+#include "mandoc.h"
+#include "mandoc_aux.h"
+#include "libroff.h"
+#include "libmandoc.h"
+
+/* Maximum number of nested if-else conditionals. */
+#define RSTACK_MAX 128
+
+/* Maximum number of string expansions per line, to break infinite loops. */
+#define EXPAND_LIMIT 1000
+
+enum rofft {
+ ROFF_ad,
+ ROFF_am,
+ ROFF_ami,
+ ROFF_am1,
+ ROFF_as,
+ ROFF_cc,
+ ROFF_ce,
+ ROFF_de,
+ ROFF_dei,
+ ROFF_de1,
+ ROFF_ds,
+ ROFF_el,
+ ROFF_fam,
+ ROFF_hw,
+ ROFF_hy,
+ ROFF_ie,
+ ROFF_if,
+ ROFF_ig,
+ ROFF_it,
+ ROFF_ne,
+ ROFF_nh,
+ ROFF_nr,
+ ROFF_ns,
+ ROFF_ps,
+ ROFF_rm,
+ ROFF_rr,
+ ROFF_so,
+ ROFF_ta,
+ ROFF_tr,
+ ROFF_Dd,
+ ROFF_TH,
+ ROFF_TS,
+ ROFF_TE,
+ ROFF_T_,
+ ROFF_EQ,
+ ROFF_EN,
+ ROFF_cblock,
+ ROFF_USERDEF,
+ ROFF_MAX
+};
+
+/*
+ * An incredibly-simple string buffer.
+ */
+struct roffstr {
+ char *p; /* nil-terminated buffer */
+ size_t sz; /* saved strlen(p) */
+};
+
+/*
+ * A key-value roffstr pair as part of a singly-linked list.
+ */
+struct roffkv {
+ struct roffstr key;
+ struct roffstr val;
+ struct roffkv *next; /* next in list */
+};
+
+/*
+ * A single number register as part of a singly-linked list.
+ */
+struct roffreg {
+ struct roffstr key;
+ int val;
+ struct roffreg *next;
+};
+
+struct roff {
+ struct mparse *parse; /* parse point */
+ int options; /* parse options */
+ struct roffnode *last; /* leaf of stack */
+ int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
+ char control; /* control character */
+ int rstackpos; /* position in rstack */
+ struct roffreg *regtab; /* number registers */
+ struct roffkv *strtab; /* user-defined strings & macros */
+ struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
+ struct roffstr *xtab; /* single-byte trans table (`tr') */
+ const char *current_string; /* value of last called user macro */
+ struct tbl_node *first_tbl; /* first table parsed */
+ struct tbl_node *last_tbl; /* last table parsed */
+ struct tbl_node *tbl; /* current table being parsed */
+ struct eqn_node *last_eqn; /* last equation parsed */
+ struct eqn_node *first_eqn; /* first equation parsed */
+ struct eqn_node *eqn; /* current equation being parsed */
+};
+
+struct roffnode {
+ enum rofft tok; /* type of node */
+ struct roffnode *parent; /* up one in stack */
+ int line; /* parse line */
+ int col; /* parse col */
+ char *name; /* node name, e.g. macro name */
+ char *end; /* end-rules: custom token */
+ int endspan; /* end-rules: next-line or infty */
+ int rule; /* current evaluation rule */
+};
+
+#define ROFF_ARGS struct roff *r, /* parse ctx */ \
+ enum rofft tok, /* tok of macro */ \
+ char **bufp, /* input buffer */ \
+ size_t *szp, /* size of input buffer */ \
+ int ln, /* parse line */ \
+ int ppos, /* original pos in buffer */ \
+ int pos, /* current pos in buffer */ \
+ int *offs /* reset offset of buffer data */
+
+typedef enum rofferr (*roffproc)(ROFF_ARGS);
+
+struct roffmac {
+ const char *name; /* macro name */
+ roffproc proc; /* process new macro */
+ roffproc text; /* process as child text of macro */
+ roffproc sub; /* process as child of macro */
+ int flags;
+#define ROFFMAC_STRUCT (1 << 0) /* always interpret */
+ struct roffmac *next;
+};
+
+struct predef {
+ const char *name; /* predefined input name */
+ const char *str; /* replacement symbol */
+};
+
+#define PREDEF(__name, __str) \
+ { (__name), (__str) },
+
+static enum rofft roffhash_find(const char *, size_t);
+static void roffhash_init(void);
+static void roffnode_cleanscope(struct roff *);
