gbranden pushed a commit to branch master
in repository groff.
commit 64711b3a875ce9ea3a023401d641b9922de61cc8
Author: G. Branden Robinson <[email protected]>
AuthorDate: Fri Nov 14 11:44:18 2025 -0600
[troff]: Make numeric expression parser stricter.
* src/roff/troff/number.cpp: Refactor numeric expression parser to
simplify and heighten severity of diagnostics. This isn't a behavior
change; character sequences that were invalid as numeric expressions
were ignored before and remain so. But since we're aborting
interpretation of an argument to a request or escape sequence, where
in at least one case AT&T troff did not--it would read through
spurious tab characters to reach a valid numeric expression
afterward--an error seems more appropriate.
(is_valid_expression_start): Drop special handling of tab and right-
brace escape sequences. These get handled anyway when the term at the
start of the expression gets interpreted...
(is_valid_term): ...thus. Raise diagnostic severity from warning (in
"number" category) to error.
Illustration:
Before
------
$ printf '.nr a A2\n.tm a=\\na\n' | ~/groff-1.23.0/bin/groff -z -ww
troff:<standard input>:1: warning: expected numeric expression, got 'A'
a=troff:<standard input>:2: warning: register 'a' not defined
0
$ printf '.nr a \t2\n.tm a=\\na\n' | ~/groff-1.23.0/bin/groff -z -ww
troff:<standard input>:1: warning: expected numeric expression, got a tab
character
a=troff:<standard input>:2: warning: register 'a' not defined
0
$ printf '.nr a \\}2\n.tm a=\\na\n' | ~/groff-1.23.0/bin/groff -z -ww
troff:<standard input>:1: warning: expected numeric expression, got
rightbrace escape sequence
a=troff:<standard input>:2: warning: register 'a' not defined
0
After
-----
$ printf '.nr a A2\n.tm a=\\na\n' | ./build/test-groff -z -ww
troff:<standard input>:1: error: ignoring invalid numeric expression
starting with character 'A'
a=troff:<standard input>:2: warning: register 'a' not defined
0
$ printf '.nr a \t2\n.tm a=\\na\n' | ./build/test-groff -z -ww
troff:<standard input>:1: error: ignoring invalid numeric expression
starting with a tab character
a=troff:<standard input>:2: warning: register 'a' not defined
0
$ printf '.nr a \\}2\n.tm a=\\na\n' | ./build/test-groff -z -ww
troff:<standard input>:1: error: ignoring invalid numeric expression
starting with an escaped '}'
a=troff:<standard input>:2: warning: register 'a' not defined
0
DWB 3.3
-------
$ printf '.nr a A2\n.tm a=\\na\n' | DWBHOME=~/dwb ~/dwb/bin/nroff >/dev/null
a=0
$ printf '.nr a \t2\n.tm a=\\na\n' | DWBHOME=~/dwb ~/dwb/bin/nroff
>/dev/null
a=2
$ printf '.nr a \\}2\n.tm a=\\na\n' | DWBHOME=~/dwb ~/dwb/bin/nroff
>/dev/null
a=0
---
ChangeLog | 17 +++++++++++++++++
src/roff/troff/number.cpp | 14 ++------------
2 files changed, 19 insertions(+), 12 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index fa634c7f4..074135e68 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,20 @@
+2025-11-14 G. Branden Robinson <[email protected]>
+
+ * src/roff/troff/number.cpp: Refactor numeric expression parser
+ to simplify and heighten severity of diagnostics. This isn't
+ a behavior change; character sequences that were invalid as
+ numeric expressions were ignored before and remain so. But
+ since we're aborting interpretation of an argument to a request
+ or escape sequence, where in at least one case AT&T troff did
+ not--it would read through spurious tab characters to reach a
+ valid numeric expression afterward--an error seems more
+ appropriate.
+ (is_valid_expression_start): Drop special handling of tab and
+ right-brace escape sequences. These get handled anyway when the
+ term at the start of the expression gets interpreted...
+ (is_valid_term): ...thus. Raise diagnostic severity from
+ warning (in "number" category) to error.
+
2025-11-14 G. Branden Robinson <[email protected]>
* src/roff/troff/input.cpp (token::description): Improve
diff --git a/src/roff/troff/number.cpp b/src/roff/troff/number.cpp
index f17629d23..78b5de285 100644
--- a/src/roff/troff/number.cpp
+++ b/src/roff/troff/number.cpp
@@ -245,16 +245,6 @@ static bool is_valid_expression_start()
warning(WARN_MISSING, "numeric expression missing");
return false;
}
- if (tok.is_tab()) {
- warning(WARN_TAB, "expected numeric expression, got %1",
- tok.description());
- return false;
- }
- if (tok.is_right_brace()) {
- warning(WARN_RIGHT_BRACE, "expected numeric expression, got right"
- " brace escape sequence");
- return false;
- }
return true;
}
@@ -512,8 +502,8 @@ static bool is_valid_term(units *u, int scaling_unit,
*u = 0;
return !is_mandatory;
default:
- warning(WARN_NUMBER, "expected numeric expression, got %1",
- tok.description());
+ error("ignoring invalid numeric expression starting with %1",
+ tok.description());
return false;
}
int divisor = 1;
_______________________________________________
groff-commit mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/groff-commit
