Reject the escape sequences \[uD800] to \[uDFFF] in the parser.

[mandoc.git] / tbl_data.c

diff --git a/tbl_data.c b/tbl_data.c
index 23a16450111f9b6104c4bd76d661340815433680..40b756a05e1aace8082f9eea83c898abf6e7a148 100644 (file)
--- a/tbl_data.c
+++ b/tbl_data.c
@@ -1,4 +1,4 @@
-/*     $Id: tbl_data.c,v 1.36 2015/01/28 17:32:07 schwarze Exp $ */
+/*     $Id: tbl_data.c,v 1.41 2015/10/06 18:32:20 schwarze Exp $ */

 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011, 2015 Ingo Schwarze <schwarze@openbsd.org>
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011, 2015 Ingo Schwarze <schwarze@openbsd.org>


@@ -42,20 +42,12 @@ getdata(struct tbl_node *tbl, struct tbl_span *dp,
 {
        struct tbl_dat  *dat;
        struct tbl_cell *cp;
 {
        struct tbl_dat  *dat;
        struct tbl_cell *cp;

-       int              sv, spans;
+       int              sv;

 
 

-       cp = NULL;
-       if (dp->last && dp->last->layout)
-               cp = dp->last->layout->next;
-       else if (NULL == dp->last)
-               cp = dp->layout->first;
+       /* Advance to the next layout cell, skipping spanners. */

 
 

-       /*
-        * Skip over spanners, since
-        * we want to match data with data layout cells in the header.
-        */
-
-       while (cp && TBL_CELL_SPAN == cp->pos)
+       cp = dp->last == NULL ? dp->layout->first : dp->last->layout->next;
+       while (cp != NULL && cp->pos == TBL_CELL_SPAN)

                cp = cp->next;
 
        /*
                cp = cp->next;
 
        /*


@@ -63,7 +55,7 @@ getdata(struct tbl_node *tbl, struct tbl_span *dp,
         * cells.  This means that we have extra input.
         */
 
         * cells.  This means that we have extra input.
         */
 

-       if (NULL == cp) {
+       if (cp == NULL) {

                mandoc_msg(MANDOCERR_TBLDATA_EXTRA, tbl->parse,
                    ln, *pos, p + *pos);
                /* Skip to the end... */
                mandoc_msg(MANDOCERR_TBLDATA_EXTRA, tbl->parse,
                    ln, *pos, p + *pos);
                /* Skip to the end... */


@@ -72,25 +64,21 @@ getdata(struct tbl_node *tbl, struct tbl_span *dp,
                return;
        }
 
                return;
        }
 

-       dat = mandoc_calloc(1, sizeof(struct tbl_dat));
+       dat = mandoc_calloc(1, sizeof(*dat));

        dat->layout = cp;
        dat->pos = TBL_DATA_NONE;
        dat->layout = cp;
        dat->pos = TBL_DATA_NONE;

-
-       assert(TBL_CELL_SPAN != cp->pos);
-
-       for (spans = 0, cp = cp->next; cp; cp = cp->next)
-               if (TBL_CELL_SPAN == cp->pos)
-                       spans++;
+       dat->spans = 0;
+       for (cp = cp->next; cp != NULL; cp = cp->next)
+               if (cp->pos == TBL_CELL_SPAN)
+                       dat->spans++;

                else
                        break;
 
                else
                        break;
 

-       dat->spans = spans;
-
-       if (dp->last) {
+       if (dp->last == NULL)
+               dp->first = dat;
+       else

                dp->last->next = dat;
                dp->last->next = dat;

-               dp->last = dat;
-       } else
-               dp->last = dp->first = dat;
+       dp->last = dat;

 
        sv = *pos;
        while (p[*pos] && p[*pos] != tbl->opts.tab)
 
        sv = *pos;
        while (p[*pos] && p[*pos] != tbl->opts.tab)


@@ -102,16 +90,12 @@ getdata(struct tbl_node *tbl, struct tbl_span *dp,
         * until a standalone `T}', are included in our cell.
         */
 
         * until a standalone `T}', are included in our cell.
         */
 

-       if (*pos - sv == 2 && 'T' == p[sv] && '{' == p[sv + 1]) {
+       if (*pos - sv == 2 && p[sv] == 'T' && p[sv + 1] == '{') {

                tbl->part = TBL_PART_CDATA;
                return;
        }
 
                tbl->part = TBL_PART_CDATA;
                return;
        }
 

-       assert(*pos - sv >= 0);
-
-       dat->string = mandoc_malloc((size_t)(*pos - sv + 1));
-       memcpy(dat->string, &p[sv], (size_t)(*pos - sv));
-       dat->string[*pos - sv] = '\0';
+       dat->string = mandoc_strndup(p + sv, *pos - sv);

 
        if (p[*pos])
                (*pos)++;
 
        if (p[*pos])
                (*pos)++;


@@ -127,14 +111,12 @@ getdata(struct tbl_node *tbl, struct tbl_span *dp,
        else
                dat->pos = TBL_DATA_DATA;
 
        else
                dat->pos = TBL_DATA_DATA;
 

-       if (TBL_CELL_HORIZ == dat->layout->pos ||
-           TBL_CELL_DHORIZ == dat->layout->pos ||
-           TBL_CELL_DOWN == dat->layout->pos)
-               if (TBL_DATA_DATA == dat->pos && '\0' != *dat->string)
-                       mandoc_msg(MANDOCERR_TBLDATA_SPAN,
-                           tbl->parse, ln, sv, dat->string);
-
-       return;
+       if ((dat->layout->pos == TBL_CELL_HORIZ ||
+           dat->layout->pos == TBL_CELL_DHORIZ ||
+           dat->layout->pos == TBL_CELL_DOWN) &&
+           dat->pos == TBL_DATA_DATA && *dat->string != '\0')
+               mandoc_msg(MANDOCERR_TBLDATA_SPAN,
+                   tbl->parse, ln, sv, dat->string);