+static void roffnode_pop(struct roff *);
+static void roffnode_push(struct roff *, enum rofft,
+ const char *, int, int);
+static enum rofferr roff_block(ROFF_ARGS);
+static enum rofferr roff_block_text(ROFF_ARGS);
+static enum rofferr roff_block_sub(ROFF_ARGS);
+static enum rofferr roff_cblock(ROFF_ARGS);
+static enum rofferr roff_cc(ROFF_ARGS);
+static void roff_ccond(struct roff *, int, int);
+static enum rofferr roff_cond(ROFF_ARGS);
+static enum rofferr roff_cond_text(ROFF_ARGS);
+static enum rofferr roff_cond_sub(ROFF_ARGS);
+static enum rofferr roff_ds(ROFF_ARGS);
+static int roff_evalcond(const char *, int *);
+static int roff_evalnum(const char *, int *, int *, int);
+static int roff_evalpar(const char *, int *, int *);
+static int roff_evalstrcond(const char *, int *);
+static void roff_free1(struct roff *);
+static void roff_freereg(struct roffreg *);
+static void roff_freestr(struct roffkv *);
+static size_t roff_getname(struct roff *, char **, int, int);
+static int roff_getnum(const char *, int *, int *);
+static int roff_getop(const char *, int *, char *);
+static int roff_getregn(const struct roff *,
+ const char *, size_t);
+static int roff_getregro(const char *name);
+static const char *roff_getstrn(const struct roff *,
+ const char *, size_t);
+static enum rofferr roff_it(ROFF_ARGS);
+static enum rofferr roff_line_ignore(ROFF_ARGS);
+static enum rofferr roff_nr(ROFF_ARGS);
+static void roff_openeqn(struct roff *, const char *,
+ int, int, const char *);
+static enum rofft roff_parse(struct roff *, const char *, int *);
+static enum rofferr roff_parsetext(char **, size_t *, int, int *);
+static enum rofferr roff_res(struct roff *,
+ char **, size_t *, int, int);
+static enum rofferr roff_rm(ROFF_ARGS);
+static enum rofferr roff_rr(ROFF_ARGS);
+static void roff_setstr(struct roff *,
+ const char *, const char *, int);
+static void roff_setstrn(struct roffkv **, const char *,
+ size_t, const char *, size_t, int);
+static enum rofferr roff_so(ROFF_ARGS);
+static enum rofferr roff_tr(ROFF_ARGS);
+static enum rofferr roff_Dd(ROFF_ARGS);
+static enum rofferr roff_TH(ROFF_ARGS);
+static enum rofferr roff_TE(ROFF_ARGS);
+static enum rofferr roff_TS(ROFF_ARGS);
+static enum rofferr roff_EQ(ROFF_ARGS);
+static enum rofferr roff_EN(ROFF_ARGS);
+static enum rofferr roff_T_(ROFF_ARGS);
+static enum rofferr roff_userdef(ROFF_ARGS);
+
+/* See roffhash_find() */
+
+#define ASCII_HI 126
+#define ASCII_LO 33
+#define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
+
+static struct roffmac *hash[HASHWIDTH];
+
+static struct roffmac roffs[ROFF_MAX] = {
+ { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "as", roff_ds, NULL, NULL, 0, NULL },
+ { "cc", roff_cc, NULL, NULL, 0, NULL },
+ { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "ds", roff_ds, NULL, NULL, 0, NULL },
+ { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
+ { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
+ { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
+ { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
+ { "it", roff_it, NULL, NULL, 0, NULL },
+ { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "nr", roff_nr, NULL, NULL, 0, NULL },
+ { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "rm", roff_rm, NULL, NULL, 0, NULL },
+ { "rr", roff_rr, NULL, NULL, 0, NULL },
+ { "so", roff_so, NULL, NULL, 0, NULL },
+ { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
+ { "tr", roff_tr, NULL, NULL, 0, NULL },
+ { "Dd", roff_Dd, NULL, NULL, 0, NULL },
