-/* $Id: mandocdb.c,v 1.89 2013/12/27 20:35:51 schwarze Exp $ */
+/* $Id: mandocdb.c,v 1.96 2014/01/02 22:44:10 schwarze Exp $ */
/*
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
};
struct str {
- char *utf8; /* key in UTF-8 form */
+ char *rendered; /* key in UTF-8 or ASCII form */
const struct mpage *mpage; /* if set, the owning parse */
uint64_t mask; /* bitmask in sequence */
- char key[]; /* the string itself */
+ char key[]; /* may contain escape sequences */
};
struct inodev {
static void hash_free(void *, size_t, void *);
static void *hash_halloc(size_t, void *);
static void mlink_add(struct mlink *, const struct stat *);
+static int mlink_check(struct mpage *, struct mlink *);
static void mlink_free(struct mlink *);
static void mlinks_undupe(struct mpage *);
static void mpages_free(void);
static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_In(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_St(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
static void putkey(const struct mpage *,
const char *, uint64_t);
const char *, size_t, uint64_t);
static void putmdockey(const struct mpage *,
const struct mdoc_node *, uint64_t);
+static void render_key(struct mchars *, struct str *);
static void say(const char *, const char *, ...);
static int set_basedir(const char *);
static int treescan(void);
static size_t utf8(unsigned int, char [7]);
-static void utf8key(struct mchars *, struct str *);
static char *progname;
static int use_all; /* use all found files */
static int nodb; /* no database changes */
static int verb; /* print what we're doing */
static int warnings; /* warn about crap */
+static int write_utf8; /* write UTF-8 output; else ASCII */
static int exitcode; /* to be returned by main */
static enum op op; /* operational mode */
static char basedir[PATH_MAX]; /* current base directory */
{ parse_mdoc_Fn, 0 }, /* Fn */
{ NULL, TYPE_Ft }, /* Ft */
{ NULL, TYPE_Ic }, /* Ic */
- { parse_mdoc_In, TYPE_In }, /* In */
+ { NULL, TYPE_In }, /* In */
{ NULL, TYPE_Li }, /* Li */
{ parse_mdoc_Nd, TYPE_Nd }, /* Nd */
{ parse_mdoc_Nm, TYPE_Nm }, /* Nm */
{ NULL, 0 }, /* Ot */
{ NULL, TYPE_Pa }, /* Pa */
{ NULL, 0 }, /* Rv */
- { parse_mdoc_St, 0 }, /* St */
+ { NULL, TYPE_St }, /* St */
{ NULL, TYPE_Va }, /* Va */
{ parse_mdoc_body, TYPE_Va }, /* Vt */
{ parse_mdoc_Xr, 0 }, /* Xr */
path_arg = NULL;
op = OP_DEFAULT;
- while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW")))
+ while (-1 != (ch = getopt(argc, argv, "aC:d:nT:tu:vW")))
switch (ch) {
case ('a'):
use_all = 1;
case ('n'):
nodb = 1;
break;
+ case ('T'):
+ if (strcmp(optarg, "utf8")) {
+ fprintf(stderr, "-T%s: Unsupported "
+ "output format\n", optarg);
+ goto usage;
+ }
+ write_utf8 = 1;
+ break;
case ('t'):
CHECKOP(op, ch);
dup2(STDOUT_FILENO, STDERR_FILENO);
ohash_delete(&mlinks);
return(exitcode);
usage:
- fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
- " %s [-anvW] dir ...\n"
- " %s [-nvW] -d dir [file ...]\n"
+ fprintf(stderr, "usage: %s [-anvW] [-C file] [-Tutf8]\n"
+ " %s [-anvW] [-Tutf8] dir ...\n"
+ " %s [-nvW] [-Tutf8] -d dir [file ...]\n"
" %s [-nvW] -u dir [file ...]\n"
" %s -t file ...\n",
progname, progname, progname,
FTSENT *ff;
struct mlink *mlink;
int dform;
- char *fsec;
- const char *dsec, *arch, *cp, *path;
+ char *dsec, *arch, *fsec, *cp;
+ const char *path;
const char *argv[2];
argv[0] = ".";
continue;
} else
fsec[-1] = '\0';
+
mlink = mandoc_calloc(1, sizeof(struct mlink));
strlcpy(mlink->file, path, sizeof(mlink->file));
mlink->dform = dform;
- if (NULL != dsec)
- mlink->dsec = mandoc_strdup(dsec);
- if (NULL != arch)
- mlink->arch = mandoc_strdup(arch);
- mlink->name = mandoc_strdup(ff->fts_name);
- if (NULL != fsec)
- mlink->fsec = mandoc_strdup(fsec);
+ mlink->dsec = dsec;
+ mlink->arch = arch;
+ mlink->name = ff->fts_name;
+ mlink->fsec = fsec;
mlink_add(mlink, ff->fts_statp);
continue;
} else if (FTS_D != ff->fts_info &&
* Try to infer this from the name.