 }
 
 int
 }
 
 int


@@ -150,11 +132,12 @@ tbl_cdata(struct tbl_node *tbl, int ln, const char *p, int pos)
                if (p[pos] == tbl->opts.tab) {
                        tbl->part = TBL_PART_DATA;
                        pos++;
                if (p[pos] == tbl->opts.tab) {
                        tbl->part = TBL_PART_DATA;
                        pos++;

-                       getdata(tbl, tbl->last_span, ln, p, &pos);
-                       return(1);
-               } else if ('\0' == p[pos]) {
+                       while (p[pos] != '\0')
+                               getdata(tbl, tbl->last_span, ln, p, &pos);
+                       return 1;
+               } else if (p[pos] == '\0') {

                        tbl->part = TBL_PART_DATA;
                        tbl->part = TBL_PART_DATA;

-                       return(1);
+                       return 1;

                }
 
                /* Fallthrough: T} is part of a word. */
                }
 
                /* Fallthrough: T} is part of a word. */


@@ -162,7 +145,7 @@ tbl_cdata(struct tbl_node *tbl, int ln, const char *p, int pos)
 
        dat->pos = TBL_DATA_DATA;
 
 
        dat->pos = TBL_DATA_DATA;
 

-       if (dat->string) {
+       if (dat->string != NULL) {

                sz = strlen(p + pos) + strlen(dat->string) + 2;
                dat->string = mandoc_realloc(dat->string, sz);
                (void)strlcat(dat->string, " ", sz);
                sz = strlen(p + pos) + strlen(dat->string) + 2;
                dat->string = mandoc_realloc(dat->string, sz);
                (void)strlcat(dat->string, " ", sz);


@@ -170,11 +153,11 @@ tbl_cdata(struct tbl_node *tbl, int ln, const char *p, int pos)
        } else
                dat->string = mandoc_strdup(p + pos);
 
        } else
                dat->string = mandoc_strdup(p + pos);
 

-       if (TBL_CELL_DOWN == dat->layout->pos)
+       if (dat->layout->pos == TBL_CELL_DOWN)

                mandoc_msg(MANDOCERR_TBLDATA_SPAN, tbl->parse,
                    ln, pos, dat->string);
 
                mandoc_msg(MANDOCERR_TBLDATA_SPAN, tbl->parse,
                    ln, pos, dat->string);
 

-       return(0);
+       return 0;

 }
 
 static struct tbl_span *
 }
 
 static struct tbl_span *


@@ -182,22 +165,20 @@ newspan(struct tbl_node *tbl, int line, struct tbl_row *rp)
 {
        struct tbl_span *dp;
 
 {
        struct tbl_span *dp;
 

-       dp = mandoc_calloc(1, sizeof(struct tbl_span));
+       dp = mandoc_calloc(1, sizeof(*dp));

        dp->line = line;
        dp->opts = &tbl->opts;
        dp->layout = rp;
        dp->line = line;
        dp->opts = &tbl->opts;
        dp->layout = rp;

-       dp->head = tbl->first_head;

        dp->prev = tbl->last_span;
 
        if (dp->prev == NULL) {
                tbl->first_span = dp;
                tbl->current_span = NULL;
        dp->prev = tbl->last_span;
 
        if (dp->prev == NULL) {
                tbl->first_span = dp;
                tbl->current_span = NULL;

-               dp->flags |= TBL_SPAN_FIRST;

        } else
                dp->prev->next = dp;
        tbl->last_span = dp;
 
        } else
                dp->prev->next = dp;
        tbl->last_span = dp;
 

-       return(dp);
+       return dp;

 }
 
 void
 }
 
 void


@@ -214,11 +195,11 @@ tbl_data(struct tbl_node *tbl, int ln, const char *p, int pos)
         * (it doesn't "consume" the layout).
         */
 
         * (it doesn't "consume" the layout).
         */
 

-       if (tbl->last_span) {
-               assert(tbl->last_span->layout);
+       if (tbl->last_span != NULL) {

                if (tbl->last_span->pos == TBL_SPAN_DATA) {
                        for (rp = tbl->last_span->layout->next;
                if (tbl->last_span->pos == TBL_SPAN_DATA) {
                        for (rp = tbl->last_span->layout->next;

-                                       rp && rp->first; rp = rp->next) {
+                            rp != NULL && rp->first != NULL;
+                            rp = rp->next) {

                                switch (rp->first->pos) {
                                case TBL_CELL_HORIZ:
                                        dp = newspan(tbl, ln, rp);
                                switch (rp->first->pos) {
                                case TBL_CELL_HORIZ:
                                        dp = newspan(tbl, ln, rp);


@@ -236,7 +217,7 @@ tbl_data(struct tbl_node *tbl, int ln, const char *p, int pos)
                } else
                        rp = tbl->last_span->layout;
 
                } else
                        rp = tbl->last_span->layout;
 

-               if (NULL == rp)
+               if (rp == NULL)

                        rp = tbl->last_span->layout;
        } else
                rp = tbl->first_row;
                        rp = tbl->last_span->layout;
        } else
                rp = tbl->first_row;


@@ -255,6 +236,6 @@ tbl_data(struct tbl_node *tbl, int ln, const char *p, int pos)
 
        dp->pos = TBL_SPAN_DATA;
 
 
        dp->pos = TBL_SPAN_DATA;
 

-       while ('\0' != p[pos])
+       while (p[pos] != '\0')

                getdata(tbl, dp, ln, p, &pos);
 }
                getdata(tbl, dp, ln, p, &pos);
 }