = n->next) { + if (MDOC_TEXT != n->type) + continue; + if (first) + dbt_appendb(rval, rsz, n->string, strlen(n->string) + 1); + else + dbt_append(rval, rsz, n->string); + + first = 0; + } +} + +/* ARGSUSED */ +static void pmdoc_Nm(MDOC_ARGS) { uint32_t fl; @@ -597,6 +689,25 @@ pmdoc_Nm(MDOC_ARGS) memcpy(val->data, &fl, 4); } +static void +dbt_put(DB *db, const char *dbn, DBT *key, DBT *val) +{ + + if (0 == key->size) + return; + + assert(key->data); + assert(val->size); + assert(val->data); + + if (0 == (*db->put)(db, key, val, 0)) + return; + + perror(dbn); + exit((int)MANDOCLEVEL_SYSERR); + /* NOTREACHED */ +} + /* * Call out to per-macro handlers after clearing the persistent database * key. If the macro sets the database key, flush it to the database. @@ -622,38 +733,129 @@ pmdoc_node(MDOC_ARGS) break; dbt_init(key, ksz); - (*mdocs[n->tok])(db, dbn, key, ksz, val, n); + (*mdocs[n->tok])(db, dbn, key, ksz, val, rval, rsz, n); - if (0 == key->size) - break; - if (0 == (*db->put)(db, key, val, 0)) - break; - - perror(dbn); - exit((int)MANDOCLEVEL_SYSERR); - /* NOTREACHED */ + dbt_put(db, dbn, key, val); + break; default: break; } - pmdoc_node(db, dbn, key, ksz, val, n->child); - pmdoc_node(db, dbn, key, ksz, val, n->next); + pmdoc_node(db, dbn, key, ksz, val, rval, rsz, n->child); + pmdoc_node(db, dbn, key, ksz, val, rval, rsz, n->next); } +static int +pman_node(MAN_ARGS) +{ + const struct man_node *head, *body; + const char *start, *sv; + const char nil = '\0'; + size_t sz; + uint32_t fl; + + if (NULL == n) + return(0); + + /* + * We're only searching for one thing: the first text child in + * the BODY of a NAME section. Since we don't keep track of + * sections in -man, run some hoops to find out whether we're in + * the correct section or not. + */ + + if (MAN_BODY == n->type && MAN_SH == n->tok) { + body = n; + assert(body->parent); + if (NULL != (head = body->parent->head) && + 1 == head->nchild && + NULL != (head = (head->child)) && + MAN_TEXT == head->type && + 0 == strcmp(head->string, "NAME") && + NULL != (body = body->child) && + MAN_TEXT == body->type) { + + fl = MANDOC_NAME; + memcpy(val->data, &fl, 4); + + assert(body->string); + start = sv = body->string; + + /* + * Go through a special heuristic dance here. + * This is why -man manuals are great! + * Conventionally, one or more manual names are + * comma-specified prior to a whitespace, then a + * dash, then a description. Try to puzzle out + * the name parts here. + */ + + for ( ;; ) { + sz = strcspn(start, " ,"); + if ('\0' == start[(int)sz]) + break; + + dbt_init(key, ksz); + dbt_appendb(key, ksz, start, sz); + dbt_appendb(key, ksz, &nil, 1); + + dbt_put(db, dbn, key, val); + + if (' ' == start[(int)sz]) { + start += (int)sz + 1; + break; + } + + assert(',' == start[(int)sz]); + start += (int)sz + 1; + while (' ' == *start) + start++; + } + + if (sv == start) { + dbt_init(key, ksz); + dbt_append(key, ksz, start); + return(1); + } + + while (' ' == *start) + start++; + + if ('\\' == *start && '-' == *(start + 1)) + start += 2; + else if ('-' == *start) + start++; + + while (' ' == *start) + start++; + + dbt_appendb(rval, rsz, start, strlen(start) + 1); + } + } + + if (pman_node(db, dbn, key, ksz, val, rval, rsz, n->child)) + return(1); + if (pman_node(db, dbn, key, ksz, val, rval, rsz, n->next)) + return(1); + + return(0); +} + static void -pmdoc(DB *db, const char *dbn, - DBT *key, size_t *ksz, DBT *val, - const char *path, struct mdoc *m) +pman(DB *db, const char *dbn, DBT *key, size_t *ksz, + DBT *val, DBT *rval, size_t *rsz, struct man *m) { - pmdoc_node(db, dbn, key, ksz, val, mdoc_node(m)); + pman_node(db, dbn, key, ksz, val, rval, rsz, man_node(m)); } + static void -version(void) +pmdoc(DB *db, const char *dbn, DBT *key, size_t *ksz, + DBT *val, DBT *rval, size_t *rsz, struct mdoc *m) { - printf("%s %s\n", progname, VERSION); + pmdoc_node(db, dbn, key, ksz, val, rval, rsz, mdoc_node(m)); } static void @@ -661,7 +863,6 @@ usage(void) { fprintf(stderr, "usage: %s " - "[-V] " "[-d path] " "[file...]\n", progname);