+ { "TH", roff_TH, NULL, NULL, 0, NULL },
+ { "TS", roff_TS, NULL, NULL, 0, NULL },
+ { "TE", roff_TE, NULL, NULL, 0, NULL },
+ { "T&", roff_T_, NULL, NULL, 0, NULL },
+ { "EQ", roff_EQ, NULL, NULL, 0, NULL },
+ { "EN", roff_EN, NULL, NULL, 0, NULL },
+ { ".", roff_cblock, NULL, NULL, 0, NULL },
+ { NULL, roff_userdef, NULL, NULL, 0, NULL },
+};
+
+/* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
+const char *const __mdoc_reserved[] = {
+ "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
+ "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
+ "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
+ "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
+ "Dt", "Dv", "Dx", "D1",
+ "Ec", "Ed", "Ef", "Ek", "El", "Em",
+ "En", "Eo", "Er", "Es", "Ev", "Ex",
+ "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
+ "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
+ "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
+ "Oc", "Oo", "Op", "Os", "Ot", "Ox",
+ "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
+ "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
+ "Sc", "Sh", "Sm", "So", "Sq",
+ "Ss", "St", "Sx", "Sy",
+ "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
+ "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
+ "%P", "%Q", "%R", "%T", "%U", "%V",
+ NULL
+};
+
+/* not currently implemented: BT DE DS ME MT PT SY TQ YS */
+const char *const __man_reserved[] = {
+ "AT", "B", "BI", "BR", "DT",
+ "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
+ "LP", "OP", "P", "PD", "PP",
+ "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
+ "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
+ NULL
+};
+
+/* Array of injected predefined strings. */
+#define PREDEFS_MAX 38
+static const struct predef predefs[PREDEFS_MAX] = {
+#include "predefs.in"
+};
+
+/* See roffhash_find() */
+#define ROFF_HASH(p) (p[0] - ASCII_LO)
+
+static int roffit_lines; /* number of lines to delay */
+static char *roffit_macro; /* nil-terminated macro line */
+
+
+static void
+roffhash_init(void)
+{
+ struct roffmac *n;
+ int buc, i;
+
+ for (i = 0; i < (int)ROFF_USERDEF; i++) {
+ assert(roffs[i].name[0] >= ASCII_LO);
+ assert(roffs[i].name[0] <= ASCII_HI);
+
+ buc = ROFF_HASH(roffs[i].name);
+
+ if (NULL != (n = hash[buc])) {
+ for ( ; n->next; n = n->next)
+ /* Do nothing. */ ;
+ n->next = &roffs[i];
+ } else
+ hash[buc] = &roffs[i];
+ }
+}
+
+/*
+ * Look up a roff token by its name. Returns ROFF_MAX if no macro by
+ * the nil-terminated string name could be found.
+ */
+static enum rofft
+roffhash_find(const char *p, size_t s)
+{
+ int buc;
+ struct roffmac *n;
+
+ /*
+ * libroff has an extremely simple hashtable, for the time
+ * being, which simply keys on the first character, which must
+ * be printable, then walks a chain. It works well enough until
+ * optimised.
+ */
+
+ if (p[0] < ASCII_LO || p[0] > ASCII_HI)
+ return(ROFF_MAX);
+
+ buc = ROFF_HASH(p);
+
+ if (NULL == (n = hash[buc]))
+ return(ROFF_MAX);
+ for ( ; n; n = n->next)
+ if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
+ return((enum rofft)(n - roffs));
+
+ return(ROFF_MAX);
+}
+
+/*
+ * Pop the current node off of the stack of roff instructions currently
+ * pending.
+ */
+static void
+roffnode_pop(struct roff *r)
+{
+ struct roffnode *p;
+
+ assert(r->last);
+ p = r->last;
+
+ r->last = r->last->parent;
+ free(p->name);
+ free(p->end);
+ free(p);
+}
+
+/*
+ * Push a roff node onto the instruction stack. This must later be
+ * removed with roffnode_pop().