* If we're not in use_all, enforce it.
*/
- dsec = NULL;
- dform = FORM_NONE;
cp = ff->fts_name;
if (FTS_DP == ff->fts_info)
break;
} else if (0 == strncmp(cp, "cat", 3)) {
dform = FORM_CAT;
dsec = cp + 3;
+ } else {
+ dform = FORM_NONE;
+ dsec = NULL;
}
if (NULL != dsec || use_all)
* Possibly our architecture.
* If we're descending, keep tabs on it.
*/
- arch = NULL;
if (FTS_DP != ff->fts_info && NULL != dsec)
arch = ff->fts_name;
+ else
+ arch = NULL;
break;
default:
if (FTS_DP == ff->fts_info || use_all)
*p++ = '\0';
if (0 == strncmp(start, "man", 3)) {
mlink->dform = FORM_SRC;
- mlink->dsec = mandoc_strdup(start + 3);
+ mlink->dsec = start + 3;
} else if (0 == strncmp(start, "cat", 3)) {
mlink->dform = FORM_CAT;
- mlink->dsec = mandoc_strdup(start + 3);
+ mlink->dsec = start + 3;
}
start = p;
if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
*p++ = '\0';
- mlink->arch = mandoc_strdup(start);
+ mlink->arch = start;
start = p;
}
}
if ('.' == *p) {
*p++ = '\0';
- mlink->fsec = mandoc_strdup(p);
+ mlink->fsec = p;
}
/*
mlink->name = p + 1;
*p = '\0';
}
- mlink->name = mandoc_strdup(mlink->name);
-
mlink_add(mlink, &st);
}
assert(NULL != mlink->file);
- if (NULL == mlink->dsec)
- mlink->dsec = mandoc_strdup("");
- if (NULL == mlink->arch)
- mlink->arch = mandoc_strdup("");
- if (NULL == mlink->name)
- mlink->name = mandoc_strdup("");
- if (NULL == mlink->fsec)
- mlink->fsec = mandoc_strdup("");
+ mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
+ mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
+ mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
+ mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
if ('0' == *mlink->fsec) {
free(mlink->fsec);
char *bufp;
mpage->form = FORM_CAT;
- for(prev = &mpage->mlinks; *prev; prev = &(*prev)->next) {
- mlink = *prev;
+ prev = &mpage->mlinks;
+ while (NULL != (mlink = *prev)) {
if (FORM_CAT != mlink->dform) {
mpage->form = FORM_NONE;
- continue;
+ goto nextlink;
}
if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
if (warnings)
say(mlink->file, "Filename too long");
- continue;
+ goto nextlink;
}
bufp = strstr(buf, "cat");
assert(NULL != bufp);
strlcat(buf, mlink->dsec, PATH_MAX);
if (NULL == ohash_find(&mlinks,
ohash_qlookup(&mlinks, buf)))
- continue;
+ goto nextlink;
if (warnings)
say(mlink->file, "Man source exists: %s", buf);
if (use_all)
- continue;
+ goto nextlink;
*prev = mlink->next;
mlink_free(mlink);
- mlink = *prev;
+ continue;
+nextlink:
+ prev = &(*prev)->next;
+ }
+}
+
+static int
+mlink_check(struct mpage *mpage, struct mlink *mlink)
+{
+ int match;
+
+ match = 1;
+
+ /*
+ * Check whether the manual section given in a file
+ * agrees with the directory where the file is located.
