From: Kirill A. Korinsky Subject: Re: patch: simplify fetchname() To: "Omar Polo" Cc: tech@openbsd.org Date: Sat, 07 Nov 2026 00:23:50 +0100 On Thu, 05 Nov 2026 23:53:31 +0100, "Omar Polo" wrote: > > "Omar Polo" wrote: > > This is a preparatory step to get more hardening being put into > > fetchname(). The goal is to make fetchname() a bit more humanly > > understandable before adding things to it. > > > > While here, also add a regress for file names with spaces on them. > > > > okay? > > forgot to 'cvs add' a test file, and also added \f\v to the character > set as discussed with kirill@ to match completely the previous > behvaiour of isspace(3). > It simplifies a code and as I read it, it doesn't change the logic. Did it? > Index: regress/usr.bin/patch/Makefile > =================================================================== > RCS file: /cvs/src/regress/usr.bin/patch/Makefile,v > diff -u -p -r1.14 Makefile > --- regress/usr.bin/patch/Makefile 3 Aug 2022 07:35:04 -0000 1.14 > +++ regress/usr.bin/patch/Makefile 5 Nov 2026 19:52:58 -0000 > @@ -2,10 +2,11 @@ > > PATCH?= patch > PATCHFLAGS= -sN > -CLEANFILES= *.copy *.orig *.rej t5 d19/* > +CLEANFILES= *.copy *.orig *.rej t5 d19/* t21?file > > REGRESS_TARGETS= t1 t2 t3 t4 t5 t6 t7 t8 t9 \ > - t10 t11 t12 t13 t14 t15 t16 t17 t18 t19 t20 > + t10 t11 t12 t13 t14 t15 t16 t17 t18 t19 \ > + t20 t21 > > # .in: input file > # .diff: patch > @@ -29,6 +30,7 @@ REGRESS_TARGETS= t1 t2 t3 t4 t5 > # t18: diff in ed format that fully replaces input content. > # t19: git-produced unified diff. > # t20: reversal application of a patch to create a file with a single line. > +# t21: diff contains a file name with whitespaces in it > > .SUFFIXES: .in > > @@ -80,6 +82,12 @@ t20: > @cmp -s ${*}.copy ${.CURDIR}/${*}.out || \ > (echo "XXX ${*} failed" && false) > > -.PHONY: t5 > +t21: > + @echo ${*} > + @echo one > "${*} file" > + @${PATCH} ${PATCHFLAGS} < ${.CURDIR}/${*}.diff > + @echo two | cmp -s "${*} file" - || (echo "XXX t21 failed" && false) > + > +.PHONY: t5 t21 > > .include > Index: regress/usr.bin/patch/t21.diff > =================================================================== > RCS file: regress/usr.bin/patch/t21.diff > diff -N regress/usr.bin/patch/t21.diff > --- /dev/null 1 Jan 1970 00:00:00 -0000 > +++ regress/usr.bin/patch/t21.diff 5 Nov 2026 17:46:31 -0000 > @@ -0,0 +1,5 @@ > +--- t21 file Thu Nov 5 11:43:45 CST 2026 > ++++ t21 file Thu Nov 5 11:43:45 CST 2026 > +@@ -1 +1 @@ > +-one > ++two > Index: usr.bin/patch/util.c > =================================================================== > RCS file: /cvs/src/usr.bin/patch/util.c,v > diff -u -p -r1.47 util.c > --- usr.bin/patch/util.c 5 Nov 2026 11:27:33 -0000 1.47 > +++ usr.bin/patch/util.c 5 Nov 2026 22:51:46 -0000 > @@ -341,7 +341,7 @@ char * > fetchname(const char *at, bool *exists, int strip_leading) > { > char *fullname, *name, *t; > - int sleading, tab; > + int sleading; > struct stat filestat; > > if (at == NULL || *at == '\0') > @@ -355,19 +355,21 @@ fetchname(const char *at, bool *exists, > /* So files can be created by diffing against /dev/null. */ > if (strnEQ(at, _PATH_DEVNULL, sizeof(_PATH_DEVNULL) - 1)) > return NULL; > - name = fullname = t = savestr(at); > + name = fullname = savestr(at); > if (out_of_mem) > return NULL; > > - tab = strchr(t, '\t') != NULL; > - /* Strip off up to `strip_leading' path components and NUL terminate. */ > - for (sleading = strip_leading; *t != '\0' && ((tab && *t != '\t') || > - !isspace((unsigned char)*t)); t++) { > + if ((t = strchr(fullname, '\t')) != NULL) > + *t = '\0'; > + else > + fullname[strcspn(fullname, " \f\v\r\n")] = '\0'; > + > + /* Strip off up to `strip_leading' path components. */ > + for (sleading = strip_leading, t = fullname; *t != '\0'; t++) { > if (t[0] == '/' && t[1] != '/' && t[1] != '\0') > if (--sleading >= 0) > name = t + 1; > } > - *t = '\0'; > > /* > * If no -p option was given (957 is the default value!), we were > -- wbr, Kirill