+ */
+static void
+roffnode_push(struct roff *r, enum rofft tok, const char *name,
+ int line, int col)
+{
+ struct roffnode *p;
+
+ p = mandoc_calloc(1, sizeof(struct roffnode));
+ p->tok = tok;
+ if (name)
+ p->name = mandoc_strdup(name);
+ p->parent = r->last;
+ p->line = line;
+ p->col = col;
+ p->rule = p->parent ? p->parent->rule : 0;
+
+ r->last = p;
+}
+
+static void
+roff_free1(struct roff *r)
+{
+ struct tbl_node *tbl;
+ struct eqn_node *e;
+ int i;
+
+ while (NULL != (tbl = r->first_tbl)) {
+ r->first_tbl = tbl->next;
+ tbl_free(tbl);
+ }
+
+ r->first_tbl = r->last_tbl = r->tbl = NULL;
+
+ while (NULL != (e = r->first_eqn)) {
+ r->first_eqn = e->next;
+ eqn_free(e);
+ }
+
+ r->first_eqn = r->last_eqn = r->eqn = NULL;
+
+ while (r->last)
+ roffnode_pop(r);
+
+ roff_freestr(r->strtab);
+ roff_freestr(r->xmbtab);
+
+ r->strtab = r->xmbtab = NULL;
+
+ roff_freereg(r->regtab);
+
+ r->regtab = NULL;
+
+ if (r->xtab)
+ for (i = 0; i < 128; i++)
+ free(r->xtab[i].p);
+
+ free(r->xtab);
+ r->xtab = NULL;
+}
+
+void
+roff_reset(struct roff *r)
+{
+
+ roff_free1(r);
+ r->control = 0;
+}
+
+void
+roff_free(struct roff *r)
+{
+
+ roff_free1(r);
+ free(r);
+}
+
+struct roff *
+roff_alloc(struct mparse *parse, int options)
+{
+ struct roff *r;
+
+ r = mandoc_calloc(1, sizeof(struct roff));
+ r->parse = parse;
+ r->options = options;
+ r->rstackpos = -1;
+
+ roffhash_init();
+
+ return(r);
+}
+
+/*
+ * In the current line, expand escape sequences that tend to get
+ * used in numerical expressions and conditional requests.
+ * Also check the syntax of the remaining escape sequences.
+ */
+static enum rofferr
+roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
+{
+ char ubuf[24]; /* buffer to print the number */
+ const char *start; /* start of the string to process */
+ char *stesc; /* start of an escape sequence ('\\') */
+ const char *stnam; /* start of the name, after "[(*" */
+ const char *cp; /* end of the name, e.g. before ']' */
+ const char *res; /* the string to be substituted */
+ char *nbuf; /* new buffer to copy bufp to */
+ size_t maxl; /* expected length of the escape name */
+ size_t naml; /* actual length of the escape name */
+ int expand_count; /* to avoid infinite loops */
+ int npos; /* position in numeric expression */
+ int irc; /* return code from roff_evalnum() */
+ char term; /* character terminating the escape */
+
+ expand_count = 0;
+ start = *bufp + pos;
+ stesc = strchr(start, '\0') - 1;
+ while (stesc-- > start) {
+
+ /* Search backwards for the next backslash. */
+
+ if ('\\' != *stesc)
+ continue;
+
+ /* If it is escaped, skip it. */
+
+ for (cp = stesc - 1; cp >= start; cp--)
+ if ('\\' != *cp)
+ break;
+
+ if (0 == (stesc - cp) % 2) {
+ stesc = (char *)cp;
+ continue;
+ }
+
+ /* Decide whether to expand or to check only. */
+
+ term = '\0';
+ cp = stesc + 1;
+ switch (*cp) {
+ case '*':
+ res = NULL;
+ break;
+ case 'B':
+ /* FALLTHROUGH */
+ case 'w':
+ term = cp[1];
+ /* FALLTHROUGH */
+ case 'n':
+ res = ubuf;
+ break;
+ default:
+ if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
+ mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
+ ln, (int)(stesc - *bufp), NULL);
+ continue;
+ }
+
+ if (EXPAND_LIMIT < ++expand_count) {
+ mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
+ ln, (int)(stesc - *bufp), NULL);
+ return(ROFF_IGN);
+ }
+
+ /*
+ * The third character decides the length
+ * of the name of the string or register.
+ * Save a pointer to the name.
+ */
+
+ if ('\0' == term) {
+ switch (*++cp) {
+ case '\0':
+ maxl = 0;
+ break;
+ case '(':
+ cp++;
+ maxl = 2;
+ break;
+ case '[':
+ cp++;
+ term = ']';
+ maxl = 0;
+ break;
+ default:
+ maxl = 1;
+ break;
+ }
+ } else {
+ cp += 2;
+ maxl = 0;
+ }
+ stnam = cp;
+
+ /* Advance to the end of the name. */
+
+ for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
+ if ('\0' == *cp) {
+ mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
+ ln, (int)(stesc - *bufp), NULL);
+ break;
+ }
+ if (0 == maxl && *cp == term) {
+ cp++;
+ break;
+ }
+ }
+
+ /*
+ * Retrieve the replacement string; if it is
+ * undefined, resume searching for escapes.