+ * Some manuals have suffixes like (3p) on their
+ * section number either inside the file or in the
+ * directory name, some are linked into more than one
+ * section, like encrypt(1) = makekey(8).
+ */
+
+ if (FORM_SRC == mpage->form &&
+ strcasecmp(mpage->sec, mlink->dsec)) {
+ match = 0;
+ say(mlink->file, "Section \"%s\" manual in %s directory",
+ mpage->sec, mlink->dsec);
}
+
+ /*
+ * Manual page directories exist for each kernel
+ * architecture as returned by machine(1).
+ * However, many manuals only depend on the
+ * application architecture as returned by arch(1).
+ * For example, some (2/ARM) manuals are shared
+ * across the "armish" and "zaurus" kernel
+ * architectures.
+ * A few manuals are even shared across completely
+ * different architectures, for example fdformat(1)
+ * on amd64, i386, sparc, and sparc64.
+ */
+
+ if (strcasecmp(mpage->arch, mlink->arch)) {
+ match = 0;
+ say(mlink->file, "Architecture \"%s\" manual in "
+ "\"%s\" directory", mpage->arch, mlink->arch);
+ }
+
+ if (strcasecmp(mpage->title, mlink->name))
+ match = 0;
+
+ return(match);
}
/*
struct ohash title_table;
struct ohash_info title_info, str_info;
struct mpage *mpage;
+ struct mlink *mlink;
struct mdoc *mdoc;
struct man *man;
struct title *title_entry;
mparse_reset(mp);
mdoc = NULL;
man = NULL;
- match = 1;
/*
* Try interpreting the file as mdoc(7) or man(7)
mandoc_strdup(mpage->mlinks->name);
}
- /*
- * Check whether the manual section given in a file
- * agrees with the directory where the file is located.
- * Some manuals have suffixes like (3p) on their
- * section number either inside the file or in the
- * directory name, some are linked into more than one
- * section, like encrypt(1) = makekey(8). Do not skip
- * manuals for such reasons.
- */
- if (warnings && !use_all && FORM_SRC == mpage->form &&
- strcasecmp(mpage->sec, mpage->mlinks->dsec)) {
- match = 0;
- say(mpage->mlinks->file, "Section \"%s\" "
- "manual in %s directory",
- mpage->sec, mpage->mlinks->dsec);
- }
+ for (mlink = mpage->mlinks; mlink; mlink = mlink->next)
+ putkey(mpage, mlink->name, TYPE_Nm);
- /*
- * Manual page directories exist for each kernel
- * architecture as returned by machine(1).
- * However, many manuals only depend on the
- * application architecture as returned by arch(1).
- * For example, some (2/ARM) manuals are shared
- * across the "armish" and "zaurus" kernel
- * architectures.
- * A few manuals are even shared across completely
- * different architectures, for example fdformat(1)
- * on amd64, i386, sparc, and sparc64.
- * Thus, warn about architecture mismatches,
- * but don't skip manuals for this reason.
- */
- if (warnings && !use_all &&
- strcasecmp(mpage->arch, mpage->mlinks->arch)) {
+ if (warnings && !use_all) {
match = 0;
- say(mpage->mlinks->file, "Architecture \"%s\" "
- "manual in \"%s\" directory",
- mpage->arch, mpage->mlinks->arch);
- }
- if (warnings && !use_all &&
- strcasecmp(mpage->title, mpage->mlinks->name))
- match = 0;
-
- putkey(mpage, mpage->mlinks->name, TYPE_Nm);
+ for (mlink = mpage->mlinks; mlink;
+ mlink = mlink->next)
+ if (mlink_check(mpage, mlink))
+ match = 1;
+ } else
+ match = 1;
if (NULL != mdoc) {
if (NULL != (cp = mdoc_meta(mdoc)->name))
if (end > start)
putkeys(mpage, start, end - start + 1, TYPE_In);
- return(1);
-}
-
-static int
-parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL != n->child && MDOC_TEXT == n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_In);
- return(1);
+ return(0);
}
static int
return(0);
}
-static int
-parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL == n->child || MDOC_TEXT != n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_St);
- return(1);
-}
-
static int
parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
{
parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
{
- if (SEC_NAME == n->sec)
- return(1);
- else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
- return(0);
-
- return(1);
+ return(SEC_NAME == n->sec ||
+ (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
}
static int
}
/*
- * Store the UTF-8 version of a key, or alias the pointer if the key has
- * no UTF-8 transcription marks in it.