+ */
+
+ switch (stesc[1]) {
+ case '*':
+ res = roff_getstrn(r, stnam, naml);
+ break;
+ case 'B':
+ npos = 0;
+ irc = roff_evalnum(stnam, &npos, NULL, 0);
+ ubuf[0] = irc && stnam + npos + 1 == cp
+ ? '1' : '0';
+ ubuf[1] = '\0';
+ break;
+ case 'n':
+ (void)snprintf(ubuf, sizeof(ubuf), "%d",
+ roff_getregn(r, stnam, naml));
+ break;
+ case 'w':
+ (void)snprintf(ubuf, sizeof(ubuf), "%d",
+ 24 * (int)naml);
+ break;
+ }
+
+ if (NULL == res) {
+ mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
+ ln, (int)(stesc - *bufp), NULL);
+ res = "";
+ }
+
+ /* Replace the escape sequence by the string. */
+
+ *stesc = '\0';
+ *szp = mandoc_asprintf(&nbuf, "%s%s%s",
+ *bufp, res, cp) + 1;
+
+ /* Prepare for the next replacement. */
+
+ start = nbuf + pos;
+ stesc = nbuf + (stesc - *bufp) + strlen(res);
+ free(*bufp);
+ *bufp = nbuf;
+ }
+ return(ROFF_CONT);
+}
+
+/*
+ * Process text streams:
+ * Convert all breakable hyphens into ASCII_HYPH.
+ * Decrement and spring input line trap.
+ */
+static enum rofferr
+roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
+{
+ size_t sz;
+ const char *start;
+ char *p;
+ int isz;
+ enum mandoc_esc esc;
+
+ start = p = *bufp + pos;
+
+ while ('\0' != *p) {
+ sz = strcspn(p, "-\\");
+ p += sz;
+
+ if ('\0' == *p)
+ break;
+
+ if ('\\' == *p) {
+ /* Skip over escapes. */
+ p++;
+ esc = mandoc_escape((const char **)&p, NULL, NULL);
+ if (ESCAPE_ERROR == esc)
+ break;
+ continue;
+ } else if (p == start) {
+ p++;
+ continue;
+ }
+
+ if (isalpha((unsigned char)p[-1]) &&
+ isalpha((unsigned char)p[1]))
+ *p = ASCII_HYPH;
+ p++;
+ }
+
+ /* Spring the input line trap. */
+ if (1 == roffit_lines) {
+ isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
+ free(*bufp);
+ *bufp = p;
+ *szp = isz + 1;
+ *offs = 0;
+ free(roffit_macro);
+ roffit_lines = 0;
+ return(ROFF_REPARSE);
+ } else if (1 < roffit_lines)
+ --roffit_lines;
+ return(ROFF_CONT);
+}
+
+enum rofferr
+roff_parseln(struct roff *r, int ln, char **bufp,
+ size_t *szp, int pos, int *offs)
+{
+ enum rofft t;
+ enum rofferr e;
+ int ppos, ctl;
+
+ /*
+ * Run the reserved-word filter only if we have some reserved
+ * words to fill in.
+ */
+
+ e = roff_res(r, bufp, szp, ln, pos);
+ if (ROFF_IGN == e)
+ return(e);
+ assert(ROFF_CONT == e);
+
+ ppos = pos;
+ ctl = roff_getcontrol(r, *bufp, &pos);
+
+ /*
+ * First, if a scope is open and we're not a macro, pass the
+ * text through the macro's filter. If a scope isn't open and
+ * we're not a macro, just let it through.
+ * Finally, if there's an equation scope open, divert it into it
+ * no matter our state.
+ */
+
+ if (r->last && ! ctl) {
+ t = r->last->tok;
+ assert(roffs[t].text);
+ e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
+ assert(ROFF_IGN == e || ROFF_CONT == e);
+ if (ROFF_CONT != e)
+ return(e);
+ }
+ if (r->eqn)
+ return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
+ if ( ! ctl) {
+ if (r->tbl)
+ return(tbl_read(r->tbl, ln, *bufp, pos));
+ return(roff_parsetext(bufp, szp, pos, offs));
+ }
+
+ /*
+ * If a scope is open, go to the child handler for that macro,
+ * as it may want to preprocess before doing anything with it.
+ * Don't do so if an equation is open.
+ */
+
+ if (r->last) {
+ t = r->last->tok;
+ assert(roffs[t].sub);
+ return((*roffs[t].sub)(r, t, bufp, szp,
+ ln, ppos, pos, offs));
+ }
+
+ /*
+ * Lastly, as we've no scope open, try to look up and execute
+ * the new macro. If no macro is found, simply return and let
+ * the compilers handle it.
+ */
+
+ if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
+ return(ROFF_CONT);
+
+ assert(roffs[t].proc);
+ return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
+}
+
+void
+roff_endparse(struct roff *r)
+{
+
+ if (r->last)
+ mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
+ r->last->line, r->last->col, NULL);
+
+ if (r->eqn) {
+ mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
+ r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
+ eqn_end(&r->eqn);
+ }
+
+ if (r->tbl) {
+ mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
+ r->tbl->line, r->tbl->pos, NULL);
+ tbl_end(&r->tbl);
+ }
+}
+
+/*
+ * Parse a roff node's type from the input buffer. This must be in the
+ * form of ".foo xxx" in the usual way.
+ */
+static enum rofft
+roff_parse(struct roff *r, const char *buf, int *pos)
+{
+ const char *mac;
+ size_t maclen;
+ enum rofft t;
+
+ if ('\0' == buf[*pos] || '"' == buf[*pos] ||
+ '\t' == buf[*pos] || ' ' == buf[*pos])
+ return(ROFF_MAX);
+
+ /* We stop the macro parse at an escape, tab, space, or nil. */
+
+ mac = buf + *pos;
+ maclen = strcspn(mac, " \\\t\0");
+
+ t = (r->current_string = roff_getstrn(r, mac, maclen))
+ ? ROFF_USERDEF : roffhash_find(mac, maclen);
+
+ *pos += (int)maclen;
+
+ while (buf[*pos] && ' ' == buf[*pos])
+ (*pos)++;
+
+ return(t);
+}
+
+static enum rofferr
+roff_cblock(ROFF_ARGS)
+{
+
+ /*
+ * A block-close `..' should only be invoked as a child of an
+ * ignore macro, otherwise raise a warning and just ignore it.
+ */
+
+ if (NULL == r->last) {
+ mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
+ return(ROFF_IGN);
+ }
+
+ switch (r->last->tok) {
+ case ROFF_am:
+ /* FALLTHROUGH */
+ case ROFF_ami:
+ /* FALLTHROUGH */
+ case ROFF_am1:
+ /* FALLTHROUGH */
+ case ROFF_de:
+ /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
+ /* FALLTHROUGH */
+ case ROFF_dei:
+ /* FALLTHROUGH */
+ case ROFF_ig:
+ break;
+ default:
+ mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
+ return(ROFF_IGN);
+ }
+
+ if ((*bufp)[pos])
+ mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
+
+ roffnode_pop(r);
+ roffnode_cleanscope(r);
+ return(ROFF_IGN);
+
+}
+
+static void
+roffnode_cleanscope(struct roff *r)
+{
+
+ while (r->last) {
+ if (--r->last->endspan != 0)
+ break;
+ roffnode_pop(r);
+ }
+}
+
+static void
+roff_ccond(struct roff *r, int ln, int ppos)
+{
+
+ if (NULL == r->last) {
+ mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
+ return;
+ }
+
+ switch (r->last->tok) {
+ case ROFF_el:
+ /* FALLTHROUGH */
+ case ROFF_ie:
+ /* FALLTHROUGH */
+ case ROFF_if:
+ break;
+ default:
+ mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
+ return;
+ }
+
+ if (r->last->endspan > -1) {
+ mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
+ return;
+ }
+
+ roffnode_pop(r);
+ roffnode_cleanscope(r);
+ return;
+}
+
+static enum rofferr
+roff_block(ROFF_ARGS)
+{
+ int sv;
+ size_t sz;
+ char *name;
+
+ name = NULL;
+
+ if (ROFF_ig != tok) {
+ if ('\0' == (*bufp)[pos]) {
+ mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
+ return(ROFF_IGN);
+ }
+
+ /*
+ * Re-write `de1', since we don't really care about
+ * groff's strange compatibility mode, into `de'.