+ * Store the rendered version of a key, or alias the pointer
+ * if the key contains no escape sequences.
*/
static void
-utf8key(struct mchars *mc, struct str *key)
+render_key(struct mchars *mc, struct str *key)
{
size_t sz, bsz, pos;
char utfbuf[7], res[5];
int len, u;
enum mandoc_esc esc;
- assert(NULL == key->utf8);
+ assert(NULL == key->rendered);
res[0] = '\\';
res[1] = '\t';
* pointer as ourselvse and get out of here.
*/
if (strcspn(val, res) == bsz) {
- key->utf8 = key->key;
+ key->rendered = key->key;
return;
}
/* Read past the slash. */
val++;
- u = 0;
/*
* Parse the escape sequence and see if it's a
* predefined character or special character.
*/
+
esc = mandoc_escape
((const char **)&val, &seq, &len);
if (ESCAPE_ERROR == esc)
break;
-
if (ESCAPE_SPECIAL != esc)
continue;
- if (0 == (u = mchars_spec2cp(mc, seq, len)))
- continue;
/*
- * If we have a Unicode codepoint, try to convert that
- * to a UTF-8 byte string.
+ * Render the special character
+ * as either UTF-8 or ASCII.
*/
- cpp = utfbuf;
- if (0 == (sz = utf8(u, utfbuf)))
- continue;
+
+ if (write_utf8) {
+ if (0 == (u = mchars_spec2cp(mc, seq, len)))
+ continue;
+ cpp = utfbuf;
+ if (0 == (sz = utf8(u, utfbuf)))
+ continue;
+ sz = strlen(cpp);
+ } else {
+ cpp = mchars_spec2str(mc, seq, len, &sz);
+ if (NULL == cpp)
+ continue;
+ if (ASCII_NBRSP == *cpp) {
+ cpp = " ";
+ sz = 1;
+ }
+ }
/* Copy the rendered glyph into the stream. */
- sz = strlen(cpp);
bsz += sz;
-
buf = mandoc_realloc(buf, bsz);
-
memcpy(&buf[pos], cpp, sz);
pos += sz;
}
buf[pos] = '\0';
- key->utf8 = buf;
+ key->rendered = buf;
}
/*
* Flush the current page's terms (and their bits) into the database.
* Wrap the entire set of additions in a transaction to make sqlite be a
* little faster.
- * Also, UTF-8-encode the description at the last possible moment.
+ * Also, handle escape sequences at the last possible moment.
*/
static void
dbindex(const struct mpage *mpage, struct mchars *mc)
key = ohash_find(&strings,
ohash_qlookup(&strings, mpage->desc));
assert(NULL != key);
- if (NULL == key->utf8)
- utf8key(mc, key);
- desc = key->utf8;
+ if (NULL == key->rendered)
+ render_key(mc, key);
+ desc = key->rendered;
}
SQL_EXEC("BEGIN TRANSACTION");
for (key = ohash_first(&strings, &slot); NULL != key;
key = ohash_next(&strings, &slot)) {
assert(key->mpage == mpage);
- if (NULL == key->utf8)
- utf8key(mc, key);
+ if (NULL == key->rendered)
+ render_key(mc, key);
i = 1;
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
- SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->utf8);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
SQL_STEP(stmts[STMT_INSERT_KEY]);
sqlite3_reset(stmts[STMT_INSERT_KEY]);
- if (key->utf8 != key->key)
- free(key->utf8);
+ if (key->rendered != key->key)
+ free(key->rendered);
free(key);
}
git.ckatri.com / mandoc.git / blobdiff