+ */
+
+ if (ROFF_de1 == tok)
+ tok = ROFF_de;
+ if (ROFF_de == tok)
+ name = *bufp + pos;
+ else
+ mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
+ roffs[tok].name);
+
+ while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
+ pos++;
+
+ while (isspace((unsigned char)(*bufp)[pos]))
+ (*bufp)[pos++] = '\0';
+ }
+
+ roffnode_push(r, tok, name, ln, ppos);
+
+ /*
+ * At the beginning of a `de' macro, clear the existing string
+ * with the same name, if there is one. New content will be
+ * appended from roff_block_text() in multiline mode.
+ */
+
+ if (ROFF_de == tok)
+ roff_setstr(r, name, "", 0);
+
+ if ('\0' == (*bufp)[pos])
+ return(ROFF_IGN);
+
+ /* If present, process the custom end-of-line marker. */
+
+ sv = pos;
+ while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
+ pos++;
+
+ /*
+ * Note: groff does NOT like escape characters in the input.
+ * Instead of detecting this, we're just going to let it fly and
+ * to hell with it.
+ */
+
+ assert(pos > sv);
+ sz = (size_t)(pos - sv);
+
+ if (1 == sz && '.' == (*bufp)[sv])
+ return(ROFF_IGN);
+
+ r->last->end = mandoc_malloc(sz + 1);
+
+ memcpy(r->last->end, *bufp + sv, sz);
+ r->last->end[(int)sz] = '\0';
+
+ if ((*bufp)[pos])
+ mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
+
+ return(ROFF_IGN);
+}
+
+static enum rofferr
+roff_block_sub(ROFF_ARGS)
+{
+ enum rofft t;
+ int i, j;
+
+ /*
+ * First check whether a custom macro exists at this level. If
+ * it does, then check against it. This is some of groff's
+ * stranger behaviours. If we encountered a custom end-scope
+ * tag and that tag also happens to be a "real" macro, then we
+ * need to try interpreting it again as a real macro. If it's
+ * not, then return ignore. Else continue.
+ */
+
+ if (r->last->end) {
+ for (i = pos, j = 0; r->last->end[j]; j++, i++)
+ if ((*bufp)[i] != r->last->end[j])
+ break;
+
+ if ('\0' == r->last->end[j] &&
+ ('\0' == (*bufp)[i] ||
+ ' ' == (*bufp)[i] ||
+ '\t' == (*bufp)[i])) {
+ roffnode_pop(r);
+ roffnode_cleanscope(r);
+
+ while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
+ i++;
+
+ pos = i;
+ if (ROFF_MAX != roff_parse(r, *bufp, &pos))
+ return(ROFF_RERUN);
+ return(ROFF_IGN);
+ }
+ }
+
+ /*
+ * If we have no custom end-query or lookup failed, then try
+ * pulling it out of the hashtable.
+ */
+
+ t = roff_parse(r, *bufp, &pos);
+
+ /*
+ * Macros other than block-end are only significant
+ * in `de' blocks; elsewhere, simply throw them away.
+ */
+ if (ROFF_cblock != t) {
+ if (ROFF_de == tok)
+ roff_setstr(r, r->last->name, *bufp + ppos, 2);
+ return(ROFF_IGN);
+ }
+
+ assert(roffs[t].proc);
+ return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
+}
+
+static enum rofferr
+roff_block_text(ROFF_ARGS)
+{
+
+ if (ROFF_de == tok)
+ roff_setstr(r, r->last->name, *bufp + pos, 2);
+
+ return(ROFF_IGN);
+}
+
+static enum rofferr
+roff_cond_sub(ROFF_ARGS)
+{
+ enum rofft t;
+ char *ep;
+ int rr;
+
+ rr = r->last->rule;
+ roffnode_cleanscope(r);
+ t = roff_parse(r, *bufp, &pos);
+
+ /*
+ * Fully handle known macros when they are structurally
+ * required or when the conditional evaluated to true.
+ */
+
+ if ((ROFF_MAX != t) &&
+ (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
+ assert(roffs[t].proc);
+ return((*roffs[t].proc)(r, t, bufp, szp,
+ ln, ppos, pos, offs));
+ }
+
+ /*
+ * If `\}' occurs on a macro line without a preceding macro,
+ * drop the line completely.
+ */
+
+ ep = *bufp + pos;
+ if ('\\' == ep[0] && '}' == ep[1])
+ rr = 0;
+
+ /* Always check for the closing delimiter `\}'. */
+
+ while (NULL != (ep = strchr(ep, '\\'))) {
+ if ('}' == *(++ep)) {
+ *ep = '&';
+ roff_ccond(r, ln, ep - *bufp - 1);
+ }
+ ++ep;
+ }
+ return(rr ? ROFF_CONT : ROFF_IGN);
+}
+
+static enum rofferr
+roff_cond_text(ROFF_ARGS)
+{
+ char *ep;
+ int rr;
+
+ rr = r->last->rule;
+ roffnode_cleanscope(r);
+
+ ep = *bufp + pos;
+ while (NULL != (ep = strchr(ep, '\\'))) {
+ if ('}' == *(++ep)) {
+ *ep = '&';
+ roff_ccond(r, ln, ep - *bufp - 1);
+ }
+ ++ep;
+ }
+ return(rr ? ROFF_CONT : ROFF_IGN);
+}
+
+/*
+ * Parse a single signed integer number. Stop at the first non-digit.
+ * If there is at least one digit, return success and advance the
+ * parse point, else return failure and let the parse point unchanged.
+ * Ignore overflows, treat them just like the C language.
+ */
+static int
+roff_getnum(const char *v, int *pos, int *res)
+{
+ int myres, n, p;
+
+ if (NULL == res)
+ res = &myres;
+
+ p = *pos;
+ n = v[p] == '-';
+ if (n)
+ p++;
+
+ for (*res = 0; isdigit((unsigned char)v[p]); p++)
+ *res = 10 * *res + v[p] - '0';
+ if (p == *pos + n)
+ return 0;
+
+ if (n)
+ *res = -*res;
+
+ *pos = p;
+ return 1;
+}
+
+/*
+ * Evaluate a string comparison condition.
+ * The first character is the delimiter.
+ * Succeed if the string up to its second occurrence
+ * matches the string up to its third occurence.
+ * Advance the cursor after the third occurrence
+ * or lacking that, to the end of the line.
+ */
+static int
+roff_evalstrcond(const char *v, int *pos)
+{
+ const char *s1, *s2, *s3;
+ int match;
+
+ match = 0;
+ s1 = v + *pos; /* initial delimiter */
+ s2 = s1 + 1; /* for scanning the first string */
+ s3 = strchr(s2, *s1); /* for scanning the second string */
+
+ if (NULL == s3) /* found no middle delimiter */
+ goto out;
+
+ while ('\0' != *++s3) {
+ if (*s2 != *s3) { /* mismatch */
+ s3 = strchr(s3, *s1);
+ break;
+ }
+ if (*s3 == *s1) { /* found the final delimiter */
+ match = 1;
+ break;
+ }
+ s2++;
+ }
+
+out:
+ if (NULL == s3)
+ s3 = strchr(s2, '\0');
+ else
+ s3++;
+ *pos = s3 - v;
+ return(match);
+}
+
+/*
+ * Evaluate an optionally negated single character, numerical,
+ * or string condition.
+ */
+static int
+roff_evalcond(const char *v, int *pos)
+{
+ int wanttrue, number;
+
+ if ('!' == v[*pos]) {
+ wanttrue = 0;
+ (*pos)++;
+ } else
+ wanttrue = 1;
+
+ switch (v[*pos]) {
+ case 'n':
+ /* FALLTHROUGH */
+ case 'o':
+ (*pos)++;
+ return(wanttrue);
+ case 'c':
+ /* FALLTHROUGH */
+ case 'd':
+ /* FALLTHROUGH */
+ case 'e':
+ /* FALLTHROUGH */
+ case 'r':
+ /* FALLTHROUGH */
+ case 't':
+ (*pos)++;
+ return(!wanttrue);
+ default:
+ break;
+ }
+
+ if (roff_evalnum(v, pos, &number, 0))
+ return((number > 0) == wanttrue);
+ else
+ return(roff_evalstrcond(v, pos) == wanttrue);
+}
git.ckatri.com / mandoc.git / blobdiff